Integrative Analysis Revealed Common Genes, Brain Cell-Type and Pathways Between Stiripentol Targets and Risk Genes of Dravet Syndrome and Epilepsy

Stiripentol is an anti-epileptic drug used for treating Dravet syndrome and epilepsy. To explore common molecular mechanism between antiepileptic effect of stiripentol and genetic etiology of Dravet syndrome and epilepsy, we retrieved target genes of stiripentol through DrugBank database, as well as risk genes of Dravet syndrome and epilepsy from related Database and literature research. Then we performed genetic overlap analysis, Expression Weighted Cell type Enrichment (EWCE) analysis based on single-cell RNA-sequencing (scRNA-seq) data of brain, as well as pathway enrichment analysis. A total of 23, 19 and 118 genes were retrieved for stiripentol targets, risk genes of Dravet syndrome and epilepsy respectively. For stiripentol targets and risk genes of Dravet syndrome, three genes (GABRA1, GABRB3 and GABRG2) were overlapped with P-value of 1.265×10 −6 ; hippocampal CA1 pyramidal cells and interneurons were common brain cell types that were signicantly enriched by EWCE; and 10 common pathways were identied. For stiripentol targets and risk genes of epilepsy, ve genes (GABRA1, GABRA2, GABRB2, GABRB3, and GABRG2) were overlapped with P-value of 1.963 × 10 −7 ; hippocampal CA1 pyramidal cells and interneurons were also common brain cell types that were signicantly enriched and 22 common pathways were identied. Our results revealed that stiripentol might exert its anti-epileptic effect by regulating GABA A receptors on hippocampal CA1 pyramidal cells and interneurons.


Introduction
Stiripentol is an antiepileptic drug with a unique structure, which is currently approved in Europe, Japan, and Canada in combination with clobazam and sodium valproate for treating refractory seizures associated with Dravet syndrome. For Dravet syndrome, the evidence for its e cacy is the strongest. Mostly, adding stiripentol to clobazam and sodium valproate can reduce the frequency and severity of seizures. The small case series also shows the bene ts of malignant migratory partial seizures in infancy, super-refractory status epilepticus, and stubborn focal epilepsy, although these diseases require larger prospective studies [1,2].
Epilepsy is a collective term for a group of syndromes caused by abnormal discharges of the nervous system, and can cause varying degrees of damage to behavior, cognition, and memory. Previous studies have shown that epilepsy genetic factors play a major role in the pathogenesis of epilepsy. During these decades, genetic association studies, genome wide association studies, as well as researches using next generation sequencing technology have identi ed several genetic variations of genes such as sodium channel, potassium channel and GABA A receptor that were clearly related to a variety of epileptic phenotypes [3]. Similarly, Dravet syndrome is a case in epileptic encephalopathy caused by genetic variation. Dravet syndrome is a severe myoclonic epilepsy characterized by the onset of prolonged febrile and afebrile seizures in infancy, and behavioral, cognitive, and movement disorders appear after approximately 2 years old. Most cases are currently deemed to be caused by pathogenic variants in the sodium channel gene SCN1A, but numerous other genes have also been implicated [4]. Genes that have been recorded to cause DS-like phenotypes include SCN2A, SCN8A, SCN9A, SCN1B, PCDH19, GABRA1, GABRG2, STXBP1, HCN1, CHD2, and KCNA2 [5].
As hundreds of risk genes have been identi ed in both epilepsy and Dravet syndrome, their regulating roles on brain cells were worthy of being explored. As gene expression is heterogeneous among various cell types in brain tissue, a more accurate understanding of transcriptome in a single cell is essential to clarify their role in cellular function and to understand how gene expression promotes bene cial or harmful states [6]. Emerging advances using single-cell RNA sequencing (scRNA-seq) in the central nervous system (CNS) have already begun to provide exciting molecular insights into the complexity of the brain by identifying novel cellular subtypes based on transcriptional pro les as well as possible disease-relevant mechanisms [7]. By applying knowledge of the cellular taxonomy of the brain from single-cell RNA sequencing, previous researchers have evaluated whether the genomic loci implicated in schizophrenia map onto speci c brain cell types [8].
For stiripentol, the molecular and cellular mechanisms of its antiepileptic effect are still not fully understood, therefore, in our research, we explore the molecular mechanism and cellular location of the target genes of stiripentol from a multi-omics perspective and investigate common molecular mechanism between antiepileptic effect of stiripentol and genetic etiology of Dravet syndrome and epilepsy. Our analysis may provide insights for molecular mechanism of anti-epileptic of stiripentol on Dravet syndrome and epilepsy.

Materials And Methods
Identi cation of stiripentol targets DrugBank (https://go.drugbank.com/) is a web-based database containing comprehensive molecular information about drugs, their mechanisms of action, interactions and their targets [9]. To obtain target genes of stiripentol, we searched DrugBank database (Version 5.0) with the searching term "stiripentol", and all types of targets listed with supported publications in stiripentol were collected.

Identi cation of risk genes of Dravet syndrome and epilepsy
As stiripentol is used for treatment of both Dravet syndrome and epilepsy, we retrieved risk genes of Dravet syndrome and epilepsy respectively. Genes with convincing evidences from both comprehensive literature search and disease-gene related databases (OMIM Database (https://omim.org) [12], DisGeNET Database (http://www.disgenet.org) [13] and MalaCards Database (http://www.malacards.org/) [14]) were considered as risk genes. Therefore, we retrieved Dravet syndrome risk genes from genetic studies or reviews with convincing evidences [5,10,11]; meanwhile, we obtained risk genes of epilepsy identi ed by both common variants and rare variants, of which common variants are extracted from the largest genome-wide association study currently [15], which included 15,212 individuals with epilepsy and 29,677 controls and identi ed 16 genome-wide signi cant loci with P-value<5.0×10 -8 . Rare variants were identi ed by exome-sequencing under the largest sample size with 1165 cases and 3877 controls and their mapped genes were considered as monogenic epilepsy genes [15,16].

Genetic overlap of stiripentol targets with risk genes of Dravet syndrome and epilepsy
To investigate whether there was genetic overlap between stiripentol targets and risk gene of Dravet syndrome and epilepsy, we analyzed the overlap genes between the stiripentol targets and Dravet syndrome genes, and the overlap genes between the stiripentol targets and epilepsy genes, then Fisher's exact test were performed through R language ,using whole human genome (~20,000 genes) as background. P-value < 0.05 was considered as a statistical signi cance.
Expression Weighted Cell type Enrichment analysis of stiripentol targets, risk genes of Dravet syndrome and epilepsy To explore whether target genes of stiripentol, risk genes of Dravet syndrome and epilepsy could map on speci c brain cell types, we implemented Expression Weighted Cell type Enrichment (EWCE) method, which used single-cell transcriptome dataset to calculate whether the average expression levels of input gene list was signi cantly stronger than that in randomly generated gene list with the same size as input in each annotated cell type [17].
Moreover, we utilized a superset of brain scRNA-seq data from the Karolinska Institutet (KI), which included a total of 9,970 cells annotated with 24 cell types from mouse brain regions of neocortex, hippocampus, hypothalamus, striatum and midbrain, as well as samples enriched for oligodendrocytes, dopaminergic neurons and cortical parvalbuminergic interneurons [8,[18][19][20]. Taking into account the existence of species differences, we validated the results in another human single-nucleus RNA sequencing dataset from the Allen Institute for Brain Science (AIBS), which included a total of 4,401 cells with 6 cell types from human mid-temporal cortex [8]. Since we used the mouse single-cell transcriptome sequencing data set as the background gene set, we rst converted the human stiripentol target genes into mouse gene form, then we perform EWCE to calculate that signi cance of expression enrichment for stiripentol target genes in each brain cell type with 100,000 permutations and P-value <0.05 and false discovery rate (FDR) <0.2 was considered as signi cance. R Package EWCE was used to perform the analysis and ggplot2 was used to generate graphs [21]. Then common brain cell-type between stiripentol targets and risk genes of Dravet syndrome and epilepsy were obtained.

Pathway enrichment analysis of stiripentol targets , Dravet syndrome and epilepsy
We used DAVID (https://david.ncifcrf.gov/) to perform functional enrichment analysis for stiripentol targets, risk genes of Dravet syndrome and epilepsy [22], in which GO functional annotation and KEGG annotation were used and Hypergeometric test was performed, with false discovery rate (FDR) < 0.05 as signi cance. Then common pathways between stiripentol targets and risk genes of Dravet syndrome and epilepsy were obtained.

Identi cation of stiripentol targets
After searching in DrugBank, we retrieved a total of 23 target genes for stiripentol, of which 16 were GABA A receptor subunits, 2 were lactate dehydrogenase and 5 were Cytochrome P450. The targets of stiripentol were shown in Table 1.

Identi cation of risk genes of Dravet syndrome and Epilepsy
Through literature and database searches, risk genes of Dravet syndrome and epilepsy were obtained, and the results are listed in Supplementary Table 1 and Supplementary Table 2. A total of 19 Dravet syndrome risk genes and 118 epilepsy risk genes have been summarized.
Genetic overlap of stiripentol targets with risk genes of Dravet syndrome and epilepsy Among 23 target genes of stiripentol, there are 3 overlapping genes of GABA A receptor subunits between the targets of stiripentol and risk genes of Dravet syndrome, which were GABRA1, GABRB3 and GABRG2 genes and its Fisher's exact test P-value was 1.265×10 −6 ; meanwhile, there were 5 genes of GABA A receptor subunits (GABRA1, GABRA2, GABRB2, GABRB3, and GABRG2) overlapped with 118 risk genes of epilepsy. By Fisher's exact test, the total overlap P-value achieved 1.963 × 10 −7 . Common brain cell-type enrichment between stiripentol targets and risk genes of Dravet syndrome and epilepsy To evaluated whether expressions of stiripentol targets and risk genes of Dravet syndrome and epilepsy were signi cantly enriched in speci c brain cell types, we performed EWCE in mouse brain scRNA-seq of Karolinska Institute (KI) dataset and human brain scRNA-seq of the Allen Institute for Brain Science (AIBS) dataset [8,[18][19][20].
For KI dataset, among 24 cell types, stiripentol target genes, risk genes of Dravet syndrome and epilepsy were signi cantly enriched in six, three and ve brain cell types respectively, with P-value < 0.05 and FDR < 0.2 (Figure1, Supplementary Table 3). Among these signi cant cell types, hippocampal CA1 pyramidal cells, interneurons, and somatosensory pyramidal cells were common brain cell types that were enriched for stiripentol target genes, risk genes of Dravet syndrome and epilepsy. Besides, striatal medium spiny neurons were also common in stiripentol target genes and risk genes of epilepsy. For AIBS dataset, among 6 cell types, stiripentol target genes, risk genes of Dravet syndrome and epilepsy were all signi cantly enriched in interneurons (GABAergic) and pyramidal neurons (Glutamatergic), with P-value < 0.05 and FDR < 0.2 (Figure2, Supplementary Table 4).

Common pathways between stiripentol targets and risk genes of Dravet syndrome and epilepsy
By pathway enrichment analysis of target genes of stiripentol, risk genes of Dravet syndrome and epilepsy, a total of 44, 23 and 87 pathways were signi cantly enriched respectively (Supplementary Table 5-7). Among these pathways, we identi ed 10 signi cantly pathways that were common between stiripentol targets and risk genes of Dravet syndrome, which were involved in cellular response to histamine, GABA A receptor complex, plasma membrane, chloride channel complex, GABA A receptor activity and extracellular ligand-gated ion channel activity for GO terms, as well as nicotine addiction, GABAergic synapse, morphine addiction and retrograde endocannabinoid signaling for KEGG. Meanwhile, we identi ed 22 signi cantly pathways that were common between stiripentol targets and risk genes of epilepsy, which were involved in gamma-aminobutyric acid signaling pathway, ion transmembrane transport, cellular response to histamine, GABAergic synaptic transmission for GO terms, as well as nicotine addiction, GABAergic synapse, morphine addiction, retrograde endocannabinoid signaling and neuroactive ligand-receptor interaction for KEGG (Table2, Supplementary Table 8).

Discussion
Stiripentol, as a novel antiepileptic drug, it was originally used in Dravet syndrome [23]. When stiripentol was used in combination with valproic acid and clobazam, it could signi cantly reduce the frequency of seizures and the incidence of status epilepticus in patients with Dravet syndrome, and protect the brain nerve injury of the patients [24]. Meanwhile, stiripentol also has e cacy in refractory focal seizures, particularly when combined with carbamazepine, and in super-refractory status epilepticus [2]. However, there is still a lack of research on the antiepileptic mechanism of stiripentol, we carried out an integrated analysis based on bioinformatics methods for providing a better idea for the study of the antiepileptic mechanism of stiripentol.
In our research, we have identi ed the target genes of stiripentol and the risk genes of Dravet syndrome and epilepsy through literature search and database search. Next, we used the hypergeometric distribution test method to determine the signi cant overlap between the stiripentol targets and the Dravet risk gene or the epilepsy risk gene. As a result, we observed 5 target genes (GABRA1, GABRA2, GABRB2, GABRB3, and GABRG2) were signi cantly overlapped with 118 epilepsy risk genes, and 3 target genes (GABRA1, GABRB3 and GABRG2) were signi cantly overlapped with 19 Dravet syndrome risk genes. Our results suggested stiripentol might exert its anticonvulsant effect by regulating these genes. For epilepsy, previous studies have reported that SNPs on GABRA2 [25], GABRG2 [26], GABRR2 [27] were involved in risk of epilepsy. Meanwhile, for rare variants, it was also reported that mutations in GABRA1, GABRA6, GABRB3, GABRG2, and GABRD could cause various forms of epilepsy [28,29]. For Dravet syndrome, GABRA1 [30], GABRB3 [10], GABRG2 [31] gene mutations occupy an important role in its genomic variation. It is the genomic variation of GABA A receptor caused by various causes that leads to the dysfunction of the GABA A receptor so that it can not be bind well with GABA ligand and reduce the inhibitory effect. As an agonist of the GABA A receptor, stiripentol strengthens the inhibitory effect of GABA and plays an antiepileptic mechanism. Quilichini et al also proved this point that stiripentol can increase GABA A receptormediated signal transduction by patch-clamp methodology, which is mainly achieved by increasing the frequency of micro-inhibitory postsynaptic current and prolonging decay-time constant [32].
Brain cell types enrichment analysis from two independent scRNA-seq data of brain cells both demonstrated hippocampal CA1 pyramidal cells and interneurons were common brain cell-type enriched by target genes of stiripentol, risk genes of Dravet syndrome and epilepsy. Increasing evidence showed that epilepsy was due to an imbalance between neuronal excitation and inhibition [33]. The normal function of the cerebral cortex depends on two types of neurons: a) excitatory, pyramidal neurons, using glutamate as their neurotransmitter; b) inhibitory local-circuit interneurons, using GABA as a neurotransmitter [34]. Excitatory pyramidal cells mainly releasing glutamate and inhibitory interneurons mainly releasing GABA work together to maintain the excitability of nerve cells [35]. In our research, by analyzing Dravet syndrome susceptibility genes and epilepsy susceptibility genes, we found that they were both enriched in interneurons and pyramidal cells, which supported that the imbalance of excitation and inhibition played an important role in the pathogenesis of epilepsy from a genetic perspective. In a case-control study, spontaneous excitatory postsynaptic currents(sEPSCs) was found in 42% and 62% of hippocampal CA1 pyramidal cells in the control group and kainate-induced epilepsy group respectively, and the frequency of sEPSCs in the kainate-induced epilepsy group was signi cantly higher than that in the control group [36]. These ndings strongly proved that there was a signi cant correlation between abnormal discharges of pyramidal neurons in the CA1 region and epilepsy. Of course, interneurons also play an important role in the pathogenesis of epilepsy [37]. As hippocampal CA1 pyramidal cells and interneurons were also signi cantly enriched by target genes of stiripentol , our results indicated that stiripentol may exert its antiepileptic effect by acting on interneurons or hippocampal CA1 pyramidal cells.
Pathway enrichment analysis also identi ed that GABA A receptor activity and multi-substance addiction related pathways were common pathways between stiripentol targets and risk genes of Dravet syndrome and epilepsy.
Previous experimental studies have also proved that the GABA A receptor is linked to nicotine addiction and alcohol addiction, which may involve the mechanism of reward and punishment [38]. GABA A receptors are heteromeric GABA-gated chloride channels. The transmembrane ion channel is opened pursuant to a stimulus generated by GABA, which allows an in ux of chloride ions. This results in a decrease of the depolarizing effects of excitatory input, depressing excitability [39]. Therefore, we imply that stiripentol might exert its antiepileptic mechanism mainly by heightening the inhibitory effect of GABA. Previous study reported that the relative expression levels of two inhibitory neurotransmitter receptors GABRA1 and GABRA2 in CA1 neurons were markedly higher than those in CA3 neurons [40]. The pyramidal cells in the CA1 region of the hippocampus receive 92% of the GABAergic input into the dendrites. GABAergic input mainly comes from local interneurons, which promote the formation of cell aggregates by regulating the activity of pyramidal cells. Local GABAergic interneurons control the ring frequency of pyramidal cells, adjust their spike timing, and synchronize their activities [41]. Based on previous research results and our results, we speculate that the target genes of stiripentol may still function mainly by acting on pyramidal cells in the hippocampal CA1 area, and this effect may be maintained mainly by binding to GABA A receptors.
In our study, we explored common molecular mechanism between antiepileptic effect of stiripentol and genetic etiology of Dravet syndrome and epilepsy by systematic data collection and integrative analysis. However, there are still some limitations of our analysis. First, since the number of cells taken in the single-cell data sets accounts for only a small portion of the whole brain tissue, they may not represent all types of brain cells. Second, our results are obtained from public databases based on bioinformatics, which might only reveal underlying mechanisms with currently existing information and needs further experimental validation.

Conclusions
Based on our analysis, we identi ed there were common genes, brain cell-type and pathways between target genes of stiripentol and genetic basis of epilepsy, indicating antiepileptic mechanism of stiripentol may be achieved by acting on the GABA A receptors in the pyramidal cells of the CA1 region and interneurons.