2.1. Subjects

The patient was recruited from Guangdong Sanjiu Brain Hospital. Epileptic seizures and epilepsy syndromes were diagnosed according to the criteria of the Commission on Classification and Terminology of the International League Against Epilepsy (2022). TLE was diagnosed according to the following features: (1) focal original seizures or focal seizures manifesting automatic symptoms, mental symptoms, special sensory symptoms such as vision and auditory hallucinations, and/or gastrointestinal symptoms and (2) interictal focal dischargers in the temporal lobe detected by electroencephalogram (EEG).

This study received approval from the Ethics Committee of the Guangdong Sanjiu Brain Hospital. Written informed consents were provided by the patient’s legal guardians.

2.2. WES

Genomic DNAs were extracted from blood samples of the probands and the parents to perform WES by a NextSeq500 sequencing instrument (Illumina, San Diego, California, United States) following the standard procedures previously described [11–14]. High-quality sequencing data were acquired through massively parallel sequencing, achieving an average depth of over 125× and more than 98% coverage of the capture region on the chip. These reads were then successfully mapped to the Genome Reference Consortium Human Build 37 using Burrows–Wheeler alignment. The Genome Analysis Toolkit was employed to identify single-nucleotide point variants and indels.

2.3. Damaging Effect Analysis

The consequences of all the missense variants were predicted by 18 common in silico tools, including PolyPhen2_HVAR [15], LRT [16], MutationTaster [17], FATHMM [18], PROVEAN [19], CADD [20], MetaSVM [21], MetaLR, M_CAP [22], FATHMM_MKL [23], Eigen [24], GenoCanyon [25], fitCons [26], GERP [27], phyloP [28], phastCons [29], REVEL [30], and ReVe [31]. Protein modeling was performed by using the three web tools, including the SWISS model software (April 2024 version, https://swissmodel.expasy.org/), the Phyre2 model (http://www.sbg.bio.ic.ac.uk/phyre2/html/page.cgi?id=index), and the AlphaFold model (monomer v2.0, https://alphafold.com/). PyMOL Molecular Graphics System (version 2.3.2; Schrödinger, LLC; New York, United States) was used for three-dimensional protein structure visualization and analysis. Molecular docking was conducted by the ZDOCK Server (https://zdock.umassmed.edu/). The hydrophobicity changes of variants were analyzed by the VarSite web server [32]. The pathogenicity of variants was evaluated by the ACMG guideline [33].

2.4. Assessment of ACTB Expression Profile

The expression profile of ACTB was assessed by the method of our previous studies [34–36]. Expression in adult 54 nondisease tissues was analyzed by the data from the Genotype-Tissue Expression (GTEx) database (https://www.gtexportal.org/home/). Human RNA-seq data of developmental stages (from eight postconceptional weeks to 40 years) for multiple brain areas were obtained from the BrainSpan database (http://www.brainspan.org/). RNA expression was normalized to reads per kilobase million (RPKM). The expression spline is fitted by the locally weighted scatterplot smoothing (LOWESS) algorithm for interpreting the expression pattern of ACTB.

2.5. Protein–Protein Interaction (PPI) Network Analysis

The PPI network of the ACTB protein was analyzed using the STRING database (https://string-db.org/; version: 12.0) [37, 38]. The interactive genes with a confidence score ≥ 0.9 were taken into analysis, with basic settings including (1) “physical subnetwork” for network type, (2) “evidence” for the meaning of network edges, (3) “experiments” and “databases” for active interaction sources, and (4) “500 max interactors (max)” for max number of interactors to show.

2.6. Statistical Analysis

R (version 4.0.3) was used for data processing. p value and/or Q‑value (false discovery rate) < 0.05 was considered statistically significant.

3.1. Identification of ACTB Variant and Clinical Feature

In this study, one variant in the ACTB gene (c.275A>G/p.N92S) was identified in a patient (Figures 1(a) and 1(b)). The variant was of de novo and absent in any controls, consistent with the Mendelian-dominated inherited pattern (Table 1).

The patient was born to nonconsanguineous parents, was delivered normally, experienced neurodevelopment consistent with peers, and had no facial dysmorphism. When the patient was 17 years old, the first seizure occurred, manifesting focal features including nausea, headaches, and subsequent loss of consciousness. The seizures were repeatedly exhibited two or three times per year. Two years later, the patient exhibited another type of seizure, focal original tonic–clonic seizures, manifesting loss of consciousness, upward rolling of the eyes, and convulsions of the limbs during sleep. Then, he was referred to our hospital and received the treatment of levetiracetam tablets (1 g/day). After 3 months of treatment with levetiracetam, seizures still attacked. Another antiseizure medicine, lamotrigine (50 mg/day), was thus added. Subsequently, the patient achieved a seizure-free status for more than 2 years, until now. The patient also reported a decline in memory and some irritability since the first seizure, which has improved slightly recently. Focal discharges were detected in the bilateral temporal lobes (Figure 1(c)) by EEG before being treated with lamotrigine, but normal in the past 2 years. Magnetic resonance imaging scans detected bilateral hippocampal sclerosis (Figure 1(d)). Taken together, the clinical features and epileptic discharges, the patient was diagnosed with TLE.

3.2. The Pathogenicity of the Identified ACTB Variant

The identified variant, c.275A>G/p.N92S, was located in the third exon of the ACTB gene (Figure 2(a)). According to the Missense Tolerance Ratio (MTR) Gene Viewer developed to assess the pathogenicity of missense variants [39], the variant affects a residue of a missense tolerance ratio of 0% (Figure 2(b)). Cross-species sequence alignment suggests that this variant occurs in a residue with high conservation (Figure 2(c)). The variant is also predicted to decrease the hydrophobicity, calculated by the Fauchère and Pliska hydrophobicity scale (Figure 2(d)). Notably, the variant is predicted to be “damaging” or “conserved” by almost all in silico tools (17/18 tools, Table 2). According to the guidelines of the ACMG, this variant is classified as “likely pathogenic,” due to the origination of de novo (PS2), not presented in populations (PM2), and predicted to be damaging by multiple in silico tools (PP3).

Additionally, no “likely pathogenic” or “pathogenic” variants in other genes are identified in this case, according to the ACMG guidelines (Table S1). The identified ACTB variant is thus considered to be the genetic explanation of this case.

3.3. The Interacted Protein of Beta-Actin and Their Associated Function, Pathways, and Phenotypes

The ACTB gene encodes beta-actin protein, a highly conserved protein that polymerizes to produce filaments that form cross-linked networks that are ubiquitous in the cytoplasm of diverse cells. Actin exists in both monomeric (G-actin) and polymeric (F-actin) forms, both forms interact with numerous proteins, playing key functions, such as cell motility and contraction. In addition to their role in the cytoplasmic cytoskeleton, G- and F-actin also localize in the nucleus and regulate gene transcription and motility and repair of damaged DNA. To explore the underlying pathogenic mechanism, we investigated the PPI network of the beta-actin protein via the STRING database. The PPI network of ACTB included 100 nodes, 530 edges, 10.6 of average node degree, 0.907 of average local clustering coefficient, and the PPI enrichment p value < 1.0 × 10⁻¹⁶ (Figure 3(a)). The PPI included numerous genes associated with seizures, such as ARID2, ARID1A, CDC42, CTNNA2, DYNC1H1, TRRAP, and SMARCA4 (Online Mendelian Inheritance in Man (OMIM) database, https://omim.org/).

Meanwhile, lots of neurodevelopmental-related terms were significantly enriched on genes of the PPI network. The top 10 cellular components of Gene Ontology (GO) terms showed that these genes are involved in chromatin structure regulation (male-specific lethal (MSL) complex, NuA4 histone acetyltransferase complex, Swr1 complex, etc.) or participate in cell movement and morphogenesis (myosin II filament, arp2/3 protein complex, etc.), which contribute to proper neuronal differentiation and synapse formation (Figure 3(b)) [40, 41]. The top 10 biological processes of GO terms showed similar localization of function in the above components, namely, chromatin remodeling or actin cytoskeleton activity (Figure 3(c)). The top 10 molecular functions of GO terms showed that 30% of pathways are directly related to synapses, and 60% indirectly affect the formation of synapses or other neuronal structures (Figure 3(d)). The top 10 Kyoto Encyclopedia of Genes and Genomes (KEGG) terms showed that these node genes are involved in the regulation of actin cytoskeleton, which may be underlying pathogenic pathways of actin causing epilepsy and/or neurodevelopmental disorders (Figure 3(e)) [42, 43]. Similarly, phenotypic enrichment analysis of these genes from the Monarch database revealed several significantly enriched Human Phenotype Ontology (HPO) terms associated with various abnormalities of the nervous system, including seizure (Figure 3(f)). The enrichment analysis revealed the vital roles of ACTB in epilepsy/neurodevelopment. Variants in the ACTB gene may disrupt its interaction with these genes to cause the phenotypes of neurodevelopmental disorders and epilepsy.

3.4. Protein Modeling and Molecular Docking

Given the crucial role of beta-actin as a scaffolding protein, we evaluated the structural implications of the identified variant c.275A>G/p.N92S. This assessment was conducted by protein modeling, through three advanced artificial intelligence algorithms: the SWISS model, the Phyre2 model, and the AlphaFold model. Notably, the substitution of asparagine with serine at Residue 92 resulted in significant alterations to the hydrogen bonding patterns, calculated by all three models. Specifically, in all three models, we observed altered hydrogen bonds previously established between Residue 92 and the histidine residues at Position 88 (Figure 4(a)).

The consequence of the variant on the monomeric (G-actin) and polymeric (F-actin) forms of beta-actin was further investigated, by two representative structures of experimental verification. One of the structures is the actin-fimbrin actin-binding domain 2 (ABD2) complex (Protein Data Bank (PDB): 3byh) [44], formed by a beta-actin and a fimbrin ABD2 and representing the monomeric (G-actin) forms. Another structure is the nonmuscle F-actin decorated with nonmuscle tropomyosin complex (PDB: 7ztd) [45], formed by eight beta-actin and four nonmuscle tropomyosin 3.2 and representing the polymeric (F-actin) forms. This variant was predicted to alter the hydrogen bonds that were previously linked to the residue of fimbrin ABD2 (Figure 4(b)) but showed less impact on the formation of polymeric beta-actin (Figure 4(c)), indicating the damaging effects of molecular docking with other proteins but not self-polymerization. We thus investigated the impact of this variant in molecular docking. It is shown that the variant altered molecular docking between beta-actin and fimbrin ABD2 (Figure 4(d)), but not molecular docking of self-polymerization (Figure 4(e)). This finding underscores the impact of this variant on the structural integrity of beta-actin and its potential functional consequences, which may be evolved in the pathogenesis.

3.5. The Spatial–Temporal Expression of ACTB

TLE is generally intractable epilepsy with onset age mostly in adults [1], while the patient of this study had the age of onset in juvenile and achieved seizure-free status in adulthood. The spatiotemporal expression stands as one of the pivotal characteristics of genes, intricately linked to the phenotypes they cause. The spatiotemporal expression of ACTB was thus investigated. The ACTB gene is ubiquitously expressed in multiple adult tissues, including the brain cortex, temporal lobe, hippocampus, and other brain regions, showed by the data obtained from the GTEx database (Figure 5(a)). ACTB showed fluctuating expression at different development stages, exhibiting the highest expression in juveniles, according to the NCBI-UniGene database (Figure 5(b)). According to the specific brain expression database BrainSpan, in most brain regions, the expression of ACTB is highest in fetal but decreased after birth with minimum expression at approximately 4 years old, exhibiting a development-dependent feature (Figure 5(c)). A slightly increased expression of ACTB was observed in juveniles. Further investigation of regions associated with the phenotypes of this study showed that the second peak of ACTB expression in the temporal lobe and anterior cingulate cortex was in the stage of juvenile but decreased in adulthood (Figures 5(d), 5(e), and 5(f)), consisting of the phenotypic feature of this patient. The second peak of ACTB expression in the hippocampus is in the stage of later childhood (Figure 5(g)), which may be the explanation for bilateral hippocampal sclerosis detected in juveniles. Overall, the expression of ACTB consisted of the clinical and imaging features of this patient.