Chromodomain helicase DNA-binding protein 2 (CHD2, MIM: 615369) is a member of the CHD family, which is only known to cause the brain-restricted phenotypes when disrupted in humans. It is mapped to chromosome 15q26.1 and is considered an ATP-dependent chromatin remodeling that regulates the transcription expression of many genes (Lamar and Carvill, 2018; Carvill and Mefford, 2015). Some studies have demonstrated that pathogenic variants of the CHD2 gene are associated with childhood-onset developmental and epileptic encephalopathy (DEE), which is a severe form of neurodevelopmental disorder with a wide range of phenotypic variability, including autism spectrum disorder (ASD), intellectual disability (ID), developmental delay, microcephalus, behavioral anomalies, facial dysmorphisms, and several types of epilepsy (Carvill and Mefford, 2015; Thomas et al., 2015; Verhoeven et al., 2016; Chen et al., 2020). Recently, Chen et al. presented the largest single case series of patients with CHD2-related epilepsy and then comprehensively reviewed 53 published cases in the literature through seizure onset age, seizure types, developmental outcome, electroencephalogram (EEG), brain magnetic resonance imaging (MRI), and diagnoses, among others, which improved the understanding on the relationship between genotype and phenotype (Chen et al., 2020). In 2021, De Maria et al. reported 18 new patients and reviewed 84 previously reported patients, getting the results that the median age of seizures onset in 92% of patients was 2.5 years, and there was no clear association between genotypes and phenotypes (De Maria et al., 2021).

This study reported a novel variant in the CHD2 gene in a Chinese girl with developmental delay and myoclonic epilepsy through whole-exome sequencing (WES). We also systematically reviewed the published literature, provided a thorough overview of clinical, neuroimaging, physical, and genetic findings in CHD2-related epilepsy patients, and built the possible relationship between genotypes and phenotypes.

The CHD protein family (containing CHD1–CHD9) is mainly involved in the ATP-dependent chromatin remodelers that contribute to the reorganization of chromatin structure and deposition of histone variants necessary for regulating gene transcription expression. Among the nine CHD family members, the CHD2 pathogenic variants only lead to a brain-restricted phenotype when disrupted in humans, which indicates a unique role for this gene in neurodevelopment (Lamar and Carvill, 2018).

CHD2 mutation-related epilepsy was first published in 2009 (Veredice et al., 2009). This paper described a 30-month-old girl with refractory myoclonic epilepsy associated with mental retardation, growth delay, peculiar facial appearance, minor physical anomalies, and photosensitivity. In the patient, a 15q26.1–15q26.2 deletion including CHD2 was detected by CGH-array. To date, there are 144 reported individuals encompassing 126 single-nucleotide or indel variants (non-CNVs) and 18 copy number variants (CNVs) with CHD2-related epileptic encephalopathy (Table 1 and Supplementary Table S1). From Table 1 and Supplementary Table S1, almost all patients were diagnosed at over 1 year or even over 6 years of age. Up to now, most CHD2 variants have been reported de novo in patients with epilepsy rather than in an inherited mode. Nevertheless, in 2015, the first patients diagnosed with eyelid myoclonia with absence (EMA) inherited it from an unaffected mother, who carried the CHD2 variant c.653C>T (p. Pro218Leu) (Galizia et al., 2015). In 2018, the patient who inherited a pathogenic CHD2 variant c.628G>T (p. Glu218*) from the affected mother was published (Petersen et al., 2018). The distressing thing is that compared with her daughter’s refractory epilepsy, her seizures have always been well controlled (Petersen et al., 2018). In 2020, Chen et al. reported a sporadic case inherited from an unaffected father with c.5153+2T>C variant of CHD2 gene and dizygotic twins inherited from the affected father with c.5232G>A (p. Met1744Ile) (Chen et al., 2020). To our surprise, the five mutations in patients did not occur again in other cases. In CHD2 CNVs patients, the inherited patients were not reported. These findings indicate that CHD2 variants, with the exception of CHD2 CNVs, can be inherited (Table 1).

Previously, CHD2 variants were generally considered to be childhood-onset epileptic encephalopathy. However, more and more individuals with CHD2 mutants were published, and the ages of seizure onset were different from infancy to childhood (Table 1 and Supplementary Table S1). We had also found that different patients had different development in pre-seizure onset. For example, Carvill et al. reported that two out of six patients with epileptic encephalopathies caused by CHD2 mutations had normal development and another four were abnormal (Carvill et al., 2013). Interestingly, all the patients with CHD2 CNVs could lead to developmental delays in pre-seizure onsets, such as motor and speech developmental delays (Table 1 and Supplementary Table S1). In addition, the inducing factors, including photosensitivity and fever sensitivity, EEG results, and facial dysmorphisms, were also various among CHD2 variants’ patients. Galizia et al. performed photosensitivity tests on zebrafish larvae with CHD2 gene knockout and found that the CHD2 gene knockout significantly enhanced the photosensitivity of congenital zebrafish larvae. Their results had confirmed that this gene was related to photosensitive epilepsy (Galizia et al., 2015). Except for the twin patients mentioned above, most patients had abnormal EEG results (Chen et al., 2020). We also found that the reported patients had behavior problems such as aggressive behavior. In other aspects of patients, the findings showed diversity, such as ASD, height, and MRI (Table 1 and Supplementary Table S1).

The patients have variable seizure types, including absence seizures (AbS), atonic seizures, tonic-clonic seizures, and myoclonic absence seizures. Importantly, an individual can have multiple seizure types; even the first seizure and the later seizure could be of different types (Supplementary Table S1). Based on these phenomena, we summarized and analyzed the seizure types of patients with CHD2 non-CNVs and CNVs (Table 2 and Supplementary Table S2). In the CHD2 non-CNVs patients, both the myoclonic seizures (MS, 27.3%, 24/88) and generalized tonic-clonic seizures (GTCS, 19.3%, 17/88) were the top two types of the first seizures, which presented the difference than average value (5.87 people/seizure types). Some seizure types, including atypical myoclonic-absence seizures (aMAS, 0%) and febrile clonic seizures (FCS, 0%), did not exist. In contrast, MS (50.8%, 30/59), GTCS (50.8%, 30/59), atonic seizures (AtS, 25.4%, 15/59), focal seizures (FoS, 25.4%, 15/59) and andother three seizure types were the main types of second and later seizures compared to average value (2.95 people/seizure types). In general, MS, GTCS, AtS, febrile seizures (FS), absence seizures (AbS), FoS, atypical absence seizures (aAS), and eyelid myoclonia with absence (EMA) were the main seizure types in CHD2 non-CNVs (Table 2 and Supplementary Table S2). Interestingly, the main type for CHD2 CNVs patients could be GTCS, which did not exist using the Bonferroni correction. This phenomenon could be related to the small samples. Furthermore, the patients’ numbers in different epileptic encephalopathy were analyzed using the above methods (Figure 3 and Supplementary Table S3). The results showed that Lennox–Gastaut syndrome (LGS) was usually diagnosed in CHD2 non-CNVs patients, which presented the difference compared to the average value (2.94 people/epileptic encephalopathy) using the Pearson χ2 test, and the adjusted difference with the Bonferroni correction was disappearance (Supplementary Table S3). The CHD2 CNVs individuals did not exist in these results (Supplementary Table S3).

Though a series of studies indicated that there was no clear correlation between genotypes and phenotype in CHD2-related epileptic encephalopathy (Epi et al., 2013; Chen et al., 2020; De Maria et al., 2021), we thought that it was necessary to further research the association between genotypes and phenotypes. We referred to the mRNA sequences NM_001271.4 and protein sequences NP_001262.3 to analyze all the CHD2 non-CNVs related epilepsy patients (Thomas et al., 2015; Carvill and Mefford, 2015) (Figure 4 and Supplementary Tables S4, S5). The whole protein sequences were divided into 11 fragments (D1–D11), including four conserved domains (Thomas et al., 2015). There were 102 types of CHD2 non-CNVs in 126 patients, and almost one mutational type corresponded to one patient. Among them, the c.4173dupA (p. Gln1392Thrfs*17) and c.5035C>T (p. Arg1679*) could be the hotspot variants but only occurred five times, respectively (Supplementary Table S4). The CHD2 variants were distributed in the whole protein sequences and focused on the D9 fragments (906-1458aa, 33.92%, 37/126) (Supplementary Table S4), which led to a serious shortcoming because the longer the domain, the more the number. In order to address this shortcoming, the incidence rate was applied. Though the D11 fragments (1554-1828aa, 3.05%, 25/825) tended to host the CHD2 variants, the difference value was not shown than the average value (2.30%) (Figure 4A and Supplementary Table S4). Meanwhile, The D3 fragment was too short and was not further analyzed. Interestingly, the incidence rate presented the difference between D1 and D11 (Supplementary Table S4). We also analyzed the relationship between CHD2 genotypes and seizure control, inducing factors including photosensitivity and fever sensitivity in all the CHD2 non-CNVs patients, and found a significant correlation between genotypes and phenotypes (Figures 4B,C and Supplementary Table S5). The patients with CHD2 variants in D11 tended to have better seizure control, and patients with CHD2 variants in D6 and D9 tended to have worse seizure control. Similarly, there also was a difference in fever sensitivity among D6, D8, D9, and D11. The patients in D8 tended to have more serious fever sensitivity compared to D6, D9, and D11. Though the difference for photosensitivity did not present with CHD2 variants in D6, D9, and D11, patients with CHD2 variants in D9 suffered photosensitivity more easily than CHD2 variants in D6 and D11. These results suggested that there could be associations between genotypes and phenotypes, which were further demonstrated by the dizygotic or monozygotic twins (Pinto et al., 2016; Wang et al., 2017; Chen et al., 2020). Pinto et al. reported that the monozygosity cases had the same seizure onset at 2 years and 6 months old, carried the c.4173dupA (p. Gln1392Thrfs*17) mutations, and presented a set of similar clinical features phenotypes, including autism spectrum disorder, hypotonia, postnatal microcephaly, stereotypic movements, circadian rhythm alterations, and AbS (Pinto et al., 2016). Wang et al. also reported that the monozygotic twins had the same mutation c.5035C>T (p. Arg1679*) and showed similar clinical features (Wang et al., 2017). In another paper, the dizygotic twins were diagnosed with febrile seizure plus (FS+), carried the c.5232G>A (p. Met1744Ile) mutation inherited from the affected father, showed similar clinical descriptions such as fever introduce, MRI normal, EEG normal, first seizure type GTCS, and seizure control except for seizure onset and further seizure type (Chen et al., 2020).

A large number of studies demonstrated the phenotypic heterogeneity of CHD2-associated epilepsy. So far, 144 patients have been reported to have epileptic seizures and associated CHD2 variants (Table 1 and Supplementary Table S1). However, not all patients with CHD2 variants have seizures. Up to now, a total of 9 patients with CHD2 variants were reported to have no epilepsy (Bhakta et al., 2005; Kulkarni et al., 2008; Chenier et al., 2014; Hamdan et al., 2014; Pinto et al., 2014; Chen et al., 2017; Cabrera-Salcedo et al., 2019), For example, Hamdan et al. found that a severe ID girl with motor and speech development delays and CHD2 variant c.335C>G (p. Ser112*) did not have the history of epilepsy (Hamdan et al., 2014). Interestingly, patients without epilepsy tended to be CHD2 CNVs patients (Bhakta et al., 2005; Kulkarni et al., 2008; Chenier et al., 2014; Hamdan et al., 2014; Pinto et al., 2014). So far, five patients carrying CHD2 CNVs but without epilepsy have been reported. For example, Pinto et al. reported a patient with a de novo deletion in the CHD2 gene, with ASD, mild ID, and dysmorphic features including micrognathia and protruding ears, but no seizures (Pinto et al., 2014). However, his ASD-affected brother carried the same deletion of the CHD2 gene, had similar dysmorphic features and mild ID, and experienced an epilepsy onset at 9 years of age. Kulkarni et al. described a de novo translocation t(X; 15) (p22.2; q26.1) dn disrupting CHD2 in a child with developmental delay, scoliosis, and hirsutism (Kulkarni et al., 2008).

According to the data, seizures in about 1/4 of patients are controlled through single or multiple anti-epileptic drugs (Table 1 and Supplementary Table S1). Interestingly, the ages of seizure onset of patients who were seizures control and carried CHD2 non-CNVs were from 1 to 12 years, and generalized tonic-clonic seizures (GTCS, 61.1%, 11/18) and myoclonic seizures (MS, 38.9%, 7/18) were the top two seizure types (Supplementary Table S1). However, most patients with CHD2-related neurodevelopmental disorders remain refractory to treatment. Although in some epilepsy syndromes, a ketogenic diet could have vital benefits (Daci et al., 2018; Zhou et al., 2018), a total of four patients with CHD2 variants were treated with the ketogenic diet, which did not produce significant effects (Thomas et al., 2015; Costain et al., 2019). Therefore, a large sample size and follow-up studies would be helpful to define the treatment and prognosis of CHD2-related epilepsy.

In this study, a girl with developmental delay and myoclonic epilepsy caused by a new mutation c.4318C>T (pArg1440*) in the CHD2 gene was studied using WES. The mutation produced a nonsense variant and disrupted the CHD2 protein structure.

Subsequently, all patients with reported neurodevelopmental disorder and variants in the CHD2 gene were systematically reviewed and analyzed. A total of 144 patients, including 126 non-CNVs and 18 CNVs with CHD2 pathogenic variants, were analyzed. We found that, except for CHD2 CNV, a small portion of patients obtained CHD2 variants by inheritance rather than de novo. The ages of seizure onset with CHD2 mutants varied from infancy to childhood, even adults. The patients with CHD2 non-CNVs had a certain tendency toward the seizure types, such as MS and GTCS. Meanwhile, not all patients with CHD2 variants had seizures. Although almost one mutational type corresponded to one person, there may be an association between genotypes and phenotypes, especially for seizure control and fever sensitivity. The above results can provide a theoretical basis for better research on CHD2-related neurodevelopmental disorders.