Abstract
Childhood epilepsy is a common and devastating condition, for which many children still do not have adequate treatment. Some children with drug-resistant epilepsy require surgical excision of epileptogenic brain tissue for seizure control, affording the opportunity to study this tissue ex vivo to interrogate human epileptic neurons for potentially hyperexcitable perturbations in intrinsic electrophysiological properties. In this study, we characterized the diversity of layer L2/3 (L2/3) pyramidal neurons (PNs) in ex vivo brain slices from pediatric patients with epilepsy. We found a remarkable diversity in the firing properties of epileptic L2/3 PNs: five distinct subpopulations were identified. Additionally, we investigated whether the etiology of epilepsy influenced the intrinsic neuronal properties of L2/3 PNs when comparing tissue from patients with epilepsy due to malformations of cortical development (MCDs), other forms of epilepsy (OEs), or with deep-seated tumors. When comparing epileptic with control L2/3 PNs, we observed a decrease in voltage sag and lower maximum firing rates. Moreover, we found that MCD and OE L2/3 PNs were mostly similar indicating that epilepsy etiology may not outweigh the influences of epileptiform activity on L2/3 PN physiology. Lastly, we show that the proconvulsant drug, 4-aminopyridine (4-AP), leads to increased AP half-width, reduced firing rate accommodation, and slower AHPs. These changes imply that 4-AP induces an increase in [K+]o and a resultant increase in AP duration, leading to the release of more excitatory neurotransmitters per action potential, thereby promoting network hyperexcitability.
- DNET
- focal cortical dysplasia
- gliosis
- layer 2/3 pyramidal neurons
- intrinsic properties
- tuberous sclerosis
Significance Statement
This study characterizes the diversity of L2/3 PNs within human epileptic loci ex vivo. We identify significant differences in L2/3 PN intrinsic properties between epileptic subtypes and control L2/3 PNs, and these differences promote increased synaptic summation and neurotransmission. Furthermore, we document that AHP kinetics do not dictate epileptic L2/3 PN firing rates. We also find that L2/3 PNs demonstrate commensurate properties regardless of the etiology of epilepsy. Finally, we document the effects of the convulsant drug 4-AP, on epileptic L2/3 PN intrinsic properties. This study contributes to the understanding of human neocortical epilepsy, the effects on neuron subtypes using pro-ictal drugs to generate seizures ex vivo, and the neuronal abnormalities associated with epileptogenesis.
Introduction
Childhood epilepsy is a common disorder, with a reported prevalence ranging from 0.3 to 4.4% (Camfield and Camfield, 2015; Aaberg et al., 2017). Antiseizure medications (ASMs) remain the first-line treatment for children with epilepsy, yet approximately one-third of children develop drug-resistant epilepsy and continue to have seizures (Golyala and Kwan, 2017). Malformations of cortical development (MCDs), including focal cortical dysplasia (FCD), hemimegalencephaly (HMG), and tuberous sclerosis complex (TSC), are the most common causes of surgically treated epilepsy in the pediatric population (Blumcke et al., 2017). Although surgery is curative for many patients with MCDs, not all patients are eligible for surgery and 5 year seizure freedom rates after surgery are <70% (Lamberink et al., 2020). Thus, there remains a substantial proportion of patients who continue to have seizures despite optimal medical and surgical therapy.
Over the last few decades, many studies have characterized human epileptic cortical neurons and investigated the mechanisms that underlie interictal and ictal activity (Hwa et al., 1991; D'Antuono et al., 2004; Avoli et al., 2005; Cepeda et al., 2006; Kohling and Avoli, 2006; Jacobs et al., 2008; Huberfeld et al., 2011; Talos et al., 2012; Abdijadid et al., 2015; Blauwblomme et al., 2019; Levinson et al., 2020; Rossini et al., 2021). A common neocortical layer of focus is layer 2/3 (L2/3), which controls the gain of cortical output (Quiquempoix et al., 2018). Prior pediatric MCD ex vivo studies have demonstrated that L2/3 generates ictal discharges, abnormal bursting activity, and pathogenic high-frequency oscillations (HFOs) and has shown that these epileptiform signatures are significantly regulated by NMDA receptors, GABAergic neurotransmission, and dopaminergic modulation (Wuarin et al., 1990; Cepeda et al., 1992; Mattia et al., 1995; Avoli et al., 1999; D'Antuono et al., 2004; Calcagnotto et al., 2005; Blauwblomme et al., 2019; Levinson et al., 2020). However, few of these prior studies include nonepileptic control patients or compared results between epileptic subtypes.
For many MCDs, genetic alterations in the mammalian target of rapamycin (mTOR) underlie the developmental malformation and presumably drive epileptogenesis (Jozwiak et al., 2006; Cepeda et al., 2012; O'Dell et al., 2012; Heinemann and Staley, 2014; Lipton and Sahin, 2014; Crino, 2015; Guerrini et al., 2015; D'Gama et al., 2017; Iffland and Crino, 2017; Nakagawa et al., 2017; Curatolo et al., 2018; Park et al., 2018; Wu et al., 2021). Other common pathologies leading to surgically treatable focal-onset epilepsy include perinatal stroke, often pathologically characterized by gliosis or encephalomalacia, and idiopathic forms of epilepsy, often characterized as low-grade tumors. Based on the mechanistic, developmental, and cellular differences between MCD epilepsies and other epilepsies, we hypothesized that L2/3 PNs of these subclasses differ in their intrinsic properties.
Despite the intrinsic hyperexcitability of human brain tissue resected during epilepsy surgery, ex vivo ictal activity is only induced via pharmacologic manipulations or alterations in the concentration of extracellular cations (Chang et al., 2019). One reproducible method for the generation of epileptiform activity in rodent or human tissue is wash on of 4-aminopyridine (4-AP), an A-type K+ channel blocker (Chesnut and Swann, 1988; Avoli and Jefferys, 2016; Chang et al., 2019). Addition of 4-AP has been shown to increase action potential (AP) duration, elicit a large increase in extracellular [K+] and intracellular [Ca2+]; generate a GABAergic, long-lasting depolarizing potential, reduce firing rate accommodation, and augment neurotransmitter release leading to interictal and ictal discharges (Lorenzon and Foehring, 1992; Kohling and Avoli, 2006; Wu et al., 2009; Abdijadid et al., 2015; Williams and Hablitz, 2015). Moreover, 4-AP allows for the examination of how voltage-gated K+ channels (members of the Kv1, Kv3, and Kv4 families) play a role in hyperexcitability, as 4-AP has no effect on Na+ or Ca2+ channels (Wulff and Zhorov, 2008). Although 4-AP can elicit ictal activity in brain slices from control and epileptic tissue, there has been little prior study of the differing effects of 4-AP on these respective tissue types. Therefore, we examined the effects of 4-AP on L2/3 PN intrinsic properties to gain a better understanding of how this potent convulsant elicits ictal activity ex vivo.
Here, we present a systematic analysis of the diversity of L2/3 PNs in human epileptic tissue. This study, to the best of our knowledge, is the first to characterize the diversity of L2/3 PNs from human epileptic foci including the differences in firing rate properties, passive membrane properties, and AP kinetics. In addition, it is the first to show differences in L2/3 PN intrinsic properties between MCD, other epilepsies, and control tissues and determines the direct effect of 4-AP on epileptic L2/3 PN intrinsic properties.
Materials and Methods
Acute human brain slice preparation for electrophysiology
Resected neocortical tissue was obtained from Children's Hospital Colorado. The decision to recommend epilepsy surgery is always made by a multidisciplinary team including board-certified physicians in pediatric epileptology, pediatric neurosurgery, pediatric neuroradiology, and a pediatric neuropsychologist specializing in patients with epilepsy. All patients, or their legal guardians, provided written informed consent for the study before surgery. If appropriate, the patient also provided assent. This process was performed in accordance with our institution's Institutional Review Board. For the present study, cortical tissue samples were collected from 33 patients in three experimental groups: MCD epilepsy, other epilepsy, and control. Control patients were patients with deep-seated neoplastic lesions, which required removal of normal (nonepileptogenic, non-neoplastic) superficial cortex to resect the tumor. No additional cortex was removed beyond that which was needed for resection of the deep-seated lesion. In this study, we used tissue resected from 28 epilepsy and five tumor control patients. Patients ranged from 1 to 21 years of age (mean, 9.9). Twenty were male and 13 were female. The brain samples analyzed in the present experiments were from the temporal (n = 15), frontal (n = 13), parietal (n = 4), or occipital (n = 1) neocortex. Patient information including age, sex, diagnosis, and experimental group is presented in Table 1. For all epilepsy patients, the resected neocortical tissue used for recordings was located in the epileptic focus.
Immediately following the surgical resection, resected tissue was submerged in 0–4°C carbogenated (95% O2–5% CO2) N-methyl-d-glucamine (NMDG) substituted artificial cerebrospinal fluid [ACSF; in mM: 92 NMDG, 2.5 KCl, 1.25 NaH2PO4, 30 NaHCO3, 20 4-(2-hydroxyethyl)-1-pip-erazineethanesulfonic acid (HEPES), 25 d-glucose, 2 thiourea, 5 Na-ascorbate, 3 Na-pyruvate, 0.5 CaCl2·4H2O, and 10 MgSO4·7H2O, pH adjusted to 7.3–7.4 with concentrated hydrochloric acid (HCl), osmolality ∼305 mOsm/kg]. The total duration from operating room to slicing was 15–20 min. The tissue was then placed in a carbogenated petri dish with 0–4°C carbogenated NMDG ACSF. Approximately 1 cm3 tissue blocks were prepared with a scalpel, if the specimen was larger than that size. The arachnoid, superficial blood vessels, and cauterized tissue were carefully removed from the specimen. No effort was made to remove the pia mater due to the risk of damaging the underlying gray matter. The ∼1 cm3 tissue block was then superglued on to a vibratome stage (Leica Biosystems) and immersed in 0–4°C carbogenated NMDG ACSF. Acute coronal slices (400 µm) were prepared by slicing perpendicular to the pial surface to preserve the laminated cortical structure as well as the primary dendrites of PNs. Slices were then incubated in carbogenated NMDG ACSF at ∼35°C for 12 min and were then transferred to a room temperate (∼23°C) carbogenated modified HEPES ACSF (in mM: 92 NaCl, 2.5 KCl, 1.2 NaH2PO4, 30 NaHCO3, 20 HEPES, 25 d-glucose, 2 thiourea, 5 Na-ascorbate, 3 Na-pyruvate, 2 CaCl2·4H2O, 2 MgSO4·7H2O, pH adjusted to 7.3–7.4 with HCl, osmolality ∼305 mOsm/kg). The slices remained in the ∼23°C modified HEPES ACSF for at least 30 min before being transferred to the recording chamber for electrophysiology experiments.
Electrophysiology
For whole-cell patch-clamp recordings, slices were placed in a submerged slice chamber and perfused at a rate of 2 ml/min with heated (∼33°C) recording ACSF (in mM: 124 NaCl, 2.5 KCl, 1.2 NaH2PO4, 24 NaHCO3, 5 HEPES, 12.5 d-glucose, 2 CaCl2·4H2O, 2 MgSO4·7H2O, pH adjusted to 7.3–7.4 with HCl, osmolality ∼305 mOsm/kg). Slices were visualized using a moving stage microscope (Scientifica, SliceScope Pro 2000) equipped with 4× (0.10 NA) and a 40× water immersion objective lens (0.80 NA) objectives, differential interference contrast (DIC) optics, a SciCam Pro camera (Scientifica), and Micro-Manager 1.4 (Open Imaging). L2/3 PNs were visualized under DIC. Whole-cell patch-clamp recordings were performed with pipettes pulled to 3–6 MΩ and filled with the following internal solution (in mM: 135 potassium gluconate, 20 KCl, 10 HEPES,0.1 EGTA, 2 Mg-ATP, 0.3 Na2-GTP, pH adjusted to 7.3–7.4 with KOH, osmolality ∼290 mOsm/kg). For most recordings, 0.4% biocytin was added to the internal solution.
After achieving whole-cell configuration, L2/3 PNs were recorded from rest in current-clamp mode (Ihold = 0 pA). Following a 0.5 s baseline period, the holding current was linearly ramped from 0 to 1,000 pA over 2 s followed by a 2.5 s recovery period. Before initiation of the series of current injections, holding current was adjusted so that the resting membrane potential of neurons was approximately −60 mV. Each cell was subjected to two series of 600 ms square current injections: −100 to +100 pA at 10 pA intervals and −250 to +1,000 pA at 50 pA intervals. The following intrinsic properties were measured in each cell using the above current injections: resting membrane potential (mV), input resistance (MΩ), membrane decay (τ, ms), voltage sag (%), AP threshold, AP amplitude, AP half-width, magnitude of afterhyperpolarization (AHP, mV), AHP latency, ΔAHP, AP phase plot, AP broadening ratio, AP amplitude adaptation ratio, max firing rate (FR), FR ratio, instantaneous frequency, and frequency versus time. Extended Data Table 2-2 describes how each of the above parameters were calculated. In a subset of experiments, 100 µM of 4-aminopyridine (4-AP) was washed on to induce ictal activity. For the analysis of AHP kinetics in Figure 3, the AHPs were measured at rheobase and 2× rheobase.
Electrical recordings were acquired with a MultiClamp 700B amplifier and were sent through a Hum Bug Noise Eliminator (Quest Scientific) to then be converted to a digital signal with the Axon Digidata 1440A digitizer using pCLAMP 10.7 software (Molecular Devices). Access resistance was monitored throughout the experiments, and data were discarded if access resistance exceeded 25 MΩ. No junction potential compensation or series resistance compensation was performed. In current-clamp mode, compensation for voltage variations was achieved using a bridge balance circuit. Data were sampled at 10 kHz.
Biocytin reconstructions
After the recordings were completed, the slice was then left to rest in the recording chamber for 4–5 min to ensure transport of biocytin to distal dendrites and axon processes (Swietek et al., 2016). The slice was then fixed in 4% PFA for 24–48 h. Biocytin was visualized with streptavidin conjugated to Alexa Fluor 488 or 594 at 1:1,000 (Jackson ImmunoResearch). Biocytin-filled neurons were visualized using a Zeiss Axio Imager M2 microscope. 3D slide scanning and neuron tracing was performed using Neurolucida software (MBF Bioscience).
Statistical analyses
All data analysis was performed using MATLAB R2018a, GraphPad Prism 9.3.1, or Easy Electrophysiology v4.2.0 (http://www.easyelectrophysiology.com/). If data sets contained at least 10 data points, a D’Agostino and Pearson omnibus K2 normality test was performed to assess normality. For statistical tests between three or more groups with normally distributed data, 10 or more data points, and equal variance, an ANOVA test was performed. Equal variance was determined using the Brown–Forsythe test. If data was normally distributed but had unequal variance (Brown–Forsythe test p < 0.05), then a Welch's ANOVA test was performed. Following an ANOVA, a Tukey honestly significant difference (HSD) post hoc test was performed to find which group(s) differed. For statistical tests between three or more groups of non-normally distributed data, or data sets with <10 data points, a Kruskal–Wallis (KW) test was performed. Differences between groups were determined using Dunn's multiple-comparisons post hoc test and the H (degrees of freedom) statistic was checked against the critical chi-square value.
When comparing two groups with normally distributed data and standard deviations (SDs) between groups being <2×, an unpaired t test was applied. Two groups with non-normal data and/or data with SD being >2× between groups were compared using a Mann–Whitney U (MWU) test. For paired data, a paired t test was performed on normally distributed data with SD between groups being <2×, and a Wilcoxon matched-pairs signed rank test was performed on non-normal data and data where SD between groups was >2×.
All statistical tests were two-tailed. Unless otherwise stated, experimental numbers are reported as n = x, y, where x is the number of neurons and y is the number of patients. Statistical significance is notated in figures as the number of asterisks with the corresponding p values:*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Data visualizations were created in MATLAB, GraphPad Prism, Adobe Illustrator, and OriginPro 2022. Data in tables and figure scatterplots are presented as mean ± SD, and figure data presented as line graphs are presented as mean ± SEM for ease of comparison. Frequency and FR ratio line graphs indicate smoothed data using the smoothing spline method in GraphPad Prism. Phase plots were created using the B-spline connect interpolation in OriginPro 2022.
Code accessibility
The MATLAB scripts used for intrinsic property analysis are from a previous publication (Guthman et al., 2020).
Results
Subjects enrolled
For this study, human neocortical tissue samples were surgically obtained from a total of 33 patients aged 1–21 years, in three histopathological groups: patients with congenital malformations of cortical development (MCD, n = 13), patients with other epilepsies (OE, n = 15), and patients with deeply located tumors and no history of seizures (control, n = 5). Final pathologic diagnoses were based upon examination of epileptic tissue by a board-certified pediatric neuropathologist. Classification of focal cortical dysplasias (FCDs) was based on the ILAE classification system (Blumcke et al., 2011). The experimental tissue used in this study and the tissue sent to pathology were sent from immediately adjacent areas of lesional tissue, as determined by the attending neurosurgeon. Demographic and histopathological details are presented in Table 1. We analyzed the intrinsic properties of 108 L2/3 PNs (MCD n = 37; other epilepsies n = 45; control n = 26). For the initial analysis of firing patterns, intrinsic properties and AHPs, presented in Figures 1⇓–3, L2/3 PNs from both MCD and OE subgroups are combined.
Patient information
Putative L2/3 PNs from human epileptic foci have diverse morphology and spiking properties. Ai–Ei, Depolarizing current steps at rheobase (pA; dark trace) and rheobase + 100 pA (2× rheobase, light trace) elicit action potentials (APs) of various shapes and frequencies. Each group was determined based on firing property and AHP shape. Aii–Eii, Representative trace of each AP from neuron subtype (scale bars: 10 mV, 10 ms). Aiii–Eiii, Corresponding 3D reconstruction of neuron morphology (x-axis scale bar: 200 µm). Each neuron's location in the representative cortical layers was determined by the respective distance of their soma from pial surface (dendrites: subtype color scheme, axon branches: blue).
Putative L2/3 PN subtypes show subtle differences in AP kinetics and firing properties. A, Neuron subtype percentage (%) of the entire population of recorded neurons (ACC, accommodating neurons; RS, regular spiking neurons; Not, notch neurons; ST, stuttering neurons; ES, early spiking neurons). B, Representative traces of L2/3 PN subtype AP (scale bars: 10 mV, 10 ms). C, Overlay of average AP of each neuron subtype. D, Overlay of average phase plot (dV/dT vs voltage) indicating differences in subtype AP kinetics. E, Notch neurons showed the shortest AP half-width with a significant difference compared with accommodating neurons. F, Stutter neurons showed a significant reduction in AP spike amplitude. G, Notch neurons have the shortest AHP latency with a significantly shorter AHP latency compared with RS and ACC neurons. H, FR accommodation ratio of each neuron subtype taken at 2× rheobase indicated ES neurons had a significantly lower chance of accommodating compared with stutter neurons. I, Initial instantaneous frequency ± SEM versus injected current (pA). J, Initial instantaneous frequency was taken at the rheobase + 100 pA current step and indicated significant differences between subtype initial instantaneous firing frequency. K, Final instantaneous frequency ± SEM versus injected current (pA). Scatterplots include mean values ± SD.
Figure 2-1
Other intrinsic properties of putative L2/3 PN subtypes. Other analyzed intrinsic properties were graphed to show spread and variability as follows: A) resting membrane potential (RMP, mV), B) input resistance (MΩ), C) voltage sag (%), D) membrane decay, E) AHP amplitude, F) Mean FR ± SEM vs injected current (pA), G) Max FR, H) final instantaneous frequency, and I) FR accommodation ratio ± SEM vs injected current (pA). Scatter plots include mean values ± SD. Download Figure 2-1, TIF file.
Figure 2-2
PCA and K-means clustering of epileptic putative L2/3 PNs. A) Scree Plot used to determine the number of clusters for K-means clustering. B) Silhouette Value for each epileptic L2/3 PN based on cluster assignment. C) PCA 3D plot with neurons clustered based on 6 intrinsic properties with cluster centroids marked by an ‘X’. Download Figure 2-2, TIF file.
Table 2-1
Post hoc test p-values for statistically significant properties based on putative L2/3 PN subtype. The correlated post-hoc test for comparing intrinsic properties from Table 2. Download Table 2-1, DOCX file.
Table 2-2
Definitions of electrophysiologic parameters. How each intrinsic property was calculated based on the square 600ms current steps. Download Table 2-2, DOC file.
Epileptic L2/3 PNs with fAHP or mAHP do not differ in their firing rates or firing rate accommodation. A, Representative traces of L2/3 PN AP shapes showing neurons with a (Ai) fAHP followed by mAHP, (Aii) fAHP followed by ADP and mAHP, and (Aiii) mAHP only. B, Overlay of average AP of neuron split by AHP shape. C, Overlay of average phase plots (dV/dT vs voltage) of neuron split by AHP shape. D, There is no significant differences in the maximum depolarization slope of the AP (dV/dT) between subtypes. E, mAHP PNs have a significantly slower repolarization slope (dV/dT) when compared with fAHP PNs. F, AHP latency is shorter for fAHP and ADP PNs as expected compared with mAHP neurons. G, fAHP and ADP PNs show a more positive ΔAHP. H, mAHP PNs have more AP amplitude adaptation compared with ADP PNs. I, mAHP PNs have higher overall voltage sag (%) compared with fAHP PNs. We observe no differences in J, initial instantaneous frequency ± SEM versus injected current (pA); K, mean instantaneous frequency ± SEM versus injected current (pA); L, final instantaneous frequency ± SEM versus injected current (pA); or M, FR accommodation ratio ± SEM versus injected current (pA). N, Percentage (%) of AHP PN based on cell type characterized in Figure 2. f, fAHP; ADP, fAHP-ADP; m mAHP. Scatterplots include mean values ± SD.
Figure 3-1
Other intrinsic properties of L2/3 PN subtypes based on AHP shape. Other analyzed intrinsic properties were graphed to show spread and variability as follows: A) AP half-width, B) input resistance, C) initial instantaneous frequency, D) max instantaneous frequency, E) final instantaneous frequency, and F) FR accommodation ratio. Scatter plots include mean values ± SD. Download Figure 3-1, TIF file.
Table 3-1
Group comparison values with post hoc adjusted p-values for statistically significant properties based on AHP subtype. The correlated post-hoc test for comparing intrinsic properties from Table 3. Download Table 3-1, DOCX file.
Physiological diversity of L2/3 human epileptic neocortical PNs
In rodent neocortex, L2/3 PNs are a homogenous group of cells that exhibit a characteristic regular spiking phenotype. In contrast, human L2/3 PNs display diversity in their passive and active intrinsic properties, firing patterns, dendritic morphologies, and transcriptomes (DeFelipe and Farinas, 1992; Deitcher et al., 2017; Kalmbach et al., 2018; Berg et al., 2021). However, the diversity of human L2/3 PN physiological properties has not been examined in the setting of pediatric epilepsy. Therefore, we performed whole-cell patch-clamp electrophysiology on putative L2/3 PNs located within the epileptic focus from resected brain tissue of pediatric patients. We observed a considerable diversity in the firing patterns and action potential (AP) kinetics of these L2/3 PNs (Fig. 1). To confirm that we were successfully targeting L2/3 PNs in our epileptic tissue slices, 3D reconstructions of biocytin-filled neurons were performed on a subset of neurons (n = 20 reconstructions from 40 attempted fills). Although there was some variability in the extent of the dendritic arborization among the filled neurons, all L2/3 PNs filled with biocytin exhibited dendritic spines (data not shown) and morphologies consistent with PNs (Fig. 1; Mohan et al., 2015; Deitcher et al., 2017).
To determine if there were subtypes of epileptic putative L2/3 PNs, we first grouped them based on their action potential (AP) shape, afterhyperpolarization kinetics, and firing patterns at rheobase and 2× rheobase. Based on this initial categorization, we observed five distinct subtypes: accommodating neurons, regular spiking (RS) neurons, stuttering neurons, early spiking (ES) neurons, and notch neurons (Fig. 1). Accommodating neurons display significant adaptation of AP frequency and AP amplitude (Fig. 1A). RS neurons demonstrate the canonical RS AP shape and lack of accommodation in their firing pattern (Fig. 1B; Markram et al., 2015; Neske et al., 2015; Varga et al., 2015; He et al., 2016; Stedehouder et al., 2019; Berg et al., 2021; Moradi Chameh et al., 2021). Notch neurons have a fast AHP followed by an afterdepolarization (ADP; Fig. 1C; Foehring et al., 1991; Lorenzon and Foehring, 1992; Varga et al., 2015; Eyal et al., 2018). Stuttering neurons exhibit irregular firing patterns more commonly exhibited by interneurons (Fig. 1D; Druckmann et al., 2013; Stiefel et al., 2013; Komendantov et al., 2019). ES neurons exhibit a relatively hyperpolarized rheobase (Fig. 1E). Of the 82 recorded epileptic L2/3 PNs, we observed 7 accommodating neurons, 44 RS neurons, 18 notch neurons, 6 stuttering neurons, and 7 ES neurons (Fig. 2A).
After our initial classification, we analyzed the active and passive membrane properties of epileptic L2/3 PNs using a series of hyperpolarizing and depolarizing current steps (summary data and statistical analyses are presented in Table 2 and post hoc test p values in Extended Data Table 2-1). We observed no significant differences between PN subtypes in resting membrane potential (RMP, mV), input resistance (MΩ), or voltage sag (%) and found a significant difference in membrane decay time (ms) between accommodating and notch neurons (Extended Data Fig. 2-1A–D). We also observed substantial differences in AP kinetics between PN subtypes (Fig. 2B–D). More specifically, notch neurons showed the shortest AP half-width (ms) with a significant difference compared with accommodating neurons (Fig. 2E), and stuttering neurons showed the lowest spike amplitude (mV), with significant differences compared with accommodating, notch and ES neurons (Fig. 2F). Notch neurons displayed the shortest AHP latencies, with significant differences compared with RS and ES neurons, with RS neurons having a wide range of AHP latencies (Fig. 2G). We observed no subtype differences in AHP magnitude (Extended Data Fig. 2-1E). Analysis of the firing pattern within a train of APs revealed quantifiable differences between subtypes, as expected. In general, compared with other subtypes, we found that stuttering neurons were more likely to fire quickly initially and then accommodated their firing rates while ES neurons did not initially fire quickly but were the least likely to accommodate their firing rates (Fig. 2H–J). Surprisingly, we did not find differences in final instantaneous frequency (Fig. 2K, Extended Data Fig. 2-1H) or maximum firing rate (Extended Data Fig. 2-1G).
Intrinsic properties of putative L2/3 PN subtypes based on firing properties and AP shape
Our manual classification scheme presented in Figures 1 and 2 demonstrates a high degree of physiological variability within L2/3 PNs. To determine whether these cell types could be separated mathematically, we performed an unbiased multivariable analysis using principal component analysis and K-means clustering using a custom-written MATLAB code. We included the six intrinsic properties that showed meaningful differences between the subtypes characterized above, including AHP latency, AP half-width, AP spike amplitude, initial ISI, FR accommodation ratio, and final ISI. We then performed a PCA using z-score normalized values for each property and found that three components were responsible for 76% of the total variation. Following PCA, we performed K-means clustering which indicated either two or five optimal clusters (Extended Data Fig. 2-2A). However, with five clusters, the majority of neurons (43/82) had silhouette values <0.5 (Extended Data Fig. 2-2B). These results indicate that PCA with K-means clustering did not quantifiably separate L2/3 PNs into distinct categories similar to our manually differentiated categories described above. In addition, no particular property dictated a principal component (Extended Data Fig. 2-2C). There are several possible explanations for the different results between manual and unsupervised clustering analyses. First, it is possible that we did not measure other variables which would separate the L2/3 PNs into classes presented in Figure 1, such as transcriptomic differences. Additionally, it is possible that the observed differences in L2/3 epileptic PN intrinsic properties have minimal physiologic significance, in terms of cellular function.
Prior studies have reported that firing pattern, firing frequency, and frequency adaptation are very highly regulated by voltage-gated potassium and calcium channels that dictate AHPs (Lorenzon and Foehring, 1992; Perez-Garci et al., 2003; Phelan et al., 2005; Higgs and Spain, 2009; Rizzo et al., 2014; Mendez-Rodriguez et al., 2021). In specific relevance to the study of epilepsy, AHP duration and/or presence of an ADP has been used to classify neurons into subtypes that differ in intrinsic excitability (Phelan et al., 2005; Mendez-Rodriguez et al., 2021). Fast AHPs (fAHPs) and medium AHPs (mAHPs) can be observed following a single action potential and during repetitive firing. In some L2/3 PNs, the fAHP is followed by an ADP which drives neuronal bursting and decreases the initial interspike interval (ISI), allowing for quicker and more powerful signal transduction (Foehring et al., 1991; Higgs and Spain, 2009). However, due to its slow kinetics, the slow AHP component (sAHP, >1 s) is only observed following a period of repetitive firing, such as a burst of APs (Rizzo et al., 2014). Thus, we did not analyze the sAHP given that the differences seen in our manual analysis were seen within the train of APs generated by a depolarization.
Given that our PCA and K-means clustering did not separate neurons into distinct groups based on a set of six properties, we examined whether AHP or ADP kinetics contributed to the generation of specific L2/3 PN subtypes. For this analysis, we separated neurons into three groups (Fig. 3A): (1) neurons with a fAHP followed by an mAHP [fAHP: 16 neurons (20.25%)], (2) neurons with a fAHP- ADP-mAHP pattern [fAHP-ADP: 28 neurons (35.44%)], and (3) neurons with only an mAHP (mAHP: 35 neurons, 44.30%). We then compared the intrinsic properties of these three subsets of neurons (mean values and statistics presented in Table 3 and post hoc test p values in Extended Data Table 3-1). AP kinetics indicated no significant difference in the depolarization slope (dV/dT), but we observed a significantly higher repolarization slope in fAHP neurons compared with mAHP neurons suggesting alterations in voltage-gated potassium channel conductances (Fig. 3D,E). We also observed no significant difference in AP half-widths (Extended Data Fig. 3-1A). As expected, mAHP neurons displayed the longest AHP latencies and also displayed the largest ΔAHP magnitudes compared with fAHP and fAHP-ADP neurons (Fig. 3F,G). These results suggest that the fAHP group of neurons may have more functional BK channels while neurons without fAHPs have less BK channels and more functional SK channels, which are known to dictate mAHP amplitude (Kohling and Wolfart, 2016).
Intrinsic properties of epileptic L2/3 PNs separated based on AHP shape and latency
Surprisingly, we found that there were no significant differences between AHP subtypes in terms of firing patterns. More specifically, there were no significant differences in initial instantaneous frequency (p = 0.94, KW test), final instantaneous frequency (p = 0.43), mean instantaneous frequency (p = 0.66), or FR accommodation ratio (p = 0.08; Fig. 3J–M, Extended Data Fig. 3-1C–F, Table 3, Extended Data Table 3-1). We also note that accommodating, RS, and ES neurons included neurons that had all three types of AHP subsets. Notch neurons were made up of 83% fAHP-ADP PNs and 17% of fAHP PNs and stuttering neurons were all in the fAHP subtype (Fig. 3N). Together, our findings suggest that AHP and ADP kinetics do not play a significant role in determining the firing patterns of human epileptic L2/3 PNs, and therefore, the observed diversity of action potential and spiking properties are intrinsic to human epileptic L2/3 PNs and do not suggest specific subtypes of PNs with specific roles in epileptiform activity.
L2/3 PN intrinsic properties differ between histopathologic groups
In the above experiments, we grouped L2/3 PNs from patients with MCDs and other epilepsies (OEs) to demonstrate the variability of PN subtypes, given that some of the PN subtypes were relatively rare. However, we also wanted to determine if epileptic L2/3 PN intrinsic properties would differ between histopathologic groups (i.e., control, MCD, and other epilepsy). In this context, we found substantial differences in passive membrane properties and AP kinetics (mean values and statistics presented in Table 4 and post hoc test p values in Extended Data Table 4-1). Compared with control, we observed no significant difference in membrane decay but observed a significant decrease in voltage sag ratio (%) for both MCD and OE L2/3 PNs (Table 4, Extended Data Table 4-1, Fig. 4K), indicating that there may be possible alterations in Ih current. Interestingly, the decrease in epileptic L2/3 PN voltage sag compared with control L2/3 PNs correlates with extensive research demonstrating altered HCN channel expression in human and rodent epilepsies (Strauss et al., 2004; Albertson et al., 2011; Brennan et al., 2016; Lin et al., 2020; Concepcion et al., 2021). In addition, we found that MCD, but not OE, L2/3 PNs had higher input resistances compared with control (Fig. 4J). Since we observed significant variability in epileptic L2/3 PN input resistances, and previous studies have indicated that L2/3 PN morphologies change based on distance from the pial surface, we wanted to determine if the recording distance from the pial surface could potentially be contributing to this variability (Deitcher et al., 2017). Interestingly, we observed that the input resistance did not correlate with the recorded neuron's distance from the pial surface (r = 0.0002, p = 0.8522, n = 18). Therefore, this variability in input resistances is most likely due to the abnormal cell types found in MCD epilepsies, such as cytomegalic/dysmorphic neurons and immature PNs (Crino, 2015; Levinson et al., 2020).
Epileptic L2/3 PNs show minor differences between etiology, but crucial differences compared with control neurons. A, Representative traces from current (pA) ramp protocol (1,000 pA, 1 s) of L2/3 PNs from control (blue), MCD epileptic (purple), and other epileptic (OE) tissue (green). B, Representative traces from L2/3 PNs from control and epileptic subtype showing the corresponding −250 pA hyperpolarizing step (darker negative trace), rheobase step (darker positive trace), and 2× rheobase (lighter positive trace) depolarizing current step (600 ms steps). C, Percentage (%) of PNs with certain AHP cell type based on control and epilepsy subtype. D, Overlay of average AP of control and epileptic subtype L2/3 PNs. E, Overlay of average phase plots (dV/dT vs voltage) of control and epileptic subtype L2/3 PNs. F, MCD L2/3 PNs have a depolarized AP threshold compared with control. G, OE L2/3 PNs have significantly longer AP half-widths compared with MCD with trends toward longer half-widths in both epileptic subtype PNs compared with control PNs. H, Depolarization slopes (dV/dT) were significantly slower in both epileptic subtypes compared with control. I, AHP magnitude (mV) is significantly larger in both epileptic subtypes compared with control. J, Epileptic L2/3 PNs show a wide variability in input resistance. MCD L2/3 PNs on average have significantly higher input resistances compared with control and OE L2/3 PNs. K, Voltage sag is significantly smaller in both epileptic subtypes compared with control. L, Left, Mean FR ± SEM versus injected current (pA). L, Right, We observed lower max firing rate for both epileptic subtypes compared with control. M, Left, FR accommodation ratio ± SEM versus injected current (pA). M, Right, FR accommodation ratios indicate a lack of FR accommodation by OE L2/3 PNs compared with MCD L2/3 PNs with no differences compared with control. Although, on average, control PNs accommodate their frequency the most compared with the epileptic subtype PNs. Scatterplots include mean values ± SD.
Table 4-1
Group comparison values with post hoc adjusted p-values for statistically significant properties based on epilepsy subtype. The correlated post-hoc test for comparing intrinsic properties from Table 4. Download Table 4-1, DOCX file.
Intrinsic properties of L2/3 PNs split by control and epilepsy subtype
In terms of AP kinetics and firing patterns, we found no differences in AHP latency, ΔAHP, AP amplitude adaptation, or FR accommodation between control L2/3 PNs and those from epileptic foci (Table 4, Extended Data Table 4-1). However, epileptic L2/3 PNs had the propensity to generate longer AP half-widths and had significantly depolarized thresholds, slower AP depolarization slopes, and larger AHP magnitudes (Fig. 4F–I). These data suggest alterations in the expression and function of perisomatic voltage-gated sodium and potassium channels including Kv1, Kv2, Kv4, SK channels, and BK channels in epileptic L2/3 PNs. Moreover, we observed a significant decrease in epileptic L2/3 PN max FRs compared with control (Fig. 4L, right). We postulate that although epileptic foci L2/3 PNs show signs of what is canonically considered hypoexcitable, they display trends toward longer APs and reduced activation of HCN channels. Together, this may cause heightened input summation, and therefore, more excitatory neurotransmission.
Epilepsy etiology influences L2/3 PN properties
When comparing epileptic subtypes to one another, we noted few differences in intrinsic properties. Namely, the OE L2/3 PNs had lower input resistances, longer AP half-widths, larger AP amplitudes, and a lack of FR accommodation when compared with MCD L2/3 PNs (Table 4, Extended Data Table 4-1, Fig. 4G,J,M). This indicates that OE L2/3 PNs fire just as frequently (still less than control PNs) yet sustain firing with longer AP half-widths and larger AP amplitudes compared with MCD L2/3 PNs. When comparing the maximum depolarization and repolarization slopes of the AP (dV/dT), we found no significant difference between epileptic subtypes. We did find that OE L2/3 PNs had slightly faster depolarizations and slower repolarizations compared with MCD L2/3 PNs. Therefore, we postulate that these intrinsic differences may be due to dysregulated transcription, post-translational modifications, or potentially somatic mutations leading to altered function of the voltage-gated channels that regulate AP kinetics (Guerrini et al., 2015; Brennan et al., 2016; Iffland and Crino, 2017; Subramanian et al., 2019).
4-AP influences L2/3 PN intrinsic properties
Acute brain slices resected from epileptic foci do not generate spontaneous seizure-like activity ex vivo. Therefore, prior studies have relied on the addition of convulsant drugs to induce epileptiform activity. 4-aminopyridine (4-AP) has been used for decades as one of these convulsant drugs and works by blocking A-type potassium currents arising from the Kv1, Kv3, and Kv4 families of potassium channels (Kohling and Avoli, 2006; Guerrini et al., 2015; Williams and Hablitz, 2015; Cepeda et al., 2018; Chang et al., 2019; Levinson et al., 2020). 4-AP has been shown to block fast AHPs, increase AP half-widths, and modulate neuron firing rates and FR adaptation (Lorenzon and Foehring, 1992; Faber and Sah, 2002). However, to our knowledge, few prior studies have investigated the direct effect of 4-AP on the intrinsic properties of L2/3 PNs from epileptic human tissue. Since 4-AP is so widely used in ex vivo epilepsy electrophysiology recordings, we believed it was critical to determine whether 4-AP affects epileptic PNs differently than nonepileptic PNs, in order to adequately interpret studies (ours and others’) utilizing 4-AP to induce ictal activity in human epileptic tissue.
To this end, we performed whole-cell current-clamp recordings on L2/3 PNs and elicited hyperpolarizing and depolarizing current steps before and after 4-AP (100 μM) wash on. When comparing L2/3 PNs before and after 4-AP wash on (n = 17, 8), we found that 4-AP induced several alterations in epileptic L2/3 PN passive membrane properties, AP properties, and firing properties (Fig. 5, Table 5). As expected, 4-AP induced an increase in AP half-width, an increase in AHP latency, a decrease in AHP magnitude, and a hyperpolarization of the AP threshold, most likely through blockade of Kv1 and Kv4 channels located at the perisomatic region and axon initial segment (Guan et al., 2006, 2007, 2013, 2015; Pathak et al., 2016; Nguyen and Anderson, 2018; Gonzalez Sabater et al., 2021; Fig. 5E–G). 4-AP had no significant effect on voltage sag (%; Table 5), AP amplitude (Fig. 5H), or max FR (Fig. 5I, right). However, 4-AP did cause a subjective increase in EPSP frequency (data not shown), induced a significant decrease in L2/3 PN input resistance, and caused significant fluctuation of FR accommodation at more depolarizing potentials (400, 500, 600, 750, 800, 850 pA steps) due to ∼1-s-long sporadic hyperpolarizations (Fig. 5C,J). Lastly, we found no significant difference in the response to 4-AP between epileptic subtype L2/3 PNs [MCD (n = 6) and OE (n = 11), data not shown] suggesting that while both MCD and OE L2/3 PNs have dysfunctional Kv channel function compared with control (Fig. 4), they share similar A-Type Kv channel function (Fig. 5).
4-AP increases L2/3 PN AP half-width and AHP latency and decreases AHP magnitude leading to sustained firing. A, Overlay of a representative AP before and after 4-AP wash on (scale bars: 10 mV, 10 ms). Bi, Overlay of average AP before and after 4-AP wash on. Bii, Overlay of phase plots. C, Input resistance (MΩ) decreased after 4-AP wash on. D, AP half-width increased after 4-AP wash on. E, AP threshold became hyperpolarized after 4-AP wash on. F, AHP magnitude was reduced after 4-AP wash on G, AHP latency got slower after 4-AP wash on. H, AP amplitude was not different after 4-AP wash on. I, Left, Mean FR ± SEM versus injected current (pA). I, Right, No significant difference was found for mean Max FR after 4-AP wash on. J, FR accommodation ratio ± SEM versus injected current (pA) before and after 4-AP wash on at 400, 500, 600, 750, 800, and 850 pA depolarizing current steps indicates a significant absence of accommodation after 4-AP wash on. Scatterplots include mean values ± SD.
4-AP induces changes of epileptic L2/3 PN intrinsic properties
Discussion
To our knowledge, this is the first systematic study focused on characterizing intrinsic and morphological properties of human L2/3 neocortical PNs from within the epileptic focus and comparing these to control brain tissue taken from patients without any history of epilepsy. In the present study, we demonstrate clearly distinct subtypes of L2/3 PNs in human pediatric epileptic neocortex. We demonstrate that the presence of fAHPs or mAHPs do not alone dictate the excitability or firing patterns of epileptic L2/3 PNs. Additionally, we present the new finding that human epileptic L2/3 PNs exhibit reduced voltage sag, reduced max FR, and many PNs with prolonged AP half-widths compared with control PNs. These changes in intrinsic properties suggest that L2/3 PNs may allow for more synaptic input summation and increased glutamate release from a single AP. We demonstrated that there are subtle but significant differences in the intrinsic properties of L2/3 PNS from MCD and OE groups, suggesting distinct mechanisms of epileptogenesis in MCDs. Finally, we demonstrated that 4-AP leads to an increased AP half-width, a hyperpolarized AP threshold, and increased AHP magnitude in human epileptic neurons and that epileptic subtype does not dictate L2/3 PN response to 4-AP.
Human L2/3 PNs are intrinsically diverse
We observed that human L2/3 PNs from within the epileptic focus show a large variety of firing patterns and AP kinetics, as previously described in control L2/3 human cortical PNs (Moradi Chameh et al., 2021). Specifically, we found L2/3 PNs that exhibit regular spiking, early spiking, stuttering, and accommodating firing patterns. The last group which we termed “notch” neurons are of particular interest as they exhibit firing patterns similar to low-threshold spiking (LTS) interneurons, and yet they appear to be a subclass of PNs. Additionally, we have not encountered this cell type in mouse neocortex. In mouse cortex, LTS interneurons possess a fAHP followed by an ADP and usually express cholecystokinin (CCK), calretinin (CR), or somatostatin (SST; Beierlein et al., 2003; Tremblay et al., 2016). However, we demonstrate that morphologically, these L2/3 notch neurons are PNs. Thus, notch neurons may be a class of human L2/3 PNs not observed in mouse cortex. We postulate that the notch neuron may play a specific role in L2/3 as a network activator with the ability to vigorously respond and then adapt to a similar firing rate as RS neurons. Future studies investigating the transcriptomic profiles of the epileptic L2/3 PN subtypes, that we reported here, would help elucidate if the differences in intrinsic properties relate to gene expression profiles. Transcriptomic analysis may also provide insights into different proportions of the L2/3 PN subtypes in various subtypes of epilepsy (Berg et al., 2021).
AHP kinetics alone do not dictate firing properties of human epileptic L2/3 PNs
AHP latency and shape has previously been thought to play a critical role in dictating firing rate and firing patterns, in association with the expression of neuronal potassium channels (Avoli and Olivier, 1989; Foehring et al., 1991; Lorenzon and Foehring, 1992; Faber and Sah, 2002; Higgs and Spain, 2009; Rizzo et al., 2014; Mendez-Rodriguez et al., 2021). Similar to previous reports on human PNs, we observed epileptic L2/3 PNs with several AHP shapes and some neurons with ADPs (Foehring et al., 1991). The differences in AHP imply differences in voltage-gated potassium channels (Kv1, Kv2, and Kv4) and/or calcium-dependent potassium channels (BK and SK channels). However, contrary to previous findings, we observed no differences in initial, maximum, or final instantaneous firing frequencies between AHP subtypes (Dodson et al., 2002; Toledo-Rodriguez et al., 2004; Guan et al., 2007, 2013; Shruti et al., 2008; Pathak et al., 2016; Rathour et al., 2016). Furthermore, we observed no differences in AHP magnitude, input resistance, or AP half-width. Our data suggest that the potassium channels that play a role in AHP shape and the corresponding firing pattern are compensatory in epileptic L2/3 PNs.
Epilepsy etiology influences L2/3 PN properties
In this study, we included specimens from patients with TSC, HME, and FCD in the MCD epilepsies group. These epileptic subtypes are characterized by disruption of the mTOR signaling pathway leading to cortical dyslamination, neuronal heterotopia, and dysmorphic neurons (Cepeda et al., 2010, 2012; Crino, 2015). Thus, we included these pathologies within a single group to determine if PI3K-AKT-mTOR pathway mutations lead to significant changes in excitatory neuron intrinsic properties (Crino, 2011; Lasarge and Danzer, 2014; Lipton and Sahin, 2014; Blumcke and Sarnat, 2016; Hanai et al., 2017; Nguyen and Anderson, 2018; Park et al., 2018). The OEs in this study included gliosis and tumor-related epilepsies. These epilepsies are not specifically related to enhanced mTOR signaling but instead, encompass a wide range of idiopathic epilepsies with most having unclear origins (Losi et al., 2012). Gliosis is a common, but nonspecific pathological finding in patients with drug-resistant epilepsies and is characterized by the presence of reactive astrocytes, which are known to release trophic factors that promote axonal sprouting and synaptic formation (O'Dell et al., 2012). In addition, astrocytes play a crucial role in neurotransmitter concentrations and the clearance of the [K+]o via inward rectifying K+ channels (Losi et al., 2012). Therefore, reactive astrocytes can facilitate improper synaptic transmission, shift neuronal excitability, and thereby, decrease seizure thresholds. The other, less common, OE cases were epileptic cases in which tumors caused seizures, such as glioma or dysembryoplastic neuroepithelial tumor (DNET). To our knowledge, there is no evidence that suggests that embryonic upregulation of the PI3K-AKT-mTOR pathway, which causes MCDs, is related to these cases, and in Type III FCDs, whether the congenital FCD occurs first or is a result of the neoplasia (Najm et al., 2018).
With the considerable differences in epilepsy subtypes, we determined the electrophysiological differences between their L2/3 PNs in order to further understand how epilepsy etiology can differ so significantly yet cause the same phenotype. With this in mind, we surprisingly observed only a few differences in L2/3 PN intrinsic properties between MCD and OEs. Specifically, we found OE L2/3 PNs had increased AP amplitudes, AP half-widths, lower input resistances, and less FR accommodation compared with MCD L2/3 PNs. This indicates that OE L2/3 PNs are effectively more hyperexcitable compared with MCD L2/3 PNs. We believe the differences compared with MCD L2/3 PNs may be due to maturation differences and the consequent abnormalities of morphologic and electrophysiologic properties. Nonetheless, our results suggest that epileptic L2/3 PNs across epileptic etiologies are more similar than different, and therefore, the cause of epilepsy may not outweigh the influences of seizure activity on excitatory neuron properties.
Epileptic L2/3 PNs are significantly different from tumor control L2/3 PNs
Compared with control L2/3 PNs, we observed a significant decrease in max FRs in both epileptic subtype L2/3 PNs. This was surprising considering hyperexcitable neurons are usually thought to have higher FRs. We postulate this may be a homeostatic response due to seizure activity or may result from improper neuronal development, migration, and synaptic plasticity (Andreae and Burrone, 2014; Curatolo et al., 2018; Jarero-Basulto et al., 2018; Park et al., 2018; Buchsbaum and Cappello, 2019; Medvedeva and Pierani, 2020). Furthermore, we observed that many L2/3 PNs in both epileptic subtypes possessed markedly longer AP half-widths and heightened input resistances with a considerable overlap in FRs with control PNs up to 20 Hz. When taken together, we postulate that epileptic PNs may fire less when maximally stimulated but may still release more excitatory neurotransmitters (Yang and Wang, 2006).
We also observed a reduction of voltage sag in both epileptic subtypes compared with control leading us to suspect that both MCDs and OEs lead to dysregulation of HCN channel function. In neocortical PNs, HCN channels localize primarily at the dendrites and are responsible for the Ih current, a nonspecific cation current that regulates the integration and summation of synaptic inputs (Brennan et al., 2016). Ih has previously been shown to reduce temporal summation of synaptic inputs, and therefore, reduction of Ih could lead to increased synaptic summation and contribute to network hyperexcitability (Magee, 1998; Kalmbach et al., 2018). Interestingly, dysregulated HCN channel expression and reduction in Ih current have been demonstrated in both human epilepsies and rodent epilepsy models and this dysregulation is widely considered to be proepileptogenic (Strauss et al., 2004; Brennan et al., 2016).
When comparing intrinsic properties of L2/3 PNs from MCD and control cortical tissue, we found that the MCD L2/3 PNs demonstrate a more depolarized threshold and larger AHP magnitude. One possible explanation for these findings is that an increased extracellular potassium concentration has been shown to lead to a depolarized AP threshold and increased AHP magnitude (Powell and Brown, 2021). Hyperexcitable networks often lead to increased extracellular potassium due to increased numbers of APs during ictal or interictal activity (D'Antuono et al., 2004; Avoli et al., 2005; Powell and Brown, 2021). This may also occur in OE slices as their PNs also demonstrated an increased AHP magnitude. Another explanation for these AP changes would be changes in the functionality or expression of potassium channel subunits. AP threshold is modulated by Kv1 and Kv4, whereas AHP magnitude is regulated by BK and SK channels (Lorenzon and Foehring, 1992; Pineda et al., 1992; Faber and Sah, 2002; Fernandez de Sevilla et al., 2006; Guan et al., 2007; Pathak et al., 2016).
One limitation of these comparisons is the variability in recorded cortical regions. Our control tissue included parietal and temporal lobe L2/3 PN recordings while epileptic tissue recordings were made across the cerebral cortex. Previous research has found significant differences between neocortical regions in terms of PN cytoarchitectural structure and PN intrinsic properties (Jacobs et al., 2001; DeFelipe, 2011; Mohan et al., 2015; Berg et al., 2021; Moradi Chameh et al., 2021; Benavides-Piccione et al., 2024). In this study, we utilized control tissue from patients with no history of epilepsy or use of ASMs in order to decipher the influence of ictal activity on neuronal intrinsic properties. Even at our busy neurosurgical center which performs over 800 cases per year, these samples are extremely rare. Although there are differences in the cortical origin of the control, OE, and MCD groups, we believe that the variability among cortical regions is less than the variability between phenotypes. Further research is needed to gain higher resolution of PN intrinsic properties based on cortical location and epilepsy phenotype. This limitation highlights the need for multicenter collaboration and online sharing of data in a standardized form.
4-AP influences L2/3 PN active and passive membrane properties
Prior studies on the influence of 4-AP on human neocortical neurons have shown that 4-AP abolishes the fast AHP and increases AP half-width through blockade of perisomatic Kv1 and Kv4 channels. We observed a similar effect, as 4-AP caused an increase in AP half-widths, increased AHP latencies, and decreased AHP magnitudes. To our surprise, we observed a significant decrease in input resistances after 4-AP application, which is contrary to previous reports of increased input resistance following blockade of A-type potassium channels (Rathour et al., 2016). The mechanism behind the decreased input resistances following 4-AP wash on is currently unclear but may be due to the hyperpolarizing events that occurred during current step recordings. Overall, with the increase in AP half-width, and no significant change in AP amplitude, 4-AP seemingly elicits L2/3 PNs to fire longer APs at the same efficiency most likely leading to an increase in glutamatergic neurotransmission, thereby promoting network hyperexcitability (D'Antuono et al., 2004; Avoli et al., 2005; Yang and Wang, 2006; Abdijadid et al., 2015; Dossi et al., 2024).
Potassium channels: antiseizure medications (ASMs)
Our data indicate that epileptic L2/3 PNs during pro-ictal, 4-AP administration display increased AP half-widths, hyperpolarized AP threshold, slower AHPs, and increased sustained firing at depolarized potentials. Therefore, increasing the activity of voltage-gated potassium channels at resting membrane potentials and/or stabilizing extracellular potassium could be beneficial therapeutic goals to reduce firing probability, decrease AP half-widths, and mitigate a hyperpolarized shift in AP threshold. Obviously, finding a single drug that will restore dynamic potassium channel function will be tricky. For example, there are over 80 genes for potassium channels in the human genome and previous reports indicate that both gain-of-function and loss-of-function mutations can cause seizures (Kohling and Wolfart, 2016; Zhao et al., 2024). However, as genetic testing becomes standard-of-care in patients with drug-resistant epilepsy, potassium channels should not be overlooked. Therapies aimed at correcting specific potassium channelopathies may provide significant benefit to patients.
Footnotes
The authors declare no competing financial interests.
These studies were supported by grants from the National Institutes of Health: R56 MH127213 (M.M.H.), R01NS095311 (M.M.H.), AES Junior Investigator Award (A.L.A.); NINDS Award 1K08NS131572-01A1 (A.L.A.).
This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International license, which permits unrestricted use, distribution and reproduction in any medium provided that the original work is properly attributed.
References
In this issue
