1 INTRODUCTION

Neurodevelopmental disorders (NDDs) include a spectrum of conditions characterized by psychomotor delay and a variable range of intellectual disability (ID) signs, with/without autistic traits and behavioral problems. Among NDDs, there are developmental and epileptic encephalopathies (DEEs) characterized by early-onset and refractory seizures. Chromodomain helicase DNA-binding protein 2 (CHD2) has been recently associated with DEEs (Lamar & Carvill, 2018), and pathogenic variants in this gene have been reported in relation to childhood-onset epilepsy in almost all the 144 surveyed cases (Zhu et al., 2022). Myoclonic, atonic, absence, or generalized tonic–clonic seizures (GTCS) are prevalent in the second year of life, associated with photosensitivity in a variable percentage of cases (Caputo et al., 2018; Chen et al., 2020). Psychomotor delay and ID are observed in most cases, while autistic traits and behavior disorders, in particular aggression (Zhu et al., 2022), may be additional features.

We describe a family with two brothers characterized by mild ID and only the elder presented a history of pharmacologically controlled childhood epilepsy and behavior problems associated with sleep disorders. Based on his clinical manifestations, the first diagnostic suspect has been Smith–Magenis syndrome (SMS; MIM #182290). The identification of a pathogenic CHD2 variant, shared with his differently symptomatic brother, led to the detection by next-generation sequencing (NGS) on parents' tissues of a low-rate gonosomal mosaicism in the mother, then confirmed by digital polymerase chain reaction (dPCR).

2 MATERIALS AND METHODS

3 RESULTS

Proband 1 was addressed to the medical genetics laboratory with a clinical suspicion of SMS to investigate by NGS analysis the four main genes responsible for SMS and SMS-like syndrome: RAI1 (MIM *607642), HDAC4, MBD5, and TCF20. A custom panel of 35 genes involved in epilepsy and ID, including these four genes and, among others, the CHD2 gene, was sequenced. This analysis ruled out pathogenic variants in SMS genes, while unexpectedly a novel CHD2 heterozygous nonsense variant, NC_000015.9 g.93482836 C>T, NM_001271.3 c.580 C>T, p.(Gln194*) was identified and confirmed by Sanger sequencing. According to predictions, the variant should elicit mRNA degradation by nonsense-mediated mRNA decay (NMD). As CHD2 haploinsufficiency has been associated with a variable phenotype (Chénier et al., 2014) comprising developmental delay, ID, epilepsy (ranging from mild febrile seizures to severe epileptic encephalopathy) (Chen et al., 2020), and behavioral problems, the variant appeared compatible with the patient's and his younger brother's phenotypes. We did not identify any other variants of uncertain significance (VUS) consistent with the clinical picture. The subsequent Sanger sequencing analysis detected the same variant in the affected patient 2 and, conversely, its absence in the DNA from the peripheral blood of the healthy brother and parents, supporting its pathogenicity (Figure 1). To distinguish between a parental condition of germline mosaicism from low-grade gonosomal mosaicism, which escapes detection by Sanger sequencing, NGS was performed on the DNA from peripheral blood and buccal swabs of the parents and the healthy brother. This analysis highlighted only in the mother's buccal swab the presence of the CHD2 variant, with a variant allele frequency (VAF) of ~1% (Table 1, left). To confirm or exclude this finding, dPCR was carried out on parents' DNA samples obtained from peripheral blood and buccal swabs. The presence of the pathogenic variant was detected only in the cells of the mother's buccal epithelium, with a similar percentage to that calculated by the Mutect2 algorithm used for NGS data analysis (Table 1, right). These data highlight the presence of somatic gonosomal mosaicism in the patients’ mother.

4 DISCUSSION

SMS is typically due to haploinsufficiency of RAI1, coding for a transcriptional factor involved in multiple functions, including embryonic neurodevelopment and neuronal differentiation, behavioral functions, and circadian activity. The SMS phenotype is characterized by distinctive facial features, developmental delay, mild-to-moderate ID, behavioral abnormalities (including severe sleep disturbance, with disruption of the circadian rhythm, stereotypies, temper tantrum, impulsivity, aggression, primarily self-inflicted or directed toward others, injurious behaviors, with particularly early onset and childhood onset) and abdominal obesity (Falco et al., 2017; Rinaldi et al., 2022). Since childhood, proband 1 displayed a severe alteration of the sleep–wake cycle, as well as aggressive behavior, mainly directed against others, clinical features that summed up to psychomotor retardation and mild ID, oriented clinicians toward the diagnostic suspicion of SMS. GTCS epilepsy, which occurred in childhood, completed the clinical picture, although epilepsy is not so frequent in RAI1-deficient patients (Carmona-Mora & Walz, 2010; Rinaldi et al., 2022). Conversely, the younger brother, later evaluated by clinicians, never displayed epilepsy, or other peculiar characteristics of SMS while suffering from relational disorders and panic attacks (anxiety). CHD2-related syndrome is typically a DEE: although CHD2 pathogenic variants have been reported in patients with NDD without epilepsy, 92% of patients with CHD2 manifest, during their lifetime, different seizure types (De Maria et al., 2022) such as GTCS, myoclonic, absences, focal-onset, myoclonic jerks (eyelid myclonia), atonic, myoclonic–atonic, tonic, and epileptic spasms. Furthermore, seizures have a wide onset time, ranging from 3 months to as long as 22 years (median age 2.5 years), and are drug-resistant in about three-fourths of the patients (Zhu et al., 2022); the nonresistant ones are generally responsive to sodium valproate (Chen et al., 2020). Photosensitivity, generally rare in other forms of epilepsy, is reported in the literature with variable percentages, ranging from 45% (De Maria et al., 2022) to 70% (Thomas et al., 2015). The epileptic picture of proband 1 overlaps with the phenotype typical of the CHD2-related syndromes. In the majority of the reported cases, a developmental delay mostly affecting speech (as in proband 2) is noticed before seizure onset, as well as aggressive and impulsive behavior, repetitive behaviors, more rarely ADHD and autism spectrum disorder (ASD) and severe psychiatric profile (Bernardo et al., 2017). These behavioral characteristics are compatible with the clinical picture of proband 1 and partially overlap those of RAI1-deficient patients. Unlike RAI1-deficient patients, to the best of our knowledge, alteration of the sleep pattern is reported in the literature only in two CHD2-mutated cases: a patient who carries a CHD2 stop mutation and, like proband 1, presents psychotic traits (Poisson et al., 2020) and two monozygotic twins both showing alteration of the circadian rhythm (A. M. Pinto et al., 2016). The overall findings suggest the importance of considering CHD2 involvement in children with developmental delay, ID with or in the absence of epilepsy, even at late onset, behavioral disturbances, in particular aggressiveness and sleep disorder reminiscent of RAI1 defects.

Finally, the two brothers carrying the same pathogenic variant, express a different phenotype, not only as regards epilepsy but also behavioral and sleep disturbances. This discrepancy has been similarly observed in reported family cases where affected family members show differences in clinical manifestation (Feng et al., 2022; Lebrun et al., 2017; Petersen et al., 2018) as well as in unrelated patients with the same pathogenic variant, but discrepant phenotypes (De Maria et al., 2022). The overall evidence points to the difficulty of establishing a genotype–phenotype correlation for CHD2-caused disorder and to the existence of genetic and environmental modifiers.

The CHD2 gene codes for a chromatin assembly ATPase, which catalyzes the assembly of chromatin into periodic arrays of nucleosomes (Liu et al., 2015) and modulates gene expression via the ability to remodel chromatin structure (Chénier et al., 2014). To date (July 2023) more than 140 CHD2 variants associated with clinical phenotype are reported in the Human Gene Mutation Database HGMD professional (https://www.hgmd.cf.ac.uk/ac/index.php), including partial or complete deletion of the gene, as well as point mutations with a prevalence of nonsense/truncating variants, likewise, the one here described. These variants generally lead to mRNA degradation by NMD (De Maria et al., 2022). Our data confirm that the pathogenic mechanism determining the CHD2-associated phenotype is haploinsufficiency (Lamar & Carvill, 2018) as demonstrated by studies in zebrafish models (Suls et al., 2013).

The mutations are almost exclusively de novo: only five variants were inherited from asymptomatic (Chen et al., 2020; Feng et al., 2022; Galizia et al., 2015), mildly affected (Chen et al., 2020) or affected parents (Petersen et al., 2018). Six cases with more than one affected family members are described (Chen et al., 2020; Feng et al., 2022; Galizia et al., 2015; Lebrun et al., 2017; Petersen et al., 2018; A. M. Pinto et al., 2016) including two cases of germline mosaicism (Lebrun et al., 2017; D. Pinto et al., 2014) (see Table 2).

Our siblings inherited a pathogenic CHD2 variant from their unaffected mother, and although the presence of the pathogenic variant has not been directly assessed in gonadal tissue, our family represents the third case of germline mosaicism and the first case of gonosomal mosaicism involving CHD2. The NGS analysis for somatic variants on the DNA from the mother's buccal swab allowed us to suspect a condition of gonosomal mosaicism. Parental gonosomal mosaicism with a higher number of cells has been described for other genes responsible for epileptic encephalopathies (Stosser et al., 2018). Our data allow to hypothesize that gonosomal mosaicism may underlie germline mosaicism, relatively frequent in the CHD2 gene, although its real frequency is yet unestimated. Dedicated NGS somatic analyses and dPCR may represent a powerful tool to explore the real occurrence of this phenomenon, which may have a relevant impact on genetic counseling, considering that in the presence of gonosomal mosaicism the recurrence risk is significantly higher (Bai et al., 2023). Additional molecular investigation by high-coverage NGS/dPCR on at least two parental tissues should be routinely performed to detect parental mosaicism and define a proper recurrence risk, considering also that low levels of mosaicism in the blood could be predictive of higher levels in other tissues (Rahbari et al., 2016). Even if it is challenging, it could be proposed also to couples with a single child carrying a de novo pathogenic variant in the CHD2 gene.

AUTHOR CONTRIBUTIONS

Francesca Cogliati and Silvia Russo conceived the study. Francesca Cogliati, Sara Perego, and Maura Masciadri carried out the experiments and analyzed the NGS genetic data. Berardo Rinaldi, Donatella Milani, and Maria Francesca Bedeschi clinically evaluated the patients and hypothesized the clinical suspicion. Letizia Straniero, Valeria Rimoldi, and Rosanna Asselta performed and interpreted the dPCR experiment. Davide Gentilini handled bioinformatics data analysis. Francesca Cogliati drafted the manuscript. Silvia Russo and Lidia Larizza read and revised the manuscript with the help of Maria Francesca Bedeschi and Rosanna Asselta. Silvia Russo recruited the funding for the study. All authors read and approved the final manuscript.

FUNDING INFORMATION

This work was funded by BIBLIOSAN.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

CONSENT STATEMENT

Written informed consent and consent to publish were obtained from the family.