Based on a clinical course of viral infection, fever, status epilepticus (early seizure), clustering focal seizures (late seizure), and cortical neuronal damage, AESD may be regarded as another syndrome of “acute encephalopathy with inflammation-mediated status epilepticus” 16. Previous MR spectroscopic studies demonstrated the pathogenetic role of glutamate in cerebral cortical lesions after initial status epilepticus 17. The appearance of characteristic lesions on MRI is as late as around the late seizure, delaying the diagnosis of AESD 18. To enable an early diagnosis immediately after its onset, candidate biomarkers need to be identified.
Candidate gene analyses in Japan previously identified several susceptibility genes for AESD, such as common variants of carnitine palmitoyltransferase 2 (CPT2) 19, 20 and adenosine A2A receptor (ADORA2A) 21, and rare variants of the sodium voltage-gated channel alpha subunit 1 (SCN1A) and SCN2A 22, 23. However, AESD has not yet been studied using a genome-wide approach.
In the present study, GWAS for AESD patients identified 7 candidate loci reaching the genome-wide suggestive level. Among the 8 representative variants in these susceptibility loci, 3 SNPs, rs1850440, rs12656207, and rs60651483, showed odds ratios of the same direction between the GWAS and replication study. Regarding the first SNP, rs1850440, we found its location in the enhancer region of the STK39 gene, and its regulation of STK39 expression using the RegulomeDB database. cis-eQTL revealed a relationship between disease-risk allele T and the stronger expression of STK39 in peripheral blood. STK39 encodes a serine/threonine kinase mediating cellular stress-activated signals 24. STK39 is widely expressed in the brain, including the cerebral cortex in GTEx. In response to hypotonic stress with cell swelling, STK39 is activated and phosphorylates several cation–chloride cotransporters (CCCs). Based on the important roles of CCCs in the regulation of ion and water homeostasis in the mammalian brain, STK39 has been implicated in cerebral edema 25. STK39 activates the p38 mitogen-activated protein kinase (MAPK) pathway. A previous study reported that heat stress triggered the activation of p38 MAPK, leading to an increase in reactive oxygen species and the apoptosis of glial cells 26. Proinflammatory cytokines, such as interleukin-1 and tumor necrosis factor-alpha (TNF-α), also activate the p38 MAPK pathway and induce cellular apoptosis. Status epilepticus up-regulates the expression of these cytokines in brain astrocytes and microglial cells 27. Therefore, we speculate that the rs1850440-associated strong expression of STK39 predisposes children to AESD because the onset of AESD is preceded by a high fever and status epilepticus.
Regarding the second disease risk SNP, rs12656207, our single-tissue eQTL analysis revealed disease-risk allele G correlated with higher expression levels of FBXO38 in the blood. FBXO38, a ubiquitin ligase of programmed cell death 1(PD-1), is a negative regulator of T cell-mediated immunity 28, 29. The expression of PD-1 is up-regulated during acute viral infection 30, a triggering factor of AESD. On the other hand, the third SNP, rs60651483, had a protective allele T for AESD. In the eQTL analysis, the T allele of rs60651483 correlated with the weaker expression of GIPC3 in the blood. GIPC3, a PDZ domain protein, belongs to the GIPC family, which regulates a number of cellular processes, such as proliferation, planar cell polarity, cytokinesis, and migration 31. Mutations in GIPC3 have previously been reported in sensorineural hearing loss and audiogenic seizures 32. The potential involvement of FBXO38 and GIPC3 in AESD warrants further study.
Using GWAS summary statistics, we conducted a miRNA enrichment analysis to identify miRNA and miRNA-target gene networks associated with AESD, which may provide additional insights into its pathogenesis as well as candidate biomarkers for an early diagnosis 8, 33. In the present study, we obtained 3 candidate miRNAs, hsa-mir-34c, hsa-mir-449b, and hsa-mir-449c, with high confident annotation in miRBase. These miRNAs belong to the mir-34/449 family, have similar sequences to each other, and are reportedly involved in immune responses and viral infections 34. For example, hsa-mir-34c is expressed in human peripheral blood mononuclear cells following inflammation-associated endogenous damage 35. Previous in vitro studies demonstrated that hsa-mir-34c derived from astrocyte exosomes exerted neuroprotective effects against cerebral ischemia-reperfusion injury by down-regulating the MAPK pathway 36. On the other hand, hsa-mir-449b enhanced the activation of the interferon-β promoter induced by influenza A virus infection 37. Therefore, we speculated that febrile status epilepticus caused by viral infection may provoke immune responses and up-regulate the expression of hsa-mir-34c and hsa-mir-449b, thereby inducing proinflammatory cytokines in AESD patients. The mir-34/449 family plays an essential role in the brain, especially in the development of forebrain, which is implicated in reward pathways, feeding, and social behaviors 38. As the target gene of hsa-mir-449b, our miRNA analysis detected the ASB4 gene encoding ankyrin repeat and suppressor of cytokine signaling box containing 4 (ASB4), which plays a role in proinflammatory responses up-regulated by TNF-α in endothelial cells 39. The present results implicate these miRNAs of the mir-34/449 family, as well as the target gene ASB4, in the pathogenesis of AESD. Since they are all detectable in peripheral blood mononuclear cells, they have potential as biomarkers for the diagnosis of AESD.
There are several limitations in the present study. Firstly, due to the low incidence of AESD, the sample size was too small to find a locus of genome-wide significance and confirm reproducibility between the GWAS and replication study. The expected power for our GWAS was up to 42% at the genome-wide significant threshold under the additive model, assuming a genotype relative risk ranging between 1.7 and 2.0 and disease allele frequency of higher than 40%. GWAS achieved 84.3% to detect common alleles with a minor allele frequency ≥ 5%, genotype relative risk > 2.0, and disease allele frequency > 40% at a significant p-value threshold of 5.0 × 10− 8 under the additive model when the number of cases was more than 450 (Supplementary Fig. S3). Secondly, the present study did not replicate previous findings on the susceptibility loci of AESD using a candidate gene approach (Supplementary Table S5) 19–21. The reason for this discrepancy may be the small effect sizes of the variants reported previously and the insufficient sample size of the present study. Nevertheless, by using genome-wide approach, the present study revealed the pathogenetic roles of common genetic variants in AESD, a rare disease, as had previously been shown for other rare neurodevelopmental disorders formerly considered to be monogenic 40. Thirdly, the present study could have detected variants and miRNA-target gene networks of febrile status epilepticus rather than those of AESD because most of the AESD cases have febrile status epilepticus at the onset. However, none of the SNPs and miRNAs found in this study have ever been described in previous studies on the genetic predisposition of febrile seizures. To directly address this question, another study using disease controls of febrile seizures is warranted.
In the present study, despite these limitations, we reported 3 variants with a suggestive association with AESD, including rs1850440 in the STK39 gene. By integrating GWAS summary statistics and miRNA prediction software, we found the enrichment of GWAS signals on the networks of miRNAs and its target genes. These results may provide additional insights into the pathophysiology, earlier diagnosis, and better treatment of AESD.