Network and Pathway-Based Analysis of Single-Nucleotide Polymorphism of Mirna in Temporal Lobe Epilepsy

Molecular Neurobiology (2019) 56:7022–7031 https://doi.org/10.1007/s12035-019-1584-4

Network and Pathway-Based Analysis of Single-Nucleotide Polymorphism of miRNA in Temporal Lobe Epilepsy

Wenbiao Xiao1 & Yanhao Wu2 & Jianjian Wang3 & Zhaohui Luo1 & Lili Long1 & Na Deng1 & Shangwei Ning4 & Yi Zeng5 & Hongyu Long1 & Bo Xiao1

Received: 18 September 2018 /Accepted: 21 March 2019 /Published online: 9 April 2019 # Springer Science+Business Media, LLC, part of Springer Nature 2019

Abstract Temporal lobe epilepsy (TLE) is a complex disease with its pathogenetic mechanism still unclear. Single-nucleotide polymorphisms (SNPs) of miRNA (miRSNPs) are SNPs located on miRNA genes or target sites of miRNAs, which have been proved to be associated with neuropsychic disease development by interfering with miRNA-mediated regulatory function. In this study, we integrated TLE–related risk genes and risk pathways multi-dimensionally based on public data resources. Furthermore, we systematically screened candidate functional miRSNPs for TLE and constructed a TLE-associated pathway-based miRSNP switching network, which included 92 miRNAs that target 12 TLE risk pathways. Moreover, we dissected thoroughly the correlation between 5 risk genes of 4 risk pathways and TLE development. Additionally, the biological function of several candidate miRSNPs were validated by luciferase reporter assay. In silico approach facilitates to select potential BmiRSNP- miRNA-risk gene-pathway^ axis for experimental validation, which provided new insights into the mechanism of miRSNPs as potential genetic risk factors of TLE.

Keywords Temporal lobe epilepsy (TLE) . Risk gene . miRSNP . Pathway . Network

Introduction necessitating the evaluation for surgical resection of the epilep- tic focus [2]. As a complex disease, the contribution of genetic Temporal lobe epilepsy (TLE) is the most common form of susceptibility to TLE has been widely studied, and a number of focal epilepsy and often refractory to antiepileptic drugs [1], single-nucleotide polymorphisms (SNPs) have been supposed to be implicated in the pathogenesis mechanism of TLE [3–6]. Although previous studies have provided insights into the sig- Electronic supplementary material The online version of this article (https://doi.org/10.1007/s12035-019-1584-4) contains supplementary nificant impacts of genetic factors on TLE, the molecular mech- material, which is available to authorized users. anism underlying TLE remains largely indeterminate. A genome-wide association study (GWAS) has become a power- * Hongyu Long ful tool to study the genetic architecture of complex diseases [email protected] and had uncovered tens of thousands of disease-associated * Bo Xiao SNPs [7, 8]. However, the majority of identified SNPs locate [email protected] in the non-coding regions of the genome [9], which remains a challenge to explain the biological significance. 1 Department of Neurology, Xiangya Hospital, Central South MicroRNAs (miRNAs) are a class of endogenous small University, Changsha 410008, China non-coding RNAs (~ 22 nucleotides) that regulate gene ex- 2 Department of Respiratory Medicine, Xiangya Hospital, Central pression at the post-transcriptional level by translational re- South University, Changsha 410008, China pression and/or mRNA deadenylation/decay [10, 11]. 3 Department of Neurology, the Second Affiliated Hospital, Harbin Recently, expression profiling studies in both animal models Medical University, Harbin 150081, China and human epilepsy revealed select changes to brain miRNA 4 College of Bioinformatics Science and Technology, Harbin Medical that mainly impact inflammation, neuronal excitation, and ap- University, Harbin 150081, China optosis, which suggest that miRNAs may regulate certain key 5 Department of Geriatrics, Second Xiangya Hospital, Central South processes broadly altering pathophysiology in epilepsy [12, University, Changsha 410011, China Mol Neurobiol (2019) 56:7022–7031 7023

13]. Moreover, modulating individual miRNA can reduce the abstract was read carefully, and the risk genes which met occurrence of spontaneous seizures and mitigate the attendant the following criteria were recorded: (i) the number of TLE pathological features (neuron loss, gliosis, and rearrangement samples is greater than 5 (mainly includes temporal lobe of mossy fibers) in a TLE model; therefore, miRNAs are novel samples), (ii) there were significant difference between potential therapeutic targets for the treatment of epilepsy [14]. TLE patient samples and the controls in expression levels MiRNAs elicit their functions mainly through sequence- (mRNA level or protein level), (iii) genes in possession of specific binding within the 3′UTR of mRNA transcripts [10, SNPs that prominently associated with TLE patients or 15], so it is believable that miRNA-associated SNP (miRSNP) subgroups. All of the identified risk genes were experimen- could affect miRNA regulation. According to their locations, tally verified by reliable biological methods (PCR, Western miRSNPs have been classified into SNPs within miRNA blotting, et al.) and statistically significant. genes and miRNA target sites [16]. The evidence is mounting that miRSNPs play a vital role in diseases of the central nervous system (CNS) including Alzheimer’s disease (AD), Function Enrichment Analysis of TLE Risk Genes Parkinson’s disease (PD), and multiple sclerosis [17]. For in- stance, a functional polymorphism rs1056628 at miR-491-5p To figure out the potential functions of TLE risk genes, we binding site in the 3′UTR of MMP-9 gene confers increased carried out Kyoto Encyclopedia of Genes and Genomes MMP-9 protein expression, which then increased the risk for (KEGG) [23] pathway enrichment analysis by applying func- atherosclerotic cerebral infarction in a Chinese population tional annotation tools in DAVID Bioinformatics Resources [18]. SNP rs662702 of miRNA-328 binding site in the 3′ 6.7 (http://david.abcc.ncifcrf.gov/)[24]. In addition, we UTR of PAX6, which is known to result in increased PAX6 employed DAVID to describe the gene ontology (GO) anno- expression, conferred the increased risk of centrotemporal tation for TLE risk genes [25]. Adjusted P using the spikes of Rolandic epilepsy [19]. A functional rs57095329 Benjamini and Hochberg false discovery rate (FDR) < 0.01 of MIR146A gene is associated with susceptibility to drug- was set as the threshold of significance for KEGG pathway resistant epilepsy and seizure frequency [20]. The miR-124 and GO term. rs531564 polymorphism has no major role in genetic susceptibility to mesial TLE in an Italian sample [21], nor does the rs2910164 variant in MIR146A gene [22]. However, to date, MiRNAandmiRNATargetGeneData connecting these miRSNPs to specific genes or to molecular pathways that may be implicated in TLE has not been We downloaded annotation files of human miRNA informa- elaborated. tion from miRBase (http://www.mirbase.org/)[26]. Human In this study, we conducted a bioinformatic genome-wide miRNA target gene data were obtained from 10 miRNA survey of human miRSNPs in miRNA target sites and miRNA target predicting tools, namely RNA22, mirSVR, DIANA- genes themselves and systematically identified candidate microT, PicTar5, RNAhybrid, TargetScan, PITA, functional miRSNPs for TLE based on integrating TLE- MirTarget2, TargetMiner, and miRanda. The target gene related risk genes and risk pathways multi-dimensionally. assemblage of every miRNA was picked out when miRNA Our study helps to unravel miRSNPs that affect the miRNA- target gene pairs were predicted by at least four tools. KEGG mRNA interaction and study the etiology and pathogenesis of pathway enrichment analysis was used in the identification of TLE. In addition, clues for the follow-up studies and function- pathway, in which target genes of individual miRNA were al verification experiments were provided, which contribute to significantly enriched. building the evidence for novel therapeutic targets.

MiRSNP Data Materials and Methods The miRSNPs within miRNA target sites were obtained Human TLE Risk Gene Collection from five databases, namely PolymiRTS Database (v.3.0), miRdSNP (v.11.03), MirSNP, miRNASNP (v.2.0), SNP Risk gene data were acquired by browsing the GeneCards effects on microRNA targeting. The miRSNPs within databases (https://www.genecards.org/). Existing studies miRNA genes were obtained from miRvar (v.2.0 Build that focused on genes and TLE were reviewed from 22), miRNASNP (v.2.0), PolymiRTS Database (v.3.0), PubMed (http://www.ncbi.nlm.nih.gov/pubmed)by miRNA-SNiPer. As to the two kinds of miRSNPs, we manually exploring literature published before May 1, screened for the candidate miRSNPs that may potentially 2017 using the terms B(epilepsy, temporal lobe [MeSH impact miRNA-mRNA interactions predicted by at least Terms]) AND English [Language]^.Then,thefulltextor two databases. 7024 Mol Neurobiol (2019) 56:7022–7031

Cumulative Hypergeometric Distribution (candidate SNP, miRNA, and target gene prime were shown in Table S1). All of the constructs were confirmed by Sanger The significant correlations between TLE risk pathways were sequencing. For reporter assays, the 293T cell was co- identified by performing a cumulative hypergeometric test transfected with wild-type (or mutant) reporter plasmid and [27]. The P value was computed using the following formula: miR-Ribo™ mimics (or miR-Ribo™ negative control) using Lipofectamine 3000 (Invitrogen). Luciferase activity was j m−j measured 48 h post-transfection using the dual-luciferase re- x i n−i P ¼ ∑ m porter system (Promega, Madison, WI, USA) and a Wallac i¼0 Victor Luminometer. Firefly luciferase units were normalized n against Renilla luciferase units to control for transfection effi- where m denotes the number of whole human genome genes, j ciency. Student’s t test was used for statistical analysis of dif- and n represent the number of genes in the two given pathway ferences and results were considered statistically significant at respectively, and x denotes the number of genes overlapping P value < 0.05. between the two given pathways. Adjusted P values using the Benjamini and Hochberg FDR < 0.01 were considered to be significant. Results

Luciferase Reporter Assay Manually Compiled Human TLE Risk Gene Set

To confirm the hypothesis that miRSNP could affect By means of known database searching and manual litera- miRNAs’ regulatory function, the dual-luciferase reporter as- ture mining, 157 TLE risk genes were screened (detailed say was used to determine whether miRNA targets directly the information outlined in Table S2). The risk genes were 3′UTR of candidate gene and miRSNP interferes with the verified by classical experimental methods such as PCR, miRNA-mRNA interaction in human-type cells. The 3′UTR WB, immunohistochemistry, immunofluorescence, et al. of target genes were amplified from human genomic DNA The GO annotation of TLE risk gene set was significantly and individually inserted into the carrier vector pmiR-RB- enriched into the following categories: glutamate receptor REPORT™ (Ribobio, Guangzhou, China) using the XhoI activity, transmission of nerve impulse, and regulation of and NotI sites. Similarly, the fragments of 3′UTR mutant were apoptosis (Table S3), which was in accord with the existing inserted into the vector at the same sites as control group knowledge of TLE pathogenesis.

Table 1 Enriched KEGG pathway of TLE risk genes

Term Benjamini Pathway maps hsa05014: amyotrophic lateral sclerosis (ALS) 1.05E−14 Human disease (neurodegenerative diseases) hsa04210: apoptosis 3.63E−10 Cellular processes (cell growth and death) hsa04080: neuroactive ligand-receptor interaction 1.15E−07 Environmental information processing (signaling molecules and interaction) hsa05010: Alzheimer’s disease 3.73E−07 Human disease (neurodegenerative diseases) hsa05200: pathways in cancer 1.41E−06 Human disease (cancers: overview) hsa04010: MAPK signaling pathway 3.35E−06 Environmental information processing (signal transduction) hsa05020: prion diseases 6.12E−05 Human disease (neurodegenerative diseases) hsa04720: long-term potentiation 8.08E−05 Organismal systems (nervous system) hsa05212: pancreatic cancer 1.16E−04 Human disease (cancers: specific types) hsa05220: chronic myeloid leukemia 1.47E−04 Human disease (cancers: specific types) hsa05222: small cell lung cancer 3.36E−04 Human disease (cancers: specific types) hsa04620: Toll-like receptor signaling pathway 0.00131867 Organismal systems (immune system) hsa04621: NOD-like receptor signaling pathway 0.00154441 Organismal systems (immune system) hsa05215: prostate cancer 0.00248985 Human disease (cancers: specific types) hsa04510: focal adhesion 0.00389858 Cellular processes (cellular community-eukaryotes) hsa04722: neurotrophin signaling pathway 0.00458217 Organismal systems (nervous system) hsa05210: colorectal cancer 0.00756884 Human disease (cancers: specific types) hsa05219: bladder cancer 0.00730862 Human disease (cancers: specific types) Mol Neurobiol (2019) 56:7022–7031 7025

Fig. 1 TLE-associated pathway crosstalk network. a Pathway crosstalk rhombuses stands for the significant correlation between two pathways. b network demonstrates the notably correlated pathways among TLE risk Barplotofpathway’s degree distribution in the pathway crosstalk pathways. The orange rhombus stands for pathway, the line between two network

TLE Risk Pathways Enrichment Analysis pathway-based network to dissect the roles of miRNAs in TLE. Two hundred and sixty-seven significant We obtained 18 TLE risk pathways (Table 1)byKEGGpath- miRNA-pathway pairs including 92 miRNAs and 12 way enrichment analysis of risk genes. About 66.9% of the risk TLE risk pathways were explored (Table S5). On account genes (103/157) were statistically related with TLE risk path- of the potential roles of miRSNPs in TLE susceptibility, a ways (FDR < 0.01), which provided an overview of TLE path- pathway-based miRSNP switching network (PMSN) was ogenesis. Moreover, the identified pathways in the KEGG da- also built to objectively demonstrate the underlying im- tabase enriched into the category of Bneurodegenerative pact of miRSNPs on TLE at the pathway level (Fig. 2a). diseases^ highlighting the fundamental characteristics of neu- We screened for candidate functional miRSNPs associated rodegenerative diseases similar to TLE and Bimmune system^ with these 92 miRNAs in seven relevant miRSNP data- highlighting the inflammation role in the pathogenesis of TLE. bases. As a consequence, miRSNPs within miRNA target Furthermore, we identified the correlation among TLE sites may disrupt the regulation of 29 miRNAs in 11 TLE risk pathways through exploring the intersection of every risk pathways and miRSNPs within miRNA genes which two pathways and sequentially established the TLE- may change sequence binding or spatial structure of 56 associated pathway crosstalk network (Fig. 1a, miRNAs. These miRSNPs could impact their regulatory Table S4). The pathway crosstalk network suggested that function to the target pathway. most of the enriched pathways were notably correlated To acquire more detailed information, we dissected with the rest part. There were 15 pathways correlated with thoroughly the PMSN. At first, we counted up the degree more than 10 other pathways. Biological pathways for the pathway (Fig. 2b)andmiRNA(Fig.2c) respec- Bhsa04010: MAPK signaling pathway,^ Bhsa04210: tively to highlight the specific miRNAs or pathways with apoptosis^ extensivelycorrelatedwith16otherpathways, top linkage. The 4 pathways Bhsa05200: pathway in and Bhsa04722: neurotrophin signaling pathway^ exten- sively correlated with 14 other pathways (Fig. 1b). Analysis of crosstalk network suggested that the TLE risk Fig. 2 The PMSN and its topological property. a The PMSN is pathways may have a synergistic role in TLE comprised of 12 risk pathways, 92 screened miRNAs, and identified pathogenesis. miRSNPs. Orange rhombus and blue circle represent pathway and miRNA respectively, and their sizes accord with their degrees. The miRSNPs within miRNA target genes are represented with red lines Constructing and Dissecting the Pathway-Based between the miRNA and the pathway which contains the target gene, miRSNP Switching Network while miRSNPs within miRNA genes are shown with red circle around the blue circle. b Bar plots of miRNAs’ degree distribution. c Bar plots of pathways’ degree distribution. d Node degree distribution of PMSN and Since the accumulating evidences hinted the involvement the fitted curve. e Betweenness centrality distribution of PMSN and the of miRNAs in epilepsy pathogenesis, we constructed a fitted curve 7026 Mol Neurobiol (2019) 56:7022–7031 Mol Neurobiol (2019) 56:7022–7031 7027 cancer,^, Bhsa04010: MAPK signaling pathway,^ The Dysregulatory Role of miRNASNP Was Bhsa04510: focal adhesion,^ Bhsa05215: prostate cancer^ Demonstrated by the Luciferase Reporter Assay were identified to higher linkage with the majority of the miRNAs, which indicated above 4 risk pathways were Using the luciferase reporter system, we found that co- more susceptibility to be regulated by the miRNAs. transfection of let-7b mimics along with the 3′UTR of Thirty-eight miRNAs regulated more than 3 TLE risk wild-type BCL2L1 caused a significant decrease in rela- pathways, and miR-424 and miR-520b had interactions tive luciferase activity compared with the negative con- with 9 TLE risk pathways (Fig. 1a). Moreover, we also trols. Moreover, compared with the co-transfected with analyzed the network topological property of the PMSN. let-7b mimics of wild-type and mutated BCL2L1, the rel- The degree distribution (Fig. 2d) and the betweenness ative luciferase activity was significantly suppressed in centrality distribution (Fig. 2e) showed that the PMSN cells co-transfected with wild-type BCL2L1. In contrast, have features both of small network and scale-free net- no change in relative luciferase activity was observed in work. That is to say, hub nodes (miRNA and risk path- cells transfected with let-7b mimic groups of wild-type way) with higher degree and betweenness centrality have and mutated (Fig. 4a). HTR2A and miR-203a mimic more vital role in the PMSN. groups gave similar results (Fig. 4b). Significantly sup- Based on the existing knowledge of TLE pathogene- pression in relative luciferase activity was observed in sis,wescreened5high-riskgenesand4enrichedrisk cells transfected with miR-200c mimic or negative control pathways to in-deep dissection (Fig. 3). We supposed the groups of DNMT3A wild-type and mutated respectively. biological mechanisms of potential BmiRSNP-miRNA- Moreover, compared with the co-transfected with miR- risk gene-pathway^ axis, which hints that the miRSNPs 200c mimics of wild-type and mutated DNMT3A, the might perturb the TLE pathway by affecting miRNA’s relative luciferase activity was significantly suppressed regulation to target gene. Five high-risk genes were in cells co-transfected with wild-type DNMT3A (Fig. found to be regulated by 19 miRNAs, while 8 of them 4c). These results suggested that miRNAs target their tar- had been experimentally reported relevant to TLE patient get genes and miRSNPs impair the ability of miRNA in PubMed database (Fig. 3). Meanwhile, the identified binding. Exceptionally, miR-548d and let-7b can target 19 miRSNPs potentially influenced high-risk gene ex- BDNF and TNFRNF1B respectively, but their miRSNPs pression and function. had no influence on the ability of miRNA binding (Fig. 4d, f).

Discussion

TLE is a complex disease with its pathogenetic mechanism still unclear. In this study, we took miRSNP as the breakthrough point, integrated TLE–related risk genes, and risk pathways multi-dimensionally based on public data resources and focused on BmiRSNP-miRNA- mRNA-risk pathway^ axis. We have systematacially screened candidate functional miRSNPs and elaborated their potential mechanisms on the basis of the limited existing knowledge of TLE. For the first time, we constructed and dissected TLE-associated PMSN, which could contribute to elucidating their underlying pathogenesis in TLE from the post-transcriptional regulation level and genetic variation as well. Moreover, the dual- luciferase reporter assay was completed for further biological validation. Fig. 3 The schematic diagram of miRSNP-miRNA-gene-pathway axis. The 18 TLE risk pathways including neurodegenerative The green rectangle and blue circle represent the high-risk gene and their diseases, apoptosis, long-term potentiation (LTP), immune regulatory miRNA respectively. The lines between them indicate the system, et al. could be used to systematically characterize regulation of miRNA to genes. The switch symbol sited on the line means the pathogenesis of TLE. There is mounting evidence for the miRSNP. The large peripheral circles denote the pathways that risk genes are enriched. Orange stars beside miRNAs stand for literature re- the direct involvement of mitochondrial dysfunction in ported miRNAs that are relevant to TLE mesial TLE and neurodegenerative disorders, such as PD 7028 Mol Neurobiol (2019) 56:7022–7031

Fig. 4 The luciferase reporter assay. Relative luciferase activity of reporters, including WT or mutant target gene 3′UTR co-transfected with NC or miRNA mimics. Asterisk indicates significant difference at P < 0.05. NS no significance

and ALS [28]. Pathological features common to AD and were reported to be significantly dysregulated in TLE pa- TLE include hippocampal sclerosis and tau pathology, tients. Taken together, epileptogenesis is a complex dy- which may be related with neurotoxicity, neuronal loss, namic biological process that may implicate molecular and accelerated cognitive decline [29–31]. Although these and pathological alterations. Our findings indicated that neurodegenerative disorders differ in their underlying miRNAs might result in the aberrant expression of genes causes, they share a common feature that each disorder and, subsequently, might cooperatively mediate pathway implicates in certain neuronal apoptosis or death through dysregulation, finally leading to the development and pro- specific pathway. LTP, a long-lasting increase in synaptic gression of TLE. efficacy, is the cellular and molecular basis for learning We had experimentally confirmed that SNP rs3208684 and memory [32]. Structural remodeling of synapses and (A > C) variation in 3′UTR of BCL2L1 impairs the ability of formation of aberrant synaptic contacts, similar to those let-7b binding affinity with BCL2L1, which is consistent with described for LTP, are common features in TLE [33]. the findings of MA et al. in hepatocellular carcinoma cells. Activation of the Toll-like receptor (TLR) pro- They also demonstrated SNP rs3208684 enhanced the expres- inflammatory pathway may contribute to neuronal hyper- sion of bcl2l1 protein by Western blot analysis and significant- excitability and neurotoxicity, and an increasing evidence ly decreased chemotherapy sensitivity [37]. BCL2L1, which has demonstrated that neuroinflammation might constitute belongs to the anti-apoptotic member of the Bcl-2 family, a common and vital mechanism in the pathophysiology of plays a vital role in regulating cell survival and apoptosis epilepsy and seizure [34, 35]. and was found to be overexpressed in human TLE, therefore MiRNAs mainly function as key regulators of gene conferring a survival property to neural cells [38]. Yan et al. expression at the post-transcriptional level. Since single also reported that let-7b could act as a key regulator in the miRNA can target several hundred genes, miRNA could intrinsic apoptotic pathway by demonstrating let-7b targeting regulate various biological pathways simultaneously; the of BCL2L1 [39]. Alterations of apoptosis-associated signaling same is true for pathway targeted by diverse miRNAs pathways are widely reported in patients and animal models of [36]. We selected 92 miRNAs that have the potential to TLE. Let-7b is downregulated in TLE [40]. In a rat model of regulate TLE risk pathways, and the 33 among which TLE, long non-coding RNA (lncRNA) H19 is involved in SE- Mol Neurobiol (2019) 56:7022–7031 7029

Fig. 5 Schematic drawing of miRSNPs taking effect through their roles in miRNAs’ regulation to target genes. MiRSNPs potentially interfere with molecular pathway and cell phenotype, which may be implicated in the development and progression of TLE induced neuronal damage by functioning as competing en- apoptosis [41]. It is speculated that these apoptosis-associated dogenous RNA to sponge let-7b in the regulation of cellular genes and pathways targeted by dysregulated miRNA may 7030 Mol Neurobiol (2019) 56:7022–7031 imply a dynamic balance between pro- and anti-apoptotic genetic risk factors for complex disorders and shed light on mechanisms in TLE pathogenesis. These results suggest that future investigation and validation. rs3208684 (let-7 family)-BCL2L1-apoptosis may be potential molecular pathological mechanisms of TLE and personalized Funding Information This study was supported financially by Omics- therapeutic targets (Fig. 5). based precision medicine of epilepsy being entrusted by Key Research Project of the Ministry of Science and Technology of China (No. Using bioinformatics analysis, it was predicted that the 2016YFC0904400), the National Natural Science Foundation of China 3′UTR of the DNMT3A gene contains a potential miRNA (No. 81671299 to B.X and No. 81401078 to H.Y.L), the Science and binding site for miR-200c and that SNP rs35163679 re- Technology Department Funds of Hunan Province Key Project (No. sides within this binding site. Luciferase report assays 2016JC2057 to B.X. and No. 2018JJ3822 to H.Y.L), and Independent Exploration and Innovation project for postgraduate of Central South demonstrated that miR-200c targeted DNMT3A gene ex- University (No. 2018zzts248 to W.B.X). pression and rs35163679 variation influenced the ability of miR-200c binding affinity with DNMT3A. These results Compliance with Ethical Standard suggest that SNP (rs35163679) may increase DNMT3A expression. DNMT3A is a member of the DNA methyl- Conflict of Interest The authors declare that they have no conflict of transferase (DNMT) enzyme family, which promotes de interest. novo methylation during development and regulate synaptic function in mature CNS neurons [42]. Whole-exome sequencing identified de novo DNMT3A mutations could References cause epilepsy phenotype [43], and an increased expression of Dnmt3a was found in patients with intractable 1. Tellez-Zenteno JF, Hernandez-Ronquillo L (2012) A review of the TLE [44]. Meanwhile, our previous study revealed that epidemiology of temporal lobe epilepsy. Epilepsy Res Treat 2012: the hypermethylation of mRNAs and non-coding RNAs 630853 2. Blumcke I, Thom M, Aronica E, Armstrong DD, Bartolomei F, (lncRNAs and miRNAs) gene promoters was the predom- Bernasconi A et al (2013) International consensus classification of inant difference between TLE patients and healthy controls hippocampal sclerosis in temporal lobe epilepsy: a task force report [45]. Hypermethylation of gene promoters was also the from the ILAE commission on diagnostic methods. Epilepsia 54: – major effect in rodent models of TLE, which is consistent 1315 1329 ’ 3. Gambardella A, Manna I, Labate A, Chifari R, La Russa A, Serra P with the increased expression of Dnmt3a [46]. What s et al (2003) GABA(B) receptor 1 polymorphism (G1465A) is as- more, hsa-miR-200c was expressed at higher levels in sociated with temporal lobe epilepsy. Neurology 60:560–563 TLE brain samples than controls [40], which may also 4. Lv RJ, He JS, Fu YH, Shao XQ, Wu LW, Lu Q, Jin LR, Liu H contribute to the pathogenesis of human TLE. (2011) A polymorphism in CALHM1 is associated with temporal lobe epilepsy. Epilepsy Behav 20:681–685 Increasing evidence indicates that genetic variation in reg- 5. Li Z, Ding C, Gong X, Wang X, Cui T (2016) Apolipoprotein E ulatory regions, such as the 3′UTR, could equally be the foun- epsilon4 allele was associated with nonlesional mesial temporal dation of the etiology of a variety of human diseases [17]. The lobe epilepsy in Han Chinese population. Medicine (Baltimore) investigation of miRSNPs as potential genetic risk factors for 95:e2894 6. Kasperaviciute D, Catarino CB, Matarin M, Leu C, Novy J, complex diseases is increasingly gaining momentum. Tostevin A et al (2013) Epilepsy, hippocampal sclerosis and febrile However, a very low frequency of SNPs is located in genetic seizures linked by common genetic variation around SCN1A. Brain elements that code for miRNAs [16, 47]. A study being able to 136:3140–3150 link a miRNA gene SNP to disease requests for an extremely 7. Welter D, MacArthur J, Morales J, Burdett T, Hall P, Junkins H, Klemm A, Flicek P et al (2014) The NHGRI GWAS catalog, a large population, which highlights the critical role of sample curated resource of SNP-trait associations. Nucleic Acids Res 42: size for detecting low-frequency variants in the genetic study D1001–D1006 [48]. In addition, rather than preliminary dual-luciferase re- 8. Jia P, Zhao Z (2014) Network-assisted analysis to prioritize GWAS porter assay, more thorough in vivo and vitro experiments results: principles, methods and perspectives. Hum Genet 133:125– 138 approaches to validate these miRSNPs’ biological function 9. Alexander RP, Fang G, Rozowsky J, Snyder M, Gerstein MB are also needed. (2010) Annotating non-coding regions of the genome. Nat Rev Genet 11:559–571 10. Bartel DP (2004) MicroRNAs: genomics, biogenesis, mechanism, Conclusion and function. Cell 116:281–297 11. Djuranovic S, Nahvi A, Green R (2012) miRNA-mediated gene silencing by translational repression followed by mRNA Consequently, this study revealed that a computational ap- deadenylation and decay. Science 336:237–240 proach to unravel miRSNPs that affect miRNA-mRNA inter- 12. Alsharafi WA, Xiao B, Abuhamed MM, Luo Z (2015) miRNAs: action enables the course of selecting potential BmiRSNP- biological and clinical determinants in epilepsy. Front Mol Neurosci 8:59 ^ miRNA-risk gene-pathway axis. This study explored the 13. Jimenez-Mateos EM, Henshall DC (2013) Epilepsy and possible molecular mechanisms of miRSNPs as potential microRNA. Neuroscience 238:218–229 Mol Neurobiol (2019) 56:7022–7031 7031

14. Jimenez-Mateos EM, Engel T, Merino-Serrais P, McKiernan RC, 32. Lynch MA (2004) Long-term potentiation and memory. Physiol Tanaka K, Mouri G et al (2012) Silencing microRNA-134 produces Rev 84:87–136 neuroprotective and prolonged seizure-suppressive effects. Nat 33. Leite JP, Neder L, Arisi GM, Carlotti CJ, Assirati JA, Moreira JE Med 18:1087–1094 (2005) Plasticity, synaptic strength, and epilepsy: what can we learn 15. Bartel DP (2009) MicroRNAs: target recognition and regulatory from ultrastructural data? Epilepsia 46(Suppl 5):134–141 functions. Cell 136:215–233 34. Vezzani A, Friedman A, Dingledine RJ (2013) The role of inflam- 16. Saunders MA, Liang H, Li WH (2007) Human polymorphism at mation in epileptogenesis. Neuropharmacology 69:16–24 microRNAs and microRNA target sites. Proc Natl Acad Sci U S A 35. Rana A, Musto AE (2018) The role of inflammation in the devel- – 104:3300 3305 opment of epilepsy. J Neuroinflammation 15:144 17. Saba R, Booth S (2013) Polymorphisms affecting miRNA regula- 36. Xu J, Li CX, Li YS, Lv JY, Ma Y, Shao TT, Xu LD, Wang YYet al tion: a new level of genetic variation affecting disorders and dis- (2011) MiRNA-miRNA synergistic network: construction via co- – eases of the human CNS. Future Neurol 8:411 431 regulating functional modules and disease miRNA topological fea- 18. YuanM,ZhanQ,DuanX,SongB,ZengS,ChenX,YangQ,XiaJ tures. Nucleic Acids Res 39:825–836 (2013) A functional polymorphism at miR-491-5p binding site in 37. Ma J, Guo R, Wang T, Pan X, Lei X (2015) Let-7b binding site the 3′-UTR of MMP-9 gene confers increased risk for atheroscle- polymorphism in the B-cell lymphoma-extra large 3′UTR is asso- rotic cerebral infarction in a Chinese population. Atherosclerosis ciated with fluorouracil resistance of hepatocellular carcinoma. Mol 226:447–452 Med Rep 11:677–681 19. Panjwani N, Wilson MD, Addis L, Crosbie J, Wirrell E, Auvin S, 38. Henshall DC, Clark RS, Adelson PD, Chen M, Watkins SC, Simon Caraballo RH, Kinali M et al (2016) A microRNA-328 binding site RP (2000) Alterations in bcl-2 and caspase gene family protein in PAX6 is associated with centrotemporal spikes of rolandic epi- expression in human temporal lobe epilepsy. Neurology 55:250– lepsy. Ann Clin Transl Neurol 3:512–522 257 20. Cui L, Tao H, Wang Y, Liu Z, Xu Z, Zhou H, Cai Y, Yao L et al (2015) A functional polymorphism of the microRNA-146a gene is 39. Yan Y, Xie R, Zhang Q, Zhu X, Han J, Xia R (2017) Bcl-xL/Bak associated with susceptibility to drug-resistant epilepsy and seizures interaction and regulation by miRNA let-7b in the intrinsic apopto- – frequency. Seizure 27:60–65 tic pathway of stored platelets. Platelets:1 6 21. Manna I, Labate A, Borzi G, Mumoli L, Cavalli SM, Sturniolo M 40. McKiernan RC, Jimenez-Mateos EM, Bray I, Engel T, Brennan GP, et al (2016) An SNP site in pri-miR-124, a brain expressed miRNA Sano T et al (2012) Reduced mature microRNA levels in associa- gene, no contribution to mesial temporal lobe epilepsy in an Italian tion with dicer loss in human temporal lobe epilepsy with hippo- sample. Neurol Sci 37:1335–1339 campal sclerosis. PLoS One 7:e35921 22. Manna I, Labate A, Mumoli L, Pantusa M, Ferlazzo E, Aguglia U, 41. Han CL, Ge M, Liu YP, Zhao XM, Wang KL, Chen N, Hu W, Quattrone A, Gambardella A (2013) Relationship between genetic Zhang JG et al (2018) Long non-coding RNA H19 contributes to variant in pre-microRNA-146a and genetic predisposition to tem- apoptosis of hippocampal neurons by inhibiting let-7b in a rat mod- poral lobe epilepsy: a case-control study. Gene 516:181–183 el of temporal lobe epilepsy. Cell Death Dis 9:617 23. Kanehisa M, Goto S (2000) KEGG: Kyoto encyclopedia of genes 42. Feng J, Zhou Y, Campbell SL, Le T, Li E, Sweatt JD et al (2010) and genomes. Nucleic Acids Res 28:27–30 Dnmt1 and Dnmt3a maintain DNA methylation and regulate syn- 24. Huang DW, Sherman BT, Lempicki RA (2009) Systematic and aptic function in adult forebrain neurons. Nat Neurosci 13:423–430 integrative analysis of large gene lists using DAVID bioinformatics 43. Liu Z, Li Z, Zhi X, Du Y, Lin Z, Wu J (2018) Identification of de resources. Nat Protoc 4:44–57 novo DNMT3A mutations that cause west syndrome by using 25. Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, whole-exome sequencing. Mol Neurobiol 55:2483–2493 Davis AP, Dolinski K et al (2000) Gene ontology: tool for the 44. Zhu Q, Wang L, Zhang Y,Zhao FH, Luo J, Xiao Z, Chen GJ, Wang unification of biology. The Gene Ontology Consortium. Nat XF (2012) Increased expression of DNA methyltransferase 1 and Genet 25:25–29 3a in human temporal lobe epilepsy. J Mol Neurosci 46:420–426 26. Kozomara A, Griffiths-Jones S (2011) miRBase: integrating 45. Xiao W, Cao Y, Long H, Luo Z, Li S, Deng N, Wang J, Lu X et al microRNA annotation and deep-sequencing data. Nucleic Acids (2018) Genome-wide DNA methylation patterns analysis of non- Res 39:D152–D157 coding RNAs in temporal lobe epilepsy patients. Mol Neurobiol 55: 27. Li X, Jiang W, Li W, Lian B, Wang S, Liao M, Chen X, Wang Y 793–803 et al (2012) Dissection of human MiRNA regulatory influence to 46. Kobow K, Kaspi A, Harikrishnan KN, Kiese K, Ziemann M, – subpathway. Brief Bioinform 13:175 186 Khurana I, Fritzsche I, Hauke J et al (2013) Deep sequencing re- 28. Baron M, Kudin AP, Kunz WS (2007) Mitochondrial dysfunction veals increased DNA methylation in chronic rat epilepsy. Acta – in neurodegenerative disorders. Biochem Soc Trans 35:1228 1231 Neuropathol 126:741–756 29. Tai XY, Koepp M, Duncan JS, Fox N, Thompson P, Baxendale S, 47. Muinos-Gimeno M, Montfort M, Bayes M, Estivill X, Espinosa- Liu JYW, Reeves C et al (2016) Hyperphosphorylated tau in pa- Parrilla Y (2010) Design and evaluation of a panel of single- tients with refractory epilepsy correlates with cognitive decline: a nucleotide polymorphisms in microRNA genomic regions for asso- – study of temporal lobe resections. Brain 139:2441 2455 ciation studies in human disease. Eur J Hum Genet 18:218–226 30. Bernasconi N (2016) Is epilepsy a curable neurodegenerative dis- 48. The Schizophrenia Psychiatric Genome-Wide Association Study ease? Brain 139:2336–2337 (GWAS) Consortium (2011) Genome-wide association study iden- 31. Davidson YS, Raby S, Foulds PG, Robinson A, Thompson JC, tifies five new schizophrenia loci. Nat Genet 43:969–976 Sikkink S, Yusuf I, Amin H et al (2011) TDP-43 pathological changes in early onset familial and sporadic Alzheimer’s disease, late onset Alzheimer’s disease and Down’s syndrome: association Publisher’sNoteSpringer Nature remains neutral with regard to with age, hippocampal sclerosis and clinical phenotype. Acta jurisdictional claims in published maps and institutional affiliations. Neuropathol 122:703–713