The KCNQ gene subfamily consist of five members (KCNQ1–5), all encoding voltage-gated potassium (K⁺) channel subunits (Kv7.1–5). The Kv7.1 subunit is predominantly expressed in the heart tissue, while Kv7.2-5 subunits are expressed most abundantly in the nervous system; Kv7.2 and Kv7.3 subunits expression is mostly restricted to the nervous system, whereas that of Kv7.4 and Kv7.5 subunits are more widespread, also including smooth and skeletal muscles (Sanguinetti et al., 1996; Schroeder et al., 2000; Cooper et al., 2001; Yeung et al., 2007). In adult neurons, Kv7 channels consist most frequently of tetramers containing Kv7.2 and Kv7.3, and more rarely Kv7.5 subunits (Wang et al., 1998; Schroeder et al., 2000). Independently of subunit composition, all Kv7 channels generate voltage-dependent K⁺ currents and represent the molecular basis of the M-current, a voltage-gated and K⁺-selective current which derives its name from its suppression upon activation of M₁ muscarinic receptors (Brown and Adams, 1980). Muscarinic-dependent M-current suppression is caused by M₁ receptor-dependent activation of phospholipase C, which hydrolyses phosphatidylinositol-4,5-bisphosphate (PIP₂), a molecule which is necessary for Kv7 channel opening (Suh and Hille, 2002). The M-current is a slowly activating, non-inactivating, time‐ and voltage-dependent potassium current, which regulates the resting membrane potential (RMP) and dampens repetitive neuronal firing, thereby controlling neuronal excitability (Wang et al., 1998; Brown and Passmore, 2009).

During the last decade, an emerging number of studies pointed out that severe disruption of the function of Kv7.2, Kv7.3, and Kv7.5 subunits due to de pathogenic variants in the encoding gene leads to developmental and epileptic encephalopathy (DEE; Weckhuysen et al., 2012; Lehman et al., 2017; Lauritano et al., 2019). DEEs are a heterogeneous group of mostly neonatal-, infantile‐ or childhood-onset disorders characterized by severe, drug-resistant seizures, characteristic electroencephalographic (EEG) signatures, and different levels of developmental delay or regression, often with a poor prognosis. Although anti-seizure drugs can have some effect on seizure control in KCNQ-related DEEs, and several patients become seizure free after a few weeks to months, these patients invariably suffer from severe developmental delay (Pisano et al., 2015). A complex relationship is known to exist between epilepsy and neurodevelopment; before the most recent International League Against Epilepsy classification introducing the DEE term (Fisher et al., 2017), the most-commonly used term of “epileptic encephalopathy” was meant to highlight the classical belief that epileptic activity by itself is the primary cause of cognitive and developmental deterioration. However, some patients can have developmental impairment before seizure onset, or cognitive deterioration may progress despite complete seizure control; given the variable relationship between seizures and cognition, these conditions have been more recently referred to as DEEs. Thus, the possibility that the control exerted over neuronal excitability by Kv7.2, Kv7.3, and Kv7.5 subunits may play a role in neurodevelopment deserves to be investigated. In this review, we will focus on what is currently known about the role of Kv7.2 channels in neurodevelopment.

The best proof for a role of the M-current in neurodevelopment is the observation that pathogenic variants in KCNQ2 are responsible for a spectrum of neurodevelopmental disorders in humans. Already in 1998, inherited pathogenic variants in KCNQ2 were identified as a cause of benign familial neonatal epilepsy (BFNE; KCNQ2-B; Singh et al., 1998). This disorder is characterized by neonatal seizures, but development in these patients is generally normal (Maljevic and Lerche, 2014). However, the risk of seizure recurrence in BFNE individuals is 10–20 times higher than the general population (Plouin, 1994), and a few BFNE families with individuals affected with various degrees of developmental disability have been described (Dedek et al., 2003; Borgatti et al., 2004). In 2012, some heterozygous pathogenic variants in KCNQ2 were surprisingly shown to result in DEE, with a developmental delay ranging from mild to profound, nowadays referred to as “KCNQ2-encephalopathy” (KCNQ2-E; Weckhuysen et al., 2012). Pathogenic variants in KCNQ2 are shown to be the most common genetic cause of neonatal-onset DEE. A recent prospective cohort study showed that 83% of newborns with a DEE had an identifiable genetic etiology, of whom 40% had a pathogenic variant in KCNQ2 (Shellhaas et al., 2017). The incidence of KCNQ2-E is reported to be 1–3/100.000 births (Lopez-Rivera et al., 2020). By contrast, only a handful of DEE patients with pathogenic variants in KCNQ3 or KCNQ5 have been described (Rauch et al., 2012; Epi4k Consortium et al., 2013; Lehman et al., 2017; Ambrosino et al., 2018; Kothur et al., 2018; Lauritano et al., 2019; Rosti et al., 2019; Sands et al., 2019).

Interestingly, the degree of developmental delay does not completely correlate with the frequency or severity of the epileptic seizures (Weckhuysen et al., 2013). Furthermore, there seems to be a clear distinction between pathogenic variants leading to either KCNQ2-B or KCNQ2-E (Figure 1). Pathogenic variants associated with KCNQ2-B are truncating (splice, nonsense, and frameshift), whole gene deletions, or heterozygous missense variants resulting in haploinsufficiency with a 20–30% reduction of the M-current density when mutant subunits are expressed together with wild-type subunits in heterologous cell systems (Maljevic and Lerche, 2014). KCNQ2-E, on the other hand, is caused by heterozygous missense or in-frame indel mutations shown to have a dominant negative (DN; >50% reduction of the M-current density), or more rarely gain of function effect (GOF; >100% of the M-current density) when co-expressed with wild-type subunits (Miceli et al., 2013, 2015). Interestingly, KCNQ2 GOF variants can lead to distinct neurodevelopmental phenotypes. For example, patients with the mild GOF variant R198Q present infantile spasms without prior neonatal seizure (Millichap et al., 2017), whereas patients with the strong GOF variants R201C/H showed severe neonatal-onset encephalopathy without neonatal seizures but prominent startle-like myoclonus and a burst-suppression EEG pattern (Mulkey et al., 2017). KCNQ2 has also gained interest in the field of autism spectrum disorder (ASD). First of all, many of the reported KCNQ2-E patients show autistic features or have ASD as a behavioral comorbidity (Milh et al., 2013; Millichap et al., 2016). Second, KCNQ2 variants are also commonly identified in genetic studies on ASD cohorts (Jiang et al., 2013; Long et al., 2019). The mechanism by which two opposite effects on the M-channel, either a reduction of more than 50% of the M-current (DN effect) or an increase of the M-current density (GOF effect), both lead to developmental delay, is not known yet. However, this shows that the level of the M-current has to be regulated in a very precise manner to maintain homeostasis of the neuronal network.

Heterologous cellular systems have been used abundantly to study the functional consequences of human pathogenic KCNQ2 variants, and electrophysiological assays have been used as a read-out to assess genotype-phenotype correlations (Orhan et al., 2014; Abidi et al., 2015; Miceli et al., 2015; Millichap et al., 2017; Kim et al., 2018; Soldovieri et al., 2019). These systems have the advantage of being relatively straightforward, cost-efficient, and relatively high-throughput (Maljevic et al., 2017). The disadvantage of this approach is that the lack of native-neuronal environment may influence channel behavior in such a way that those variant-induced functional changes that affect channel subcellular localization or interaction with neuronally-expressed proteins may not become apparent. This was demonstrated by two recent studies on two different recurrent KCNQ2-E variants, A294V and R560W. Both generated similar currents as seen for KCNQ2-B variants when overexpressed in heterologous systems (Orhan et al., 2014; Abidi et al., 2015). The DN effect could only be observed in transfected neurons; the A294V variant altered distribution of Kv7.2 containing channels to the somato-dendritic compartment, while the R560W variant reduced Kv7.2 axonal surface expression and impaired gating regulation by phosphoinositides (Kim et al., 2018). This data underline the importance of a neuronal environment to fully understand the disease pathomechanism(s) triggered by each pathogenic KCNQ2 variant.

An interesting side note is that the Kv7.2 channel has several accessory proteins, many of which are also associated with neurodevelopmental disorders. SCN1B encodes for Na_Vβ1, a multifunctional molecule that is involved in modulation of sodium and potassium channels (including Kv7.2 channels), neurite outgrowth, axon pathfinding, and cell migration (Brackenbury and Isom, 2011; Nguyen et al., 2012). Bi-allelic pathogenic variants in the SCN1B gene are linked with DEE with phenotypic properties similar to Dravet syndrome, whereas heterozygous pathogenic variants lead to generalized epilepsy with febrile seizures plus (Scheffer et al., 2007; Aeby et al., 2019). Another important accessory protein of the Kv7.2 channel is Ankyrin-G, which is encoded by the ANK3 gene. Ankyrin-G has the essential role to anchor the Kv7.2 subunit at the axon initial segment (AIS) and nodes of Ranvier (Devaux et al., 2004; Pan et al., 2006). Heterozygous pathogenic missense variants have been linked to intellectual disability, ASD, bipolar disorder, and schizophrenia (Ferreira et al., 2008; Bi et al., 2012; Iqbal et al., 2013). Lastly, syntaxin-1A is a member of the soluble N-ethylmaleimide-sensitive fusion protein attachment protein receptors (SNARE) superfamily. The SNARE protein complex facilitates the docking and fusion of synaptic vesicles at presynaptic membrane to enable neurotransmission (Rizo, 2018). Syntaxin-1A can bind to the C-terminus of Kv7.2 subunit, thereby decreasing open probability via facilitation of the interaction between the C-terminus and the N-terminus (Regev et al., 2009). To date, malfunction of syntaxin-1A itself is not known to be directly associated with neurodevelopmental disorders, however, pathogenic variants in the gene that encodes for the binding partner syntaxin binding protein 1 (STXBP1) are known to cause several DEE forms including a similar phenotype to KCNQ2-encephalopathy (Stamberger et al., 2016). STXBP1-Loss-of-function (LOF) mutations results in an increased binding of syntaxin-1A to the Kv7.2 channel, and thus, a decrease in M-current density (Devaux et al., 2017). Therefore, alterations in M-current density could be a convergence point between STXBP1-E and KCNQ2-E.

In this section, we will focus on Kv7.2 channel expression in the central nervous system. However, it should be noted that Kv7 channels are also expressed in the peripheral nervous system (PNS) (Passmore et al., 2003). To date, only few studies are available regarding the Kv7 subunit distribution in the PNS; nonetheless, most of those document a high Kv7.2 and Kv7.3 subunits expression in rat small-diameter dorsal root ganglionic neurons and non-myelinated C fibers (Passmore et al., 2003; King and Scherer, 2012; Huang et al., 2016). In rat peripheral sciatic nerves, most small‐ and medium-sized axons were shown to express Kv7.2/Kv7.3 heteromers, while in the largest sciatic nerve axons, mostly Kv7.2 homomers are observed (Cooper, 2011).

A common method to study the function of M-channels is the use of pharmacological blockers or openers in in vitro or ex vivo cell models. Retigabine and flupirtine are the most commonly used openers and have a specificity for Kv7.2-5 subunits (Gribkoff, 2003; Yeung et al., 2008). In 2011, retigabine was approved as a treatment for epilepsy by the Food and Drug Administration and the European Medicines Agency; however, it was withdrawn from the market in June 2017 due to side effects including blue discoloration of skin and retina (Tompson et al., 2016). 10,10-bis(pyridin-4-ylmethyl)anthracen-9-one (XE991) and linopirdine are the most commonly used Kv7 channel blockers, both inhibiting all members of the Kv7 family. When interpreting these pharmacological studies, it is, thus, important to realize that none of these drugs are truly Kv7.2 subunit specific.

A clear distinction should also be made between studies using acute and chronic interventions. Acutely blocking the M-current in neurons is known to increase the intrinsic excitability, whereas chronic blocking gradually reduces the intrinsic excitability again. The chronic block induces a compensatory distal shift of the AIS and a relocation away from the soma of Nav and Kv7-channels, thus, again stabilizing network excitability (Lezmy et al., 2017). Chronical blocking of the M-current does not only influence excitability properties but has also been shown to alter neurodevelopmental processes, such as neurite outgrowth and synaptogenesis in different neuronal cell models (an overview of the studies is given in Table 1).

Chronic blocking of the M-current with XE991 in PC12 and P0 rat cortical neurons was shown to result in an increased neurite outgrowth, while retigabine had the opposite effect (Zhou et al., 2016). Additionally, Kv7.2 channels have been described to play a key role in the regulation of synaptic plasticity through its interaction with neuronal pentraxin 1 (NP1) at presynaptic terminals of excitatory synapses and the axonal growth cones of fetal cortical neurons. Chronic blocking of Kv7.2 channels in these neurons increases excitatory synaptogenesis (Figueiro-Silva et al., 2015). Remarkably, another study using chronic XE991 treatment in differentiating mESC and embryonic hippocampal neurons was shown to decrease synaptic and vesicular proteins in inhibitory neurons but not in excitatory neurons (Zhou et al., 2011). A possible explanation for these opposite results could be the difference in studied cell type and cell maturity between the two studies, as Kv7 subunits and transcripts show distinct expression levels, spatial expression, and interactors depending on cell type and maturity stage.

To date, animal models have led to many fundamental discoveries of brain function and have given us a better understanding of the regulation of neurodevelopmental processes. Both knockout (KO) or knock-in (KI) mice models have been generated to study the role of Kv7.2 channels in vivo.

The role of the Kv7.2 channel in neurodegeneration is not the focus of this review; however, it is interesting to note that common genetic variation in KCNQ2 has recently been associated with risk of cognitive decline in healthy elderly (Bonham et al., 2018). Furthermore, some evidence exists for an (indirect) role of Kv7.2 in neurodegeneration disorders. Amyloid beta (Aβ) oligomers, involved in the pathogenesis of Alzheimer’s disease (AD), were reported to reduce Kv7.2 expression, resulting in a decrease of the M-current (Colom et al., 2011; Leao et al., 2012; Duran-Gonzalez et al., 2013; Mayordomo-Cava et al., 2015). In the medial septal neurons in mice, Aβ seems to reduce the M-current only in the glutamatergic neurons, resulting in increased firing and altered medial septal rhythmicity (Leao et al., 2012). As the medial septal nucleus is a key to generate the theta waves in the hippocampus, which is important for memory formation, alterations in the rhythmicity might underlie the cognitive deficits in AD patients (Winson, 1978; Seager et al., 2002). If the hypothesis is correct that hyperexcitability contributes to cognitive decline, specific Kv7.2 channels openers could be of benefit in AD patients. This could also be part of the explanation why no significant improvements in cognition were observed during a clinical trial of linopirdine, a Kv7.2 blocker, in AD patients (Rockwood et al., 1997). Furthermore, Kv7.2 channels openers could also be beneficial for Huntington’s disease (HD), as retigabine has been reported to rescue the electrophysiological and behavioral phenotype of transgenic HD model (Cao et al., 2015). Moreover, retigabine was described to block the hyperexcitability and to improve motor neuron survival of iPSC-derived motor neurons from amyotrophic lateral sclerosis patients (Wainger et al., 2014). Lastly, for Parkinson’s disease, several studies have shown that blocking of Kv7 channels with XE991 increased excitability of the dopaminergic neurons and has a neuroprotective effect toward dopaminergic degeneration in a mouse model of Parkinson’s disease (Shi et al., 2013; Liu et al., 2018). Altogether, those results support the idea that Kv7 subunits, possibly including Kv7.2 subunits, are interesting therapeutic target for different neurodegenerative diseases and cognitive impairment.

Since the discovery of the KCNQ2 gene in 1998, KCNQ2-encoded Kv7.2 subunits have been studied extensively (Singh et al., 1998). Although multiple lines of evidence show that Kv7.2 channels are involved in neurodevelopment, most studies have focused on the function of the Kv7.2 channels in a mature neuronal environment. Therefore, in this review, we highlight the importance of Kv7.2 in (early) neurodevelopment. Thanks to the increasing use of RNA-sequencing and the availability of transcriptome databases, a large amount of information on KCNQ2 expression has become available, leading to the consensus that an increasing expression of this gene is consistently observed during neuronal development (Safiulina et al., 2008; Telezhkin et al., 2018). Interestingly, multiple transcriptome datasets show evidence for the expression of a shorter fetal KCNQ2 transcript (transcript e) in iPSCs and neuronal derivatives, in addition to the longer transcript transcript c (Wang et al., 2005; Jiang et al., 2010; Linta et al., 2013). The presence of the short transcript was already reported in 2001 in brain, both at the transcript and protein levels; however, no follow up studies investigating this transcript were performed (Cooper et al., 2001; Smith et al., 2001). Unfortunately, functional evidence for the presence of Kv7.2 protein at stem cell and neuronal progenitor stages is currently lacking. Future studies confirming or negating the presence of Kv7.2 isoforms in stem cells will be needed to clarify this issue. In the meantime, and based on the available data, we introduce a model dividing the M-current density in three phases: (1) very low density important to maintain cell proliferation, (2) increasing M-current density to hyperpolarize the membrane and support neuronal differentiation, and (3) a stable high M-current density present in maturing or mature neurons to dampen repetitive firing and control neurotransmitter release.

Regarding human disease related to Kv7.2 channel dysfunction, more specifically KCNQ2-E, it is clear that pathology goes beyond a simple increased open or closed state of the channel, but includes impact on temporo-spatial expression levels, binding partners, and widespread homeostatic network changes that are difficult to study in simplified models. The recent publication of a KCNQ2-E KI mouse model is certainly promising and will be an important tool for further in-depth studies (Milh et al., 2020). Another promising model that would contribute to our understanding of the role of Kv7.2 channels in neurodevelopment is neuronal cultures derived from human iPSCs, as they provide the neuronal environment which is lacking in heterologous cell systems. Their ability to differentiate in several different brain cell types allows us to look not only at network changes but also to the contribution of different cell types to pathology. They are especially useful in the study of neurodevelopmental disorders, since they largely recapitulate fetal development (Tidball and Parent, 2016; Logan et al., 2019). Compared to monolayer neuronal cultures, brain organoids have the ability to foster even more complex cell interactions that stimulate cellular maturation, better recapitulating potential pathological cascades (Logan et al., 2019). Both 2D iPSC-derived neuronal cultures and 3D brain organoids have already successfully been used for the study of several neurodevelopmental disorders to identify disease mechanisms and potential new therapeutic targets (Fink and Levine, 2018; Russo et al., 2019). To date, no iPSC-derived neuronal model for KCNQ2-E has been reported in the literature yet. Furthermore, the current research performed to understand the role of the Kv7.2 channel in neurodevelopment has focused mainly on LOF of Kv7.2 channels, whereas KCNQ2 GOF variants give rise to an at least equally severe neurodevelopmental phenotype in human patients (Miceli et al., 2015). How both DN and GOF variants lead to neurodevelopmental delay is not understood yet. Future efforts comparing the effects of both DN and GOF variants on the neuronal network in vitro and in vivo are necessary to understand how Kv7.2 regulates neurodevelopment and can potentially open new doors for innovative therapeutic tools. Lastly, the development of Kv7 channel blockers and openers overcoming some of the pharmacokinetic and pharmacodynamic limitations shown by the available drugs (all of which have been synthesized and started to be developed well before the discovery of the KCNQ gene subfamily and the description of their multiple pathophysiological roles) could not only be beneficial for patients suffering from KCNQ2-B and KCNQ2-E but could also positively impact patients affected with multiple neurodegenerative disorders (Wainger et al., 2014; Kumar et al., 2016; Ostacolo et al., 2020).

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication. ND and SW designed the scope and structure of the review. ND, SW, FM, and MT wrote and edited the final manuscript.