Targeting Kv7 Potassium Channels for Epilepsy

Abstract

Voltage-gated Kv7 potassium channels, particularly Kv7.2 and Kv.7.3 channels, play a critical role in modulating susceptibility to seizures, and mutations in genes that encode these channels cause heterogeneous epilepsy phenotypes. On the basis of this evidence, activation of Kv7.2 and Kv.7.3 channels has long been considered an attractive target in the search for novel antiseizure medications. Ezogabine (retigabine), the first Kv7.2/3 activator introduced in 2011 for the treatment of focal seizures, was withdrawn from the market in 2017 due to declining use after discovery of its association with pigmentation changes in the retina, skin, and mucosae. A novel formulation of ezogabine for pediatric use (XEN496) has been recently investigated in children with KCNQ2-related developmental and epileptic encephalopathy, but the trial was terminated prematurely for reasons unrelated to safety. Among novel Kv7.2/3 openers in clinical development, azetukalner has shown dose-dependent efficacy against drug-resistant focal seizures with a good tolerability profile and no evidence of pigmentation-related adverse effects in early clinical studies, and it is now under investigation in phase III trials for the treatment of focal seizures, generalized tonic-clonic seizures, and major depressive disorder. Another Kv7.2/3 activator, BHV-7000, has completed phase I studies in healthy subjects, with excellent tolerability at plasma drug concentrations that exceed the median effective concentrations in a preclinical model of anticonvulsant activity, but no efficacy data in patients with epilepsy are available to date. Among other Kv7.2/3 activators in clinical development as potential antiseizure medications, pynegabine and CB-003 have completed phase I safety and pharmacokinetic studies, but results have not been yet reported. Overall, interest in targeting Kv7 channels for the treatment of epilepsy and for other indications remains strong. Future breakthroughs in this area could come from exploitation of mechanistic differences in the action of Kv7 activators, and from the development of molecules that combine Kv7 activation with other mechanisms of action.

Key Points

Voltage-gated type 7 potassium channels (Kv7), encoded by the KCNQ gene family and comprising five different families (Kv7.1 to Kv7.5), are important modulators of the function of excitable cells.

In the central nervous system, Kv7.2 and Kv7.3 play a particularly relevant role in controlling neuronal excitability and susceptibility to epileptic seizures.

Ezogabine (retigabine), the first clinically effective Kv7.2/3 activator introduced for the treatment of focal seizures, was withdrawn from the market due to declining use after discovery of off-target pigmentation-related adverse effects. A recent trial of a pediatric formulation of ezogabine (XEN496) in the treatment of KCNQ2-related developmental and epileptic encephalopathy was terminated prematurely for reasons unrelated to safety.

Among novel Kv7.2/3 openers, azetukalner (XEN 1101) is currently in phase III development for the treatment focal seizures, generalized tonic-clonic seizures and major depressive disorder. Initial clinical data suggest robust antiseizure activity and good tolerability, with no pigmentary changes or other major safety issues reported.

Other Kv7 activators in clinical development include BHV-7000, pynegabine, and CB-003. Publicly available information on the preclinical and clinical pharmacology of these compounds is still very limited.

Introduction

Establishing neuronal Kv7 channels as a validated target for antiseizure medications (ASMs) is the destination of a fascinating journey through many disciplines. In this article, we review the most relevant milestones of this journey. After describing the role of disrupted potassium ion (K⁺) homeostasis in seizure activity and the main K⁺ channels involved, we highlight the role of a specific neuronal K⁺ current (the M-current) in regulating brain excitability, and Kv7 channels as the molecular basis of the M-current. We subsequently discuss the structure of neuronal Kv7 channels and their functioning, as well as human epilepsy phenotypes associated with variants in the genes encoding these channels. This information provides the conceptual basis for understanding the effects of medications targeting Kv7 channels for the treatment of seizure disorders, which include ezogabine (retigabine), the first-in-class Kv7 opener approved as an ASM, and the investigational drugs azetukalner (XEN1101), BHV-7000, and pynegabine. For each of these compounds, we summarize information on their preclinical pharmacology and, when available, results of human studies completed to date.

Role of K⁺ Channels in Seizure Activity

Role of K⁺ Channels in Modulating Neuronal Excitability

An epileptic seizure is defined as “a transient occurrence of signs and/or symptoms due to abnormal excessive or synchronous neuronal activity in the brain” [1]. Although the cellular and network mechanisms underlying epileptic seizures are still unclear, seizures ultimately result from an imbalance between excitation and inhibition, leading to hyperexcitability of selected neuronal subpopulations. Alterations in K⁺ homeostasis are a recognized mechanism underlying neuronal excitability since they have been associated with seizures in animal models and have an etiological role in some human epilepsies [2]. Evidence for an increased extracellular K⁺ concentration ([K⁺]_o) in the vicinity of neurons involved in seizure activity has suggested a correlation between network activation and [K⁺]_o changes during seizure generation. The “K⁺ accumulation hypothesis” postulates that a high [K⁺]_o is sufficient to trigger neuronal depolarization coupled with an increased firing rate, leading in turn to further increase in [K⁺]_o [3]. To maintain physiological neuronal function, K⁺ must be cleared from the extracellular space following neuronal activation, a process in which astrocytes play critical roles [4].

Channels selectively permeable to K⁺ (K⁺ channels) are critical for K⁺ homeostasis at distinct cellular locations in the brain [5, 6], and also at distinct subcellular regions (dendrites, soma, axons, axon initial segment) within specific neurons [7]. Pathological changes in the expression and function of these channels facilitate seizure activity [5–7]. In neurons, K⁺ channels mostly exert an inhibitory membrane-stabilizing action because their opening drives the membrane potential closer to the K⁺ equilibrium potential, away from the action potential threshold. Activation of K⁺ channels reduces firing frequency and decreases the efficiency of excitatory inputs. Under certain conditions, however, enhanced K⁺ channel activity may lead to seizure activity, as revealed by the existence of gain-of-function variants in K⁺ channel-encoding genes associated with human epilepsy syndromes. This “unresolved paradox” [8] could be due to (a) disinhibition (i.e., reduction in inhibitory interneuron activity) or (b) a recurrent neuron–astrocyte–neuron excitatory loop [9], (c) faster action potentials repolarization and sodium (Na⁺) channels repriming caused by hyperactivation of BK channels [10], or (d) hyperpolarization-activated nonselective cation currents [11], resulting in secondary depolarization.

Many K⁺ channel subtypes are convergently regulated by membrane receptors, intracellular signal pathways, and second messengers, thereby integrating neuronal metabolism and excitability. Beside these roles in neurons, K⁺ channels participate in solute transport across endothelial cells [12] and to glial cells K⁺ clearance from brain interstitial spaces [13].

Heterogeneity of K⁺ Channels

Each cell type at various differentiation stages expresses a specific pattern of K⁺ currents, each with distinct subcellular localization, biophysical properties, modulation, and pharmacological profile. Further, electrophysiological studies have revealed that K⁺ channels are the largest and most functionally heterogeneous class of ion channels, and that they are expressed in all eukaryotic cells and in prokaryotes [14]. Such functional diversity parallels a remarkable genetic and structural heterogeneity [15, 16].

More than 70 genes encoding K⁺ channel proteins have been identified in humans. On the basis of their presumed topology deduced from primary sequences of the pore-forming α-subunits, these channels have been classified into those with six, four, or two putative transmembrane segments [5]. Each K⁺ channel family, in turn, includes different channel subtypes. In addition to α-subunits, many K⁺ channels contain nonconductive, auxiliary β-subunits that modulate gating properties, inactivation, cell surface expression, or trafficking of the ion channel complex, and may serve as binding sites for endogenous and exogenous ligands.

Six-transmembrane (6TM) channels include voltage-gated K⁺ channels (Kv), calcium (Ca²⁺)-dependent K⁺ channels (KCa), and Na⁺-dependent K⁺ channels (KNa) [17]. Functional channels are formed by the tetrameric association of identical (homotetramers) or compatible (heterotetramers) pore-forming α-subunits. Each α-subunit consists of a core region with six or seven transmembrane segments (from S0/S1 to S6) and a single-pore region. Kv channels are activated by membrane depolarization and play a major role in determining membrane excitability and duration of the action potential. These channels include several subfamilies consisting of different α-subunits (Kv1–Kv12) [18], and show marked heterogeneity depending on their subunit composition, presence of modifier subunits, accessory proteins, and post-translational modifications. KCa channels, in addition to voltage, depend on intracellular Ca²⁺ for opening, and are subdivided into three main subgroups according to their single-channel conductance (SK, small conductance; IK, intermediate conductance; BK, large conductance). KNa channels, are voltage-independent and open in response to increased intracellular Na⁺ concentrations. Given that changes in intracellular Ca²⁺ and Na⁺ concentration often occur during metabolic stress (e.g., ischemia, intense neuronal activity, energy deprivation), KCa and KNa channels link cell excitability with cell signaling and metabolism.

Four-transmembrane (4TM) channels, also known as two-pore channels (K2P), assemble as dimers of α-subunits consisting of four transmembrane segments and two pore domains. These channels have a critical role in setting the resting membrane potential and contribute to modulation of action potential duration and responsiveness to synaptic stimuli.

Two-transmembrane (2TM) channels consist of tetramers of α-subunits containing two transmembrane segments (M1 and M2, homologous to the S5 and S6 transmembrane segments of Kv channels), and an amphipathic loop in between where critical determinants for conduction and selectivity are located. The 2TM channels are voltage independent. Inward-rectifier channels belong to this group, which includes at least seven gene families (KIR1–KIR7). Inward-rectifying channels drive currents inward (into the cell) more easily than outward because of a high-affinity block by endogenous intracellular polyamines and magnesium ions that plug the channel pore at positive potentials [19, 20]. At membrane potentials more negative than the resting potential, inward-rectifying K⁺ channels (Kir) support the flow of K⁺ into the cell, pushing the membrane potential back to the resting potential. Kir channels mediate essential functions such as resting membrane potential stabilization, coupling of energy metabolism with membrane excitability, and maintenance of K⁺ homeostasis. Important examples include the adenosine triphosphate (ATP)-gated K⁺ channel (K_ATP), the G protein-coupled inward rectifier K⁺ (GIRK) channels that mediate effects of several neuromodulators, and the Kir 4.1 channel expressed in the end-foot processes of astrocytes.

The Rationale for Targeting Kv7 Channels to Control Seizure Activity

Physiology of the M-Current (I_KM)

The M-current (I_KM) is a voltage-gated K⁺ current first characterized in bullfrog sympathetic neurons in the 1980s. In these neurons, acetylcholine released from preganglionic cells triggers fast nicotinic excitatory postsynaptic potentials (fEPSPs) as well as slow excitatory membrane responses (sEPSPs) [21]. While fEPSPs are typically generated by a transient increase in postsynaptic membrane conductance, sEPSPs are associated with a decrease in membrane conductance [22] caused by inactivation of a K⁺ current [23]. When compared with fEPSPs, these slow responses showed long synaptic delays (tenths to hundredths of ms), were long-lasting (many seconds) and were due to acetylcholine binding to M1, M3, or M5 muscarinic receptors, hence the name “M-current (I_KM)” [24] for the K⁺ current whose suppression generated sEPSPs [23]. Conductances with biophysical characteristics similar (if not identical) to those of I_KM were later identified in other neurons, including mammalian hippocampal and cortical pyramidal cells [25].

I_KM is a noninactivating, voltage-dependent K⁺ current, which activates in a time- and voltage-dependent manner at about − 60 mV, close to the resting membrane potential. I_KM opposes cell depolarization by incoming stimuli, therefore inhibiting neuronal hyperexcitability and causing spike frequency adaptation during sustained depolarizations. Suppression of I_KM by muscarinic receptor activation has a depolarizing effect and enhances membrane excitability, causing tonic firing, whereas activation of I_KM leads to hyperpolarization and reduced neuronal excitability [26]. I_KM activates during spike afterdepolarization and limits spike duration, thereby preventing the spike from progressing into a burst [27]. In addition to muscarinic receptor agonists, ligands binding at several G_q/11-coupled receptors, by activating phospholipase C_β and depleting membrane phosphatidylinositol-(4,5)-bisphosphate (PI_(4,5)P2) levels, can suppress I_KM. Additional I_KM modulatory pathways, possibly exploited by other neurotransmitters, may also exist [28].

Neuronal Kv7 Channels Underlie I_KM

The recognition that Kv7 channels determine I_KM occurred after discovery of epileptogenic variants in two neuronally expressed genes, KCNQ2 [29] and KCNQ3 [30], respectively encoding Kv7.2 and Kv7.3 voltage-gated K⁺ channel subunits. When co-expressed in frog oocytes, KCNQ2 and KCNQ3 yield voltage-gated K⁺ currents with biophysical and pharmacological properties identical to those of the M-current [31, 32], suggesting that the native ganglionic I_KM is carried by an heterotetramer of Kv7.2 and Kv7.3 subunits [33]. However, I_KM composition may not necessarily be fixed, but may be dynamic and flexible, allowing some neurons to express Kv7.2 and Kv7.3 homomers depending on factors such as developmental stage, brain regions, cell types, subcellular region, and disease states [34, 35].

In neuronal cells, Kv7.2 and Kv7.3 subunits are localized at key subcellular sites, including the perisomatic region, the axon initial segment, nodes of Ranvier, and synaptic terminals. According to their subcellular localization, Kv7 channels regulate several aspects of neuronal excitability.

Somatodendritic Kv7 channels are strongly activated by the backpropagating action potential and attenuate afterdepolarization and repetitive firing [36]. Kv7 channels in the perisomatic region contribute to the medium and the slow components of the afterhyperpolarization current that follows the action potential and determines the refractory period and neuronal discharge frequency in response to sustained stimuli, a phenomenon known as spike-frequency adaptation [37]. By contrast, axonal Kv7 channels are minimally recruited by forward-propagating action potentials; in fact, Kv7.2/7.3 channels in nodal domains increase voltage-gated Na⁺ channel availability and action potential amplitude by stabilizing the resting membrane potential. On the basis of these data, it has been suggested that Kv7 clustering near voltage-dependent Na⁺ channels at the axon initial segment [38] or at Ranvier nodes [39] serves specific and context-dependent roles, both restraining initiation and enhancing conduction of the action potential [36]. Finally, Kv7 channels also modulate neurotransmitter release by controlling presynaptic terminal membrane potential, which in turn controls presynaptic Ca²⁺ influx mediated by voltage-gated Ca²⁺ channels [40, 41].

Kv7 Channels Subtypes

In addition to neuronal Kv7.2 and Kv7.3, the Kv7 channel family includes Kv7.1, Kv7.4, and Kv7.5 subunits. Kv7.1 subunits, encoded by the KCNQ1 gene, are expressed mainly in cardiomyocytes, where, together with auxiliary KCNE1 subunits, they underlie the slow component of the delayed rectifying K⁺ current (I_KS) critical for ventricular repolarization [42]. KCNQ1 mutations are responsible for the most common form of long QT syndrome [43], an arrhythmogenic disorder predisposing to torsades de pointes and ventricular fibrillation. Kv7.1 has also been detected in noncardiac tissues, such as the inner ear, thyroid gland, lung, gastrointestinal tract, small intestine, pancreas, ovaries, forebrain neuronal networks, and brainstem nuclei [26]. Kv7.4 subunits, encoded by the KCNQ4 gene, regulate the excitability of sensory outer hair cells within the cochlea, playing an important role in hearing, and KCNQ4 pathogenic variants are responsible for autosomal dominant type 2 deafness, a progressive form of sensorineural hearing loss [44]. Expression of Kv7.4 transcripts and subunits has also been detected in vascular and visceral smooth muscle (including detrusor smooth muscle cells), with profound effects on basal tone and response to myogenic stimuli [45, 46], and in skeletal muscle cells, where they control proliferation, differentiation, and response to myotoxic stimuli [47, 48]. Finally, Kv7.5 subunits, encoded by the KCNQ5 gene [49, 50], are preferentially expressed in neurons where they contribute to I_KM heterogeneity in selected brain regions or neuronal subpopulations [51]. De novo heterozygous missense mutations in KCNQ5 have been described in patients with neurodevelopmental disorders with or without epilepsy [52, 53]. Compared with Kv7.2 and Kv7.3, Kv7.4 and Kv7.5 subunits show a more widespread expression, which includes non-neuronal tissues such as skeletal and smooth muscle cells [45, 46].

Structural Aspects of Neuronal Kv7 Channels

Each Kv7 subunit contains a core region with six transmembrane segments (S1–S6). Segments 1–4 (S1–S4) form one voltage sensing domain (VSD) whereas S5, S6, and the S5–S6 interconnecting loops form the interlocking structure of the pore domain (Fig. 1). Within this region, a Gly–Tyr–Gly–Asp (GYGD) motif coordinates K⁺ ions flow through the pore and determines the ion selectivity of the channel [17]. Within the VSD, S4 contains four to six positively charged arginines (Rs), each separated by nonpolar residues; this segment is crucial for channel gating as it can move outwardly in response to changes membrane potential depolarization. This conformational shift is transmitted to the pore module via the S4–S5 linker and other interactions between elements of the pore module and S4 [54]. In addition to the transmembrane core, Kv7 channels exhibit a long intracellular C-terminus, where four α-helical regions (helices A–D) provide the molecular determinants necessary for tetramerization and for binding and transductional regulation by critical modulators [26].

Fig. 1.

Hypothetical mechanism of activation of Kv7 channels by ezogabine. Only two of the four subunits forming the functional channel are shown for simplicity. (A) Under resting conditions, the S4 segment of the voltage-sensing domain (VSD) is positioned toward the intracellular side of the membrane. This configuration is stabilized by calmodulin bridging the core region at the S2–S3 linker [where phosphatidylinositol-4,5-biphosphate (PIP2) also binds] and the C-terminus at two α-helical regions denominated HA and HB. The inner pore gate (IG) is closed, the central cavity (CC) above the IG and below the selectivity filter (SF), is narrow. Although the SF is in a conductive state, no ion can flow through the pore as the IG is closed. (B) Membrane depolarization induces a vertical shift in the S4 segment, which also induces a lateral rotation and an upward motion of the S4–S5 linker. (C) S4 repositioning, and the consequent upward motion of the S4–S5 linker, bends the S6 C-terminal half of an adjacent subunit and leads to widening of the IG and an increased hydration of the CC, triggering K⁺ conduction. Such processes are also accompanied, and possibly facilitated, by the fact that, upon depolarization, the HA and HB helices in the proximal C-terminus, where calmodulin binds in the closed channel configuration, undergo an almost 180° rotation, causing the HA and S6 helices to join in a continuous helix and calmodulin to be released from its attachment site in the S2–S3 linker, as well as by PIP2 repositioning from the S2–S3 linker to the S4–S5 linker. The positions of the ezogabine binding sites (one per subunit) are indicated relative to the overall structure of the channel. (D) A flat ribbon representation of the ezogabine binding site in a Kv7.2 subunit. The illustration shows the ezogabine molecule and the S5 and S6 transmembrane regions from a single Kv7.2 subunit contributing to the binding site. Ionized hydrogen bonds are shown as dashed-yellow lines. The Trp236 residue, which is critical for ezogabine binding, is indicated, together with Leu299 and Ser303 also contributing to drug-channel interactions. Data are from the Protein Data Bank (7CR2) [56, 60]

Advances in cryogenic-electron microscopy techniques and atom-detailed molecular dynamic simulations applied to neuronal Kv7.2 [55, 56], and Kv7.4 [57, 58] have enhanced our understanding of key biophysical mechanisms such as voltage sensing, electromechanical coupling between the VSD and the gating machinery, permeation, and selectivity. They also provided an atomistic resolution of channel regulation by mutations [59], endogenous ligands (PIP2 and calmodulin, in particular), and drugs, including the Kv7 activator ezogabine (Fig. 1) [60]. Of note, some Kv7 activators such as ezogabine interact with the pore region of the channel, whereas others such as ICA-069673 target the VSD region [61, 62].

Epilepsy Phenotypes Associated with Dysfunction of Neuronal Kv7 Channels

Consistent with the crucial role of Kv7 channels in controlling neuronal excitability, mutations in the KCNQ2, KCNQ3, and KCNQ5 genes can cause a spectrum of rare channelopathies in which epilepsy is often present. Some of the most common epilepsy phenotypes are described below.

Mutations in KCNQ2 [29] and KCNQ3 [30] were first identified in individuals with self-limited (familial) neonatal epilepsy (SeLNE), previously known as benign familial neonatal convulsions (BFNC), a rare autosomal-dominant syndrome characterized by seizures that start in otherwise healthy neonates between 2 and 7 days after full-term birth and typically disappear by 6 months of age, mostly in the first 6 weeks [63]. Rarer, closely related syndromes with favorable prognostic features occasionally associated with KCNQ2 variants include self-limited familial neonatal–infantile epilepsy (SeLFNIE), in which seizure onset is between the neonatal and the infantile period, and self-limited (familial) infantile epilepsy (SeLIE), with seizure onset usually at around 6 months of age. At the opposite end of the phenotypic spectrum, KCNQ2 and, more rarely, KCNQ3 variants can cause moderate to severe developmental and epileptic encephalopathies (DEEs) associated with abnormal neurodevelopment and intellectual disability. The most notable example is KCNQ2-DEE, an etiology-specific syndrome with onset of mostly focal tonic seizures within the first few days of life and a severe encephalopathy with abnormal neurological examination and behavior [63]. Seizures may respond to sodium channel blockers [64, 65]; although epilepsy may remit over time, moderate to severe neurodevelopment impairment typically persists. KCNQ2 pathogenic variants may also be responsible for some cases of early infantile DEE (EIDEE), typically associated with frequent drug-resistant seizures (mostly with a tonic component) appearing in the first 3 months of life; abnormalities of posture, tone, or movement; developmental impairment; and an abnormal interictal electroencephalogram (EEG), which may include a burst-suppression or multifocal pattern [63]. Variants in KCNQ5 have been recently identified in patients with intellectual disability or DEEs [52, 53]. Except for KCNQ2-DEE, all syndromes listed above can have other etiologies. For example, most cases of SeLFNIE are caused by SCN2A mutations, and PRRT2 mutations are most commonly involved in SeLIE [63].

With respect to pattern of inheritance, KCNQ2 and KCNQ3 mutations in individuals with SeLNE are invariably found in heterozygosity, and are transmitted with an autosomal dominant inheritance pattern. For both KCNQ2 and KCNQ3 variants causing SeLNE, penetrance is incomplete, with about 77–85% of heterozygous individuals showing neonatal or early infantile seizures [66].

Pathogenic KCNQ2 variants associated with SeLNE include missense (33%), splice site (14%), nonsense (18%), submicroscopic deletions (11%), and frameshift (21%) variants, as well as exon and whole-gene deletions [67]; by contrast, variants associated with more severe phenotypes are mostly de novo and they are all missense. No homozygous frameshift mutation in KCNQ2 has been described in humans as minimal KCNQ2 residual activity is likely essential under penalty of potential lethality. By contrast, homozygous frameshift mutations in KCNQ3 may be compatible with life, although associated with severe phenotypes with neonatal-onset epilepsy and non syndromic intellectual disability [68, 69].

With respect to functional changes, most SeLNE-causing KCNQ2 missense variants display a mild loss of function [70], pointing to haploinsufficiency as the main pathogenetic mechanism. By contrast, pathogenic variants in KCNQ2-DEE are all missense and affect more severely channel function, suggesting that a dominant-negative effect causes the epileptic phenotype [71, 72].

Specifically, subunits carrying these missense variants decrease channel function by more than 25%. Together with other genetic and epigenetic factors, these data suggest that the extent of I_KM channel function is an important predictor of disease severity. In addition to the loss-of-function mechanisms described for both KCNQ2-SeLNE and KCNQ2-DEE, heterozygous de novo missense gain-of-function variants have been reported in a small subset of patients with distinct neurodevelopmental disorders, including infantile spasms without prior neonatal seizures [73], or severe neonatal-onset encephalopathy with prominent startle-like myoclonus and a burst- suppression EEG pattern [74]. A similar phenotype of developmental delay, autism spectrum disorder, and frequent sleep-activated multifocal epileptiform discharges has been described in association to gain-of-function KCNQ3 variants [75]. Finally, both loss- and gain-of-function variants in KCNQ5 have been described in a small number (< 10) of children with intellectual disability and neurological abnormalities, at times associated with treatment-resistant epilepsy [52, 53]. Although the pathogenetic mechanisms by which loss- and gain-of-function KCNQ2, KCNQ3, and KCNQ5 variants affect neurodevelopment and epileptogenesis are unclear [8, 35], the relationship between clinical phenotype/severity and variant-dependent extent of functional derangement in vitro provides a conceptual framework with prognostic and therapeutic implications.

Recent findings suggest that, apart from sequence variants causing rare monogenic epilepsies, altered KCNQ2 and KCNQ3 gene expression and function could be involved in the pathogenesis of at least some cases of temporal lobe epilepsy (TLE), the most common human epilepsy. Specifically, downregulation of KCNQ2 and KCNQ3 transcript expression has been found in two publicly available databases of hippocampal samples from TLE patients [76]. The same study also showed that, in a pilocarpine mouse seizure model, KCNQ2/3 protein levels are selectively downregulated in the epileptic dentate gyrus, but not hilus GABAergic cells. A decreased spike frequency adaptation, indicative of a reduction in I_KM, was detected in dentate gyrus granule cells, strongly supporting the notion that I_KM activation has antiseizure effects.

Medications Targeting Kv7 Channels for the Treatment of Seizure Disorders

Ezogabine (Retigabine)

Historical Background

Ezogabine (N-(2-amino-4-(4-fluorobenzyl-amino)-phenyl) carbamic acid ethyl ester; Fig. 2), was identified from quantitative structure-activity relationship (QSAR) studies on derivatives of the analgesic drug flupirtine, which had demonstrated anticonvulsant activity in screening tests [77]. Ezogabine was developed for the treatment for focal seizures in adults, and was approved for this indication by the US Food and Drug Administration (FDA) and by the European Medicines Agency (EMA) in 2011. Shortly thereafter, ezogabine was associated with the development of a blue discoloration of the skin, mucosae and retina, which discouraged wide clinical use. The drug was withdrawn from the market in 2017 for commercial reasons [78]. A novel formulation of ezogabine suitable for pediatric use (XEN496) [79] has recently been investigated with KCNQ2-related DEE as targeted indication [80].

Fig. 2.

Chemical structure of ezogabine

Mechanism of Action and Antiseizure Activity in Preclinical Models

The primary mechanism of action of ezogabine involves activation of neuronal Kv7.2 and Kv7.3 [81–86]. Ezogabine also activates neuronal Kv7.4 and Kv7.5, but at clinically relevant concentrations has no effect on cardiac Kv7.1 channels [87]. The opening of Kv7 channels induced by ezogabine is associated with a shift of the voltage dependence for channel activation to more negative potentials [84, 88]. Recent studies on Kv7.2 [56] and Kv7.4 [57] channels showed that ezogabine binds in a hydrophobic pocket lined by S5, pore helix, S6, and the S6 from the adjacent subunit (Fig. 1). Although ezogabine interactions with Kv7.2 are mainly hydrophobic, ezogabine forms a hydrogen bond with the side chain of a conserved Trp in S5 (Trp236 in Kv7.2), which adopts a different rotamer upon binding with the drug. Such structural rearrangement increases the mobility of the inner pore gate when the channel opens, and reduces the energy required for the transition of the VSD from resting to activated state. Ezogabine-induced Kv7 current potentiation is retained, at least in part, in cells expressing loss-of-function KCNQ2 (Kv7.2) variants associated with seizures and epilepsy [72, 89, 90]. This is important for the investigation of ezogabine as a potential treatment for KCNQ2-related DEE.

The action of ezogabine on Kv7 channels was not identified in early preclinical studies. The seizure suppressing effects were initially ascribed to potentiation of γ-aminobutyric acid (GABA)-mediated transmission [77], with more recent data indicating that the drug can an act as a subtype-selective modulator of extrasynaptic δ subunit-containing GABA_A receptors [91]. In rat cortical neurons, GABAergic potentiation was reported to occur at ezogabine concentrations > 10 µmol/L (> 3033 ng/mL), whereas opening of K⁺ channels occurs at concentrations as low as 0.1 μmol/L (30.3 ng/mL) [92, 93]. Although the clinical relevance of ezogabine’s GABAergic action has been questioned [81–86], in other experiments GABAergic effects have also been reported to occur at ezogabine concentrations as low as 1 µmol/L [91], and the possibility of GABA_A receptor modulation contributing to the antiseizure effects of ezogabine cannot be excluded [87].

Ezogabine is effective against seizures induced by maximal electroshock (MES), pentylenetetrazole, picrotoxin and N-methyl-d-aspartate (NMDA). It is also effective in suppressing seizures in the DBA/2 mouse, the genetic epilepsy-prone rat (GEPR), and the rat amygdala kindling model [77, 92]. Other models where ezogabine is effective include spontaneous seizures after kainate-induced status epilepticus in rats [94], organophosphate-induced status epilepticus [95], acute seizures induced by Theiler's murine encephalomyelitis virus infection in mice [96], and neuronal excitability and behavioral seizures in a transgenic zebrafish larvae model [62]. In a mouse model of traumatic brain injury (TBI), ezogabine administered intraperitoneally (i.p.) 30 min after the insult inhibited the cascade of events leading to neuronal hyperexcitability, seizures, blood brain dysfunction, brain inflammation, and cell death [97]. Ezogabine also prevented biochemical and behavioral manifestations of brain damage in mice exposed to repetitive TBI [98].

In most preclinical models, ezogabine’s anticonvulsant activity is generally observed at doses that induce no or minimal motor impairment. Unlike many other ASMs, ezogabine shows relatively high potency in the amygdala-kindling model compared with other models. In particular, an increased threshold for induction of afterdischarges in kindled rats is already seen after i.p. and oral doses as low as 0.01 mg/kg, which are below those required to increase the threshold for induction of tonic hindlimb extension in the rat MES test [99]. Ezogabine given i.p. was also effective in the lamotrigine-resistant amygdala kindling rat model of drug-resistant focal epilepsy, with a median effective dose (ED50) of 3.2 [95% confidence interval (CI) 1.3–5.4] mg/kg [100]. The validity of the amygdala kindling model in predicting clinical efficacy in focal epilepsy probably played a key role in the decision to develop ezogabine for the treatment of focal seizures.

Clinical Pharmacokinetics

The pharmacokinetics of ezogabine is linear after single oral doses up to 600 mg and multiple dosing up to 1200 mg/day [101]. Peak plasma drug concentrations occur at 0.5–2.0 h after oral dosing [101, 102]. The incomplete oral bioavailability (approximately 60%) relative to intravenous administration is probably due to first-pass metabolism in the gastrointestinal tract and/or the liver, because fecal elimination as unchanged drug after oral dosing is virtually negligible [103, 104]. Co-administration with a high-fat meal does not affect the extent of absorption, but increases the peak plasma drug concentration by about 14% and 38% for the 200 and the 400 mg tablets, respectively, and delays the time to peak concentration by about 0.75 h [104, 105]. The absorption profile of the immediate release, granular dose form of ezogabine (XEN496) developed for pediatric use is comparable to that reported for tablets [106]. However, differently from tablets, co-administration of the granular formulation with food reduces the peak plasma ezogabine concentration by about 30% compared with the fasting state [106]. Similarly to the tablets, food delays by about 1 h the time to peak drug concentration and does not affect the extent of absorption of the granular formulation.

Ezogabine is about 80% bound to plasma proteins. Its steady-state volume of distribution estimated after intravenous administration is about 2–3 L/kg. Ezogabine is cleared by urinary excretion (about 35% of the dose) and by N-acetylation and subsequent N-glucuronidation [107]. The plasma concentrations of the primary metabolite N-acetyl-ezogabine, which shows weak and inconsistent activity in preclinical models, are similar to those of the parent drug [103]. In an early pharmacokinetic study conducted in six healthy subjects who received multiple doses of ezogabine 200 mg twice daily (b.i.d.), mean plasma concentrations at steady state were 440 ng/mL for ezogabine and 507 ng/mL for N-acetyl-ezogabine [101]. Ezogabine has a half-life of 6–11 h and an apparent oral clearance (CL/F) of 0.5–0.8 L h⁻¹ kg⁻¹ [102]. In elderly individuals, the half-life is prolonged by about 30% and CL/F is reduced by about one-third [102].

Drug Interactions

Ezogabine does not induce or inhibit the major cytochrome P450 (CYP) enzymes at clinically relevant concentrations, and it does not affect the plasma levels of concurrently administered CYP substrates [103, 108]. Ezogabine, however, may reduce plasma lamotrigine levels by about 20% [108].

Ezogabine (750 mg/day) did not affect the pharmacokinetics of an oral contraceptive containing ethinylestradiol and norethindrone, and there was no effect of the oral contraceptive on the pharmacokinetics of ezogabine [109]. N-Acetyl-ezogabine can inhibit P-glycoprotein (P-gp) and increase the plasma concentration of P-gp substrates such as digoxin [104]. In a drug–drug interaction study in healthy subjects, however, ezogabine at doses up to 900–1200 mg/day caused only a small, non-dose-dependent and nonclinically relevant increase in digoxin area under the plasma drug concentration–time curve (AUC) [110].

Most ASMs have no effect on ezogabine pharmacokinetics. Results from small phase II studies suggested that carbamazepine and phenytoin increase ezogabine clearance by approximately 27% and 36%, respectively, but no evidence for this interaction was detected in a population pharmacokinetic study [108].

Clinical Efficacy in Epilepsy

The efficacy of ezogabine in the treatment of focal seizures in adults was demonstrated in three randomized, double-blind, placebo-controlled, parallel-group adjunctive-therapy trials [111–113]. Each trial consisted of an 8-week baseline, a titration phase and a maintenance phase, followed by an open-label extension. In the first trial, that compared ezogabine doses of 600, 900, and 1200 mg/day, the duration of the titration phase was 8 weeks, followed by 8 week maintenance [111]. The remaining trials evaluated doses of 600 and 900 mg/day [112] or 1200 mg/day [113] and had a titration phase of 4 weeks [112] or 6 weeks [113], followed by 12 week maintenance. In all trials, ezogabine was administered in three divided daily doses (t.i.d.). The primary FDA endpoint was the percent change in seizure frequency from baseline during the double-blind treatment period, whereas the primary EMA endpoint was the responder rate, defined as the proportion of patients experiencing at least 50% reduction in seizure frequency compared with baseline.

Treatment with ezogabine was associated with a dose-dependent improvement in seizure frequency and responder rate (Figs. 3 and 4). In the two pivotal trials with a 12 week maintenance phase, the improvement in seizure frequency and responder rate was statistically significant at all doses tested [112, 113]. The earlier trial that used a shorter maintenance phase and had lower power due to smaller numbers of patients per group, the improvement was statistically significant only at 900 and 1200 mg/day (Figs. 3 and 4) [111]. Seizure-freedom rates among patients completing the double-blind period were only reported for one of the trials, with more patients on ezogabine 1200 mg/day being free of seizures during the 12 week maintenance period (5/97, or 5.2%) compared with patients on placebo (1/127 or 0.8%) [113]. In open-label extension studies, the improvement in seizure control associated with ezogabine was maintained during long-term treatment [114].

Fig. 3.

Median percent reduction from baseline in focal seizure frequency in patients included in randomized double-blind, placebo-controlled, adjunctive-therapy trials of ezogabine (EZG) in focal epilepsy. Based on data from Porter et al. [111], Brodie et al. [112], and French et al. [113]

Fig. 4.

Responder rates (percent of patients with ≥ 50% reduction in focal seizure frequency from baseline) in patients included in randomized, double-blind, placebo-controlled, adjunctive-therapy trials of ezogabine in focal epilepsy. For the study by Porter et al. [111] responder rate was calculated over the entire double-blind treatment period (titration plus maintenance), whereas for the studies by Brodie et al. [112] and French et al. [113] the responder rates shown in the figure were calculated over the maintenance dose period only

More recently, ezogabine has been proposed as a potential precision therapy for patients with KCNQ2- and KCNQ3-related epilepsies due to loss-of-function Kv7.2/3 variants [115]. Case reports suggested that patients with these disorders, mostly children with KCNQ2-related DEEs, can exhibit clinically significant improvement in seizures as well as developmental outcomes when treated with ezogabine [115–117]. A randomized double-blind adjunctive-therapy placebo- controlled trial to investigate the antiseizure effect of the investigational pediatric formulation (XEN496) in infants/children with KCNQ2-related DEEs, was initiated on 29 March 2021 [80]. The trial was terminated prematurely on 16 May 2023 after only five patients on ezogabine and three on placebo had been treated [80]. The sponsor provided no explanation for the premature termination, but indicated that this was not related to safety reasons [80]. The decision to terminate the trial might have been related to recruitment difficulties, possibly associated with the challenge of demonstrating an antiseizure effect in a condition where seizures often tend to remit spontaneously over a relatively short period [63]. A major concern in this population is impaired neurocognitive development, which may not be meaningfully assessable over a short trial period and was only included in the XEN496 trial protocol as a tertiary endpoint [118].

Adverse Effects

The adverse effects most commonly reported in the controlled trials of ezogabine in adults with focal seizures were dizziness, somnolence, fatigue, confusional state, tremor, abnormal coordination, blurred vision, and vertigo (Table 1). These effects were generally dose-related.

Table 1.

Most commonly reported adverse events reported during randomized double-blind placebo-controlled trials comparing ezogabine doses of 600, 900 and 1200 mg/day. The list includes only adverse events occurring with a frequency ≥ 5% in any ezogabine treatment group and numerically more frequent than in the placebo group. Data based on US Prescribing Information [104]

Adverse event	Proportion of patients experiencing the listed adverse event (%)
	Placebo (n = 427)	Ezogabine treatment groups (daily dose)
		600 mg (n = 281)	900 mg (n = 273)	1200 mg (n = 259)
Dizziness	9%	15%	23%	32%
Somnolence	12%	15%	25%	27%
Fatigue	6%	16%	15%	13%
Confusional state	3%	4%	8%	16%
Tremor	3%	3%	10%	12%
Abnormal coordination	3%	5%	5%	12%
Blurred vision	2%	2%	4%	10%
Vertigo	2%	8%	8%	9%
Nausea	5%	6%	6%	9%
Memory impairment	3%	3%	6%	9%
Diplopia	2%	8%	6%	7%
Dysarthria	< 1%	4%	2%	8%
Disturbance in attention	< 1%	6%	6%	7%
Aphasia	< 1%	1%	3%	7%
Asthenia	2%	4%	6%	4%
Gait disturbance	1%	2%	5%	6%
Constipation	1%	1%	4%	5%
Influenza	2%	4%	1%	5%
Balance disorder	< 1%	3%	3%	5%
Paresthesia	2%	3%	2%	5%
Anxiety	2%	3%	2%	5%
Disorientation	< 1%	< 1%	< 1%	%

Ezogabine caused a dose-related increase in body weight; in the largest trial that compared 1200 mg/day with placebo, 18.5% of ezogabine-treated patients versus 3.1% of placebo-treated patients had an increase in body weight ≥ 7% compared with baseline [113]. The proportion of patients who discontinued study medication due to treatment-emerging adverse events (TEAEs) ranged from 8 to 12% in the placebo group, 17% in the 600 mg/day group, 20–26% in the 900 mg/day group, and 27–29% in the 1200 mg/day group [111–113]. Most discontinuations due to TEAEs occurred during titration. Postmarketing investigations confirmed that the most common adverse effects were CNS-related [114, 119, 120].

Activation of Kv7 channels in sensory neurons [121] and urinary bladder smooth muscle [122] inhibits bladder contractility and may lead to urinary retention [123]. In double-blind trials, dysuria and urinary hesitation was each reported in 4% of patients receiving the 1200 mg/day dose, compared with < 1% of those on placebo [104]. In open-label studies, the proportion of patients reporting urinary system symptoms (mostly voiding problems, leading at times to urinary retention and urinary tract infections) was around 9–14% [114, 119, 120]. According to US prescribing information, urinary retention was reported as an adverse effect in 29 of 1365 (2.1%) patients included in double-blind and open-label studies, with 5 of these patients requiring catheterization [104].

The use of ezogabine declined markedly after it was found to cause retinal changes and a bluish discoloration of skin and mucosae [124–126]. On funduscopic examination, ezogabine-induced retinal abnormalities are similar to those observed in retinal pigment dystrophies known to lead to photoreceptor damage and vision loss. Macular vitelliform lesions have also been reported [104, 127]. Overall, retinal pigmentary abnormalities occurred in about one-third of patients who had eye examinations after approximately 4 years of treatment, but they may be seen earlier [104, 128]. Some patients had abnormal visual acuity, though it is unclear whether these pre-existed or developed after ezogabine treatment. After discovery of these abnormalities, the FDA and EMA restricted the use of ezogabine to patients who had not responded to other ASMs, and introduced the requirement of at least 6 monthly examinations of visual acuity, dilated fundus photography, and optical coherence tomography in all patients on treatment with ezogabine [104, 128]. The mechanisms responsible for retinal abnormalities are unclear but appear to involve reactive metabolites of ezogabine and N-acetyl-ezogabine undergoing dimerization in the presence of melanin [129–131].

A dimerization process is probably also involved in the bluish discoloration of the skin and mucosae. The discoloration affects predominantly the lips, the nail plate of fingernails or toenails, eyelids, and, more rarely, the tongue, palate, sclera, conjunctiva, and other areas [104, 127]. Skin discoloration has been reported in about 10% of patients in long-term trials, usually after at least 2 years of treatment and more frequently at doses of 900 mg/day or higher [104]. The discoloration can be slowly reversible in some patients after drug discontinuation [127, 132]. About one-quarter of patients with skin or mucous membrane discoloration have concurrent retinal pigmentary abnormalities.

Ezogabine can induce QT prolongation. In a cardiac study in healthy subjects, a mean QT prologation of 7.7 ms was observed at 3 h after dosing at a dose of 400 mg t.i.d. [104].

Azetukalner (XEN1101)

Historical Background

Azetukalner (XEN1101, XPF-008, 1OP-2198; VRX 621698) (Fig. 5) is a highly selective second-generation neuronal Kv7 channel opener, with potentially improved safety over ezogabine due to greater metabolic stability [133]. The compound was purchased in 2015 from Valeant Pharmaceuticals by 1st Order Pharmaceutics, and acquired in 2017 by Xenon Pharmaceuticals, Inc. [134]. It is currently in phase III development for the treatment of focal seizures, generalized tonic-clonic seizures, and major depressive disorder [135].

Fig. 5.

Chemical structure of azetukalner

Mechanism of Action and Antiseizure Activity in Preclinical Models

Most the available preclinical data on azetukalner have been published in the proceedings of Eilat Conferences [133, 135–139]. On the basis of these data, azetulkalner can be classified as a selective opener of neuronal Kv7.2/7.3 channels, with no effects on cardiac (Kv7.1) channels.

Azetukalner’s activity was initially assessed in Chinese hamster ovary (CHO) cells expressing Kv7.2 channels and in adrenal gland pheochromocytoma PC-12 cells [133]. In these systems, azetukalner was found to activate Kv7.2 channels with half-maximal effective concentrations (EC50) of 11 and 15 nM, respectively. In a Xenopus oocyte expression system, azetukalner activated channels Kv7.2, Kv7.5, Kv7.4, and Kv7.2/7.3 heteromer with EC50 values of 0.45, 0.35, 0.63, and 0.35 µM, respectively [133]. By comparison, ezogabine showed in the same system EC50 values of 2.6 µM (Kv7.2),15 µM (Kv7.5), 22 µM (Kv7.4), and 2.4 µM (Kv7.2/7.3) [133]. Azetukalner reportedly exerts no activity on other ion channels or the GABA system at concentrations > 100-fold greater than those active on Kv7.2 channels [133].

In in vitro studies in CHO cell lines expressing heteromeric Kv7.2/7.3 and Kv7.3/7.5 channels and homomeric Kv7.4 channels, azetukalner showed 3.4‐fold greater selectivity for Kv7.2/7.3 over Kv7.3/7.5 and Kv7.4 channels, with EC50 values of 27 nM versus 94 nM and 113 nM, respectively [136]. Assuming that the minimal shift in Kv7.2/7.3 activation required to cause a clinically significant seizure reduction is 4 mV [84], azetukalner was estimated to be approximately 50 times more potent than ezogabine [136]. In studies with human embryonic kidney (HEK) cells expressing Kv7.2/7.3 channels, azetukalner (EC50, 42 nM) was 20 times more potent than ezogabine (EC50, 920 nM) in causing an approximate 40 mV leftward (negative) shift in the half-maximal voltage of channel activation, while causing a similar increase in maximal conductance [137–140]. Experiments with a Kv7.2/Kv7.3 pore mutant suggested that azetukalner engages the same critical Trp236 residue in the core of the K⁺ channel as ezogabine [140] (Fig. 1). Azetukalner (1 μM) was also effective in rescuing Kv7 currents in CHO cells expressing KCNQ2 variants associated with DEEs [141].

In the MES test, azetukalner has ED50 values of 1.1 mg/kg in rats at 60 min after per os (p.o.) dosing, and 2.2 mg/kg in mice at 120 min after i.p. dosing. In mice, azetukalner is also effective in protecting against seizures induced by subcutaneous (s.c.) pentylenetetrazole, s.c. picrotoxin, s.c. bicuculline, and 6 Hz stimulation, with ED50 values < 10 mg/kg i.p., which correspond to a > 3–19-fold greater potency compared with ezogabine (Table 2) [133, 142]. Azetukalner did not cause motor/behavioral impairment in rats in an open field test at doses up to 5 mg/kg p.o. (highest dose assessed). In the rotarod test in mice, the median toxic dose (TD50) at 2 h after i.p. dosing was 12.6 mg/kg [133].

Table 2.

Comparative potency of azetukalner and ezogabine in mouse seizure models, based on data presented by Goldberg [142]. The data source does not specify 95% confidence intervals and whether experiments with azetukalner and ezogabine were done concurrently and at the same site

	Azetukalner ED50, i.p. (mg/kg)	Ezogabine ED50, i.p. (mg/kg)
Electrically induced seizures
Maximal electroshock (MES)	6.1	29.51
6 Hz stimulation (32 mA)	3.7	12.1
6 Hz stimulation (44 mA)	5.0	20.25
Chemically induced seizures
Pentylenetetrazole	3.9	> 50
Picrotoxin	9.86	33
Bicuculline	2.59	> 50

ED50 median effective dose, i.p. intraperitoneal

Clinical Pharmacokinetics

After single doses ranging from 5 to 30 mg (immediate-release capsules) administered in the fasting state, peak plasma azetukalner concentrations are achieved at 3.2–4.5 h on average. Thereafter, the drug concentration declines with a biphasic profile. Azetukalner AUC increases about 1.8-fold after co-administration with food [143]. In a single-dose cross-over study, the increase in plasma azetukalner concentration after intake with food was more pronounced in individuals with lower plasma azetukalner concentrations in the fasted state [137]. Absorption in the fed state was relatively slow, with peak drug concentrations occurring at 4–6 h after dosing. When azetukalner is administered in the fasting state, exposure is less than dose proportional. However, after 10 days of once-daily (q.d.) dosing with food, exposure increases proportionally with dose [143]. On day 10 of multiple dosing with 25 mg q.d. with food, peak plasma azetukalner concentration (mean ± SD) was 97 ± 9 ng/mL and AUC_0–24h was 1720 ± 198 ng h mL⁻¹ [143].

The half-life of azetukalner is about 1–2 days, which implies a modest (about 25%) peak-to-trough fluctuation in serum drug concentration on a q.d. dosing schedule [136]. After multiple dosing, azetukalner elimination shows a late terminal phase with an average half-life of 4–10 days [136, 137, 143], which has been ascribed to slow release of the drug from tissue compartments [144]. Steady state (defined as no statistical difference between trough levels on successive days) is reached after 7–10 days, even though complete steady state is expected to occur after about 3 weeks [137].

After 10 days of dosing at 25 mg q.d., only 0.01% of the administered dose is excreted in urine as unchanged drug over a 24 h collection period, indicating that azetukalner renal clearance is negligible [137]. Azetukalner is cleared by CYP3A4-mediated metabolism, with other CYP enzymes (CYP1A2, CYP2B6, CYP2C8, CYP2C9, CYP2C19, and CYP2D6) playing little or no role [138]. A major oxidative metabolite detected in plasma at concentrations of approximately 20% of those of the parent drug has been found to be inactive on Kv7 channels after preliminary testing [137]. Azetukalner lacks a secondary aniline structure and reportedly does not generate the phenazinium dimers considered to be responsible for ezogabine-induced pigmentation [129, 138, 140].

At present, information on relationships between plasma azetukalner concentration and clinical response is scarce. In a phase 1 proof-of-concept study, responses to transcranial magnetic stimulation (TMS) used as biomarkers of neuronal excitability were evaluated after single oral doses of azetukalner (20 mg) in 20 healthy subjects [145]. TMS measurements were obtained at 0 (pre-dose), 2, 4, and 6 h after dosing, and included TMS-evoked EEG potentials (TEPs) and resting motor thresholds (RMTs) as measures of cortical and corticospinal excitability, respectively. Treatment with azetukalner was associated with a reduction in TMS-evoked EEG potentials (TEPs) and with an increase in RMTs. Mean plasma azetukalner concentrations (± SD) at 2, 4, and 6 h after dosing were 15.7 ± 21.5 ng/mL, 30.2 ± 21.9 ng/mL, and 42.1 ± 19.1 ng/mL, respectively. At the time of individual peak drug concentrations, plasma azetukalner levels correlated significantly with changes in RMTs, but not with changes in TEPs.

Drug Interactions

At in vitro concentrations of 3 μmol/L, which are much higher than those anticipated to be clinically efficacious, azetukalner did not cause direct inhibition of CYP1A2, CYP2C9, CYP2C19, CYP2D6, and CYP3A4, nor did it cause time‐dependent inhibition [136]. Likewise, no clinically meaningful induction of the pregnane X nuclear receptor (PXR), which is associated with CYP3A4 expression levels, was observed at azetukalner concentrations of 0.1–1 μmol/L [136]. These data suggest a low potential for azetakulner to act as perpetrator of CYP-mediated drug–drug interactions.

In healthy subjects, a single 400 mg oral dose of the CYP3A4 inhibitor itraconazole given after 10 days of treatment with azetukalner (20 mg q.d.) was associated with an approximately 10% increase in peak plasma azetukalner concentration and a < 30% increase in azetukalner exposure (AUC_0–24h) [138]. Longer treatment with itraconazole, however, would be required to adequately assess the effect of CYP3A4 inhibition on azetukalner pharmacokinetics. Modeling pharmacokinetic data from the phase 2b efficacy and safety study (see below) predicted a 32% increase in steady-state plasma azetukalner levels in the presence of CYP3A4 inhibitors. The same model predicted a 22% decrease in azetukalner exposure in the presence of concomitant treatment with a single CYP3A4 inducer, and a 39% decrease in the presence of two or more CYP3A4 inducers [138].

Clinical Efficacy in Epilepsy

The efficacy of azetukalner in the treatment of focal seizures was assessed in a phase 2b randomized, double-blind, placebo-controlled, parallel-group, adjunctive-therapy trial conducted at 97 sites across North America and Europe [146]. A total of 325 adults with uncontrolled focal seizures were randomized to receive in a 2:1:1:2 ratio either placebo or azetukalner 10, 20, or 25 mg taken q.d. in the evening with food and without titration. Because azetukalner metabolism is subject to enzyme induction, randomization was stratified by background use of enzyme inducing ASMs. At the end of the 8-week double-blind treatment, participants could enter an open label extension at a dose of 20 mg q.d. Of the 325 randomized participants who received at least one dose of study medication, 323 had at least one seizure diary entry post-treatment and were eligible for the efficacy analysis. Of the 285 patients completing the 8-week double-blind phase, 275 (96.5%) entered the open-label phase. The primary planned analysis showed a dose-dependent reduction in seizure frequency from baseline in all azetukalner groups compared with placebo (P < 0.001). A similar dose-dependent effect was found for responder rates (Fig. 6). Among patients completing the trial, 4.5% in the 25 mg group, 3.9% in the 20 mg group, 2.2% in the 10 mg group, and 1.8% in the placebo group were seizure-free during the entire 8-week double-blind period [146]. A secondary analysis aimed at assessing the time course of response showed that for all azetukalner groups, reductions in seizure frequency and responder rates were already significant in the first week of treatment. For patients in the 20 mg and 25 mg groups, the decrease in seizure frequency versus placebo was maintained throughout the entire duration of the double-blind period. The improvement in seizure control on azetukalner treatment was associated with improvements in Global Impression–Change scales compiled separately by investigators and by participants. On a post hoc analysis, azetukalner was found to be effective irrespective of indicators of seizure severity, though reduction in seizure frequency tended to be greater among patients with lower disease severity as indicated by lower baseline seizure frequency, lower number of concomitant ASMs, and lower number of ASMs previously tried and discontinued [147].

Fig. 6.

Median percent reduction from baseline in focal seizure frequency (left panel) and responder rates (percent of participants with ≥ 50% reduction in focal seizure frequency from baseline, right panel) in patients included in the phase 2b randomized, double-blind, placebo-controlled, adjunctive-therapy trial of azetukalner (AZE) in focal epilepsy. Based on data from French et al. [146]

Analyses of seizure outcomes for the 275 patients included in the open-label extension trial suggest that efficacy is maintained during long-term treatment [135, 148, 149]. At a cutoff data of 5 September 2023, 182 patients (66%) had been treated for ≥ 12 months and 165 (60%) for ≥ 24 months [135, 149]. The most common reasons for discontinuation were lack of efficacy (14%), TEAEs (12%), and withdrawal by patient’s decision (12%). For those still on treatment, the percent reduction in seizure frequency from baseline ranged from 61% (n = 266) at month 1 to 95% (n = 124) at month 30. Among the 165 participants who received azetukalner for ≥ 24 months, 23.6% achieved seizure freedom for ≥ 12 consecutive months at any time during the treatment period. Improvements in several domains of the Quality of Life in Epilepsy Inventory-31 (QOLIE-31) were also observed during follow-up [135].

Adverse Effects

TEAEs recorded in ≥ 5% of patients during the double-blind period of the phase 2b trial [146] are summarized in Table 3. Those most commonly associated with azetukalner treatment were dizziness and somnolence, both of which were dose-related. Other TEAEs occurring more commonly in the 25 mg group compared with placebo included fatigue, balance disorder, gait disturbance, aphasia, and dysarthria. TEAEs leading to permanent discontinuation of treatment occurred in 3.5% of patients in the placebo group, 2.2% in the 10 mg group, 13.7% in the 20 mg group, and 15.8% in the 25 mg group [146]. TEAEs most commonly associated with treatment discontinuation included dizziness (4.7%), balance disorder (2.4%), dysarthria (1.9%), and gait disturbance (1.9%). The incidence of serious TEAEs was low and showed no major differences between the azetukalner groups and placebo.

Table 3.

Most commonly reported treatment-emergent adverse events (TEAEs) reported during the phase 2b randomized double-blind placebo-controlled adjunctive-therapy trial comparing azetukalner doses of 10, 20 and 25 mg/day in adults with focal seizures. The list includes TEAEs occurring with a frequency ≥ 5% in any azetukalner treatment group and numerically more frequent than in the placebo group. Data based on French et al. [146]

Adverse reaction	Proportion of patients experiencing the listed TEAE (%)
	Placebo (n = 114)	Azetukalner treatment groups (daily dose)
		10 mg (n = 46)	20 mg (n = 51)	25 mg (n = 114)
Dizziness	7%	6%	26%	32%
Somnolence	7%	11%	22%	15%
Headache	8%	13%	12%	8%
Fatigue	5%	11%	8%	12%
Balance disorder	2%	4%	8%	11%
Tremor	2%	6%	6%	10%
Urinary tract infection	4%	9%	6%	2%
Anxiety	5%	0%	10%	2%
Gait disturbance	1%	4%	4%	7%
Aphasia	1%	2%	2%	7%
Dysarthria	0%	2%	0%	7%
Confusional state	1%	2%	6%	5%
Constipation	1%	4%	6%	3%
Ataxia	1%	6%	2%	4%
Nausea	3%	2%	2%	6%
Disturbance in attention	1%	0%	6%	4%
Blurred vision	1%	0%	2%	6%
Hallucination	0%	0%	6%	0%
Memory impairment	1%	2%	4%	5%

Two patients developed urinary retention during azetukalner treatment, which in one case required dose reduction. Neither of these patients required catheterization and both remained on treatment. There were no reports of tissue discoloration [146], consistent with toxicology data in rodents and nonhuman primates [124]. TEAEs in the long-term extension study were similar to those recorded during the double-blind trial, and no new safety signals were identified [135].

Ongoing Studies

Two multicenter randomized adjunctive-therapy phase 3 trials in adults with focal seizures are ongoing (NCT05614063, NCT05716100). Both trials have the same design and evaluate the efficacy and safety of 15 and 25 mg azetukalner administered q.d. with food in comparison with placebo over a 12-week double-blind period [135]. A similar trial (NCT05667142) is being conducted in patients with generalized tonic-clonic seizures, assessing doses of 25 mg q.d. (participants aged ≥ 18 years) and 15 mg or 25 mg q.d (participants aged from 12 to < 18 years). Building on promising efficacy signals observed during an initial placebo-controlled trial (NCT05376150), a phase 3 trial is also planned in patients with major depressive disorder (MDD) [135].

BHV-7000

Historical Background

The discovery of BHV-7000 resulted from a systematic search for novel Kv7.2/7.3 channel activators differentiated from ezogabine by structural class, improved potency and tolerability, and lack of off-target effects [150]. BHV-7000 emerged from this search as a promising candidate for the treatment for KCNQ2-neonatal epileptic encephalopathy. The range of potential indications was subsequently broadened to include focal and generalized seizures as well as neuropsychiatric disorders, including MDD and bipolar disorder [139].

Mechanisms of Action and Antiseizure Activity in Preclinical Models

BHV-7000 acts by activating selectively Kv7.2/7.3 channels. In cells transfected with wildtype KCNQ2, BHV-7000 activated Kv7.2/7.3 channels with an EC50 of 0.6 μM, slowing the rate of channel deactivation and shifting to the left the voltage dependence of activation [139, 151, 152]. In rat primary cortical neuron cultures, BHV-7000 caused a concentration-dependent hyperpolarization of the resting membrane potential. The activity of BHV-7000 on the Kv7.2 channel requires presence of the Trp236 amino acid [152]. When tested in CHO cells expressing 50 pathogenic loss-of-function KCNQ2 variants, BHV-7000 was active in all the assessed variants, with current density being restored to near wildtype levels in most variants at a BHV-7000 1 µM concentration [153].

In in vitro studies, BHV-7000 (10 µM) showed little or no binding to over 50 kinase and receptor targets (including the human α1β3γ2 GABA receptor), and minimal or no activity across a cardiac panel of voltage-gated Na⁺, Ca²⁺, and K⁺ channels [139].

Antiseizure activity data have only been reported in the rat MES test. In this model, BHV-7000 protects against seizures with an ED50 of 0.5 mg/kg p.o. and a brain EC50 of 0.12 μM [139, 152]. TD50 estimated by visual assessment of spontaneous activity, ataxia and body posture was > 20 mg/kg, resulting in a protective index (PI = TD50/ED50) > 40. By contrast, ezogabine has a PI < 3 in the same system. BHV-7000 was also devoid of neurotoxic effects in the rotarod test in rats at doses up to 30 mg/kg p.o. [139].

Clinical Pharmacokinetics and Drug Interactions

Information on the pharmacokinetics and drug interaction potential of BHV-7000 has not been reported to date. An extended-release formulation suitable for once daily dosing has been developed for use in clinical trials [139].

Clinical Efficacy and Adverse Effects

No studies of BHV-7000 in patients with epilepsy have been completed to date. A pharmaco-EEG study in healthy subjects treated with single oral doses of BHV-7000 (immediate-release formulation, 10, 25, and 50 mg) showed a dose-dependent increase in EEG spectral power across all frequency bands, with a lesser impact on lower frequencies (delta and theta) [139, 154, 155]. These results are consistent with BHV-7000’s low propensity to cause somnolence.

In phase 1 studies, single doses of BHV-7000 up to 100 mg and multiple doses up to 60 mg b.i.d. for 15 days were well tolerated in healthy subjects [139, 156, 157]. The most common TEAEs after multiple dosing were headache and back pain, and no somnolence was reported. There were no clinically meaningful trends in safety laboratory tests and electrocardiogram (ECG) readings, even at plasma BHV-7000 concentrations that exceeded the EC50 in the MES test [139]. Whether these concentrations will prove to be therapeutically effective in patients with epilepsy remains to be determined.

Ongoing Studies

The ongoing phase 2/3 program includes randomized, double-blind, placebo-controlled efficacy and safety trials in patients with focal epilepsy (NCT06132893, NCT06309966), primarily generalized tonic-clonic seizures (NCT06425159), MDD (NCT06419608), and bipolar disorder (NCT06419582). Patients completing these trials can access open-label extension protocols (NCT06443463, NCT06423781, NCT06423794).

Other Potential Antiseizure Medications Targeting Kv7 Channels

ICA-105665

ICA-105665 (PF-04895162) emerged from a drug discovery program focused on Kv7 channels as a therapeutic target for seizure disorders and other indications [158–160]. ICA–105665 is an orally active, potent at selective Kv7 channel opener, with estimated EC50 values in cloned human neuronal Kv7 channels of 0.3 µmol/M (Kv7.2/7.3), 1.5 µmol/L (Kv7.3/7.5), and 3.3 µmol/L (Kv7.4) [161]. ICA-105665 is active in the MES, s.c. pentylenetetrazole, 6 Hz, and audiogenic seizure models in mice, and in the MES, hippocampal kindled and lamotrigine-resistant amygdala kindled model in rats, with a favorable PI [161]. In a proof-of-concept study, single oral doses of ICA- 105655 were effective in abolishing the photoparoxismal EEG response in patients with epilepsy [162]. Further development, however, was terminated following discovery of liver toxicity in healthy subjects, an off-target effect ascribed to inhibition of liver mitochondrial function and bile salt export protein (BSEP) transport [163, 164].

Pynegabine and CB-003

Pynegabine (HN37, Fig. 7) is an ezogabine derivative developed at the Shanghai Institute of Materia Medica [165, 166]. Pynegabine was classified as a potent neuronal Kv7 activator with a subtype selectivity similar to ezogabine [165]. The potency of pynegabine in activating Kv7.2 channels (EC50, 37 nmol/L) was 55 times greater than that of ezogabine (EC50, 2.0 µmol/L). At Kv7.2/7.3 channels, the difference in potency between pynegabine (EC50, 33 nmol/L) and ezogabine (EC50, 4.2 µmol/L) was even greater. Pynegabine was also effective in activating K⁺ currents in the presence of epileptogenic KCNQ2 mutations, when co-expressed with Kv7.3 or wildtype Kv7.2 channels [141].

Fig. 7.

Chemical structure of pynegabine

In the MES test in mice, pynegabine showed an oral ED50 of 1.9 mg/kg versus 38.6 mg/kg for ezogabine, and a greater safety margin than ezogabine [165]. Pynegabine also showed greater potency than ezogabine in the MES test in rats, the s.c. pentylenetetrazole seizure model in mice, and the 6 Hz (32 and 44 mA) mouse model [165]. Prominent anticonvulsant activity was also observed in a mouse model carrying an epileptogenic KCNQ2 mutation [167]. In preclinical studies, pynegabine showed high brain penetration, and high metabolic stability in human liver microsomes. Phase 1 trials in healthy subjects and patients with epilepsy have been completed (ChinaDrugTrials nos. CTR20222616, CTR20201676), and a randomized double-blind placebo-controlled safety, pharmacokinetic, and efficacy trial in patients with focal epilepsy is ongoing (CTR 20241138) [168].

CB-003 or CB03, also known as CB03-154 or CB04 (N-(2-(4-fluorophenyl)−5,7-dimethyl-1,2,3,4-tetrahydro isoquinolin-6-yl)−2-(1-methylcyclopropyl) acetamide) is another Kv7.2/3 activator from China [169, 170]. Details on the pharmacology of this compound do not appear to have been published. The sponsor (Shanghai Zhimeng Biopharma) has completed a single and multiple-dose phase 1 safety, tolerability, and pharmacokinetic study in healthy subjects (NCT05502549) and a double-blind placebo-controlled safety and efficacy trial in patients with focal epilepsy is due to initiate in late 2024 (NCT06612775) [171]. CB-003 also received FDA orphan drug designation for amyotrophic lateral sclerosis [172].

Compounds in Preclinical Development

As an indication of the continuing interest in targeting Kv7.2/7.3 channels for the treatment of seizure disorders and other indications, such as pain, affective disorders, and overactive bladder syndrome [45, 46, 173–175], a number of novel activators of these channels have recently been described, including SF0034 [176], ZG1732 and ZG2083 [177], RL-81 [178], Lu AA41178 (N-(2,4-dimethyl-6-morpholin-4-yl-pyridin-3-yl)−3,3-dimethylbutyramide) [179, 180], ETX-123 [181], 1025c (N,N′-{4-[(4-fluorobenzyl)(prop-2-yn-1-yl)amino]−1,2-phenylene}bis(3,3-dimethylbutanamide) [182], SCR2682 (4-(2-bromo-6,7-dihydrothieno[3, 2-c]pyridin-5(4H)-yl)−2,6-dimethylphenyl)−3,3-dimethylbutanamide) [183], ZM-003 [184], several resin acid derivatives [185], sulfide analogs of flupirtine and ezogabine [186], and other ezogabine derivatives [131].

There has also been interest in identifying/designing drugs that combine Kv7.2/7.3 activation with other mechanisms of action. One example of such compounds is GRT-X (N-[(3-fluorophenyl)-methyl]−1-(2-methoxyethyl)−4-methyl-2-oxo-(7-trifluoromethyl)−1H-quinoline-3-carboxylic acid amide), which is more potent than ezogabine in a variety of seizure models [187] and is also effective in a model of neuropathic pain [188]. GRT-X activates both Kv7.2/3 potassium channels and the mitochondrial translocator protein 18 kDa (TSPO), previously known as the peripheral benzodiazepine receptor. TSPO activators promote the synthesis of neurosteroids and exert a variety of other actions, which may translate into neuroprotective, neuroregenerative, anti-inflammatory, and possibly antidepressant effects [188, 189]; combining these properties with Kv7.2/7.3 activation is therefore an attractive strategy in the search for novel treatments for epilepsy and for pain. Other examples of compounds with a dual mode of action include BS421, BS627, and BS661. These agents combine Kv7.2/7.3 activation with modulation of the transient receptor potential vanilloid type 1 (TRPV1) ligand-gated ion channel [190]. Kv7.2/7.3 and TRPV1 channels are recognized targets being pursued in the development of novel treatments for pain, and BS421, BS627, and BS661 have all been found to be effective in various models of acute and chronic pain. Although to our knowledge these agents have not been evaluated for other indications, TRPV1 channels are also regarded as potential targets for the development of newer ASMs [191].

To date, none of the compounds listed in this section appears to have progressed to clinical development.

Summary and Conclusions

The role of changes in cell membrane permeability to ions, including K⁺, in regulating the function of excitable cells was first hypothesized by Bernstein in 1902 [192]. Since then, impressive advances have been made in understanding the genetics, structure, and function of ion channels in the brain and other organs. Among these, voltage-gated K⁺ channels stand out for their heterogeneity, their critical role in mediating physiological functions, and their pathological changes in various disease states [193]. Despite earlier evidence linking the neuronal Kv7 channels to cortical excitability and seizure disorders, selective Kv7 activators only started to be investigated as potential ASMs in the 1990s, their development being initially hindered by the complex physiology and pharmacology of these channels [173]. Ezogabine was the first opener of neuronal Kv7 channels developed as an ASM, but its actions on K⁺ channels were only discovered during late preclinical development. Ezogabine was approved for the treatment of focal seizures in 2011 but, because of the risk of pigmentary changes affecting the retina and the skin, and the consequent need for intensive ophthalmological monitoring, it was never used widely. In postmarketing studies, ezogabine led to improved seizure control in a modest subset of patients, with potential benefit in other individuals being precluded by dose-limiting adverse effects [119, 120]. One factor that probably contributed to ezogabine suboptimal tolerability was its short half-life, which results in considerable fluctuations in serum drug concentrations at steady state. The need for a t.i.d. regimen limited its acceptability in the clinical setting, and retention on treatment was generally lower than that reported for other second-generation ASMs [120]. Although ezogabine was withdrawn from the market in 2017 for commercial reasons, there has been interest in developing a pediatric formulation (XEN496) for patients with KCNQ2-related DEEs. A clinical trial of XEN496 was terminated prematurely for reasons unrelated to safety. Of note, the attractiveness of ezogabine as a potential therapy for DEEs due to loss-of-function Kv7.2 mutations is now reduced as other Kv7.2 activators with improved pharmacokinetic and safety profiles are being developed.

Among new and novel Kv7.2 openers, azetukalner has advanced most in clinical development. On the basis of the limited clinical trial data reported to date, its efficacy against drug-resistant focal seizures appears to be robust, and safety results have been reassuring, with no evidence of pigmentation-related TEAEs [194]. A mechanism of action that differs from that of currently marketed ASMs and a low potential for drug–drug interactions make azetukalner a good candidate for adjunctive therapy in focal-seizure patients receiving any type of concomitant medication. Additional advantages include the feasibility of once daily dosing without titration. Further studies are required to confirm these promising data and to determine whether efficacy extends to generalized tonic-clonic seizures. Depression is a common comorbidity in patients with epilepsy, and is also an adverse effect associated with commonly used ASMs [195]. Should azetukalner prove to be efficacious in the treatment of MDD, it would be the first ASM with demonstrated efficacy in this indication. This would enhance its attractiveness as a treatment option in epilepsy, irrespective of its potential in nonepilepsy indications.

Urinary retention has been reported in a small minority of patients receiving chronic treatment with ezogabine and azetukalner [127, 146]. Because of the inhibiting effect of Kv7 activators on detrusor smooth muscle [46, 175], as well as on bladder sensory afferent nerve activity [121, 196], urinary retention may be considered a class effect of these medications. However, there is evidence of differences in the expression of Kv7 subtypes between the brain and smooth muscle in the urinary system, with Kv7.4 being particularly abundant in the latter [197, 198]. The functional implications of these differences are not fully understood, and it unclear whether compounds with greater selectivity for specific Kv7 subtypes could reduce the risk of urinary symptoms while retaining antiseizure activity [175, 197, 198].

Available data for other Kv7 activators in clinical development, namely BHV-7000, pynegabine, and CB-003, are too limited to permit adequate assessment of their therapeutic potential. Notable features of BHV-7000 are its high protective index when using the MES ED50 in rats as measure of anticonvulsant activity, and its excellent tolerability in healthy subjects at plasma drug concentrations that exceed the MES EC50. However, whether the doses of BHV-7000 tested in phase I studies will also prove to be efficacious in patients with epilepsy remains to be determined. Interestingly, as in the case of azetukalner, phase 1/3 trials with BHV-7000 in epilepsy seem to have been initiated without preliminary dosing information from phase 2a studies. For pynegabine and CB-003, phase 1 safety and pharmacokinetic studies have been completed, but results do not appear to have been disclosed to date.

Interest in targeting Kv7 channels for the treatment of epilepsy and other disorders remains strong, as indicated by the many compounds in preclinical development that are active on these channels. Future breakthroughs in this area could come from exploitation of mechanistic differences in the action of Kv7 activators, including state dependent activity, higher selectivity for specific Kv7 subtypes, and sensitivity to KCNQ mutations [61, 62, 199], as well as from the development of molecules combining Kv7 activation with other mechanisms of action [187, 190].

Declarations

Funding

Open Access funding enabled and organized by CAUL and its Member Institutions. The preparation of this article was not supported by any funding source. The open access fee for this article was covered by The University of Melbourne.

Conflict of interest

EP received speaker fees or fees from consulting or participation in Advisory Boards/Data Safety Monitoring Board from Eisai, GRIN Therapeutics, SKL Life Science, Sintetica, Sun Pharma, Takeda, and Xenon Pharma, and royalties from Wiley, Elsevier, and Wolters Kluwers, all outside the submitted work. EP is an Editorial Board member of CNS Drugs. EP was not involved in the selection of peer reviewers for the manuscript nor any of the subsequent editorial decisions. MT received consultancy fees from Xenon Pharmaceuticals, Angelini Pharma, Clexio Bioscience. MT is the inventor of new Kv7 activators, as described in two patents entitled “Potassium channel activators and their use in treating disorders involving hyperexcitability” (EP23220110.3), and “New modulators of Kv7 potassium channels, methods of preparation and therapeutical uses” (PCT/EP2019/052067).

Ethics approval

Not applicable.

Consent to participate

Not applicable.

Consent for publication

Not applicable.

Availability of data and material

Not applicable.

Code availability

Not applicable.

Author contributions

This article is an invited review. All authors contributed to the literature search. MT drafted the initial version of sections 1 and 2, and prepared Fig. 1 (with the help of Dr. Camilla Celentano, PhD student in Neuroscience at the University of Naples Federico II, Naples, Italy). EP prepared the initial draft of Sects. 3 and 4. Both authors participated equally in the finalization of the manuscript and met ICMJE criteria for authorship. Both authors have read and approved the final submitted manuscript, and agree to be accountable for the work.