SCN2A-Related Disorders

Summary

SCN2A encodes a voltage-gated sodium channel (designated NaV1.2) vital for generating neuronal action potentials. Pathogenic SCN2A variants are associated with a diverse array of neurodevelopmental disorders featuring neonatal or infantile onset epilepsy, developmental delay, autism, intellectual disability and movement disorders. SCN2A is a high confidence risk gene for autism spectrum disorder and a commonly discovered cause of neonatal onset epilepsy. This remarkable clinical heterogeneity is mirrored by extensive allelic heterogeneity and complex genotype-phenotype relationships partially explained by divergent functional consequences of pathogenic variants. Emerging therapeutic strategies targeted to specific patterns of NaV1.2 dysfunction offer hope to improving the lives of individuals affected by SCN2A-related disorders. This Element provides a review of the clinical features, genetic basis, pathophysiology, pharmacology and treatment of these genetic conditions authored by leading experts in the field and accompanied by perspectives shared by affected families. This title is also available as Open Access on Cambridge Core.

Introduction

This is the second gene-focused Element of the Cambridge Elements series on Genetics in Epilepsy launched in September 2021 [1]. The goal of this Element is to provide an in-depth, state-of-the-art review of clinical, genetic, basic science, and family perspectives on neurological and neurodevelopmental disorders (NDD) associated with pathogenic variants in SCN2A, which encodes a major voltage-gated sodium ion channel (designated Na_V1.2) in the brain. The SCN2A-related disorders are clinically heterogenous with features ranging from neonatal and infantile onset epilepsy, late onset epileptic encephalopathy, autism spectrum disorder (ASD), and intellectual disability (ID). In addition to phenotype diversity, the widespread use of clinical genetic testing has resulted in more than 1,000 SCN2A variants deposited in ClinVar. With the explosion in genetic variant identification has come recognition of genotype–phenotype relationships, which when coupled with experimental demonstration of functional perturbations are guiding new therapeutic approaches. Investigations into the biology of SCN2A has led to fundamental discoveries about the physiology and pathophysiology of synaptic connections and neural circuits. As one of the earliest known epilepsy genes, SCN2A has emerged as an important genetic factor in neurodevelopment and NDD.

We hope this Element will provide opportunities for families, trainees, and health care professionals to learn about SCN2A-related disorders. Subsections of this Element offer complete discussions about clinical features, pathophysiology, genetics, model systems, and treatment. This Element begins with perspectives from parents and caregivers of children with these disorders, made possible by a parent-led advocacy group (the FamilieSCN2A Foundation). The following subsections are devoted to clinical features, genotype–phenotype correlations, basic science, and current and future therapeutic approaches. Thus, this Element on SCN2A-related disorders provides a comprehensive and in-depth review of the state of knowledge in this field, which should be valuable to scientists, clinicians, trainees, and families interested in the topic.

In addition to a thorough and informative narrative, this Element is augmented by video content, including interviews with parents of children with SCN2A-related disorders (Video 1); with Dr. Matthew State (Professor and Chair, Department of Psychiatry and Behavioral Sciences, University of California, San Francisco) on the genomics of ASD and the discovery of SCN2A as a major risk factor (Video 2); and with Dr. Steven Petrou (Professor of Neuroscience, University of Melbourne, and Chief Scientific Officer at Praxis Precision Medicines) on his career evolution from academic research to a new pharmaceutical company specializing in precision medicine for rare neurological diseases (Video 3).

This was a team effort that we hope provides inspiration to future clinicians, researchers, and patient advocates. We hope you enjoy learning about this important epilepsy gene.

Patient, Family, and Foundation Perspectives

Pathogenic variants in the SCN2A gene are associated with a broad spectrum of complex NDD that are collectively designated as SCN2A-related disorders. The primary clinical manifestations include epilepsy, ASD, movement disorders, and ID. The severity of these conditions varies among individuals, ranging from mild and well controlled to severe and treatment resistant. Even individuals with mild clinical phenotypes exhibit significant impairments compared to their age-matched peers. Those on the severe end of the spectrum are profoundly affected and heavily reliant on their caregivers for all aspects of daily life. Given the complexity of SCN2A-related disorders, clinical care teams are often multidisciplinary, emphasizing the importance of coordinated efforts to optimize care and minimize clinical risks (Figure 1).

Figure 1 Diagram of multispecialty care of a patient with an SCN2A-related disorder.

FamilieSCN2A Foundation

The FamilieSCN2A Foundation, founded in 2015, is the largest nonprofit patient advocacy organization representing SCN2A-related disorders and has the broadest international footprint among other SCN2A patient-advocacy groups. Focused on creating an engaged ecosystem, FamilieSCN2A Foundation acts as a central liaison between patients, caregivers, scientists, clinicians, and industry. They also work in close partnership with other large foundations such as the Simons Foundation and the Chan Zuckerberg Initiative. “Families” is part of the Foundation’s name because these rare and devastating conditions affect the entire family. The Foundation strives every day and, in every way, to improve the lives of not only the patients but also the entire family.

The missions of the Foundation are to accelerate research, foster a sense of community, and advocate for improvements in the lives of those affected. The Foundation’s vision is centered around achieving effective treatments and cures for all SCN2A-related disorders. The core values of the Foundation are urgency, integrity, collaboration, and inclusion. FamilieSCN2A strives to provide families and professionals with the information and tools needed for a rapid and accurate diagnosis as well as the resources needed to tailor treatments based on the patients’ and families’ goals informed by research knowledge.

Before 2014, there was little hope for children or families diagnosed with an SCN2A-related disorder. There were no specialists treating the condition, no support groups for families desperately seeking answers, and no researchers investigating cures. This void created a gravity that pulled together a small group of thoughtful, committed parents who sought to build a better world for their children, and thus the FamilieSCN2A Foundation was born. The Foundation grew quickly as other parents and professionals were inspired and empowered by the stated vision: a world with effective treatments and cures for all SCN2A-related disorders. As momentum built, the Foundation attracted board members that shared the core values and missions. A timeline of Foundation milestones is presented in Figure 2.

Figure 2 Timeline of major events involving the FamilieSCN2A Foundation.

Advocacy is critical to the mission of FamilieSCN2A. The tenets of their advocacy strategy are: awareness, empowerment, evidence-based research, and equity. Successful advocacy requires awareness within the patient/family community and beyond. Awareness of SCN2A-related disorders began with the first online support group using the Facebook platform, which launched in 2013 with just five members. Today, there are more than 1,000 participants. This private, robust group is a safe space for families, patients, and caregivers to share their journey, ask questions, learn from one another, and reduce feelings of isolation. The community of patient caregivers remains a pillar of support for individuals affected by SCN2A-related disorders.

To extend awareness beyond families, the FamilieSCN2A Foundation spearheads various awareness initiatives including state proclamations, listening sessions with the Food and Drug Administration (FDA), providing information to policymakers and drug developers, and organizing caregiver testimonies that are intended to raise awareness of the burden of disease. Education, which is vital for a community actively engaged in collaboration with clinicians, scientists, and industry, has been an important Foundation strategy. As such, the FamilieSCN2A Foundation organizes an annual family and professional conference that leaves participants feeling empowered, educated, and hopeful. Supplementing the educational initiatives are programs that support families both emotionally and financially. Initiatives include Family Meet Up Grants, a Birthday Club, and a Patient Assistance Grant program that has awarded more than $65,000 to families since 2015.

The accomplishments of the FamilieSCN2A Foundation include $4 million raised since 2015, 1,200 families supported globally, strategic partnerships with leading academic researchers, a voice with the FDA, representation at international conferences, and a growing attendance at the annual family and professional symposium. Its founders, board members, volunteers, and community (including researchers and clinicians) together have made significant progress in a thoughtful way and committed to changing the world for those affected with SCN2A-related disorders.

Family and Caregiver Perspectives

There is a dearth of literature that describes the burden of care and health-related quality of life (HRQoL) as they relate to SCN2A-related disorders. While Cohen and colleagues [2] summarized quality of life and its determinants in developmental and epileptic encephalopathies (DEEs), for which SCN2A pathogenic variants accounted for 24 percent of the cases (n=42/173), no peer-reviewed HRQoL studies have been published specific to SCN2A-related disorders. Given the severity of SCN2A-related disorders and their impact on individuals and their families, it is imperative that additional HRQoL studies be conducted. Further, understanding the lived experiences of patients and their caregivers facing rare neurologic diseases is critical to advancing patient-centered outcomes research and informing clinical trial design. Lived experiences also highlight unmet needs of the SCN2A-related disorders community and describe the nuances of the disease beyond clinical care. Thus, this section provides a glimpse into some of the challenges families face on a daily basis and illustrates why the patient voice is a fundamental component that complements the clinical and scientific literature on SCN2A-related disorders.

To capture the family perspective, the FamilieSCN2A Foundation interviewed caregivers of individuals with SCN2A-related disorders. Interviews were transcribed and edited into a short video that highlights various aspects of the disorder (Video 1).

I felt shattered and heartbroken. The doctors did not know anything about this diagnosis. They were new to it but luckily, they did give us all the information of the FamilieSCN2A foundation. That was helpful.

– Sofia’s mom, United States

Terrible. My world just fell apart. I didn’t know what to do. Didn’t know where to look because in Belgium, there just aren’t many cases. The doctors told me not to Google it, and that was pretty scary.

– Charlie’s mom, Belgium

I felt a mixture of emotions for the first 23 months of my son’s life. He was initially neurotypical, then seizures started. Within two months of seizures starting, we got the diagnosis. I felt overwhelmed. There was a little bit of excitement knowing that there’s something that they can possibly treat, but at the same time scary that they knew very little about it. I was scared for the future.

– Hudson’s mom, United States

I was actually thrilled. It was such a relief to finally figure out what was wrong. I knew from infancy that something was wrong. I always equated it to a juggling act: she has GI reflux, and she’s got a tic-like behavior, and she’s got apraxia of speech, and she has double hip dysplasia. I kept asking, so then what is it? There has to be something underlying. I just wanted a name for it.

– Erin’s mom, United States

The following express the difficulties of having a child affected with a SCN2A-related disorder.

Charlotte was born having over 400 seizures a day and she was very prone to illness. As her life went on, she had more and more seizures, and towards the end of her life, she was in status most of the time and her brain wasn’t functioning anymore. The hardest part for me was that I am an Intensive Care Unit nurse and knew too much.

– Charlotte’s mom, United States

It’s kind of like stripping everything away that I had envisioned for my child. I had to re-evaluate what his future might look like. I watched him suffer and was not able to do anything about it. He still pushes through with a smile, but it’s challenging to know that there’s nothing I can do about it at this moment.

– Hudson’s mom, United States

The way families feel about a cure for SCN2A-related disorders is expressed by these statements.

I believe in a cure. There are a lot of prayers as well as just the parental force of the FamilieSCN2A foundation that have come together to make me feel hopeful that there would someday be a cure.

– Sofia’s mom, United States

I always believe in a cure. That’s the main reason why I attend the conferences. That why we’re not giving up, why we’re going across the globe to find people who understand and who are willing to work with us. That’s why I’m screaming from the rooftops telling everyone, “this is what my child has!” I tell every cab driver. I’m here for this, this is my daughter, she has this, I’m telling everyone.

– Charlie’s mom, Belgium

I think there’s going to be a cure. I’m not 100 percent sure it’s going to be in her lifetime.

– Erin’s mom, United States

These quotes were in response to asking parents how they have been impacted by FamilieSCN2A Foundation.

We ask questions and we get answers that we don’t get from our doctors. We get mental as well as emotional support from each other, even coming to the conferences and learning about what the research is done or just meeting other families has been possible through the funding that is available through the FamilieSCN2A Foundation for which I am very grateful.

– Sophia’s mom, United States

It’s a huge support system knowing that I’m not alone and that there are others that understand my struggle. It’s a judgment-free zone. It’s a source of hope because I see what the foundation is doing. It gives me hope that people care and they are trying to do something about SCN2A.

– Hudson’s mom, United States

I believe that the foundation provides a community of support for our families. It gives families hope and I am really excited at the research that they initiated, to get doctors excited about researching this disease, and putting money towards finding a cure or at least a better quality of life for our kids.

– Charlotte’s mom, United States

In an effort to build upon the interview data, FamilieSCN2A disseminated a questionnaire to their community in March 2023 that asked questions related to the consequences of caregiving. Two specific questions generated data that formed visual representations of (1) how caregivers felt when their children were first diagnosed and (2) what they wished their providers knew about SCN2A-related disorders. Figure 3 illustrates word clouds representing these responses.

Video 1 Parents (Sandya Crasta, Liz Hendrickx, Amy Richards, Ashley Taylor, Tracy Umezu) discussing experiences caring for their children with SCN2A-related disorders.

A transcript of this video is available in the Appendix. The video file is available at www.cambridge.org/scn2a In addition, the following paraphrased quotes describe how families felt when they received a diagnosis of SCN2A-related disorders.

(A) Responses expressing how caregivers felt when their child was first diagnosed.

(B) Responses expressing what caregivers wished their health care providers knew about SCN2A-related disorders.

Figure 3 Word clouds generated from caregiver responses.

Given the dearth of literature describing the challenges and consequences of caring for SCN2A-related disorders for patients and their families, combined with the speed at which SCN2A-related disorders are being studied and the simultaneous growth of the FamilieSCN2A Foundation, we hope this section has provided a thorough overview that highlights various aspects of the disorder and leaves families feeling empowered and health care professionals with valuable insights to navigate the complexities associated with SCN2A-related disorders.

Clinical Spectrum and Genotype–Phenotype Correlations

Pathogenic SCN2A variants are associated with a range of NDD with or without epilepsy, having symptom onset anytime between the first day of life through later childhood (Figure 4). Due to the variability in clinical presentation, data on the incidence of SCN2A-related disorders is not complete. For example, epilepsy patient cohorts exclude cases of NDD that do not have seizures. Expanded genetic testing that includes a wider spectrum of NDD will likely reveal an even greater genotype and phenotype diversity.

Figure 4 Clinical spectrum of SCN2A-related disorders. Black-shaded boxes indicate phenotypes associated with normal development. Unshaded boxes indicate phenotypes associated with mild or moderate to severe or profound ID. Arrows indicate predominant associated functional effects of SCN2A variants in each condition. Typical age of onset is given in months (m) and generally refers to the onset of seizures.

In a meta-analysis of genetic findings discovered using next generation sequencing-based gene panels, SCN2A was the fourth most commonly implicated gene in monogenic epilepsy (after SCN1A, KCNQ2, and CDKL5) accounting for 7 percent of cases [3]. By comparison, SCN1A-associated epilepsy accounted for nearly three times as many cases (19 percent). The incidence of SCN1A-associated epilepsy was accurately estimated as 1 per 12,200 live births from a population-based cohort study in Scotland [4]. Based on the assumption that SCN1A-associated epilepsy is approximately three times more common than SCN2A-associated epilepsy, one might expect the incidence of SCN2A-related epilepsy to be in the range of 1 in 30,000 to 50,000 live births. An estimate of 1 per 78,608 live births for SCN2A-related epilepsy was based previously on observing seven cases of SCN2A-related epilepsy diagnosed in the single national Danish testing center between 2006 and 2014 [5].

A comparable incidence estimate of SCN2A-related NDD comes from the UK Deciphering Developmental Disorders (DDD) study [6,7], a national multicenter study in which participants with a wide range of developmental disorders underwent whole exome sequencing (WES). The DDD incidence estimate includes SCN2A-related NDD without epilepsy as a feature, but excludes self-limited and familial epilepsy cases. Analyzing the first 4,293 families in the DDD study, the authors estimated that 42 percent of the cohort carried a disease-causing de novo variant in any gene. They further estimated that a de novo variant can be expected to result in a developmental disorder between 1 in 213–448 births. Because 19 study participants with SCN2A-related NDD were among these 4,293 families, the incidence of SCN2A-related NDD can be estimated as 1 per 47,000 to 100,000 live births.

Nearly all individuals with SCN2A variants develop epilepsy at some point in their life. It is estimated that about half of individuals with SCN2A-related NDD will present with seizures in the neonatal period [5,8], and 80 percent will develop seizures within the first six months of life [5,8]. Focal seizures are the most common seizure type, reported in 90 percent of individuals, and epileptic spasms occur in up to 50 percent of individuals with SCN2A-related NDD [8]. Nearly 80 percent of individuals with SCN2A-related epilepsy have comorbid developmental delay [8]. Mosaicism is estimated to occur in 6.4 percent of pathogenic SCN2A variants, present at only 11.6 percent of variant reads (range: 11.6–39.5 percent) [9].

Self-limited Epilepsies

Self-limited familial neonatal-infantile epilepsy (SeLFNIE), previously known as benign familial neonatal-infantile epilepsy, was the first reported SCN2A-related disorder [10,11]. Self-limited epilepsies represent approximately 20 percent of the phenotypic spectrum of reported SCN2A-related disorders [12], and include neonatal, neonatal-infantile, and infantile subtypes, which are classified by age of seizure onset along with family history [13]. Seizure onset ranges from 2 days to 23 months with a mean of 11.2 ± 9.2 weeks and median of 13 weeks [11,12]. Approximately half of affected individuals have onset before age one month [5]. Seizures are typically focal with head and eye deviation along with tonic or clonic features and possibly apnea; some evolve to bilateral tonic-clonic seizures [10,11,14,15]. Seizures may occur in brief clusters (20 seconds to 3 minutes), are most often afebrile, and are usually controlled with single antiseizure medications such as phenobarbital, phenytoin, oxcarbazepine, clobazam, zonisamide, or valproic acid [5,8,11,14]. Electroencephalograms are generally normal or show focal epileptiform activity without severe encephalopathy patterns, and seizures typically resolve by age 12 months with rare recurrence later in life [5,8,11,14]. Developmental milestones and cognitive outcomes are typically normal, and SCN2A variants in SeLFNIE are frequently inherited with high penetrance. Distinguishing SeLFNIE from more severe DEEs at the time of seizure onset can be challenging. A representative clinical scenario is described in Box 1.

Box 1

A three-week-old infant presented with focal motor, evolving to bilateral tonic-clonic, seizures. He had been born at term with no prenatal or perinatal complications. Interictal encephalography (EEG) demonstrated normal background rhythms. Seizures were controlled with carbamazepine. He was weaned off antiseizure medication at two years. He had recurrence of a single generalized tonic-clonic seizure following a concussion at six years but had a normal 24-hour EEG. Cognitive and developmental outcome was normal at six years.

The primary differential diagnoses for SeLFNIE, assuming structural brain imaging is normal or demonstrates nonspecific changes, include the following:

1. ➢ KCNQ2-related epilepsy: This condition presents with a range from self-limited infantile epilepsy to DEE with or without burst suppression on electroencephalography (EEG). The seizure onset is more often neonatal and responds well to carbamazepine or other sodium channel blockers [11,16,17].

2. ➢ PRRT2-related epilepsy: This disorder is characterized by self-limited infantile epilepsy with focal motor seizures and an excellent response to antiseizure medication, particularly sodium channel blockers [18,19]. Seizure onset is typically at age 4–12 months with resolution by two years. Paroxysmal kinesigenic dyskinesia is also associated with pathogenic PRRT2 variants but symptoms may start after seizure onset [19,20,21].

3. ➢ SCN8A-related epilepsy: Epilepsy associated with pathogenic SCN8A variants can manifest as self-limited infantile epilepsy or DEE. Median age of onset is approximately six months. Seizure types include focal, multifocal, or bilateral tonic-clonic. In the DEE phenotype, epileptic spasms are reported. Seizures in DEE caused by gain-of-function (GOF) SCN8A variants are reported to respond favorably to sodium channel blockers, though high doses are often required [22]. Movement disorders including myoclonus, tremor, and paroxysmal dyskinesias are reported [23,24,26].

Early Infantile Developmental and Epileptic Encephalopathy

DEE represents the most frequent clinical presentation among published cases of SCN2A-related disorders. The early infantile subtype of DEE is characterized by epilepsy onset within the first three months of life, usually in the neonatal period, and represented 36 percent of participants in a study of 201 individuals with SCN2A-related disorders [5]. In this study, 43 percent (31 of 71) of the early infantile DEE subset had an identifiable epilepsy syndrome, including early infantile DEE with burst suppression on EEG (18 of 71) and epilepsy of infancy with migrating focal seizures (EIMFS) (13 of 71), while the remaining 56 percent (40 of 71) had unclassifiable epilepsies with focal or generalized seizure types including tonic, tonic-clonic, and epileptic spasms. Initial interictal EEG often shows a burst-suppression pattern or multifocal spikes [5,27,28].

Intellectual disability, often profound or severe, is present in a majority of individuals. Comorbid movement disorders are common, including dystonia, chorea, choreoathetosis, and dyskinesia [5,28,29]. Other reported comorbidities include cortical visual impairment, microcephaly, and features of dysautonomia such as temperature instability (hypo- or hyperthermia) and tachycardia [5,28,29]. Brain MRI may be normal or show cerebral atrophy, hypomyelination, and T2 hyperintensities, among other abnormalities [5,27,29,30]. These MRI findings are nonspecific, and cerebral atrophy specifically has been associated with other causes of early infantile developmental encephalopathy [13,27]. Rare individuals with SCN2A-related DEE have been found to have polymicrogyria [31,32,33], suggesting a potential role of sodium channels in neuronal migration during brain development. A representative clinical scenario is described in Box 2.

Box 2

A female infant presented on day one of life with multiple predominantly focal tonic seizures per day. EEG initially showed burst suppression evolving to generalized slowing with multifocal or bi-frontal epileptiform activity. Phenytoin was the most effective antiseizure medication but did not achieve seizure freedom. Oxcarbazepine treatment was associated with a partial response and greater alertness but was accompanied by hyponatremia. A combination of lacosamide and lamotrigine resulted in two years of near seizure-freedom, with breakthrough only in the setting of missed or late doses. A likely pathogenic, de novo variant in SCN2A was identified by epilepsy gene panel testing.

At nine years of age, the patient exhibited global developmental delay with axial hypotonia and was unable to sit independently. She showed no purposeful hand use, smiled in response to voices but expressive communication was limited to nonspecific vocalizations. She had cortical visual impairment and made limited eye contact. Other medical issues included mild scoliosis, oral phase dysphagia with poor weight gain, and irritability.

Specific early infantile DEE phenotypes associated with pathogenic SCN2A variants are described next.

Epilepsy of Infancy with Migrating Focal Seizures (EIMFS)

EIMFS is characterized by multiple types of focal seizures that “migrate” from one hemisphere to the other, usually accompanied by severe developmental impairment [29]. SCN2A is second to KCNT1 as the most frequently identified gene in EIMFS [28,29]. In these two cited studies of individuals with EIMFS and SCN2A variants (22 total), onset of seizures occurred between the first day of life and eight weeks for most (20/22), and less commonly after age one year (2/22). All had multifocal spikes on EEG.

Early Infantile DEE with Burst Suppression

SCN2A has been identified as a common gene associated with early infantile DEE with burst suppression on EEG formerly called Ohtahara syndrome [8,27,29,30,31,34,35]. Seizure types typically include epileptic spasms, tonic seizures, and/or myoclonic seizures [13,36]. In two studies of individuals with early infantile DEE, SCN2A variants were identified in 9/67 (13.4 percent) and 2/33 (7 percent) respectively [27,34]. Infantile epileptic spasms syndrome (IESS) has also been reported in individuals with SCN2A-related early infantile DEE, either independently or evolving from an initial presentation as early infantile DEE with burst suppression [8].

Prognosis is variable. With regard to epilepsy, response to sodium channel blocking antiseizure medications in SCN2A-related early infantile DEE is often favorable, but high doses are often needed and there are reports of seizure relapse when plasma drug levels drop below a certain threshold [5,29]. There have been several reported instances of sudden unexpected death in epilepsy (SUDEP) in this subset of individuals [5,8,29], which warrants consideration in discussing the prognosis with families. Other reported causes of early death include pneumonia (age 21 months) [29], respiratory failure during treatment of pneumonia [30], iatrogenic cardiorespiratory failure (age 19 days) [37], severe infection, and status epilepticus [5].

The primary differential diagnoses for this group, assuming structural brain imaging is normal or demonstrates nonspecific changes, include the following:

1. ➢ KCNT1-related disorder: As mentioned previously, KCNT1 is the gene most frequently associated with EIMFS [28,29]. Differentiating features that are more associated with SCN2A include presence of severe movement disorders and seizure response to phenytoin [29,38]. Age of seizure onset may also be slightly later in those with KCNT1-associated EIMFS (median 3.5 weeks, compared to median of 3.5 days in those with SCN2A-associated EIMFS) [28]. Prognosis appears to be less favorable in KCNT1-related EIMFS, with higher reported rates of severe to profound developmental impairment, refractory epilepsy, and SUDEP compared to those with SCN2A variants [5,28,39].

2. ➢ KCNQ2-related disorder: Variants in KCNQ2 are a frequently discovered genetic cause of early infantile DEE with burst suppression [27,40], and should be considered in the differential diagnosis of a patient with this phenotype. Neonatal onset seizures associated with pathogenic KCNQ2 variant are also responsive to sodium channel blocking antiseizure medications [17], which may resemble the response of individuals with SCN2A GOF variants.

Later Onset Developmental and Epileptic Encephalopathies

DEEs presenting after three months of age constitute the second largest group (20–30 percent) with pathogenic SCN2A variants [5,41]. Affected individuals typically present with generalized tonic-clonic, absence, and myoclonic seizures along with EEG features of generalized spike and wave or multifocal spikes. A number of distinct epilepsy syndrome presentations in this later onset group have been described. This includes individuals with IESS evolving to Lennox-Gastaut syndrome (LGS), myoclonic atonic epilepsy (MAE), and focal epilepsies with a condition resembling electrical status epilepticus during slow-wave sleep (ESES). Brain imaging is mostly normal with the exception of occasional nonspecific cortical atrophy. While some are cognitively normal before seizure onset, ID eventually develops in all affected individuals and the majority have severe cognitive impairment. Affected individuals have prominent comorbidities including ASD and motor symptoms such as hypotonia, choreiform, or dyskinetic movement disorders. A representative clinical description is illustrated in Box 3.

Box 3

A female infant presented with epileptic spasms at age 10 months, associated with developmental regression. Prior to seizure onset, her development was mildly delayed. Prior to the genetic diagnosis, she was treated with oxcarbazepine, which led to exacerbation of seizures. At age 15 months, a heterozygous, de novo likely pathogenic variant in SCN2A was identified by WES. Treatment with ketogenic diet, vigabatrin, and ACTH improved cognition and development and temporarily resolved seizures, but epileptic spasms recurred after weaning ACTH. Additional antiseizure medications (levetiracetam, valproic acid) were not effective. At age six years, while treated with clobazam, clonazepam, and phenobarbital, she experienced up to three seizures per day. She had generalized hypotonia and did not sit independently or reach for objects. She turned her head toward sounds and made nonspecific vocalizations. Additional medical issues included intermittent choreoathetosis, autonomic dysfunction (periodic flushing with Raynaud-type phenomenon in hands and feet, constipation), mouthing, and leg-crossing stereotypes.

Later onset DEE associated with pathogenic SCN2A variants has a notable genotype–phenotype relationship [42]. Typically, SCN2A variants in this cohort are reported to have loss-of-function (LOF) effects due to missense, protein truncating, and splice site variants or missense variants with mixed GOF/LOF effects [15]. For example, the mixed function p.R853Q variant is the most common recurrent SCN2A variant [42]. Individuals carrying this variant present with a characteristic phenotype consisting of epileptic spasms, hypsarrhythmia on EEG, and choreiform movement disorder [42,43,44].

In SCN2A-related disorders associated with LOF variants, seizures are typically treatment refractory and seizure freedom occurs in only 34 percent of cases [5]. Seizures are typically unresponsive to sodium channel blockers and can worsen after the introduction of this class of antiseizure medication. Persons with later onset DEE respond better to other drug classes including benzodiazepines, levetiracetam, sodium valproate, and ACTH, compared with early onset DEE [15]. Specific later onset DEE phenotypes associated with pathogenic SCN2A variants are described next.

Infantile Epileptic Spasms Syndrome (IESS)

A prominent phenotype associated with SCN2A-related later infantile onset DEE is IESS [5,42]. Typical interictal EEG features include hypsarrhythmia or multifocal or focal epileptiform discharges, which may occur soon after the onset of epileptic spasms [13]. Age of onset is typically between 3 and 24 months. Developmental delay is observed after onset of epileptic spasms, but may be absent early in the course. Abnormal neurological examination findings may be present including abnormalities of posture, tone, or movement. The family, pregnancy, and birth histories are usually normal. Cognition varies from normal to severely delayed before seizure onset; however, moderate to profound developmental impairment becomes evident with time [13,15]. Long-term prognosis is unfavorable and many subsequently evolve to LGS [5]. SCN2A-related IESS is difficult to treat and does not respond to standard antiseizure medications, steroids, or ketogenic diet [5,34].

The main differential diagnoses for late onset IESS are other genetic etiologies including pathogenic variants in ARX, IQSEC2, TSC1, TSC2, and others [13]. For early onset IESS, CDKL5 and STXBP1 should be considered [13]. In addition, a range of chromosomal abnormalities and copy number variants has been associated with IESS [45,46,47].

SCN2A-Related Disorders without Epilepsy

Individuals with pathogenic variants in SCN2A may present with NDD without epileptic seizures. The proportion of SCN2A-related disorders without epilepsy is unclear because of an ascertainment bias arising from preferential genetic testing of children presenting with seizures as compared with children having non-syndromic ID. Perhaps the best estimate of the incidence comes from the DDD study [48]. Inclusion criteria for the DDD study were broad, including children with severe undiagnosed NDD and/or congenital anomalies, abnormal growth parameters, dysmorphic features, and unusual behavioral phenotypes [7]. Richardson et al. reported the phenotypes of 22 persons including 12 presenting with neonatal or early infantile (≤3 months) onset seizures, four with ID/ASD and later onset seizures (seizure onset between two and nine years), and six with ID/ASD that had no history of epilepsy or epileptic seizures [48]. Thus, it could be estimated that approximately 20–30 percent of persons with NDD associated with SCN2A variants have no history of epileptic seizures.

In the large series of 201 individuals with SCN2A-associated disorders, 16 percent did not have epileptic seizures, although this sample may have ascertainment bias [5]. In the Richardson et al. study, all participants with either later onset seizures or no history of seizures had de novo variants and most (6 of 10) were protein truncating [48]. In contrast, all variants associated with the early onset seizure phenotypes (≤3 months) were missense. Among the six study participants without epilepsy, the degree of ID varied from mild to profound; half were formally diagnosed with ASD and no one had ataxia. Normal or nonspecific neuroimaging findings, including cerebral atrophy and hypoplasia of the corpus callosum, were reported for this subset. EEG findings were not reported by Richardson et al. [48], but a case report described the EEG in the setting of later onset developmental delay without epilepsy associated with a de novo missense SCN2A variant [49]. Findings in this case were bilateral discharges of high amplitude sharp waves and slow waves in the parietal–temporal–occipital regions that increased during sleep and were associated with right frontal-central short sequences of 5–4 Hz spike and wave complexes.

Episodic Ataxia

Episodic ataxia is observed in a subset of individuals with pathogenic SCN2A variants [50,51,52,53]. Most have epileptic seizures beginning during the first three months of life, and all have missense variants in the gene, suggesting that episodic ataxia may be a GOF disorder related to the early onset epilepsy phenotype. Onset of ataxia ranges from 10 months to 14 years. Episodes are highly variable, ranging from brief daily events to infrequent long-lasting episodes. Acetazolamide was effective in only a minority of reported cases [53]. In a systematic review of genetic epilepsies associated with paroxysmal ataxia, potential triggers for paroxysms of ataxia in SCN2A-related disorders were cited including minor head injuries, sleep deprivation, alcohol ingestion, photic stimulation, sudden noise, vibration of the body, and menstruation [54]. The majority (80 percent) with SCN2A-related episodic ataxia have good cognitive outcomes. A clinical vignette describing this syndrome is presented in Box 4.

Box 4

A female patient with an uncomplicated prenatal and perinatal history developed focal clonic seizures beginning on the fifth day of life. These were controlled with intravenous phenytoin and did not recur. Starting at age 10 months, she developed episodes of marked truncal ataxia, lasting 12–24 hours and occurring twice per month. Introduction of carbamazepine (titrated up to 10 mg/kg twice daily) and acetazolamide (titrated up to 2 mg/kg three times daily) did not change the frequency or severity of the episodes. At age two years she had a mild global developmental delay and muscular hypotonia. She was found to have a de novo pathogenic missense variant in SCN2A.

In addition to episodic ataxia, other paroxysmal movement disorders have been anecdotally associated with SCN2A variants, including paroxysmal dyskinesia and choreoathetosis occurring in the context of both early and later onset epilepsy [50,55,56]. Severe choreoathetosis has been reported in persons heterozygous for the recurrent SCN2A variant p.R853Q [12,43,44]. Familial and sporadic hemiplegic migraine has also been associated with pathogenic SCN2A missense variants either with no history of epilepsy or with SeLFNIE [57]. Alternating hemiplegia of childhood has also been associated with pathogenic SCN2A variants in cases without mutations in the main gene (ATP1A3) [58].

There is no strong evidence supporting specific therapeutic approaches for patients with non-epileptic movement disorders associated with SCN2A variants. The majority of functionally tested variants associated with episodic ataxia exhibit GOF properties in either in silico or in vitro models. This is consistent with many cases being observed in the context of prior neonatal onset epileptic seizures. There are very few case reports of effective treatment of ataxia with sodium channel blocking medications. Treatment with acetazolamide has often been reported to be effective at reducing frequency and severity of ataxia episodes in approximately half of patients. Acetazolamide has been reported as effective in cases associated with both GOF and LOF variants in SCN2A [53].

Genotype–Phenotype Correlations

General genotype–phenotype correlations are established for SCN2A-related disorders. Individuals with self-limited epilepsy are heterozygous for mostly inherited missense variants [12]. More severe epilepsy phenotypes are mainly associated with de novo variants that confer greater electrophysiological dysfunction [12,59]. There are no clear correlations between the location of the variant on the protein and phenotypic severity. Amino acid substitutions between physicochemically similar amino acids are less likely to cause severe disease and are more frequently observed in self-limited epilepsies [12,60]. Functional analyses of some variants associated with self-limited epilepsy and episodic ataxia phenotypes have demonstrated GOF effects [51,53,61].

Individuals with early onset DEE mainly have missense variants (77 percent) but can also carry truncating variants (23 percent) [5,12,62]. The missense variants tend to cause GOF effects; however, the functional properties are complex and do not strictly separate as GOF or LOF based on age of epilepsy onset [63]. Truncating and nonsense (premature termination codon) variants present exclusively in individuals with seizure onset beyond the first year of life or without epilepsy, and those with truncating variants are reported to have seizures in 50 percent of cases [5,12]. The majority of individuals with truncating variants have features of ASD or developmental delay, compared to only 20 percent of those with missense variants. Individuals with ID/ASD and later onset epilepsy or absence of epilepsy mainly have truncating variants (75 percent) that are LOF [5,12,62].

Recurrent SCN2A variants occur throughout the phenotypic spectrum (see Table 1), but individual variants may not always associate with the same phenotype, even within families [5,64]. A computational analysis of clinical features of 413 individuals with pathogenic SCN2A variants demonstrated that only 8 of 62 individual recurrent variants were associated with similar phenotypes [42].

Table 1 Recurrent missense SCN2A variants (reported in four or more affected individuals)*

Variant	Number	Clinical phenotype	Location	Functional assessment / prediction
p.R853Q	18	EE [34,43,204] [5,44,205,206,207] NE [208]	Domain 2, S4 segment	Mixed/LOF [44,63,165,209] in silico prediction: LOF [121]
p.A263V	14	EI-DEE [5,37,51,53,210,211,212] SLE [50,52,53,67]	Domain 1, S5 segment	GOF [50] in silico prediction: GOF [121]
p.R1882Q	10	EI-DEE [5,29,44,212,213] LO-DEE [44]	C-terminus	GOF [5,44] [63,214] in silico prediction: GOF [121]
p.E999K	8	EI-DEE [5,34,213,215,216] LO-DEE [212]	Domain 2−3 linker	GOF [63,91,217] in silico prediction: neutral [121]
p.L1342P	5	LO-DEE [5,55,218,219]	Domain 3, S5 segment	Mixed [62] in silico prediction: GOF [121]
p.R1319Q	5	SLE [11] EI-DEE [5]	Domain 3, S4 segment	Mixed/LOF [63,220]
p.L1650P	4	LO-DEE [221] NE [221]	Domain 4, S4−5 linker	in silico prediction: GOF [121]
p.M1545V	4	EI-DEE [5,156,222,223]	Domain 4, S1 segment	in silico prediction: GOF [121]
p.R1629H	4	EI-DEE [5,206,212] SLE [206]	Domain 4, S4 segment	in silico prediction: GOF [121]
p.V261M	4	SLE [5,50] EI-DEE [206] LO-DEE [224]	Domain 1, S5 segment	GOF [50] in silico prediction: GOF [121]

* data adapted from Crawford et al. [42]

Abbreviations: ASD = autism spectrum disorder, EI-DEE = early infantile developmental and epileptic encephalopathy, GOF = gain of function, LO-DEE = later onset developmental and epileptic encephalopathy, LOF = loss of function, NE = non-epilepsy phenotype, SLE = self-limited epilepsy

To date, the most frequently reported recurrent variants are p.R853Q (located in domain 2), p.A263V (in domain 1), and p.R1882Q (C-terminus) (Table 1). Of note, other amino acid substitutions have also been reported at the 1882 position, including p.R1882G/L/P, although these are not recurrent variants [5,51,65]. The p.R853Q variant is associated with later onset DEE characterized by IESS, severe ID, intractable seizures, and choreoathetosis [5,12,43], and less commonly with ASD [42,66]. Individuals heterozygous for p.R1882Q, which causes a strong GOF, had seizure onset on day of life one and at least one died of SUDEP [5,29]. Phenotype is more variable for other amino acid substitutions at this position. For example, p.R1882G is associated with self-limited epilepsy with later-evolving episodic ataxia [50,51,53]. The p.A263V variant is associated with early infantile DEE [5,67]. Two recurrent variants (p.E1211K, p.L1342P) are consistently associated with later onset DEE [66]. The LOF variant p.R937C is most often reported with a phenotype of developmental delay, ID, and/or ASD without seizures [66,68].

Interpretation of SCN2A Variants

Interpretation of single nucleotide or copy number variants in SCN2A should be done in collaboration with a neurologist or geneticist with expertise in genetic variant interpretation. Features to be considered include: (1) frequency in population databases such as gnomAD,¹ (2) inheritance pattern, (3) similarity to previously described variants and/or if there are nearby pathogenic variants, (4) in silico predictions, and (5) consistency with described phenotypes. The SCN Portal,² ClinVar,³ UniProt,⁴ and publications provide valuable resources to determine if a variant is previously described and what clinical phenotypes are reported to determine where it is located relative to functional domains of the protein (see Figure 5B) and whether there is information on functional impact. Diagnostic laboratories will classify the variant as pathogenic, likely pathogenic, or a variant of uncertain significance based on standard guidelines [69,70,71]. Functional impact of novel variants cannot be definitively determined with bioinformatics tools.

(A) Simplified structure of a Na_V channel highlighting major functional domains.

(B) Location by codon number of individual transmembrane segments and domains in Na_V1.2.

Figure 5 Transmembrane topology of a Na_V channel.

The Biology of SCN2A

SCN2A encodes the voltage-gated sodium (Na_V) channel formally named Na_V1.2. The primary amino acid sequence of this protein was originally deduced from rat brain by complementary DNA (cDNA) cloning. Because it was the second sequenced brain Na_V channel sequenced, it was designated as rat brain type II. Functional studies of rat Na_V1.2 in Xenopus oocytes revealed a tetrodotoxin-sensitive, rapidly activating and inactivating inward current that resembled native neuron sodium currents [72,73]. The primary amino acid sequence of human Na_V1.2 was determined in 1992 and its functional and pharmacological properties were similar to the rat channel [74]. The human SCN2A gene resides on the long arm of chromosome 2 (2q24.3) and consists of 27 exons spanning more than 150 kilobases.

Unlike voltage-gated potassium channels, which require assembly of tetramers (4 subunits) to function, Na_V channels are single polypeptides that form a pseudotetrameric arrangement of four similar domains (designated as I–IV or D1–D4) [75]. Each of the four domains consists of six transmembrane spanning segments (S1–S6), with the fourth transmembrane segment (S4) in each domain acting as the primary voltage sensor. Segments S5 and S6, along with the intervening pore (P)-loop of each domain, interact to form a conductive pathway selective for Na⁺ ions (Figure 5). Sodium ion selectivity is conferred by aspartic acid (D), glutamic acid (E), lysine (K), and alanine (A) residues individually contributed by the four domains [76,77]. The III–IV cytoplasmic linker region, along with the C-terminal domain, are involved in channel inactivation [78,79,80]. During the three decades after sequencing the first brain Na_V channels, many research groups worked to elucidate the structural, biophysical, and pharmacological properties of Na_V1.2 in great detail and investigated the underlying contribution of this channel to neuronal physiology in health and disease [81,82,83]. The protein structure of Na_V1.2 at atomic scale was determined using cryogenic electron microscopy [84].

Na_V1.2 Biophysical Properties

Na_V channels mediate the initiation and propagation of APs [85]. Neuronal Na_V channels, including Na_V1.2, have elaborate gating mechanisms that support rapid channel activation, inactivation, and recovery from inactivation, all of which are essential to sustain high-frequency AP firing. The sequence of events occurring during a neuronal AP is well understood (Box 5). The biophysical properties of Na_V1.2 that enable the AP include voltage-dependent and time-dependent (kinetic) properties. Voltage-dependence refers to the voltage range (in millivolts) over which channels activate and inactivate, while kinetic properties describe the time scales (in milliseconds) during which these events occur.

Activation of Na_V channels results in a transient, rapidly inactivating current, but in some situations a small amount of non-inactivating (persistent) current remains (Figure 6). Persistent sodium current is physiologically involved in neuronal pacemaking of CA1 pyramidal neurons driven by muscarinic stimulation, as well as amplification of synaptic signals from both excitatory and inhibitory inputs [86,87,88]. Pathophysiologically, as will be discussed below, mutations that lead to abnormally large levels of persistent current can be associated with epilepsy.

Figure 6 Representative voltage-clamp recording of wild-type (upper trace) and pathogenic variant Na_V channel with enhanced persistent sodium current (arrow).

Box 5

Synaptic signals summing to produce a localized area of membrane depolarization (excitatory postsynaptic potential) of sufficient magnitude (threshold) elicits a sudden biphasic change in the membrane potential: first depolarization, then repolarization. The initial “upstroke” of the AP (depolarization phase) is caused by the opening of Na_V channels. These channels are normally closed at the resting membrane potential but open (activate) upon a sufficiently strong local membrane depolarization. Activation of Na_V channels is voltage-dependent owing to intrinsic voltage-sensors that are part of the protein. Opening of Na_V channels allows Na⁺ to rush into the cell due to a favorable electrochemical gradient (high extracellular Na⁺ concentration and negative cell interior). The rush of Na⁺ into the cell causes the membrane potential to further depolarize toward the Na⁺ equilibrium potential (E_Na), which is approximately +60 mV in mammalian neurons. Almost immediately (1–2 msec) after opening, Na_V channels undergo a conformational change that closes the ion conducting pore by a process called inactivation. Sodium channels remain inactive until the membrane is repolarized. This phenomenon is a major reason for the refractory period – a short time window following an AP during which a second AP cannot be stimulated at all (absolute refractory period) or can only be elicited using a stronger second stimulus (relative refractory period).

Na_V1.2 During Neurodevelopment

Na_V1.2 undergoes developmentally regulated RNA splicing, resulting in two major channel isoforms with slightly different functional properties. This splicing event leads to the incorporation of an alternate exon 5 that encodes a portion of the domain I voltage sensor (S3 and S4 segments) [89,90]. The resulting neonatal and adult splice isoforms differ by a single amino acid residue at position 209 (asparagine in neonatal Na_V1.2; aspartic acid in adult Na_V1.2) and in some biophysical properties [91]. Analysis of gene expression data from rodents and primates demonstrated that during early prenatal and immediate postnatal brain development, Na_V1.2 mRNA transcripts predominantly contain exon 5N (neonatal Na_V1.2) [92,93]. Progressively through the first months of postnatal development, exon 5 splice-switching leads to predominant expression of transcripts containing exon 5 A (adult Na_V1.2). Neonatal Na_V1.2 exhibits a depolarized voltage-dependence of activation, a hyperpolarized voltage-dependence of inactivation, and slower recovery from inactivation compared to the adult isoform. Given that immature neurons have a depolarized resting membrane potential compared with mature neurons, this combination of biophysical features would suggest that neonatal Na_V1.2 may act to limit neuronal excitability during development [91,94]. To corroborate this, a mouse model engineered to express only adult Na_V1.2 throughout development showed greater neuronal excitability and increased seizure susceptibility compared to wild-type (WT) littermates [95]. Thus, variants in Na_V1.2 which show predominant GOF phenotypes may be particularly damaging during this critical period. Indeed, some variants exhibit exacerbated phenotypes in the neonatal isoform compared to that of the adult isoform [91].

Modulation and Regulation of Na_V1.2

Na_V1.2 activity is modulated by a number of factors, including protein–protein interactions, posttranslational modifications (e.g., phosphorylation, palmitoylation), and changes in intracellular Ca²⁺ concentration. Like other Na_V channels, Na_V1.2 interacts with members of a family of nonconducting accessory β subunits that appear to modulate channel properties and promote forward trafficking to the plasma membrane [96,97,98]. Heterologous co-expression of the β1 and β2 subunits with Na_V1.2 results in larger peak current amplitude, shifts in the voltage-dependence of inactivation, and faster channel activation and inactivation. In addition to β subunits, Na_V1.2 also interacts with fibroblast growth homologous factors, which bind to the channel C-terminal domain. Members of this family of proteins have a range of effects on channel function. Co-expression with FGF13-1a (also known as FHF2a) results in larger current amplitude and enhanced frequency-dependent channel rundown, whereas FGF13-1b (also known as FHF2b) causes the channel to be more resistant to frequency-dependent rundown [99]. Additionally, FGF14-1b (also known as FHF4b) attenuates Na_V1.2 current amplitude [100].

Intracellular Ca²⁺ regulates Na_V1.2 by activating Ca²⁺/calmodulin-dependent kinase II (CaMKII), which phosphorylates the channel. In the presence of activated CaMKII, the voltage-dependence of inactivation is significantly depolarized, and persistent current mediated by Na_V1.2 is larger, suggesting that CaMKII-mediated phosphorylation enhances channel function [101]. While calmodulin was shown to interact with the C-terminus of the channel, the functional consequences of this interaction are not well established [101,102].

Na_V1.2 also interacts with members of the ankyrin family, cytoskeletal proteins that promote cellular localization. In neocortical pyramidal neurons, ankyrin-G scaffolds Na_V1.2 to the axon initial segment (AIS) and nodes of Ranvier, whereas ankyrin-B scaffolds Na_V1.2 in dendrites [103,104].

Na_V1.2 Function in Cells and Circuits

Na_V1.2 channels are expressed throughout the central nervous system from the most ancient brainstem regions to the more recently evolved neocortex (Figure 7). There is an emerging understanding of Na_V1.2 function throughout development and how dysfunction of this important channel affects neuronal excitability and integrative properties of cells and circuits. Four Na_V channel subtypes are expressed in the central nervous system (Na_V1.1, Na_V1.2, Na_V1.3, Na_V1.6), which are encoded by four distinct genes (SCN1A, SCN2A, SCN3A, SCN8A, respectively). While Na_V1.6 is present on the plasma membranes of a majority of neuronal classes, Na_V1.1 and Na_V1.2 appear to function in largely nonoverlapping groups [105,106,107,108,109,110]. Furthermore, Na_V1.2 expression within neurons varies by neuron type and at different time points during development. In mature neurons, Na_V1.2 localization is correlated with axonal myelination, with unmyelinated axons containing Na_V1.2 and myelinated axons excluding Na_V1.2 [111]. In the latter population, emerging evidence indicates that Na_V1.2 is present in dendrites [112,113]. As such, Na_V1.2 function can vary markedly across the brain, affecting either input or output structures of neurons.

Figure 7 Localization of Na_V1.2 in diverse neuronal circuits. In mature circuits from mouse and rat models, Na_V1.2 expression within axons or dendrites largely corresponds to the presence of myelin. In cerebellum (left), Na_V1.2 is expressed in unmyelinated granule cell axons. In hippocampus (top), Na_V1.2 is expressed in unmyelinated axons of pyramidal cells. In neocortex (right), Na_V1.2 is expressed in somatodendritic compartments of pyramidal cells whose axons myelinated. In striatum (bottom), current data suggests that Na_V1.2 is expressed in all neuronal compartments.

Na_V1.2 function, and its relationship to other ion channel classes, has been studied most extensively in neocortical structures. SCN2A is transcriptionally expressed in a range of cell classes, including excitatory and inhibitory neurons [114]; however, this expression translates into Na_V1.2 predominantly in excitatory pyramidal neurons [109]. Within pyramidal neurons, the distribution of Na_V1.2 channels on the membrane changes during development. Early on, when neurons are still differentiating and elaborating neurites, Na_V1.2 appears to be the sole Na_V channel expressed in pyramidal neurons [95]. This period spans the first year of life in humans and the first week of life in mice, a model system commonly used to study this channel. During this time period, Na_V1.2 is found at highest density at the AIS, the site of AP initiation [115]. Because AP initiation is supported by Na_V1.2 during this developmental period, genetic variants can affect overall neuronal excitability, resulting in hyper- or hypoexcitability in developing neocortical networks. Indeed, hyperexcitability in these circuits is considered etiological to early infantile epileptic encephalopathy, which occurs within the first few months of life [5,12,41].

After this early developmental period, Na_V channels redistribute. Na_V1.6 expression increases and displaces Na_V1.2 in the distal AIS and at axonal nodes of Ranvier [116]. Na_V1.2 expression also continues to increase [103], but these channels accumulate in neocortical pyramidal neuron dendrites [112]. Here, Na_V1.2 is critical for dendritic excitability and activity-dependent synaptic maturation and plasticity. Under normal conditions, APs initiated in the AIS propagate both forward to sites of neurotransmitter release and backward into the dendrite. The latter “backpropagation” is partly a passive process, limited by the electrical properties of dendritic cables, and partly supported by Na_V channels. As discussed below, heterozygous loss of Scn2a, which is associated with ID/ASD, impairs this backpropagation process, with downstream effects on synaptic maturation and plasticity.

In contrast to neocortical pyramidal neurons, Na_V1.2 is distributed differently in unmyelinated axons where, instead of being excluded, Na_V1.2 is enriched [111]. This distribution pattern is best understood in striatal medium spiny neurons [117], hippocampal pyramidal neurons [118,119], midbrain dopaminergic neurons [110], and cerebellar granule cells [120], but likely extends to other neuron classes with unmyelinated axons. Interestingly, AP threshold does not appear affected by heterozygous loss of Na_V1.2 in either hippocampal pyramidal neurons or cerebellar granule cells [120], indicating that another Na_V channel, presumably Na_V1.6, supports AP initiation in these cells. Instead, Na_V1.2 supports AP propagation, ensuring faithful transmission of those APs from the AIS to sites of neurotransmitter release.

Dysfunction of Na_V1.2

Detailed analysis of Na_V1.2 biophysical properties provides critical insight into how SCN2A variants can result in NDD. This work is typically done in heterologous expression systems, including human embryonic kidney cells (HEK293) that express few to no sodium channels of their own. Channel biophysical properties can then be studied using patch clamp recording. Thus, researchers have been able to quantify numerous biophysical features and assign binary classifications such as GOF or LOF to individual parameters. Data from such experiments in combination with other information such as evolutionary conservation are also useful for training and validating computational (in silico) methods to predict the general functional effects of SCN2A variants [60,121,122].

One illustration of the GOF versus LOF concept is the effect of changes in voltage-dependence of activation on neuronal excitability (Figure 8). As stated earlier, Na_V1.2 channels are activated by membrane depolarization. Any mutation that hyperpolarizes the voltage-dependence of activation would increase the number of APs fired by a given stimulus, resulting in neuronal hyperactivity. By contrast, a depolarizing shift in activation voltage-dependence would result in fewer APs and neuronal hypoactivity. One can predict that mutations that enable channel opening at weaker stimulus intensities, cause channels to remain open longer, or recover faster from inactivation will potentiate neuronal excitability. Conversely, mutations that require stronger stimuli or result in channels that prefer nonconducting states would lower neuronal excitability.

(A) Activation curves illustrating WT channels and variants with either GOF or LOF.

(B) Simulated APs from a layer five cortical pyramidal neuron corresponding to the three conditions. Simulations performed with NEURON as described by Thompson et al. [91].

Figure 8 Effect of altered Na_V1.2 activation voltage-dependence on neuronal excitability.

Characterization of a limited number of variants has suggested that Na_V1.2 variants with strong GOF properties result in early onset epileptic encephalopathies, while LOF variants are associated with later onset epilepsies, ASD, or non-syndromic ID [5,66,68,91]. Some variants fall neatly into these two divergent categories. For example, T236S was identified in a child diagnosed with early infantile epileptic encephalopathy (Ohtahara syndrome) and showed a large hyperpolarizing shift in the voltage-dependence of activation, consistent with GOF [91]. Other biophysical abnormalities interpreted to be GOF include enhanced persistent sodium current and slower onset of inactivation, which result in greater magnitudes of Na⁺ influx over time. By contrast, the ASD-associated variant R937H is an example of a nonconducting Na_V1.2 variant [68]. However, many Na_V1.2 variants associated with neurodevelopmental diseases do not exhibit simple binary effects on channel function. Instead, many mutations exhibit a constellation of GOF and LOF properties, making the net effect on neuronal output difficult to predict [63].

Additional experimental strategies such as computational modeling of APs and dynamic AP clamp can help predict the net effects of Na_V1.2 dysfunction on neuronal excitability (Figure 9). In silico prediction of neuron excitability by simulating neuronal APs computationally has informed the impact of certain variants on neuronal excitability [68,91], and efforts to scale this method for higher throughput are underway [123]. Dynamic AP clamp interfaces voltage-clamp recording of a heterologously expressed Na_V1.2 variant with a computational model of a neuron or of an isolated neuronal domain such as the AIS. This approach has been used to predict effects of Na_V1.2 variants on the propensity of an AIS model to generate APs [44,66]. Beyond these approaches, expressing Na_V1.2 variants in native neurons may be the ultimate solution for determining pathogenic effects. The technology of using induced pluripotent stem cells (iPSC) to generate genetically defined human neurons is emerging as a new approach to assess the impact of Na_V1.2 variants on neuronal firing behavior [124,125]. To date, only a limited number of studies have demonstrated effects of either GOF or LOF variants [126,127,128]. Importantly, cultured iPSC-derived neurons exhibit features comparable to an embryonic state, which may be a limitation for modeling disorders with postnatal onset. Transgenic and knockin mouse models are covered in the next section.

Figure 9 Illustration of dynamic AP clamp.

SCN2A in Epilepsy

Initial reports of an association of SCN2A with human epilepsy came from family studies of benign familial neonatal-infantile seizures, which was renamed SeLFNIE in recognition that seizures are not always benign [13]. In 2002, SCN2A variants were found to segregate with epilepsy in two independent multigenerational pedigrees [10]. The clinical spectrum expanded to include SCN2A-associated DEE with the first reported associations in 2009 [129,130]. SeLFNIE is a transient disorder with seizure onset in infancy that resolves after two years of age and responds well to sodium channel blockers. Inherited SeLFNIE variants are less deleterious than de novo variants associated with DEE. SCN2A-related DEE can be separated by age of seizure onset. Early seizure onset before three months of age is associated with GOF variants that respond to sodium channel blockers, while later seizure onset after three months of age is associated with partial or complete LOF variants that are exacerbated by sodium channel blockers. More details on the genotype–phenotype relationships in SCN2A-related disorders were covered the Clinical Spectrum and Genotype–Phenotype Correlations section.

The first mouse model of SCN2A-related epilepsy preceded the association with human epilepsy and provided supportive evidence for those early reports. Scn2a^Q54 mice express a transgene in neurons with a GOF missense mutation that results in elevated persistent sodium current in vivo and a severe epilepsy phenotype [131]. Phenotype severity in Scn2a^Q54 mice is dependent on background strain, supporting a contribution of genetic modifiers even in the case of a highly penetrant driver mutation [132]. Scn2a^Q54 mice have been a useful system for identifying genetic modifiers that may contribute to variable presentation of epilepsy associated with SCN2A variants [101,133,134,135]. With the advent of genome editing technology, it has become easier to develop knockin models carrying human pathogenic variants in the corresponding position in the mouse gene. New mouse models of SCN2A-related disorders have been reported, including those with recurrent variants for early onset and later onset DEE [136,137]. Scn2a^R1882Q mice, carrying the most recurrent early onset DEE variant, exhibit a severe phenotype, with seizure onset in the neonatal period and premature lethality by postnatal day 30. Pyramidal neurons from Scn2a^R1882Q mice exhibit higher evoked AP firing frequency, higher input resistance and lower rheobase, consistent with a GOF phenotype [136]. In contrast, Scn2a^R853Q mice, carrying the most recurrent later onset DEE variant, had no spontaneous seizures or EEG abnormalities and were less prone to seizures induced by chemoconvulsants [137]. Primary cortical neurons from Scn2a^R853Q mice showed lower firing frequency and less network bursting compared to WT neurons, consistent with a LOF phenotype [137].

Pathogenic variants exhibiting mixtures of LOF and GOF properties make it challenging to translate in vitro biophysical defects into a predicted effect on neuronal excitability. One such mixed function SCN2A variant is K1422E, which exhibits altered ion selectivity allowing partial Ca²⁺ permeation and lower overall conductance [76,77,138]. A child carrying this variant presented with developmental delay, epileptic spasms with onset at 13 months of age, and features of ASD [139]. Consistent with the mixed effects on channel function, Scn2a^K1422E mice exhibit a combination of GOF and LOF phenotypes, including rare spontaneous seizures, background EEG abnormalities, and altered thresholds for induced seizures, as well altered neurobehavior that overlaps features of Scn2a haploinsufficient mice [138]. Recordings from layer 5b pyramidal neurons showed impaired AP initiation and larger intracellular Ca²⁺ transients at the AIS during the rising phase of the AP, consistent with a complex effect of the p.K1422E variant on neurons. The complex effects of mixed-effect variants suggests that therapeutic responses may be challenging to predict for these special cases. Additionally, genome background and biological sex have been shown to influence phenotype severity in Scn2a^K1422E mice [138,140,141], adding to the complexity.

SCN2A in Autism Spectrum Disorder and Intellectual Disability

SCN2A variants were first associated with ASD in 2012 in a landmark genetic analysis of exomes of 928 individuals, including 200 phenotypically discordant sibling pairs [142]. In this cohort, SCN2A was the only gene found to have disruptive de novo variants in two separate ASD cases, which was deemed extremely unlikely due to chance. In the decade following, the association of SCN2A with ASD strengthened, and the gene is now recognized as the leading genetic risk factor for ASD among all genes identified by exome sequencing [143,144]. This large-scale effort was conducted by the International Autism Sequencing Consortium. One of the lead investigators was Matthew State, PhD, Professor and Chair of Psychiatry and Behavioral Sciences, University of California, San Francisco, who was interviewed for this Element by Kevin Bender, PhD (Video 2).

Part 1

Part 2

Video 2 Two-part discussion with Matthew State, MD, PhD (Professor and Chair, Department of Psychiatry and Behavioral Sciences, University of California, San Francisco) moderated by Kevin Bender, PhD (Associate Professor of Neurology, University of California, San Francisco) on discovery of SCN2A as a risk gene for ASD, the neurobiology of this gene, and prospects for future treatments. Transcripts of these videos are available in the Appendix. The video files are available at www.cambridge.org/scn2a

Despite this strong ASD association, unraveling the contributions of SCN2A to ASD has not been straightforward. Of the 72 genes currently associated with ASD, SCN2A was initially considered an outlier, because most genes encode proteins that localized to or regulate synapses or are genes encoding chromatin modifiers in neuronal nuclei [144]. By contrast, SCN2A encodes a Na_V channel involved in AP initiation and propagation in neuronal axons, which was considered distinct from synapses and gene regulatory processes in the nucleus. It was with the discovery that Na_V1.2 localizes to neocortical pyramidal cell dendrites that some level of convergence was first understood. In dendrites, Na_V1.2 interacts functionally with synaptic genes, regulating dendritic excitability and synaptic plasticity. Heterozygous loss of Na_V1.2 in neocortical neurons results in cell-autonomous impairments in dendritic excitability. This dendritic excitability is instructive for activity-dependent refinement of excitatory circuits. In heterozygous Scn2a knockout (Scn2a^+/-) mice, excitatory synapses are less mature, with an excess of silent synapses, lower AMPA:NMDA receptor ratio, and greater proportion of NR2B-containing NMDA receptors more commonly found in developing synapses. Furthermore, impairments in Na_V-supported propagation of APs into the dendritic compartment were associated with a failure to induce synaptic plasticity [103,112]. Evidence has emerged that ankyrin-B, encoded by the ASD-risk gene ANK2, contributes to scaffolding Na_V1.2 channels to dendritic membranes. Heterozygous loss of ankyrin-B in mice phenocopies the synaptic dysfunction discovered in Scn2a^+/- mice, and this implicates dendritic dysfunction as a convergent mechanism in ASD.

Impairments in synaptic plasticity also extend to other brain regions in Scn2a haploinsufficiency. In hippocampus and cerebellum, excitatory synapses onto principal neurons (pyramidal cells and Purkinje cells, respectively) can be strengthened in response to bursts of input at high frequency. In both brain regions, the axons that convey these bursts are unmyelinated and use Na_V1.2 to propagate APs from the AIS to synaptic boutons. In Scn2a^+/- mice, this transmission may be less reliable, simply because there are fewer Na_V channels available to initiate APs [145]. As a result, synaptic plasticity that depends on burst transmission is either blunted or eliminated in both brain regions [119,120].

Synaptic plasticity deficits may be linked to behavior. Although Scn2a^+/- mice perform reasonably well in many behavioral assays [112], they may have difficulties performing more complex spatial memory tasks. Unit recordings from the hippocampus revealed that these spatial memory impairments were correlated with alterations in “replay,” which is a coordinated, sequential activation of a series of hippocampal pyramidal neurons that appears associated with the animal recalling its movement through space [146]. Moreover, recent work examining cerebellar circuitry has revealed that Scn2a^+/- mice and children heterozygous for SCN2A LOF variants exhibit hypersensitivity of the vestibulo-ocular reflex (VOR) [120]. The VOR is a conserved behavior that reflexively moves the eyes in the opposite direction of the head to help stabilize a visual scene on the retina. In mice, the combination of heightened VOR sensitivity and an inability to modulate VOR amplitude is linked specifically to heterozygous loss of Scn2a in cerebellar granule cells, which are neurons that provide excitatory input to Purkinje cells, and was found to impair synaptic plasticity between granule cells and Purkinje cells. This form of plasticity is common throughout the cerebellum [147]. As such, deficits in VOR behavior may serve as a bellwether of broader dysfunction in cerebellar plasticity, which could affect a range of behaviors, including motor coordination and social interaction [148].

Mouse genetics allows for study of not only Scn2a haploinsufficiency, mimicking the condition resulting from heterozygous LOF variants, but also genetic conditions not observed in humans (Figure 10). While constitutive knockout of Scn2a is lethal [149], 75 percent loss of Scn2a throughout the brain, or conditional knockout of Scn2a in select neuron classes, is tolerated [113,150]. These conditions have helped reveal phenotypes not easily observed in heterozygous Scn2a^+/- mice that may nevertheless inform on the human condition. To generate 75 percent loss of Scn2a, a gene knockdown approach termed gene trap was used to block most transcription from both Scn2a alleles in mice [151]. These mice display a range of behavioral deficits not observed in heterozygous Scn2a^+/- mice, including markedly elevated anxiety-related behavior, impaired nesting behavior, and fractured sleep associated with altered activity in the suprachiasmatic nucleus [151,152]. Moreover, ex vivo recordings of principal neurons in striatum and neocortex revealed an unexpected intrinsic hyperexcitability in AP generation, which was due to lower potassium channel function that normally dampens excitability. Hyperexcitability was also observed when Scn2a expression was completely knocked out using a viral approach in adolescent mice (postnatal days 28–44). In this case, hyperexcitability was due not to a change in potassium channel expression but rather to a disrupted interaction between the loss of dendritic Na_V1.2 and dendritic potassium channels, where a loss of Na_V1.2-mediated depolarization led to a loss of potassium channel-mediated repolarization. Because neurons did not hyperpolarize to the same voltages between APs, it was then easier to generate subsequent APs [113]. Such effects appear consistent in vivo, as conditional knockout in the same population of pyramidal cells, and only those pyramidal cells, is sufficient to promote seizures [153].

Figure 10 Range of Na_V1.2 dysfunction in reported mouse models of SCN2A-related disorders. Mouse models with genotypes and phenotypes associated with SCN2A-related disorders span and expand on the range of genotypes observed in the human population. GOF missense variants include those observed in children (black) and those that result in epilepsy in mice due to knockin of amino acids that would be unlikely to occur in humans (Q54). Mixed function variants, with features of both GOF and LOF, have been generated for the K1422E variant, which converts Na_V1.2 from a Na⁺ ion selective channel to a nonselective cation channel. LOF variants include variants that truncate the Na_V1.2 protein. Beyond these cases that can occur in children, mouse genetics has allowed for other approaches, including ~75 percent to 100 percent reduction of Scn2a expression, either throughout the brain or in select cells, as well as rescue alleles where full Scn2a expression can be restored in neurons using Cre recombinase.

Overall, these changes in excitability, observed when model systems are pushed beyond heterozygosity, may shed light on why some children with SCN2A protein truncating variants develop epilepsy. They suggest that Scn2a^+/- mice, on the C57BL/6 background, are near a threshold for excess excitability that can be observed if the system is pushed. Of note, different strains of inbred mice display different propensity for seizures and can be either protective or more susceptible to seizure electrogenesis if converted into an epilepsy model (e.g., by providing a pro-convulsant drug or making a genetic manipulation to a key ion channel). This variability is also present in the human population and likely contributes to variability in presentation of epilepsy in those with SCN2A LOF variants. Moving forward, a deeper understanding of factors that push or pull systems from a seizure susceptibility cliff will be critical, in both mouse models and other model species.

Treatment of SCN2A-Related Disorders

Treatment of SCN2A-related disorders is primarily aimed at seizure control along with supportive therapies for symptoms related to ASD and developmental delay. When possible, treatment should to be tailored to an individual’s pathogenic variant. Certain strategies such as the use of sodium channel blocking antiseizure medications may be more effective in persons with GOF variants and less effective in cases with LOF variants. While symptomatic treatments will continue to be important, disease-modifying therapeutic strategies targeting the underlying pathophysiology are emerging. This section will initially focus on the symptomatic treatment of epilepsy followed by a discussion of emerging disease-modifying therapies for which preclinical proof-of-concept exists for SCN2A-related disorders.

Antiseizure Therapy for SCN2A-Related Disorders

Epilepsy due to GOF variants typically presents as neonatal or early infantile onset seizures. Clinical evidence supports the notion that sodium channel blockers are more often effective in this setting, but are less effective or ineffective if epilepsy begins after three months of age [5]. The difference in epilepsy age of onset between groups with GOF versus LOF variants is a general trend but is not absolute, and some variants are difficult to categorize because they exhibit features of both GOF and LOF [63].

The efficacy of specific sodium channel blockers in the setting SCN2A GOF variant is heterogeneous. In a limited number of case series, phenytoin and carbamazepine had the best response rates [5,154,155]. High dose phenytoin may be required for efficacy [154,156], but this is accompanied by greater risk of adverse effects and requires close monitoring of plasma free drug concentration. Awareness of potential drug–drug interactions and pharmacogenomic factors is also important [157]. Genetic variants in CYP2C9 affecting phenytoin metabolism may predispose to toxicity [158]. While use of sodium channel blockers can produce significant seizure reduction in individuals with GOF variants, anecdotal evidence suggests developmental delays may persist [156,159].

Individuals with SCN2A LOF variants may also have seizures, which tend to begin later in infancy or in early childhood and do not generally respond to sodium channel blockers [5]. There is little evidence supporting specific antiseizure therapies for affected individuals with SCN2A LOF or mixed function variants. Treatment for SCN2A-related NDD is primarily supportive, with applied behavioral analysis along with physical, occupational, and speech therapies. Recognition of cortical visual impairment allows accommodations to be integrated into an individual’s care plan in a way that maximizes successful visual interaction with their environment.

Emerging Disease-Modifying Therapies for GOF SCN2A Variants

While no disease-modifying therapy is currently available for SCN2A-related disorders, many potential avenues are being investigated (Table 2), with some moving into clinical trials. The remainder of this section will focus on emerging therapeutic approaches that may have efficacy for these conditions (Figure 11). These advances are made possible by collaborations between academic laboratories and the pharmaceutical industry, as illustrated by an interview with Dr. Steven Petrou, Professor of Translational Neuroscience, University of Melbourne, and Chief Scientific Officer of Praxis Precision Medicines (Video 3).

Table 2 Potential disease-modifying strategies for SCN2A-related disorders

Type of therapy	SCN2A GOF	SCN2A LOF
Small molecule drugs	persistent current blocker*	sodium channel activator#
Antisense oligonucleotides	gapmer/RNAseH*	poison exon (TANGO)#
Transcriptional modulation	CRISPR interference RNA interference	CRISPR activation*
Gene delivery	N/A	activating transcription factor# SCN2A transgene
Nonsense suppression	N/A	tRNA suppressor (targeting premature stop codon)
Gene editing	base editing, prime editing	base editing, prime editing

* existing preclinical proof-of-concept for SCN2A;

# preclinical proof-of-concept for another sodium channel gene;

N/A, not applicable

Figure 11 Overview of potential disease-modifying therapies in SCN2A-related disorders. The left side of the figure illustrates general approaches. The triangle on the right side illustrates the ease of delivery (most difficult delivery is the top). The inverted triangle on the right side illustrates the time therapy remains in the body.

Created with BioRender.com.

Video 3 Interview of Steven Petrou, PhD (Chief Scientific Officer, Praxis Precision Medicines) conducted by Alfred L. George, Jr., MD (Professor and Chair of Pharmacology, Northwestern University) explaining his transition from academic research to leading a new pharmaceutical company developing treatments for SCN2A-related disorders. A transcript of this video is available in the Appendix. The video file is available at www.cambridge.org/scn2a

Small Molecule Drugs

Conventional drug therapy, which modulates the activity of mutant sodium channels, is the mainstay of treatment for SCN2A GOF variants. However, sodium channel blocking antiseizure medications such as phenytoin do not correct the underlying functional defect and also affect normal sodium channels. New small molecule drugs in development are attempting to target channel dysfunction more precisely with the goal of being more effective with less toxicity.

One strategy targets a specific functional anomaly, enhanced persistent sodium current, exhibited by many SCN2A GOF variants. Persistent sodium current arises when channel inactivation is not complete, leading to a small level of inward current that can affect neuronal excitability. This functional anomaly occurs in many sodium channelopathies in addition to SCN2A-related disorders. The first animal model (Q54 mouse) to test the hypothesis that enhanced SCN2A persistent current can cause epilepsy was reported years before the first human mutations were discovered [131]. Later investigations using this mouse model demonstrated the antiseizure efficacy of an approved antianginal drug, ranolazine [160], which selectively suppresses persistent current carried by cardiac sodium channels [161]. This study was motivated by in vitro evidence that ranolazine blocked persistent sodium current carried by mutant Na_V1.1 channels (encoded by SCN1A) [162,163]. Acute exposure of Q54 mice to ranolazine was effective at reducing seizure frequency, but poor brain penetration indicated that this drug would not be effective clinically [161]. However, a novel nonselective sodium channel blocker (GS-458967 or GS967; later named PRAX-330) was described as a potent and preferential inhibitor of persistent sodium current [164]. GS967 exhibits near complete seizure suppression in Q54 mice and prevented seizures following maximum electroshock [160]. GS967 was shown to suppress persistent current carried by recombinant human Na_V1.2 [165] and to prevent seizures in a mouse model of DEE caused by an SCN8A GOF variant [166,167].

Despite success in mouse models, GS967 had undesirable pharmacokinetic properties (e.g., excessive plasma and brain half-life) [164] and exhibits potent use-dependent block of peak sodium current [168]. However, the observed efficacy of preferential persistent sodium current block in mouse models of epilepsy motivated development of newer agents, including PRAX-562 [169] and NBI-921352 [170], although neither was selective for Na_V1.2. PRAX-562 is orally active and exhibits greater preference for blocking persistent current carried by mutant or neurotoxin (ATX-II)-modified Na_V1.6 channels than conventional sodium channel blocking antiseizure drugs (e.g., phenytoin, carbamazepine, lamotrigine) [169]. This compound has similar effects on other human sodium channels, including Na_V1.2. By contrast, NBI-921352 is selective for blocking Na_V1.6 channels and doesn’t specifically target persistent current [170]. In mice subjected to maximum electroshock, PRAX-562 exhibits dose-dependent anticonvulsant effects with less neurotoxicity than carbamazepine and lamotrigine [169]. Following demonstration of tolerability in healthy volunteers [171], PRAX-562 received orphan drug and rare pediatric disease designations by the FDA and entered Phase 2 clinical trials for SCN2A- and SCN8A-related DEE (NCT05818553, ClinicalTrials.gov).

While more targeted drugs show promise for controlling epilepsy, there remains the question of whether this type of therapy will be disease-modifying because sodium channel blockade has not been shown to rescue the neurodevelopmental phenotype. In theory, this could be related to the degree of selectivity for the mutant sodium channel and whether the small molecule corrects the specific pathophysiology of a particular pathogenic variant. It is conceivable that more precise sodium channel inhibition initiated earlier in the disease course may potentially improve seizure control and developmental outcomes.

Antisense Oligonucleotide (ASO)

Among the first disease-modifying therapies approved for NDD was an ASO (nusinersen) for spinal muscular atrophy [172,173]. ASOs are short stretches of synthetic nucleic acids typically comprised of deoxyribonucleic acid (DNA) or chemically modified ribonucleic acid (RNA) nucleotides. These nucleic acid drugs bind to a complementary sequence (antisense) within mRNA or pre-mRNA and act either by steric interference with enzymatic processing (e.g., splicing, translation) or by engaging ribonuclease H (RNaseH) to cleave hybrid RNA:DNA duplexes. Chemically modified RNA nucleotides (e.g., ribose-O-2-methoxyethyl or 2’-MOE) coupled with sulfur-containing (e.g., phosphorothioate) backbone linkages are employed to shield the molecule from degrading enzymes [174]. ASOs used to target mRNA or pre-mRNA for RNaseH cleavage typically use a ‘gapmer’ design consisting of a DNA core (needed to form nuclease-susceptible RNA:DNA duplexes with the target) flanked by short chemically modified RNA sequences (Figure 12). Splice-modifying ASOs can be composed of DNA or modified RNA only. These two mechanisms of action make ASOs versatile in their application (Figure 13), leading to several clinical trials for a variety of neurogenetic conditions [175,176,177,178].

(A) Gapmer design with a 10 nucleotide DNA core and flanking five nucleotide modified RNA sequences.

(B) Chemical structure of a 2’-MOE RNA base with phosphorothioate linkage.

Figure 12 Design features and chemistry of a gapmer ASO.

(A) RNase H–mediated mRNA degradation.

(B) Splice-switching ASOs that promote exon inclusion by masking silencer regions. (C) Splice-switching ASOs that promote exon skipping by masking splicing enhancer regions.

Adapted from Carvill et al. [176] and reproduced with permission from Springer Nature. Created with BioRender.com.

Figure 13 Strategies for gene regulation by ASOs.

Preclinical evidence suggests that ASOs have potential therapeutic benefits for SCN2A-related disorders associated with GOF variants. Li et al. demonstrated that immediate postnatal delivery of a gapmer ASO targeting SCN2A by the intracerebroventricular route lowered spontaneous seizure frequency and prolonged survival in mice heterozygous for a GOF missense variant (equivalent to human R1882Q) [136]. Administration of the ASO two weeks after birth also conferred a survival benefit, as did a second dose given approximately four weeks after birth in mice that received a first dose on day one of life. The ASO used in this study did not affect protein levels of other neuronal sodium channels, and treated mice showed no overt neurological toxicity. In 2023, a human SCN2A-selective ASO (PRAX-222, elsunersen) entered early stage clinical trials (NCT05737784, ClinicalTrials.gov).

While encouraging, these results need to be taken with tempered expectations as there are important considerations with ASO treatment in this context. A potential concern is over-reduction of Na_V1.2 protein levels, which may induce a LOF phenotype, a concept referred to as the goldilocks principle [179]. ASOs do not allow for quick or easy dose titration given their long half-lives (more than 100 days in brain), and this creates challenges for achieving a specific range of mRNA and protein reduction. Uptake in brain is not uniform, with superficial structures and spinal cord having higher uptake than deep structures [180]. Another consideration is whether a simple reduction in Na_V1.2 protein will result in the appropriate level of functioning protein. The ASO will down-regulate both SCN2A mRNA alleles. It is important to recognize that while we refer to variants that result in increased currents as “gain-of-function,” this is dysfunctional sodium current, and therefore reduction of both protein alleles doesn’t correct the defect specifically. The extent to which lowering the amount of Na_V1.2 channels will produce a meaningful disease improvement in humans remains an open question.

Other Potential Therapeutic Strategies for SCN2A GOF

There are other emerging approaches for lowering expression of a target gene, including RNA interference [181] and programmable transcription interference using clustered regularly interspaced short palindromic repeat (CRISPR) interference [182] that could have value for SCN2A GOF variants [183]. These approaches could work in a similar way as ASOs to diminish Na_V1.2 channel production. In addition to suppressing expression, base editing [184] and prime editing [185] are promising technologies for correcting pathogenic variants, which have been successful for other genes and may work in mammalian brain [186].

Emerging Disease-Modifying Therapies for LOF SCN2A Variants

CRISPR-Cas9 Activation (CRISPRa)

As mentioned earlier, CRISPR technology can enable targeted transcriptional regulation of genes. In the original embodiment, a CRISPR-associated nuclease (e.g., Cas9) cleaves genomic DNA at a position targeted by a single guide RNA (sgRNA), creating double-stranded breaks that are repaired by either nonhomologous end-joining or homology-dependent repair. An enzymatically inactive Cas9 (“dead” Cas9, dCas9) or a chimeric dCas9 fused to a transcriptional activator can similarly be positioned at a precise location in the genome, but this will not cut DNA. Rather, placing the chimeric dCas9 near the transcription start site will activate transcription for that gene. This strategy, called CRISPR activation (CRISPRa) [182], has been demonstrated to boost SCN2A expression in heterozygous knockout mice and human stem cell-derived neurons [187]. In that study, CRISPRa components were delivered in vivo using recombinant adeno-associated virus (AAV) intravenously or by stereotaxic injection directly into mouse brain. CRISPRa restored normal neuronal and dendritic excitability in heterozygous mice and did not induce seizures in WT mice. The latter observation suggests there are regulatory mechanisms to prevent SCN2A overexpression, which could convert a LOF disorder into a GOF disorder. Similar success has been demonstrated for SCN1A LOF in a mouse model of Dravet syndrome [188].

An alternative approach for transcriptional activation of brain sodium channel genes involves use of engineered transcription factors. This approach has been used successfully for boosting neuronal expression of SCN1A and improving survival of Dravet syndrome mice [189]. Preparations for an early phase clinical trial with the molecule used in this study (ETX101) commenced in 2022 (NCT05419492, ClinicalTrials.gov).

Other Potential Therapeutic Strategies for SCN2A LOF

Preclinical proof-of-concept exists for additional approaches for rescuing LOF variants in other sodium channel genes. Targeted augmentation of nuclear gene output (TANGO) is a strategy for suppressing unproductive mRNA splicing events arising from inclusion of “poison exons,” which evoke nonsense-mediated decay. Unproductive splicing events involve several genes associated with NDD [190], including SCN1A, which encodes the Na_V1.1 sodium channel and is the major gene responsible for Dravet syndrome. A splice-switching ASO (STK-001) prevents inclusion of a poison exon in SCN1A transcripts, boosts translation of functional Na_V1.1 protein, and improves survival in a Dravet syndrome mouse model [191,192]. Clinical trials are underway to test efficacy and safety of STK-001 in Dravet syndrome (NCT04442295, NCT04740476, ClinicalTrials.gov). Whether a similar strategy will work for SCN2A LOF variants is unknown.

Hypothetically, SCN2A-related disorders caused by LOF variants could be approached by delivering an exogenous copy of the recombinant gene or its coding region to brain using a viral vector. While conceptually appealing, this approach is not feasible with the currently available vectors that are most advanced in clinical trials (e.g., AAV), because the large size of the SCN2A coding region exceeds the capacity of these vectors. Newer vectors may overcome this size restriction as demonstrated for SCN1A [193,194,195].

For SCN2A LOF variants that create premature termination codons (PTC, nonsense), a novel approach (nonsense suppression) utilizing modified transfer RNA (tRNA) may have value. This approach seeks to trick ribosomes into incorporating the correct amino acid at the location of the PTC by using anticodon engineered tRNA molecules [196,197]. Intrinsic features of the translational machinery appear to protect native stop codons from the modified tRNA. Viral delivery of a suppressor tRNA was used successfully in vivo in mice to partially correct an engineered PTC mutation in the gene encoding α-l-iduronidase, which causes the lysosomal storage disease mucopolysaccharidosis type I in humans [198]. Nonsense suppression has garnered considerable commercial interest [199] and there is interest in applying this strategy for SCN1A and SCN2A [195,200], but major challenges must be overcome [197].

Considerations for Disease-Modifying Clinical Trial Design

Bringing a disease-modifying therapy to the clinic requires not only evidence of efficacy but also a need to consider aspects of clinical trial readiness. These include precise clinical trial design (duration, most appropriate control group), accurate patient selection (age, variant type), and outcome measures that can demonstrate meaningful changes over time. Clinical trials for conventional small molecule drugs with a primary seizure endpoint are based on established designs, such as randomized, placebo-controlled trials, but these are less feasible and appropriate for disease-modifying therapies [201]. Furthermore, for SCN2A-related disorders disease-modifying therapies may target specific subsets of affected individuals, which further subdivides an already rare disease and may compromise study power.

Variations from traditional clinical trial designs are required to achieve well-powered studies. One nontraditional trial design, stepped-wedge design (also known as a randomized start design) [201], has specific advantages such as allowing comparison both within groups and between groups, thereby increasing the power of the study with a relatively small number of trial participants. Using this approach, all groups eventually receive treatment, which is ethically superior to trials with a placebo group. The effect of treatment duration can also be investigated because each group is introduced to the treatment at different time points. So-called “n of 1” trial designs, including individual crossover studies, may also have value in rare diseases [202]. Depending on the chosen outcome measures, study duration may need to be adjusted to account for developmental outcomes that require longer study times. Careful consideration of these features is needed for clinical trial designs in SCN2A-related disorders to ensure collection of the rich and diverse data necessary for successful assessment of a novel therapy.

Another challenge to clinical trial design in SCN2A-related disorders is the heterogeneous functional effects of variants. As described previously, specific strategies can be tailored to either GOF or LOF variants. While neonatal and early infantile-onset epilepsy is typically associated with GOF variants, this is not absolute. For the aforementioned clinical trial of an ASO for SCN2A GOF variants (PRAX-222), study participants must have seizure onset before three months of age (NCT05737784, ClinicalTrials.gov). However, this could potentially include some affected individuals with LOF variants unknown at the time of enrollment. For future disease-modifying therapies, especially with approaches with permanent effects such as gene delivery and gene editing, functional testing of variants prior to treatment may need to be considered [122]. Furthermore, individuals in this and other SCN2A-related clinical trials are enrolled beginning at age two years, which is well beyond disease onset and raises the possibility that response to therapy could be affected by prior seizure burden or pharmacological treatments.

Selection of appropriate clinical trial outcome measures is challenging for SCN2A-related disorders using existing methods. This is illustrated by a recent SCN2A clinical trial readiness study in which mean standardized domain scores on the Vineland Adaptive Behavior Scales were three standard deviations below the normative average [203]. The Vineland has challenges assessing small changes at low scores, and therefore raw scores or growth scale values may be more effective in eliciting meaningful change without a floor effect. Seizure reduction, even greater than a 50 percent reduction, as the main outcome measure may not encapsulate the full array of disease-modifying effects families and clinicians are hoping to achieve with these therapies, which often involve more risk than traditional seizure medications. With invasive and potentially permanent types of treatment (Table 2), the benefit of such therapies needs to outweigh those risks. The hope is that the benefit from a disease-modifying therapy would entail more than reducing seizure frequency. Given that seizure control in SCN2A-related disorders does not necessarily rescue the developmental phenotype, it will be especially important to target other outcome measures, including developmental achievement, sleep, quality of life, and behavior, among others, to ensure that the full spectrum of disease-modifying effects is being assessed.