1. Introduction
KCNT1 encodes a sodium activated potassium channel, also known as SLACK (Sequence Like a Calcium Activated K
+ channel), K
Ca4.1 or Slo2.2 [
1]. The gene is highly expressed in the nervous system, and the KCNT1 channel is thought to regulate neuronal excitability by modulating depolarization following repetitive firing of action potentials [
1,
2,
3].
In 2012 we identified heterozygous
KCNT1 missense mutations in families and individuals with autosomal dominant nocturnal frontal lobe epilepsy (ADNFLE), now known as autosomal dominant sleep-related hypermotor epilepsy (ADSHE) [
4]. The mutations were either inherited or occurred de novo and were associated with focal seizures often arising from sleep. Concurrently, Barcia et al. reported de novo
KCNT1 missense mutations in sporadic cases of the developmental and epileptic encephalopathy (DEE) syndrome known as epilepsy of infancy with migrating focal seizures (EIMFS) [
5]. This type of infantile onset epilepsy is usually severe, with frequent focal seizures which are refractory to available epilepsy treatments. Bonardi et al. (2021) recently described 64 different
KCNT1 mutations identified in 248 individuals from around the world which provides insights into the clinical course and attempted management of KCNT1-related epilepsy [
6].
To date, all but two reported
KCNT1 mutations in epilepsy are heterozygous missense mutations, predicted to cause a single amino acid change in the KCNT1 potassium channel. The exceptions are a homozygous missense mutation, c.G2896A, A966T, reported in an individual with the epileptic encephalopathy Ohtahara Syndrome [
7] and a heterozygous in-frame deletion resulting in the deletion of a single amino acid, Gln550, causing EIMFS [
8]. KCNT1-related epilepsy affects both children and adults. Affected individuals display a spectrum of epilepsy phenotypes, ranging from intermittent focal seizures, most often in the frontal lobe and with an onset in adolescence, to highly frequent seizures beginning in the first few months of life and involving multiple brain regions. KCNT1-related epilepsy can be associated with comorbidities including intellectual disability, autism, and behavioural features and can lead to premature death [
6,
9]. Explanations for the pathologies and different phenotypes seen with
KCNT1 mutations are not known and require investigation of the biological effects of the mutations and any differential effects. Thus far, the effects of a relatively small number of
KCNT1 mutations have been analysed by electrophysiology to look at their effects on KCNT1 channel properties and to date each mutation has led to increased potassium current [
10,
11].
In this study we investigated the effects of a large series of different disease causing
KCNT1 patient mutations on the electrophysiological properties of the KCNT1 channel, analysed in the same system and at the same time, to provide consistency in making intra-experiment comparisons. We analysed the properties and kinetics of the mutated KCNT1 channel, beyond K
+ current amplitude, and looked for patterns between the observed effects and the phenotypes of the patient(s) with the particular mutation. The
KCNT1 mutations analysed in this study have all been identified in affected individuals and include nine not previously investigated by electrophysiology (T314A, N449S, L781V, E893K, M896V, F932L, S937G, L942F, and A965T) [
6]. Four patient mutations that have previously been analysed (G288S, R398Q, R928C and R961H) were included for inter-experiment comparison [
10]. One KCNT1 variant, S937G, identified in a sibling pair with severe epilepsy and co-morbidities, is previously unpublished and had been clinically classified as a variant of unknown significance. We have used electrophysiological analyses to investigate its effects on the KCNT1 channel to assess its mutational status.
2. Results
We selected 8
KCNT1 mutations that we recently published in patients with
KCNT1-related epilepsy that have not previously been functionally analysed: T314A, N449S, L781V, E893K, M896V, F932L, L942F and A965T [
6], and one
KCNT1 variant of unknown significance, S937G, for our analyses. We also included 4
KCNT1 mutations which have been previously functionally analysed by electrophysiology to allow for inter-experiment comparisons. To measure K
+ currents mediated by KCNT1 channels we employed whole cell patch clamping of HEK293T cells transfected with plasmids containing cDNA encoding either wild type (WT) or mutant YFP-tagged KCNT1 protein. WT KCNT1 currents elicited by 100 ms voltage ramps ranging from −120 to 120 mV exhibited strong outward rectification between −120 and 40 mV and saturation at voltages above 50 mV (
Figure 1A). All but one of the novel and previously described
KCNT1 mutations associated with epilepsy produced larger K
+ currents, compared to WT, with I-V plots of similar shape (
Figure 1B). The mutation that produced K
+ current with an amplitude similar to that of WT KCNT1, T314A, also abolished outward rectification of the current without affecting current saturation at highly positive potentials (
Figure 1B,C). Judging by the reversal potentials of the I-V plots, none of the mutations appreciably affected the selectivity of the channel pore (
Figure 1C). I-V plots of WT KCNT1 current recorded in inside-out patches retained non-linear characteristics of whole-cell I-V plots (
Figure 1D). Increasing Na
+ concentration on the intracellular surface of the membrane from 10 mM to 60 mM resulted in a significant increase in current amplitude, as expected, but also in a bell-shaped I-V plot, suggesting voltage dependent block of the channel pore by intracellular Na
+ at potentials above 40 mV. It is likely that whole-cell current saturation at highly positive membrane potentials accompanied by increased current noise is also a result of voltage-dependent block of the channels by Na
+ (
Figure 1A).
To compare the amplitudes of K
+ currents mediated by WT and mutant KCNT1 channels, HEK293T cells were transfected with the same amounts of plasmids carrying cDNA of ether mutant or WT
KCNT1 and the measurements were performed within a short time window between 24 and 28 h post-transfection. Employment of
KCNT1 constructs tagged with YFP allowed selection of transfected cells with similar levels of fluorescence and therefore similar levels of KCNT1 protein expression. In the control experiments all YFP-tagged
KCNT1 constructs were confirmed to produce K
+ currents indistinguishable from those of the corresponding untagged versions of
KCNT1. As expected, epilepsy-causing mutations significantly increased the amplitude of KCNT1 currents (
Figure 2). However, as stated above, there was a notable exception of the T314A mutation, which did not affect the amplitude of the current (
p = 0.6136).
We observed that some of the mutations affected not only the amplitude of the current, but also the kinetics of activation (
Figure 3), and the voltage dependence of the apparent open probability (
Figure 4). Out of 13 mutations, seven (G288S, R398Q, N449S, S937G, L942F, R961H and A965T) had little effect on the time constant of KCNT1 current activation at 600 ms steps to 60–100 mV (τ = 100 ÷ 200 ms), five (L781V, E893K, M896V, R928C and F932L) significantly accelerated activation kinetics (τ = 20 ÷ 50 ms), and one (T314A) abolished activation altogether (
Figure 3).
Using normalised tail currents, we have constructed apparent
Po curves which revealed a weak voltage dependence of KCNT1 open probability (
Figure 4). The average slope of the apparent
Po curves (48 ± 2,
n = 13) suggested a presence of a gating charge of about 0.5, however, the nature of this voltage dependence is not clear. Neither WT nor mutant KCNT1 channels were completely closed even at very negative potentials. Mutations that accelerated current activation at positive potentials also significantly increased minimum
Po at negative potentials, whereas T314A mutant exhibited virtually no voltage dependence at all (
Figure 4).
The
KCNT1 variant c.2809A > G, p.S937G is previously unpublished and functionally uncharacterised. The heterozygous variant was identified in siblings affected with epilepsy syndromes and co-morbidities (See
Supplementary Materials) previously associated with mutation of
KCNT1. Due to not having been reported previously and the inability to determine the inheritance pattern of the mutation, it was classified as a variant of unknown significance (See
Supplementary Materials for details of the mutation and the clinical phenotypes of the affected siblings). We have included the
KCNT1 S937G variant in this study to analyse its effects on the KCNT1 channel to assist in clinically classifying this variant. Our analysis showed that KCNT1 with this mutation produces a K
+ current of significantly larger amplitude compared to WT KCNT1, but with the kinetics and apparent
Po similar to that of the normal channel (
Figure 1,
Figure 2,
Figure 3 and
Figure 4,
Table 1).
To date it has been considered that the increased amplitude of the KCNT1 mediated current underlies the pathogenicity of
KCNT1 mutations. However, the T314A mutation investigated in this study produced a K
+ current of a similar amplitude to the WT. The channel with the T314A mutation lacked the voltage dependence present in WT KCNT1 and other mutants (
Figure 1,
Figure 2,
Figure 3 and
Figure 4). This mutation was identified in a child with late onset (15 months) developmental and epileptic encephalopathy, with global and severe neurodevelopmental delay. Interestingly, T314A is the only
KCNT1 mutation reported to date located in the P-loop domain of the KCNT1 channel [
6]. Considering that the
KCNT1 T314A mutation challenges the accepted paradigm of the pathobiology underlying KCNT1-epilepsy, we have investigated its properties in more detail. A crude estimation of the reversal potential using I-V plots (
Figure 1) might not have been sufficient to ascertain the subtle changes in the KCNT1 selectivity for K
+ over Na
+ caused by mutations. Therefore, we investigated T314A selectivity, and compared it to the selectivity of WT and M896V KCNT1, by replacing different amounts of NaCl in the bath solution with KCl (
Figure 5A–C) and calculating KCNT1 permeability for Na
+ relative to K
+ (P
Na/P
K) using the shifts in the reversal potentials of the I-V plots (
Figure 5D) and modified GHK equation (Equation (2),
Section 4). The data suggested that T314A mutation does not change the selectivity of KCNT1 and the relative permeability for Na
+ remains at approximately 0.01 for WT, T314A and M896V KCNT1 channels.
Another possibility that could explain pathogenic effects of T314A mutations is a steeper, compared to WT, dependence on intracellular Na
+. In the following experiment we have compared dependence of WT, T314A and M896V current amplitudes on intracellular Na
+ concentration. The amplitudes of KCNT1 currents were measured at −100 mV in the presence of 130 mM K
+ in the bath (126 mM NaCl in the control bath solution was replaced with 126 mM KCl) and using the pipette solution containing either 10 mM Na
+ (control intracellular solution) or 130 mM Na
+ (80 mM K glutamate and 40 mM KCl in the control intracellular solution were replaced with 80 mM Na glutamate and 40 mM NaCl, respectively) (
Figure 6). As expected, increasing intracellular Na
+ concentration from 10 mM to 130 mM potentiated WT KCNT1 current amplitude several fold (
Figure 6A). In contrast, the amplitude of the T314A current did not change in response to increased intracellular Na
+ concentration. For comparison, the amplitude of the M896V current did increase with higher intracellular Na
+ concentration, however, this increase was significantly smaller than in WT KCNT1 current (about 2 fold vs. 14 fold).
A shift in KCNT1 amplitude dependence on intracellular Na
+ towards lower concentrations has previously been reported for some KCNT1 mutants and has been suggested as an explanation of larger currents produced by these mutations at resting levels of intracellular Na
+ [
11]. Supporting this notion, at least for M896V mutant, the current amplitudes produced by WT and M896V mutant KCNT1 were vastly different in the presence of 10 mM Na
+, but indistinguishable at saturating levels of intracellular Na
+ (
Figure 6A,C).
Despite multiple attempts we were unable to record discernible single channel opening events in membrane patches expressing T314A KCNT1 (
Figure 7A). The noisy current traces were consistent with the presence of several ion channels with a small conductance exhibiting inward rectification in symmetrical K
+ concentration (130 mM) solutions (
Figure 7A(i)). The instantaneous I-V plots obtained by applying a ramp voltage protocol to the inside-out patches were similar to the I-V plots obtained on the whole cells expressing T314A mutant channels (
Figure 7A(ii) and
Figure 6B). For comparison, inside-out patches of cells expressing other KCNT1 construct consistently produced single channel currents under the same recording conditions (
Figure 7B, WT KCNT1 single channel recording is shown).
3. Discussion
In excitable cells, those that generate action potentials, the textbook roles of K
+ channels are to control the resting membrane potential, regulate membrane resistance, and to control the repolarisation rate of action potentials and the action potential frequency [
12]. In neuronal cells, in general, activation of K
+ channels counteracts Na
+- and Ca
2+-mediated excitation. Not surprisingly, dysfunction of K
+ channels can cause a multitude of neurological disorders [
13,
14]. Loss of function mutations in different types of K
+ channels lead to increased excitability of neuronal circuits in the brain or spinal cord [
13]. Unexpectedly, gain of function mutations in some types of K
+ channels, including KCNT1 studied here, can also lead to hyperexcitability of neuronal circuits in the brain and consequently seizures [
14,
15]. A simple explanation of such a phenomenon is that these gain-of-function (GoF) mutant K
+ channels are mainly expressed in inhibitory neurons and silencing of the inhibitory neurons due to abnormally large K
+ conductance leads to disinhibition of the neuronal circuits and therefore increased excitability [
14]. This notion is supported by the recent data obtained using a mouse model expressing human Y796H GoF
KCNT1, demonstrating that increased Na-dependent K
+ currents (Ik
Na) impair GABAergic neuron excitability and alter synaptic connectivity [
16]. Another recent study of a mouse model expressing L437F GoF
KCNT1 has revealed that KCNT1 channels are predominantly expressed in GABAergic parvalbumin-positive interneurons of the hippocampus, which exhibit a significantly reduced excitability, compared to this type of interneurons in the WT mice [
17]. In contrast, data obtained from induced pluripotent stem cells (iPSC)-derived neurons expressing homozygous
KCNT1 P924L GoF mutation suggest the possibility of a different mechanism [
18]. P924L-expressing neurons with Ik
Na increased several fold, responded to stimulation with a higher number of APs as well as with higher maximal firing rates, suggesting that
KCNT1 GoF mutations may increase excitability of the excitatory neurons [
18]. The reason for such a discrepancy is not clear, however, this may indicate that different
KCNT1 GoF mutations cause epilepsy by different mechanisms.
Analysis of a
KCNT1 mutation classified as a variant of unknown significance (S937G), and therefore not reported clinically, has shown that it produces K
+ currents with the characteristics similar to those of some other pathogenic
KCNT1 mutants investigated here (
Table 1). Based on the electrophysiological effects of this variant it would now be classed as pathogenic. This highlights the importance of functional characterisation, and demonstrates the power of electrophysiological data, to assist in the classification of pathogenicity for ion channel variants, thus aiding in the genetic diagnoses and clinical care of patients.
Seven of the 13 mutations studied in this work are located in, or adjacent to, the RCK2 domain, two are in the NAD binding domain, two in the RCK1 domain, one is in the P-loop and another one in the S5 domain (
Table 1,
Figure 8). The majority of known mutations that cause epilepsy are localised in the intracellular RCK domains and the membrane spanning S5 domain [
6]. Many of the epilepsy causing mutations increase the sensitivity of KCNT1 channel to intracellular Na
+, which explains large macroscopic K
+ currents mediated by these mutant KCNT1 channels at the resting levels of Na
+ concentration [
11]. However, there is evidence that epilepsy-causing mutations increase positive cooperativity of the KCNT1 channel opening, which can also explain the increase of the macroscopic current amplitude [
19].
The only mutation found so far within the pore-forming P-loop domain is T314A (
Figure 8). Close proximity of this mutation to the selectivity centre explains drastically reduced single channel conductance of T314A mutant and the reduced, compared to the other epilepsy-causing mutations, whole-cell K
+ conductance. All mutations investigated in this study, except for T314A, produced KCNT1 currents of larger amplitude, compared to the WT channel. Some mutations have also affected the kinetics of activation as well as the voltage dependence of the apparent open probability. To gain insights into the relation between the characteristics of the mutant KCNT1 channel with the severity of the epilepsy caused by the mutations in humans, we employed Spearman correlation analysis. Using the available clinical data, each mutation was assigned the score between 1 (less severe) and 3 (more severe) (
Table 1) and Spearman correlation coefficient, r, between the severity of the disease, KCNT1 current amplitude, relative
Po at −80 mV, and kinetics was calculated using GraphPad Prism 8 (GraphPad Software Inc., San Diego, CA, USA). The analysis suggested that there is no correlation between the severity of disease and KCNT1 current amplitude (r = 0.29;
p = 0.336). There was, however, a strong correlation between the severity and the apparent
Po around resting membrane potential (r = 0.72;
p = 0.007). There were also correlations between the disease severity and KCNT1 kinetics (r = −0.70;
p = 0.010), most likely due to a very strong correlation between the current kinetics and the
Po (r = −0.82;
p = 0.003). This suggests that for the mutations tested here, open probability of KCNT1 channels at the resting potential is a strong predictor of the severity of the disease. This also suggests that the most likely mechanism, by which these KCNT1 mutant channels affect neuronal excitability, is by increasing resting membrane K
+ conductance and therefore working as a brake for action potential firing. In such a case the effects of
KCNT1 mutations on the neuronal excitability must be more pronounced in the inhibitory than the excitatory neurons. This corresponds well with the data obtained from mouse models of KCNT1-mediated epilepsy [
16,
17]. Importantly, the effects of T314A, which abolished voltage and Na
+ dependence (at least between 10 and 130 mM) of KCNT1, suggest that the pathogenicity of this mutation is due to increased resting K
+ conductance. Increased resting K
+ conductance acts as a break for action potential firing, which strongly indicates the predominant effect of
KCNT1 GoF mutations on the inhibitory neurons [
22,
23]. Similarly, the pathogenicity of the gain-of-function mutations in Kv7.2 and Kv7.3 K
+ channels found in patients with epileptic encephalopathies, which increase the open probability of the channels at the resting membrane potential, was also suggested to be due to a decreased excitability of the inhibitory neurons and disinhibition of the neuronal circuits [
24].
Clearly, GoF mutations which do not affect the Po of the KCNT1 channel at the resting membrane potential also increase the resting K+ conductance due to the fact that KCNT1 does not completely close even at very negative potentials. To ascertain whether the KCNT1 current amplitude can be used as a predictor of the disease severity for these mutations, the mutations that increase KCNT1 apparent Po at negative potentials were removed from consideration. In this case, the analysis revealed that, indeed, KCNT1 mutations that have voltage dependence similar to WT KCNT1 do exhibit strong correlation between the current amplitude and the disease severity (r = 0.72, p = 0.033).
It should be acknowledged, however, that the scoring of the mutations’ phenotypic severity has its limitations due to variations in availability of clinical data, different numbers of patients carrying specific mutations, and possible co-morbidities unrelated to KCNT1 mutations. The epilepsy phenotypes associated with each of the KCNT1 mutations could be expanded in the future, as informed by further genetics studies, which may alter the scoring of the mutation phenotypes.
In conclusion, we investigated the effects of nine novel and four known KCNT1 mutations on the biophysical properties of KCNT1 currents, including whole cell current amplitude, apparent open probability, current kinetics, and single channel conductance (T314A mutant). The only difference that we found between WT KCNT1 and T314A that could be described as gain of function is the open probability at the resting potential, which is significantly higher in T314A. The analysis of correlation between the symptoms caused by the different KCNT1 mutations and the properties of KCNT1 current suggests for the first time, that KCNT1 Po at resting membrane potential is a strong predictor of the severity of KCNT1-related epilepsy and supports the notion that pathogenicity of KCNT1 mutations is caused by their effects on the resting K+ membrane conductance, rather than the current amplitude at positive potentials.