Biallelic

Objective Rolandic epilepsy (RE) is among the most common focal epilepsies in childhood. For the majority of patients with RE and atypical RE (ARE), the etiology remains elusive. We thus screened patients with RE/ARE in order to detect disease-causing variants.. Methods A trios-based whole-exome sequencing approach was performed in a cohort of 28 patients with RE/ARE. Clinical data and EEGs were reviewed. Variants were validated by Sanger sequencing. Results Two compound heterozygous missense variants p.Val272Ile/p.Asn3028Ser and p.Ala3657Val/p.Met4419Val of ADGRV1 were identified in two unrelated familial cases of RE/ARE. All the variants were in the calcium exchanger β domain and were suggested to be damaging by at least one web-based prediction tool. These variants are not present or are present at a very low minor allele frequency in the gnomAD database. Previously, biallelic ADGRV1 variants (p.Gly2756Arg and p.Glu4410Lys) have been observed in RE, consistent with the observation in this study and supporting the association between ADGRV1 variants and RE. Additionally, a de novo mutation, p.Asp668Asn, in GRIN2B was identified in a sporadic case of ARE, and a missense variant, p.Asn1551Ser, in RyR2 was identified in a family with RE with incomplete penetrance. These genes are all calcium homeostasis associated genes, suggesting the potential effect of calcium homeostasis in RE/ARE. Conclusions The results from the present study suggest that the genes ADGRV1, GRIN2B, and RyR2 are associated with RE/ ARE. These data link defects in neuronal intracellular calcium homeostasis to RE/ARE pathogenesis implicating that these defects plays an important role in the development of these conditions.


Introduction
Rolandic epilepsy (RE), or epilepsy with centrotemporal spikes (CTS), is one of the most frequent epilepsy in children under the age of 16 years [1]. Typical RE usually takes a benign and self-limiting course with seizures remitting spontaneously during adolescence, but it is also related to rarer and less benign epilepsy syndromes, comprising a Zhigang Liu and Xingguang Ye contributed equally to this work. phenotypic spectrum from benign epilepsy with centrotemporal spikes (BECTS) to atypical benign partial epilepsy (ABPE), Landau-Kleffner syndrome (LKS) and epileptic encephalopathy with continuous spike-and-waves during sleep (CSWS), referred to as RE-related syndromes or atypical Rolandic epilepsy (ARE) [1,2]. One of the most common characteristic features of the above diseases is their remarkable CTS discharges in electroencephalographic (EEG) recordings. In addition, various comorbidities such as preceding febrile seizures (FS), attention deficit hyperactivity disorder, language disorders, learning, and reading disabilities are also shared features between ARE and RE [3][4][5]. Given these overlapping clinical characteristics, these diseases are presumed to have a shared genetic etiology or a similar mechanism of pathogenicity [6]. Previously, mutations in KCNQ2, KCNQ3, RBFOX1, RBFOX3, DEPDC5, GABRG2, GRIN2A, PGM3, and recurrent 16p11.2 microduplication have been proven to cause RE/ARE [6][7][8][9][10][11]. Among these known RE/ARE genes, mutations in GRIN2A, encoding the N-methyl-D-aspartate receptor (NMDAR) GluN2A subunit, are the most frequently observed, accounting for approximately 20% of cases depending on the phenotype [11,12]. However, the underlying genetics causes of RE/ARE remain largely unknown.
In this study, we performed a trios-based wholeexome sequencing approach in a cohort of patients with RE/ARE without acquired causes. Two compound heterozygous missense variants in ADGRV1 (OMIM*602,851), one de novo heterozygous missense mutation in GRIN2B (OMIM*138,252), and one heterozygous missense variant in RyR2 (OMIM*180,902) were identified in four unrelated patients with RE/ ARE, suggesting their involvement as potential candidate causative genes for RE/ARE. We systematically analyzed the ADGRV1 variants and investigated their potential pathogenetic molecular mechanism.

Patients
A cohort of 28 patients of RE/ARE without acquired causes was recruited. The patients were from the Department of Pediatrics, Affiliated Foshan Maternity and Child Healthcare Hospital, Southern Medical University from July 2019 to November 2020. All individuals enrolled were unrelated ethnic Han Chinese who lived in southern China. None of the biological grandparents of the participants were from other ethnicities.
Clinical information of the affected individuals was collected, including detailed information on current age, the age of onset, sex, seizure type and frequency, general and neurological examination results, detailed family history, and response to antiepileptic drugs (AEDs). Magnetic resonance imaging (MRI) scans were performed to detect any brain structure abnormalities. Long-term video-electroencephalography (EEG) monitoring records that included hyperventilation, intermittent photic stimulation, open-close eye tests, and sleeping recordings were obtained. Epileptic seizures and epilepsies were diagnosed according to the criteria of the Commission on Classification and Terminology of the International League Against Epilepsy (1981, 1989, 2001, 2010, and 2017). RE was diagnosed based on nocturnal focal and secondarily generalized seizures, age of onset ranges from years 3 to 13, centrotemporal spikes of EEG, and sleep increase the rate of discharges. ARE was diagnosed based on a range of more severe forms of earlyonset focal childhood epilepsies or epileptic encephalopathies with various additional seizure types, neurocognitive regression, autistic features, or a regression of language with speech dyspraxia, sharing the EEG characteristic of CTS. Familial cases were defined by the existence of at least two members carrying the same mutation. Families with only one affected individual (due to incomplete penetrance) were indicated. An affected family was considered a single case in the data analysis.
This study adhered to the guidelines of the International Committee of Medical Journal Editors regarding patient consent for research or participation, and all the individuals or legal guardians undergoing testing consented to their data being used for research. This study was approved by the ethics committee of Affiliated Foshan Maternity & Child Healthcare Hospital, Southern Medical University (Approve number: FSFY-MEC-2018-016).

Whole-exome sequencing
Whole blood samples were collected from the available subjects and used for linkage and segregation analysis. A Qiagen Flexi Gene DNA kit (Qiagen, Hilden, Germany) was used to isolate genomic DNA for exome enrichment. Whole-exome sequencing was performed with NextSeq500 sequencing instruments (Illumina, San Diego, CA, USA). Sequence alignment and variant calling were performed according to standard procedures as previously described [13]. All candidate pathogenic variants were validated by Sanger sequencing. The paternity and maternity of the probands were confirmed by alignment of the segregated sequence variants. We screened the potential disease-causing mutations under five models: (1) an epilepsy-associated gene model; (2) a de novo mutation dominant model; (3) an autosomal recessive inheritance model, including homozygous and compound heterozygous variants; (4) an X-linked model; (5) a cosegregation analysis model. Classification of pathogenicity

Identification of ADGRV1, GRIN2B, and RyR2 variants
As shown in Table 1  RyR2 were located at residues highly conserved in various species. Residue Met4419 of ADGRV1 was conserved in macrovertebrates but less conserved in lower animals, variant Asn3028Ser of ADGRV1 was located at less conserved, but was predicted to be conserved by in silico tools  Uncertain significance girl (patient 1) with BECTS and FS, and variants Ala-3657Val and Met4419Val were from a boy (patient 2) with ABPE and FS. Among the 4 variants, Asn3028Ser and Ala3657Val were inherited from their parents with FS, and Val272Ile and Met4419Val were inherited from their parents without a history of seizures (Fig. 1a, b). Protein sequence alignment of the ADGRV1 missense variants showed that Val272Ile and Ala3657Val were located at residues highly conserved in various species, residue Met4419 was conserved in macrovertebrates but less conserved in lower animals, and variant Asn3028Ser was located at a less conserved residue but was predicted to be conserved by in silico tools (Fig. 1c, Table 2). Variants Val272Ile and Asn3028Ser were presented in the ExAC, 1000 Genomes, and GnomAD databases with a minor allele frequency of < 0.005. However, variants Ala3657Val and Met4419Val were not present in the ExAC, 1000 Genomes Project, or GnomAD databases. All 4 variants were located in Ca 2+ exchanger beta (Calx-β) motifs and were predicted to be damaging by at least one of the commonly used in silico prediction tools (Table 2). Based on ACMG criteria, all 4 variants were of uncertain significant. No patients had any other pathogenic or likely pathogenic variants in genes known to be associated with seizure disorders. A de novo missense mutation (c.2002G > A/p. Asp668Asn) of GRIN2B was identified in a sporadic case (patient 3) of CSWS and FS (Fig. 1a, b). Previously, a missense mutation, p.Asp668Tyr, in GRIN2B has been reported in a case of epileptic encephalopathy (EE) [14]. This mutation affected an amino acid residue that is highly conserved in various species (Fig. 1c). It was not present either in the ExAC, 1000 Genomes Project, or GnomAD databases, and was suggested to be damaging by web-based prediction tools (SIFT, PolyPhen-2, and MutationTaster) ( Table 2). It was located in the extracellular protein domain and was evaluated as likely pathogenic by ACMG assessment.
One heterozygous missense variant (c.4652A > G/p. Asn1551Ser) in RyR2 was identified in a case (patient 4) of ABPE and FS (Fig. 1a, b). This variant originated from her father and affected amino acid residues that are highly conserved in various species. This variant was present in the ExAC, 1000 Genomes, and GnomAD databases with a minor allele frequency of < 0.005 and was suggested to be possibly damaging or deleterious by PolyPhen-2 and MutationTaster (Table 2). This variant was in the N-terminal domain and was evaluated as variant of uncertain significance by the standards and guidelines of the ACMG. In the remaining genes, no definite pathogenic mutations were identified.

Clinical presentation of the patients with GRIN2B, ADGRV1, and RyR2 variants
In this cohort, four ADGRV1 variants, one GRIN2B mutation, and one RyR2 variant were identified in four unrelated families featuring focal epilepsies. All four probands had focally originated seizures and Rolandic spike-and-wave discharges on EEG recordings. Patient 3 presented a severe phenotype with partial response to antiepileptic therapy, and patient 1, patient 2, and patient 4 presented a mild phenotype with good response to antiepileptic therapy. The clinical information of the four patients is summarized in Table 1, and their representative EEGs are shown in Fig. 2a to d. Patient 1, with compound heterozygous variants Val-272Ile and Asn3028Ser in ADGRV1 (Fig. 1, Family 1: II-2) was a 6-year-old girl. Variant Val272Ile was inherited from her father, who had no history of seizures, and variant Asn3028Ser was inherited from her mother, who had a history of FS. She had first FS and aFS at 4 years of age. Her seizures were characterized by nocturnal focal and secondarily generalized seizures at the transition of sleep. Interictal EEGs revealed right-sided CTS with sleep activation c Interictal EEG of patient 3 showed bilateral centrotemporal spikes and slow waves (at the age of 5 years). d Interictal EEG of patient 4 showed bilateral centrotemporal spike and slow waves (at the age of 6 years) (Fig. 2a). MRI was normal. Seizures were controlled by valproate combined with lamotrigine for 1 year. Her younger brother with variant Asn3028Ser (Fig. 1, Family 1: II-1) had 3-5 FS before he was 3 years old. Patient 2, with compound heterozygous variants Ala-3657Val and Met4419Val in ADGRV1 (Fig. 1, Family 2: II-1), was a 7-year-old boy. Variant Ala3657Val was inherited from his father, with a history of FS, and variant p.Met4419Val was inherited from his mother, who had no history of seizures. He had 5 times FS before he was 4 years old. He had his first afebrile seizures at 7 years of age. He presented with seizures that started with blank staring, automatism, and then unilateral limb seizures, lasting 10-30 s. Interictal EEGs recorded prolonged discharges of CTS during sleep (Fig. 2b). The seizures were controlled by valproate combined with lamotrigine for 6 months. His younger sister with variant Ala3657Val (Fig. 1, Family 2: II-2) had 3 FS before she was 3 years old.
Patient 3 with the de novo mutation Asp668Asn in GRIN2B (Fig. 1, Family 3: II-1) was a 5-year-old boy with global development delay and intellectual disability. He had a single FS at 2 years of age. The FS occurred 2 times in the following 4 months and then he began to have 1-to 2-min unilateral clonic or tonic-clonic seizures or perioral sensorimotor seizures or secondary generalized tonic-clonic seizures (GTCSs), mostly during sleep. EEGs showed continuous spikes and waves during sleep or bilateral CTS (Fig. 2c). Neuroimaging was unremarkable. His seizures were improved by combined therapies (valproate, levetiracetam, and lamotrigine). His epilepsy proved to be therapy resistant. and seizure occurred 5-7 times per year. Patient 4 carrying the heterozygous missense variant Asn1551Ser of RyR2 was an 8-year-old girl (Fig. 1, Family 4: II-1). Her unaffected father also carried this mutation, consistent with incomplete penetrance. The paternal uncle had a history of FS; however, the specific diagnosis and treatment were unknown, and we were unable to obtain his peripheral blood samples for sequencing. The proband also experienced simple FS at the age of 2 years, occurring 3 times/3 years. At the age of 7 years, she began to have focal motor seizures with sensorimotor symptoms, involving the face and laryngeal muscle during sleep, although she also experienced astatic seizures or atypical absences while awake. Interictal EEGs recorded prolonged discharges of CTS, dominated sleep (Fig. 2d). The seizures were controlled by combined therapies (valproate and lamotrigine) for 11 months.

Discussion
In this study, two compound heterozygous missense mutations in ADGRV1 were identified in two unrelated families with RE/ARE, including four individuals with FS that co-segregated with the variant p.Asn3028Ser in family 1 and the variant p.Ala3657Val in family 2 (Fig. 1a). The two probands also experienced 3-5 times FS before they were 4 years old. EEGs recorded CTS with sleep activation. They were seizure-free with valproate combined with lamotrigine. The others with simple FS were seizure-free without medication. Previously, atypical childhood epilepsy with CTS was also observed in a patient with biallelic ADGRV1 mutations (p.Gly2756Arg and p.Glu4410Lys) [15], consistent with the observation in this study. Although different missense mutations have been predicted to be benign or pathogenic, the segregation of mutation (p.Asn3028Ser and p.Ala3657Val) in our families supports its pathogenicity in an autosomal dominant manner. Taken together, these data provide consistent evidence that ADGRV1 is a candidate pathogenic gene of RE/ARE. However, the exact molecular mechanisms are not understood.
ADGRV1 (previously known as GPR98, MASS1, and VLGR1) resides on chromosome 5q13 and encodes G protein-coupled receptor-1, which is highly expressed in the developing central nervous system (CNS) [16], providing an anatomic basis for the pathogenesis of epilepsy. In ADGRV1-deficient mice, homozygous truncating mutations caused audiogenic seizures, supporting the role of ADGRV1 in epilepsy [17]. In humans, homozygous or compound heterozygous ADGRV1 mutations cause severe Usher syndrome IIC, characterized by moderate to severe hearing loss, retinitis pigmentosa and normal vestibular function [18]. Of the reported mutations, most result in frameshift/truncation [18][19][20][21]. In contrast, patients carrying monoallelic heterozygous truncating mutations presented relatively mild phenotypes, such as juvenile myoclonic epilepsy and FS, with favorable outcomes [22,23], indicating a genetic dose effect (quantitative correlation). On the other hand, patients carrying the heterozygous missense mutations have a variety of epilepsy phenotypes including severe myoclonic seizures, juvenile myoclonic epilepsy, and mild FS [15,24]. A previous study demonstrated that the phenotype severity potentially depends on the degree of the damaging effect of the missense mutation [25], which is one of the explanations for phenotypical variation. From a pathophysiologic perspective, Usher IIC likely occurs when the complete loss of function of the ADGRV1 protein leads to the dysfunction of the Usher protein network [26], and epilepsy likely occurs when heterozygous mutations lead to haploinsufficiency or functional defects of ADGRV1 protein [15,23,24,27]. In the present study, heterozygous p.Val272Ile and p.Met4419Val were rare variants in the general population, and heterozygous p.Asn3028Ser and p.Ala3657Val were identified in individuals with FS. A recent study demonstrated that the damaging effects of variants usually vary and potentially present a continuing distribution with overlaps between pathogenic variants and benign variants [25].
It was therefore possible that heterozygous p.Val272Ile, p.Asn3028Ser, p.Ala3657Val, and p.Met4419Val may be less pathogenic and even overlap with rare variants in general populations. On the other hand, biallelic ADGRV1 variants were potentially associated with RE/ARE, while monoallelic ADGRV1 variants were associated with milder symptoms such as FS, suggesting a potential quantitative correlation between genetic impairment and phenotype severity, which would help understand the difference in pathogenicity between biallelic and monoallelic heterogeneous ADGRV1 missense variants. ADGRV1 protein has a large ectodomain containing 35 Calx-β repeats that resemble regulatory domains of sodiumcalcium (Na + /Ca 2+ ) exchanger proteins (NCXs) [16]. NCXs are ubiquitous membrane transporters with a key role in Ca 2+ homeostasis and neuronal processes such as memory formation, learning, and neuroprotection [28]. To date, 20 ADGRV1 mutations, including 2 null mutations and 18 missense mutations, have been identified in 20 unrelated patients with epilepsy (Table S1). All heterozygous missense variants were located in the Calx-β repeat domains, except five variants located in the extracellular space (links Calx-β repeat domains). RE/ARE-related missense mutations identified in this study were also located in the Calx-β repeat domains. These mutations potentially affect the Calx-β domain and consequently impair Ca 2+ fluxes and homeostasis. Previous studies reported that alterations in neuronal Ca 2+ homeostasis play an essential signaling role in the pathogenesis of epilepsy [29,30]. Given this, ADGRV1 mutations cause the dysfunction of NCX domains leading to alterations in neuronal Ca 2+ homeostasis and potentially playing an essential role in the generation and propagation of epilepsy. In addition, a de novo mutation in GRIN2B and a missense mutation in RyR2 were also identified in patients affected by RE/ARE and FS.
GRIN2B resides on chromosome 12p13.1 and encodes the N-methyl-D-aspartate receptor (NMDAR) subunit NR2B which acts as an agonist binding site for glutamate [31]. Heterotetrameric N-methyl-D-aspartate receptors (NMDARs) are highly Ca 2+ -permeable cation channels composed of two glycine-binding NR1 subunits (encoded by GRIN1) and two glutamate-binding NR2 subunits (encoded by GRIN2A, GRIN2B, GRIN2C, or GRIN2D) [32]. Simultaneous binding of both agonists activates NMDAR, which opens a cation-selective pore leading to an influx of Ca 2+ and depolarization. NMDARs are expressed throughout the CNS and mediate excitatory neurotransmission, thereby exerting a critical role in memory/learning, cognitive functions, synaptic plasticity, and normal neuronal development [33]. Previous studies have reported that GRIN2B mutations are associated with epileptic encephalopathy [14].
However, no EEG patterns of continuous spikes and waves during sleep or CTS were previously observed patients with GRIN2B mutations, which are predominantly associated with GRIN2A mutations [11]. In the present study, we identified a de novo GRIN2B mutation in a patient with ARE, expanding the genotype-phenotype correlation of GRIN2B mutation as a cause of ARE. A recent study demonstrated that human mutations in GRIN2B can impair calcium influx in neurons and thus lead to alterations in neuronal Ca 2+ homeostasis [31]. It may be that the imbalances in Ca 2+ homeostasis underlie the epilepsy observed in patients with GRIN2B mutations.
The ryanodine receptor-2 (RyR2) gene resides on chromosome 1q43 and encodes an intracellular Ca 2+ channel protein, which elevates cytoplasmic Ca 2+ by release from endoplasmic and sarcoplasmic stores upon activation [34]. This protein is predominantly expressed in the heart and brain, especially in the dentate gyrus of the hippocampus and cortex [35]. RyR2 mutations were initially linked only to exercise-induced catecholaminergic polymorphic ventricular tachycardia (CPVT); however, of the reported children with CPVT, the majority were also reported to have seizures [34]. In RyR2 −/+ mutant mice, heterozygous RyR2 mutations can cause epilepsy and exercise-induced sudden cardiac death, and RyR2 mutation altered intracellular Ca 2+ homeostasis, which contributes to defective neuronal and cardiac excitability and was suggested to be one cause [34,36]. In humans, RyR2 mutations were also identified in patients with sudden unexplained death in epilepsy (SUDEP) [37] and only epilepsy without cardiac arrhythmias [38]. Thus, these findings suggest that mutation in RyR2 is a cause of epilepsy and that the mechanism may be related to dysregulation of neuronal intracellular Ca 2+ homeostasis via defective RyR2 channels. In this study, we identified a missense RyR2 variant in a girl only with RE; however, the variant was inherited from her father, who had no history of seizures. Because the majority of patients with RE are sporadic, incomplete penetrance is potentially one of the explanations. Of the remaining genes, no definite pathogenic variants were identified. Thus, we suspected that RyR2 mutations can potentially cause RE/ ARE without arrhythmia.
Strikingly, there are strong similarities in the potential molecular pathogenesis of these conditions due to mutations in the studied genes as they all involve the dysregulation of Ca 2+ homeostasis and signaling. Indeed, the dysfunction of Ca 2+ homeostasis in neurons is believed to play an essential role in the generation and propagation of epileptiform events [30,39,40]. Therefore, imbalances in Ca 2+ homeostasis, in part, may underlie the RE/ARE observed in patients with the ADGRV1, GRIN2A, GRIN2B, and RyR2 mutations and may provide deep insight into the nature of pathogenicity.

Conclusion
In summary, our study suggests that ADGRV1, GRIN2B, and RyR2 variants are implicated in RE/ARE. These data link mutations in neuronal intracellular calcium homeostasis related genes to susceptibility to RE/ARE and implicate alterations in neuronal calcium homeostasis as a previously unrecognized mechanism contributing to RE/ARE. However, further studies are required to explore the exact molecular underlying mechanism, and understanding the underlying molecular processes is of critical importance for the precise treatment of childhood RE/ARE.