Possible epigenetic regulatory effect of dysregulated circular RNAs in epilepsy

Abstract

Circular RNAs (circRNAs) involve in the epigenetic regulation and its major mechanism is the sequestration of the target micro RNAs (miRNAs). We hypothesized that circRNAs might be related with the pathophysiology of chronic epilepsy and evaluated the altered circRNA expressions and their possible regulatory effects on their target miRNAs and mRNAs in a mouse epilepsy model. The circRNA expression profile in the hippocampus of the pilocarpine mice was analyzed and compared with control. The correlation between the expression of miRNA binding sites (miRNA response elements, MRE) in the dysregulated circRNAs and the expression of their target miRNAs was evaluated. As miRNAs also inhibit their target mRNAs, circRNA–miRNA-mRNA regulatory network, comprised of dysregulated RNAs that targets one another were searched. For the identified networks, bioinformatics analyses were performed. As the result, Forty-three circRNAs were dysregulated in the hippocampus (up-regulated, 26; down-regulated, 17). The change in the expression of MRE in those circRNAs negatively correlated with the change in the relevant target miRNA expression (r = -0.461, P<0.001), supporting that circRNAs inhibit their target miRNA. 333 dysregulated circRNA–miRNA-mRNA networks were identified. Gene ontology and pathway analyses demonstrated that the up-regulated mRNAs in those networks were closely related to the major processes in epilepsy. Among them, STRING analysis identified 37 key mRNAs with abundant (≥4) interactions with other dysregulated target mRNAs. The dysregulation of the circRNAs which had multiple interactions with key mRNAs were validated by PCR. We concluded that dysregulated circRNAs might have a pathophysiologic role in chronic epilepsy by regulating multiple disease relevant mRNAs via circRNA−miRNA−mRNA interactions.

About the Authors:

Woo-Jin Lee

Contributed equally to this work with: Woo-Jin Lee, Jangsup Moon

Roles Data curation, Formal analysis, Software, Writing – original draft

Affiliations Department of Neurology, Comprehensive Epilepsy Center, Laboratory for Neurotherapeutics, Biomedical Research Institute, Seoul National University Hospital, Seoul, South Korea, Program in Neuroscience, Neuroscience Research Institute of SNUMRC, College of Medicine, Seoul National University, Seoul, South Korea

Jangsup Moon

Contributed equally to this work with: Woo-Jin Lee, Jangsup Moon

Roles Data curation, Writing – review & editing

Affiliations Department of Neurology, Comprehensive Epilepsy Center, Laboratory for Neurotherapeutics, Biomedical Research Institute, Seoul National University Hospital, Seoul, South Korea, Program in Neuroscience, Neuroscience Research Institute of SNUMRC, College of Medicine, Seoul National University, Seoul, South Korea

Daejong Jeon

Roles Investigation, Methodology

Affiliation: Department of Neurology, Comprehensive Epilepsy Center, Laboratory for Neurotherapeutics, Biomedical Research Institute, Seoul National University Hospital, Seoul, South Korea

Tae-Joon Kim

Roles Methodology, Validation

Affiliations Department of Neurology, Comprehensive Epilepsy Center, Laboratory for Neurotherapeutics, Biomedical Research Institute, Seoul National University Hospital, Seoul, South Korea, Program in Neuroscience, Neuroscience Research Institute of SNUMRC, College of Medicine, Seoul National University, Seoul, South Korea

Jung-Suk Yoo

Roles Investigation, Methodology

Affiliation: Department of Neurology, Comprehensive Epilepsy Center, Laboratory for Neurotherapeutics, Biomedical Research Institute, Seoul National University Hospital, Seoul, South Korea

Dong-Kyu Park

Roles Investigation, Methodology

Affiliation: Department of Neurology, Comprehensive Epilepsy Center, Laboratory for Neurotherapeutics, Biomedical Research Institute, Seoul National University Hospital, Seoul, South Korea

Soon-Tae Lee

Roles Data curation, Formal analysis

Affiliations Department of Neurology, Comprehensive Epilepsy Center, Laboratory for Neurotherapeutics, Biomedical Research Institute, Seoul National University Hospital, Seoul, South Korea, Program in Neuroscience, Neuroscience Research Institute of SNUMRC, College of Medicine, Seoul National University, Seoul, South Korea

Keun-Hwa Jung

Roles Supervision, Validation, Writing – review & editing

Affiliations Department of Neurology, Comprehensive Epilepsy Center, Laboratory for Neurotherapeutics, Biomedical Research Institute, Seoul National University Hospital, Seoul, South Korea, Program in Neuroscience, Neuroscience Research Institute of SNUMRC, College of Medicine, Seoul National University, Seoul, South Korea

Kyung-Il Park

Roles Investigation, Resources

Affiliations Department of Neurology, Comprehensive Epilepsy Center, Laboratory for Neurotherapeutics, Biomedical Research Institute, Seoul National University Hospital, Seoul, South Korea, Program in Neuroscience, Neuroscience Research Institute of SNUMRC, College of Medicine, Seoul National University, Seoul, South Korea, Department of Neurology, Seoul National University Healthcare System Gangnam Center, Seoul, South Korea

Ki-Young Jung

Roles Data curation, Writing – review & editing

Affiliations Department of Neurology, Comprehensive Epilepsy Center, Laboratory for Neurotherapeutics, Biomedical Research Institute, Seoul National University Hospital, Seoul, South Korea, Program in Neuroscience, Neuroscience Research Institute of SNUMRC, College of Medicine, Seoul National University, Seoul, South Korea

Manho Kim

Roles Resources, Supervision

Affiliations Department of Neurology, Comprehensive Epilepsy Center, Laboratory for Neurotherapeutics, Biomedical Research Institute, Seoul National University Hospital, Seoul, South Korea, Program in Neuroscience, Neuroscience Research Institute of SNUMRC, College of Medicine, Seoul National University, Seoul, South Korea

Sang Kun Lee

Roles Conceptualization, Resources, Writing – review & editing

* E-mail: [email protected] (KC); [email protected] (SKL)

Affiliations Department of Neurology, Comprehensive Epilepsy Center, Laboratory for Neurotherapeutics, Biomedical Research Institute, Seoul National University Hospital, Seoul, South Korea, Program in Neuroscience, Neuroscience Research Institute of SNUMRC, College of Medicine, Seoul National University, Seoul, South Korea

Kon Chu

Roles Conceptualization, Investigation, Writing – review & editing

* E-mail: [email protected] (KC); [email protected] (SKL)

Affiliations Department of Neurology, Comprehensive Epilepsy Center, Laboratory for Neurotherapeutics, Biomedical Research Institute, Seoul National University Hospital, Seoul, South Korea, Program in Neuroscience, Neuroscience Research Institute of SNUMRC, College of Medicine, Seoul National University, Seoul, South Korea

ORCID logo http://orcid.org/0000-0001-5863-0302

Introduction

Epilepsy is a chronic disease resulting from a long-term process of epileptogenesis in the brain.[1–3] Regardless of the type of primary insult, common reactive processes occur around the injured brain tissue during the latent period, forming an aberrant neural network that generates synchronous hyper excitation of neurons, which manifests as spontaneous recurrent seizures.[1–5] In this regard, most of the patients with epilepsy have little chance to be treated during the period of epileptogenesis, and the treatment usually starts when the seizure events have already been evident. Therefore, elucidating the detailed genetic regulation in the brain with chronic epilepsy and finding strategies to modulate it might be fundamental to improve the efficacy of the chronic epilepsy treatment.

The genes involved in epileptogenesis are tightly regulated by various epigenetic regulatory mechanisms, and their chronic alteration in the hippocampus might have a major role of temporal lobe epilepsy.[2–4, 6] Among them, micro RNAs (miRNAs), small (20–24 bp), noncoding RNAs (ncRNAs) that regulate the expression of hundreds of target genes, have been most widely investigated.[7, 8] However, recent evidence has demonstrated that gene expression is more precisely modulated by another type of ncRNA, called circular RNA (circRNA).[9–12]

CircRNAs are a covalently closed and circular-shaped subgroup of ncRNAs.[9, 10] Predominantly, circRNAs are generated by back-splicing, a process by which downstream exons are reversely spliced to upstream exons.[9, 10] circRNAs are increasingly recognized as major epigenetic regulators in the pathogenesis of various diseases.[9–14] A circRNA might interact with the gene transcription machineries and the production of a circRNA might complete with the production of its corresponding linear form mRNA.[15] Moreover, circRNAs contains miRNA binding sites, the miRNA response elements (MREs), that enable circRNAs to sequestrate the target miRNA, which is known as the “miRNA sponge effect”.[9, 10] In this regard, circRNAs might modulate the expression of their target genes via circRNA-miRNA-mRNA regulatory networks.

Some properties of circRNAs indicate that circRNAs might have a particular role in the pathomechanism of the central nervous system (CNS) diseases. First, circRNAs are highly abundant in CNS and their expression level is tightly regulated according to the time and the region in the brain.[16] Second, because circRNAs have covalently closed loop structures without polyA tails, they are more stable in the CNS by being resistant to RNA exonucleases or RNase R-mediated degradation.[9, 11] Therefore, studying the alteration of circRNA expression and its implication on the balance between the downstream target miRNAs and mRNAs might shed light on the understanding of pathophysiologic role of circRNAs in chronic epilepsy and provide a novel therapeutic target.[9–11]

In this study, we evaluated the comprehensive profile of differentially regulated circRNAs in the hippocampus of pilocarpine epilepsy model and tried to identify their functions in epigenetic regulatory processes involved in the pathophysiology of chronic epilepsy.

Materials and methods

Tissue preparation

Our study group have been generating pilocarpine chronic epilepsy models for several years.[6, 17–23] The epilepsy model used in the current study was randomly selected from a large group of pilocarpine mouse models in our laboratory, generated according to the previously described procedures.[6, 17–23] In brief, a single intraperitoneal injection of pilocarpine (330–400 mg/kg; Sigma, St. Louis, MO, USA) was performed in 118 male C57BL/6 mice to induce status epilepticus (SE). The age of the mice was set as 5 weeks, based on our laboratory experience that the frequency of developing spontaneous recurrent seizures (SRSs) was the highest at this age.[17, 18, 20–22] Methyl-scopolamine (1 mg/kg; Sigma) was intraperitoneally administered 30 min before the pilocarpine injection to minimize muscarinic adverse effects. 86.6% (102/118 mice) developed status epilepticus (SE) and at about 40 minutes after the onset of SE, intraperitoneal diazepam (5 mg/kg) was administered to convert the form of SE from convulsive to non-convulsive.[24] 41.5% (49 mice) died during or shortly after the procedure. Sixty days after SE, all of the fifty-three mice that survived after the prolonged SE developed clinical SRSs, which is consistent with the previous reports that demonstrated a 100% frequency of developing SRSs after inducing a prolonged SE by pilocarpine injection in rodents.[25–27] 24/7 continuous video-electroencephalograph (EEG) monitoring was performed in randomly selected twenty-seven mice for the mean duration of 53.7±20.4 days, and all of the monitored mice were confirmed to have SRSs with the mean seizure frequency of 2.0±0.6/day. As the video-EEG monitoring requires craniotomy and insertion of electronic probes into cerebral hemispheres which might effect as a trauma and induce a significant alteration in the RNA expression profiles in the brain, [28, 29] four mice used in the current study were randomly selected from the remaining 26 mice that were developed SE after pilocarpine injection, but did not underwent video-EEG monitoring. To compare the expression profiles of the circRNAs in the hippocampus between the pilocarpine chronic epilepsy model and controls, four age and sex-matched mice were also allocated to the control group.

Mice were euthanized by cervical dislocation and brains were immediately removed.[6, 17–23] The hippocampus was obtained from each mouse and were immediately stored at −80°C. All animals were managed with standardized procedures approved by the Institutional Animal Care and Use Committee of Seoul National University Hospital. All procedures were approved by the institutional review boards of the Seoul National University Hospital.

Microarray

Total RNAs were extracted using TRIZOL reagent (Invitrogen, NY, USA) and purified by an RNeasy Mini Kit (Qiagen, Hilden, Germany). RNA quantity was measured with a Nanodrop ND-1000 (Thermo Fisher Scientific, MA, USA) and quality checked by an Agilent 2100 Bioanalyzer (Agilent Technologies, CA, USA). Sample preparation and microarray hybridization were performed according to the manufacturer’s protocol (Arraystar, Rockville, MD, USA). For circRNA microarray, the RNAs were treated with RNase R (Epicenter, WI, USA) to remove linear RNAs and enrich circRNA. The enriched circRNA samples were amplified and transcribed into fluorescent cRNA using a random priming method (Arraystar Super RNA Labeling Kit). The labeled cRNAs were hybridized onto the Arraystar Mouse circRNA Array V2 (8 × 15K, Arraystar). After the slides were washed, the arrays were scanned by an Agilent Scanner G2505C (Agilent Technologies). The miRNA and mRNA microarray data were also obtained using the Agilent Mouse miRNA Microarray 8X15K kit and Agilent Mouse Gene Expression Microarray 4X44K kit respectively, according to the manufacturer’s protocol (Agilent Technologies).

Agilent Feature Extraction software (version 11.0.1.1, Agilent Technologies) was used to analyze the acquired array images. Quantile normalization and subsequent data processing were performed with the R software limma package (R version 3.2.4). Low-intensity filtering was performed and RNAs with two or more out of eight samples that had flags in “P” or “M” were included in further analyses. Mann–Whitney test were used to detect differentially expressed circRNAs between the two groups by fold-changes of ≥1.5 and P-values ≤0.05.[17, 20, 21, 23] Volcano plot filtering and hierarchical clustering were also performed to identify differentially expressed circRNAs. In the analyses of miRNA and mRNA microarray, an false discovery rate (FDR) adjusted P-value of <0.05 with fold-changes of ≥1.5 was used to detect differentially expressed RNAs between the two groups.

Analysis of the regulative effect of circRNA on its linear mRNA expression

To examine whether the dysregulated circRNAs regulate the expression of their linear form mRNAs, the expression ratio of each circRNA between the epilepsy and the control groups (circRNA expression ratio) were compared with the expression ratio of its linear form mRNAs (mRNA expression ratio).[30] Pearson’s correlation analysis was performed to measure the correlations between the expression ratios of circRNA and its linear form mRNAs.

Analysis of the regulative effect of circRNA on target miRNA expression

To evaluate the hypothesis that circRNAs function as miRNA sponges, we tried to quantitatively analyze the correlation between the change in the expression of target miRNA binding sites in dysregulated circRNAs and the change in the expression of their target miRNAs (target miRNA expression ratio) according to the following steps. First, for every differentially expressed circRNA, up to five target miRNAs were identified by MRE sequence analysis.[31] MRE sequence analysis was performed using a miRNA target prediction software (Arraystar) based on TargetScan (www.targetscan.org) and miRanda (http://www.microrna.org) algorithms. Each MRE was composed of “seed site” and “complementary sites”. A seed site was designated as the ≥6 consecutive nucleotides matching to the miRNA nucleotides 2–7 from the 5′-end, whereas a complementary site was defined as the neighboring ≥4 sequences pairing to miRNA nucleotides 12–19.[32] Second, each MREs were categorized into canonical, marginal with full complement, and incomplete, where a canonical denotes a seed site with ≥7 consecutive matching nucleotides (7mer+A1, 7mer+m8, 8mer), marginal with full complement denotes a seed site with 6 matching nucleotide and a complementary site with ≥4 consecutive matching nucleotides, and incomplete denotes a seed site with <6 matching nucleotide and a complementary site with <4 matching nucleotides.[32] A circRNA could contain multiple MREs for a given target miRNA and vice versa. Third, the change in the expression of MRE was calculated for each MRE categories, by multiplying the normalized baseline expression intensity of the corresponding circRNA, circRNA expression ratio -1, and the number of MREs of the relevant category for the target miRNA. When two or more dysregulated circRNAs were targeting a common miRNA, their changes in the expression of MREs were summated to evaluate the comprehensive effect of multiple circRNAs on a given target miRNA (Fig 1A and 1B). Finally, correlations between the change in the expression of MRE and the target miRNA expression ratio were measured, separately in each MRE categories.

[Figure omitted. See PDF.]

Fig 1. Analysis of the regulative effect of circRNA on miRNA expression.

Each predicted MRE was composed of “seed site” and “complementary sites”. MREs were categorized into canonical, marginal with full complement, and incomplete, where a canonical denotes a seed site with 7–8 matching nucleotides, marginal with full complement denotes a seed site with 6 matching nucleotide with a complementary site with ≥4 canonical matching nucleotides, and incomplete denotes a seed site with 5–6 matching nucleotide or a complementary site with <4 canonical matching nucleotides (panel A). Panel B demonstrates how the relative value of MRE expression change in dysregulated circRNAs was calculated. Panel C denotes the proportion of each MRE categories. a: When two or more dysregulated circRNAs had MREs targeting a common miRNA, their change in the expression of MREs were summated. b: Calculated separately for each categories of MRE. MRE, miRNA response element.

https://doi.org/10.1371/journal.pone.0209829.g001

Specific circRNA–miRNA–mRNA regulatory network

In order to search the possible inhibitory interaction between the differentially expressed circRNAs and their target miRNAs, significantly down-regulated downstream miRNAs for the up-regulated circRNAs or significantly up-regulated downstream miRNAs for the down-regulated circRNAs were designated as circRNA-interacting miRNAs (CI-miRNAs) and were included in further analyses. Then, to predict the target mRNAs regulated by the CI-miRNAs, the up- and down-regulated CI-miRNA sets were separately entered into an integrative miRNA target prediction program miRsystem (http://mirsystem.cgm.ntu.edu.tw), along with their expression ratio data.[33] After the list of potential target mRNAs was obtained, their expression ratios were also extracted from the mRNA microarray data. Significantly down-regulated downstream target mRNAs for the up-regulated CI-miRNA sets or significantly up-regulated downstream target mRNAs for the down-regulated CI-miRNA sets were defined as circRNA-miRNA-interacting mRNAs (CMI-mRNAs).

Gene ontology and pathway analysis

To demonstrate the pathophysiologic role of the circRNA–miRNA–mRNA regulatory network in chronic epilepsy, gene ontology and pathway analyses were performed for the CMI-mRNAs. The gene ontology categories were obtained from the Gene Ontology website (http://www.geneontology.org).[23] Pathway analysis was performed using the Kyoto Encyclopedia of Genes and Genomes (http://www.genome.jp/kegg) database.[23, 34] In both analyses, categories with P < 0.05 were considered to be statistically significant.

STRING analysis

Interaction among the proteins translated by CMI-mRNAs was analyzed with STRING v10 (http://string-db.org). A network map was drawn by analyzing a protein–protein interaction database derived from multiple sources as follows: primary experimental databases, biological pathway databases, automated text-mining from Medline abstracts and a large collection of full-text articles, and de novo interactions predicted by genomic information algorithms.[35] The minimum required score for determining a significant interaction was set as 0.400, corresponding to a medium confidence.[35]

Quantitative PCR analysis

Five differentially expressed circRNAs (one up-regulated, four down-regulated) were selected for quantitative real-time reverse transcription PCR analysis to validate the microarray expression data. For PCR analysis, two hippocampal tissue samples were pooled into one RNA sample as a unit. cDNAs were synthesized from 0.5 μg of total RNA of hippocampal tissues by reverse transcription. Standard curves were prepared using 2× SuperArray PCR master mix (Arraystar) according to the manufacturer’s protocol. The relative expression ratio of each circRNA was calculated with the Rotor-Gene Real-Time Analysis Software 6.0 (Qiagen), using the housekeeping gene, Gapdh, expression for normalization. All real-time reactions were performed in triplicate.

Statistical analysis

Data were reported as number (percentage) or mean ± standard deviation. Excel 2016 (Microsoft, Redmond, WA, USA) was used for the Mann–Whitney U test to detect differentially expressed circRNAs, miRNAs, and mRNAs between the two groups by fold-changes of ≥1.5 and P-values ≤0.05.[17, 20, 21, 23] Correction for the multiple comparisons were not performed for the circRNA microarray analysis, due to that the portion of dysregulated circRNAs in microarray analysis was too small to perform the Benjamini-Hochberg procedure for p-value adjustment. In the analyses of miRNA and mRNA microarray, Benjamini-Hochberg procedure was performed to control the FDR at 0.05, and an adjusted P-value of <0.05 was used to detect differentially expressed RNAs between the two groups.

SPSS (version 22.0; SPSS Inc., Chicago, IL, USA) was used to perform Pearson’s correlation analyses to measure the correlations between the expression ratios of the differentially expressed circRNAs and their linear form mRNAs and the correlations between the change in the expression of MRE and the target miRNA expression ratio. P values < 0.05 were considered statistically significant.

Results

Overall expression profile of circRNA

Among a total of 12,984 circRNAs analyzed, the number of differentially expressed circRNAs in the hippocampus of pilocarpine model was 43 (up-regulated, 26; down-regulated, 17, Fig 2). The differentially expressed circRNAs are listed with their expression ratios in the S1 Table (Full circRNA expression profile in S1 File).

[Figure omitted. See PDF.]

Fig 2. Microarray data analysis of the differentially regulated circRNAs in the hippocampus of the pilocarpine epilepsy model.

The numbers of circRNAs that are differentially expressed between the hippocampi of models are shown in panel A. Scatter Plot (panel B) shows that for most circRNAs, their expressions in the epilepsy model and the control were comparable. Volcano plot shows differentially expressed circRNAs in the hippocampus of the pilocarpine epilepsy model by red squares (panel C). The left and right vertical lines demark two-fold up and two-fold down changes, respectively, whereas the horizontal line indicates a P-value of 0.05. Panel D shows hierarchical cluster analysis of differentially expressed circRNAs in the hippocampus of the pilocarpine epilepsy and control mice. The log2 signal intensity is reflected in the color scale, which runs from blue (low intensity) to red (strong intensity).

https://doi.org/10.1371/journal.pone.0209829.g002

circRNA expression is not associated with their linear mRNA expression

The expression ratios of 10,607 linear form mRNAs out of 12,984 circRNAs (89.3%) were available. Among them, 598 (5.6%) mRNAs were significantly upregulated and 71 (0.7%) mRNAs were significantly downregulated. However, the expression ratio of the circRNAs did not correlate with the expression ratio of their linear form mRNAs (r = 0.002, P = 0.886).

Binding site expression in dysregulated circRNA is negatively associated with their target miRNA expression

The MRE analysis identified 215 potential interactions between the 43 dysregulated circRNA and their downstream target miRNAs (five miRNAs for each dysregulated circRNA). Among them, the target miRNA expression ratio was available in 75 interactions (34.9%) from the microarray data (full data in S2 File). The total number of MREs in those 75 circRNA-miRNA interactions was 286 (3.81±5.60, range 1−46 MREs per one interaction), consisting of 110 (38.5%) canonical type, 35 (12.2%) marginal with full complement type, and 141 (49.3%) incomplete type MREs (Fig 1C).

In the correlation analyses between the change in the expression of MRE in dysregulated circRNAs and the target miRNA expression ratio, the change in the all-type MRE expression negatively correlated with the target miRNA expression ratio (r = -0.249, P = 0.032). The association was higher for the canonical or marginal with full complement types of MRE (r = -0.515, P<0.001 and r = -0.461, P<0.001, respectively), whereas no significant association was found between the incomplete type MRE and the target miRNA (r = -0.097, P = 0.407, Table 1).

[Figure omitted. See PDF.]

Table 1. Correlation analyses between the MRE change amount and the target miRNA expression ratio.

https://doi.org/10.1371/journal.pone.0209829.t001

Specific circRNA–miRNA-mRNA regulatory network

As we observed that the changes in the target miRNA binding sites in dysregulated circRNAs is negatively correlated with the change in their target miRNA expressions, we speculated that circRNAs might have an inhibitory effect on the expression of their target miRNAs, and we searched possible specific interactions between the dysregulated circRNA and CI-miRNA. Among the total of 75 potential interactions, 14 showed significant down-regulation of miRNAs with up-regulation of the upstream circRNAs and six showed significant up-regulation of miRNAs with down-regulation of the upstream circRNAs (Table 2 and S1 Fig).

[Figure omitted. See PDF.]

Table 2. Differentially regulated circRNA−miRNA interactions.

https://doi.org/10.1371/journal.pone.0209829.t002

Using the miRNA target gene prediction program, 2,844 mRNAs were predicted as potential targets of the 14 down-regulated miRNA/up-regulated circRNA interactions and 287 mRNAs were identified as potential targets of the six up-regulated miRNA/down-regulated circRNA interactions.

The mRNA microarray expression data (full data in S3 File) identified 331 (11.6%) up-regulated CMI-mRNAs of the 14 down-regulated miRNA/up-regulated circRNA interactions and 2 (0.7%) down-regulated CMI-mRNAs of the six up-regulated miRNA/down-regulated circRNA interactions. In total, 333 mRNAs were identified as differentially expressed CMI-mRNAs in the pilocarpine chronic epilepsy model (S1 Fig, see S2 Table for list).

Gene ontology and pathway analysis

Gene ontology analysis was performed to evaluate enrichment of the up-regulated CMI-mRNAs in biological processes. Stem cell division, protein K48-linked ubiquitination, regulation of the synaptic vesicle cycle, cerebral cortex cell migration, and stem cell proliferation were the five most highly enriched biological processes (Table 3). In the pathway analysis of the differentially expressed CMI-mRNAs, the top five enriched pathways were the mitogen-activated protein kinase (MAPK) signaling pathway, pathways in cancer, the PI3K-Akt signaling pathway, focal adhesion, and the Ras signaling pathway (Table 3, full pathway analysis and GO analyses results in S4 File). GO and pathway analyses for the down-regulated CMI-mRNAs were not performed due to the low number of the CMI-mRNAs.

[Figure omitted. See PDF.]

Table 3. GO biological process and pathway analysis of the up-regulated CMI-mRNAs.

https://doi.org/10.1371/journal.pone.0209829.t003

STRING analysis

STRING analysis was performed including the 333 CMI-mRNAs. A total of 458 protein–protein interactions were identified and 37 CMI-mRNAs had abundant (≥4) interactions with other differentially expressed CMI-mRNAs (see Table 4 and S2 Fig and S5 File for detailed results). Notably, some of these 37 CMI-mRNAs commonly interacted with one upstream CI-miRNA and circRNA, and some CMI-mRNAs also interacted with multiple upstream CI-miRNAs and circRNAs. For example, the up-regulated mRNAs Bdnf, Dmd, and Notch2 are commonly regulated by mmu_circRNA_002170, which is also up-regulated, via connection with the down-regulated mmu–miR–468. The down-regulated mRNA Kcnc1 is multiply regulated by the down-regulated circRNAs mmu_circRNA_016800, 004229, and 35542 via connection with the up-regulated mmu–miR–207 (Fig 3A).

[Figure omitted. See PDF.]

Fig 3. Quantitative PCR validation of differentially expressed circRNAs interacting with pathophysiologically relevant CI-miRNAs and CMI-mRNAs.

Some CMI-mRNAs had regulatory interactions with multiple differentially expressed circRNAs, and some circRNAs had interactions with multiple pathophysiologically important CMI-mRNAs. Red bars indicate inhibitory regulation on target miRNAs (panel A). In panel B, expression ratios of the circRNAs (pilocarpine model/control) from the PCR analysis and microarray are shown. Gapdh was used as a reference gene to calculate the expression ratios of the circRNAs. CI-mRNA, circRNA-interacting miRNA; CMI-mRNA, circRNA- and miRNA-interacting mRNA. *P < 0.05, **P < 0.01 for the statistical significance of altered expression.

https://doi.org/10.1371/journal.pone.0209829.g003

[Figure omitted. See PDF.]

Table 4. Proteins encoded by the differentially expressed CMI-mRNAs and with abundant protein-protein interactions, from string analysis.

https://doi.org/10.1371/journal.pone.0209829.t004

Quantitative polymerase chain reaction (PCR) validation of circRNA expression

We inferred that the abovementioned circRNAs, which have multiple regulations on the dysregulated CMI-mRNAs with abundant protein-protein interactions, can act as key molecules of epigenetic regulation in chronic epilepsy. PCR analysis of these five differentially expressed circRNAs (mmu_circRNA_002170, 016800, 35542, 004229, and 31968) was performed for validation of the microarray expression data using the primers described in the S3 Table. A general consistency was observed between the quantitative PCR and the microarray analysis data. Four out of the five (80.0%) circRNAs (mmu_circRNA_002170, 35542, 004229, and 31968) were confirmed to be differentially expressed in the relevant directions by quantitative PCR analysis (Fig 3B). The expression ratio of GAPDH in the hippocampi of pilocarpine/control mice was 1.04 (95% Confidence interval 0.71−1.37, P = 0.602, see S6 File for the PCR raw data).

Discussion

This study demonstrated the altered expression of circRNAs and their possible epigenetic regulatory role in the hippocampus of a chronic epilepsy model. The change in the expression of MREs in dysregulated circRNAs had a negative association with their target miRNA expression. This result supports the hypotheses that circRNAs function as miRNA sponges and inhibits the miRNAs via MRE matching.[15, 32] Moreover, MREs of the canonical and the marginal+ full complement categories had a higher negative correlation with the expression of their target miRNA than the incomplete type MREs, indicating that a certain level of nucleotide matching might be required for a circRNA to have a regulatory effect on a target miRNA.[32] However, the altered expression of circRNAs was not associated with a significant regulation of their linear type mRNAs.

Further analyses investigated the possible regulatory network among dysregulated circRNA, miRNA, and mRNA and their pathophysiologic role in chronic epilepsy. Fourty-three dysregulated circRNAs, 20 circRNA−CI-miRNA interactions, and 333 CMI-mRNAs were identified to be possibly involved in the disease-specific circRNA–miRNA-mRNA regulatory networks.

Gene ontology demonstrated that the up-regulated CMI-mRNAs might be closely involved in the pathophysiology of chronic epilepsy. Regarding the biological processes, “stem cell division” and “stem cell proliferation” were among the most enriched terms. These terms might represent the increased proliferation of progenitor cells in the hippocampus, resulting in aberrant neuronal and glial network formation.[1, 2, 6, 36] The enriched term “cerebral cortex cell migration” is related to the dysregulated migration of granule cells in the dentate gyrus of temporal lobe epilepsy.[2, 37] “Protein K48-linked ubiquitination” involves in degradation of disheveled 2 protein, which mediates intracellular transmission of Wnt signals,[38] resulting in the dysregulation of Wnt/β-catenin-mediated neurogenesis and synaptic remodeling.[39] In addition, enriched “regulation of synaptic vesicle cycle” may reflect altered signaling by neurotransmitters such as γ-aminobutyric acid (GABA), adenosine, and glutamate in chronic epilepsy.[40]

For the enriched pathways, the “MAPK signaling” pathway induces proliferation of progenitor cells and differentiation into neurons and astrocytes.[41] The “PI3K-Akt signaling” pathway activates mTOR, which has a crucial role in aberrant neural network formation via enhancement of neural proliferation and synaptogenesis.[42] The “Ras signaling” pathway is an upstream activator of both the MAPK and the PI3K-Akt pathways.[43] Ras also promotes cell proliferation, differentiation, adhesion, migration, and apoptosis, which are also relevant to “pathways in cancer”.[44] “Focal adhesion” if related to the remodeling of neuronal circuits in the hippocampus.[2, 45, 46] Taken together, the CMI-mRNAs are closely related to crucial mechanisms of epilepsy, indicating that circRNAs may have a substantial influence on chronic epilepsy via the circRNA–miRNA–mRNA regulatory networks.

In the STRING analyses, 37 CMI-mRNAs encoded proteins with abundant interactions with the other proteins, suggesting that these genes may have a major pathophysiologic role in chronic epilepsy. Furthermore, some of these CMI-mRNAs had multiple regulatory interactions with circRNAs and some CMI-mRNAs were commonly regulated by one circRNA, implying that these molecules may have a role as a key molecule.[3, 7, 47] For example, the down-regulated Kcnc1, which has interactions with six other dysregulated CMI-mRNAs in STING analysis, encodes the KV3.1 subunit of voltage-gated potassium channels. It mediates high-frequency neuronal firing of inhibitory GABAergic interneurons, and its loss-of-function mutation results in a spontaneous seizure.[48] Kcnc1 is regulated by the multiple down-regulated circRNAs mmu_circRNA_016800, 004229, and 35542 via the up-regulated mmu–miR–207. Runx2 is a transcription factor that is highly expressed in the hippocampus and is involved in cellular homeostasis and glutamate-mediated neuronal excitatory responses.[49, 50] Runx2 is targeted by upstream down-regulated mmu_circRNA_31968 and up-regulated mmu–miR–23a. The up-regulated mmu_circRNA_002170 interacts with multiple up-regulated CMI-mRNAs Bdnf, Dmd, and Notch2 by inhibiting down-regulated mmu–miR–468. Bdnf encodes brain-derived natriuretic factor (Bdnf), a neurotrophin that is known to enhance TrkB-mediated aberrant neurogenesis, synaptic formation, and dysregulation of neurotransmitters.[51] Dmd encodes dystrophin, which maintains the function of kainate-type glutamate receptors in the hippocampus and affects the susceptibility to seizures.[52] Notch2 encodes Notch2, which participates in aberrant remodeling of neuronal circuits.[53]

These circRNAs might serve as potential therapeutic targets as they have multiple regulatory interactions with genes highly involved in the disease pathophysiology. For up-regulated circRNAs, administration of short oligonucleotides that target the unique 3'–5' junction sequences of the circRNA or MREs for the relevant target miRNA might specifically inhibit the function of the circRNA.[47, 54] In contrast, supplementing the down-regulated circRNAs might have an antagomir-like effect on the relevant miRNAs,[20] but with improved stability in the CNS,[9, 11] possibly resulting in the restoration of the target gene function (Fig 4).

[Figure omitted. See PDF.]

Fig 4. Schematic demonstration of the impact of differentially regulated circRNAs on gene expression and strategies for therapeutic intervention.

In panel A, miRNA levels are increased as a result of their reduced sequestration by the down-regulated circRNAs. Binding of miRNAs to their target mRNAs is increased, and the expression levels of the target mRNAs are decreased. By external supplementation of circRNAs, increased sequestration of the target miRNAs results in restoration of the target gene expression (panel B). Panel C describes up-regulated circRNAs excessively inhibiting their target miRNAs, with increased expression of the target genes as a consequence. In this case, external supplementation of oligonucleotides targeting the unique 3'–5' junction sequences of the circRNA or the miRNA response element (MRE) antagonizes the function of the overexpressed circRNAs, resulting in normalization of the target mRNA expression level (panel D).

https://doi.org/10.1371/journal.pone.0209829.g004

The present study has some limitations to be addressed. First, only the association between the dysregulated circRNAs, miRNAs, and mRNAs was evaluated, and we did not directly validate the regulatory interactions among them. Second, as the miRNA microarray only partially covered the CI-miRNAs of the dysregulated circRNAs, the complete circRNA−miRNA−mRNA regulatory interactions in chronic epilepsy was not mapped in the current study. Third, correction for the multiple comparisons were not performed for the circRNA microarray analysis, due to a very small portion (43/12,984, 0.33%) of dysregulated circRNAs. However, 4/5 (80.0%) circRNAs which underwent PCR analysis were validated to be relevantly dysregulated. Using a lower P-value cutoff (<0.01) might have been an alternative option, but this cutoff value returned no circRNA–miRNA–mRNA regulatory network. Fourth, because of the large number of genes analyzed, validation of the expression ratios with quantitative PCR analyses was not performed for all differentially expressed circRNAs, CI-miRNAs, and CMI-mRNAs. Fifth, the expression of GAPDH, the house-keeping gene for qPCR, varies with inflammatory processes and might not be an ideal house-keeping gene, although the expression of GAPDH in the hippocampi of pilocarpine and control mice in this study were comparable. Future studies should endeavor to confirm the pathophysiologic role of specific circRNA−miRNA−mRNA interactions in chronic epilepsy, as well as other CNS diseases. Furthermore, to enhance the utility of circRNA as a therapeutic target, noninvasive methods of delivering therapeutic molecules into the brain, such as intranasal delivery of circRNA or its antagonists,[20] should be investigated.

Supporting information

[Figure omitted. See PDF.]

S1 Fig. Study flow to demonstrate the interactions of circRNA, miRNA, and mRNA with altered expression in the hippocampus.

MRE, miRNA response element, CMI-mRNA, circRNA- and miRNA-interacting mRNA.

https://doi.org/10.1371/journal.pone.0209829.s001

(TIF)

S2 Fig. STRING analysis of the proteins encoded by the differentially expressed CMI-mRNAs with fold changes of ≥1.5.

Visit STRING analysis site (http://string-db.org) for the detailed information of the description of the nodes (proteins) and edges (protein-protein interactions).

https://doi.org/10.1371/journal.pone.0209829.s002

(TIF)

S1 Table. Differentially expressed circRNAs in the hippocampus of pilocarpine epilepsy model.

circRNA: circular RNA and MRE: miRNA response element.

https://doi.org/10.1371/journal.pone.0209829.s003

(DOCX)

S2 Table. Differentially expressed CMI-mRNAs, with fold changes of ≥1.5.

CMI-mRNA: circRNA and miRNA interacting mRNA.

https://doi.org/10.1371/journal.pone.0209829.s004

(DOCX)

S3 Table. Primers used to validate the five differentially expressed circRNAs.

GAPDH was used as a housekeeper gene.

https://doi.org/10.1371/journal.pone.0209829.s005

(DOCX)

S1 File. Full circular RNA expression profiles at 60 days after status epilepticus.

https://doi.org/10.1371/journal.pone.0209829.s006

(XLS)

S2 File. Full micro RNA expression profiles at 60 days after status epilepticus.

https://doi.org/10.1371/journal.pone.0209829.s007

(XLS)

S3 File. Full mRNA expression profiles at 60 days after status epilepticus.

https://doi.org/10.1371/journal.pone.0209829.s008

(XLS)

S4 File. Pathway analysis and gene ontology analyses for the differentially expressed CMI-mRNAs.

https://doi.org/10.1371/journal.pone.0209829.s009

(XLSX)

S5 File. STRING analysis for the differentially expressed CMI-mRNAs.

https://doi.org/10.1371/journal.pone.0209829.s010

(XLSX)

S6 File. PCR analysis for the differentially expressed circRNAs which have multiple regulations on the dysregulated CMI-mRNAs with abundant protein-protein interactions.

https://doi.org/10.1371/journal.pone.0209829.s011

(XLSX)

Acknowledgments

This study was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIP) (No.2014R1A2A1A11052709). This study was supported by the Korean Health Technology R&D Project, Ministry of Health and Welfare, Republic of Korea (HI14C2502).

Citation: Lee W-J, Moon J, Jeon D, Kim T-J, Yoo J-S, Park D-K, et al. (2018) Possible epigenetic regulatory effect of dysregulated circular RNAs in epilepsy. PLoS ONE 13(12): e0209829. https://doi.org/10.1371/journal.pone.0209829

References

1. Pitkänen A, Lukasiuk K. Molecular and cellular basis of epileptogenesis in symptomatic epilepsy. Epilepsy Behav. 2009;14(1):16–25.

2. Pitkänen A, Lukasiuk K. Mechanisms of epileptogenesis and potential treatment targets. Lancet Neurol. 2011;10(2):173–86. pmid:21256455

3. Hauser RM, Henshall DC, Lubin FD. The Epigenetics of Epilepsy and Its Progression. Neuroscientist. 2017:1073858417705840.

4. Pitkänen A, Kharatishvili I, Karhunen H, Lukasiuk K, Immonen R, Nairismägi J, et al. Epileptogenesis in experimental models. Epilepsia. 2007;48(s2):13–20.

5. Jessberger S, Zhao C, Toni N, Clemenson GD, Li Y, Gage FH. Seizure-associated, aberrant neurogenesis in adult rats characterized with retrovirus-mediated cell labeling. J Neurosci. 2007;27(35):9400–7. pmid:17728453

6. Jung KH, Chu K, Lee ST, Park KI, Kim JH, Kang KM, et al. Molecular alterations underlying epileptogenesis after prolonged febrile seizure and modulation by erythropoietin. Epilepsia. 2011;52(3):541–50. pmid:21269282

7. Bielefeld P, Mooney C, Henshall DC, Fitzsimons CP. miRNA-Mediated Regulation of Adult Hippocampal Neurogenesis; Implications for Epilepsy. Brain Plast. (Preprint):1–17.

8. Jimenez-Mateos E, Henshall D. Epilepsy and microRNA. Neuroscience. 2013;238:218–29. pmid:23485811

9. Salzman J. Circular RNA expression: its potential regulation and function. Trends Genet. 2016;32(5):309–16. pmid:27050930

10. Memczak S, Jens M, Elefsinioti A, Torti F, Krueger J, Rybak A, et al. Circular RNAs are a large class of animal RNAs with regulatory potency. Nature. 2013;495(7441):333–8. pmid:23446348

11. Chen W, Schuman E. Circular RNAs in brain and other tissues: a functional enigma. Trends Neurosci. 2016;39(9):597–604. pmid:27445124

12. Lu D, Xu A-D. Mini review: circular RNAs as potential clinical biomarkers for disorders in the central nervous system. Front Genet. 2016;7.

13. Hansen TB, Kjems J, Damgaard CK. Circular RNA and miR-7 in cancer. Cancer Res. 2013;73(18):5609–12. pmid:24014594

14. Bachmayr-Heyda A, Reiner AT, Auer K, Sukhbaatar N, Aust S, Bachleitner-Hofmann T, et al. Correlation of circular RNA abundance with proliferation–exemplified with colorectal and ovarian cancer, idiopathic lung fibrosis, and normal human tissues. Sci Rep. 2015;5:8057. pmid:25624062

15. Qu S, Yang X, Li X, Wang J, Gao Y, Shang R, et al. Circular RNA: a new star of noncoding RNAs. Cancer Lett. 2015;365(2):141–8. pmid:26052092

16. Rybak-Wolf A, Stottmeister C, Glažar P, Jens M, Pino N, Giusti S, et al. Circular RNAs in the mammalian brain are highly abundant, conserved, and dynamically expressed. Mol Cell. 2015;58(5):870–85. pmid:25921068

17. Jang Y, Moon J, Lee S-T, Jun J-S, Kim T-J, Lim J-A, et al. Dysregulated long non-coding RNAs in the temporal lobe epilepsy mouse model. Seizure. 2018;58:110–9. pmid:29702408

18. Jung S, Yang H, Kim BS, Chu K, Lee SK, Jeon D. The immunosuppressant cyclosporin A inhibits recurrent seizures in an experimental model of temporal lobe epilepsy. Neurosci Lett. 2012;529(2):133–8. pmid:22981977

19. Jung K-H, Chu K, Lee S-T, Kim J, Sinn D-I, Kim J-M, et al. Cyclooxygenase-2 inhibitor, celecoxib, inhibits the altered hippocampal neurogenesis with attenuation of spontaneous recurrent seizures following pilocarpine-induced status epilepticus. Neurobiol Dis. 2006;23(2):237–46. pmid:16806953

20. Lee S-T, Jeon D, Chu K, Jung K-H, Moon J, Sunwoo J, et al. Inhibition of miR-203 reduces spontaneous recurrent seizures in mice. Mol Neurobiol. 2017;54(5):3300–8. pmid:27165289

21. Moon J, Lee S-T, Choi J, Jung K-H, Yang H, Khalid A, et al. Unique behavioral characteristics and microRNA signatures in a drug resistant epilepsy model. PLoS One. 2014;9(1):e85617. pmid:24454901

22. Jeon D, Chu K, Lee ST, Jung KH, Kang KM, Ban JJ, et al. A cell‐free extract from human adipose stem cells protects mice against epilepsy. Epilepsia. 2011;52(9):1617–26. pmid:21777228

23. Lee DY, Moon J, Lee S-T, Jung K-H, Park D-K, Yoo J-S, et al. Dysregulation of long non-coding RNAs in mouse models of localization-related epilepsy. Biochem Biophys Res Commun. 2015;462(4):433–40. pmid:25976677

24. Gualtieri F, Curia G, Marinelli C, Biagini G. Increased perivascular laminin predicts damage to astrocytes in CA3 and piriform cortex following chemoconvulsive treatments. Neuroscience. 2012;218:278–94. pmid:22609936

25. Curia G, Longo D, Biagini G, Jones RS, Avoli M. The pilocarpine model of temporal lobe epilepsy. Journal of neuroscience methods. 2008;172(2):143–57. pmid:18550176

26. Shibley H, Smith BN. Pilocarpine-induced status epilepticus results in mossy fiber sprouting and spontaneous seizures in C57BL/6 and CD-1 mice. Epilepsy research. 2002;49(2):109–20. pmid:12049799

27. Chen J, Larionov S, Pitsch J, Hoerold N, Ullmann C, Elger C, et al. Expression analysis of metabotropic glutamate receptors I and III in mouse strains with different susceptibility to experimental temporal lobe epilepsy. Neurosci Lett. 2005;375(3):192–7. pmid:15694259

28. Lei P, Li Y, Chen X, Yang S, Zhang J. Microarray based analysis of microRNA expression in rat cerebral cortex after traumatic brain injury. Brain Res. 2009;1284:191–201. pmid:19501075

29. Ramlackhansingh AF, Brooks DJ, Greenwood RJ, Bose SK, Turkheimer FE, Kinnunen KM, et al. Inflammation after trauma: microglial activation and traumatic brain injury. Ann Neurol. 2011;70(3):374–83. pmid:21710619

30. Chen Y, Kamat V, Dougherty ER, Bittner ML, Meltzer PS, Trent JM. Ratio statistics of gene expression levels and applications to microarray data analysis. Bioinformatics. 2002;18(9):1207–15. pmid:12217912

31. Sand M, Bechara FG, Gambichler T, Sand D, Bromba M, Hahn SA, et al. Circular RNA expression in cutaneous squamous cell carcinoma. J Dermatol Sci. 2016;83(3):210–8. pmid:27298156

32. Bartel DP. MicroRNAs: target recognition and regulatory functions. cell. 2009;136(2):215–33. pmid:19167326

33. Lu T-P, Lee C-Y, Tsai M-H, Chiu Y-C, Hsiao CK, Lai L-C, et al. miRSystem: an integrated system for characterizing enriched functions and pathways of microRNA targets. PLoS One. 2012;7(8):e42390. pmid:22870325

34. Kanehisa M, Furumichi M, Tanabe M, Sato Y, Morishima K. KEGG: new perspectives on genomes, pathways, diseases and drugs. Nucleic Acids Res. 2017;45(D1):D353–D61. pmid:27899662

35. Szklarczyk D, Franceschini A, Wyder S, Forslund K, Heller D, Huerta-Cepas J, et al. STRING v10: protein–protein interaction networks, integrated over the tree of life. Nucleic Acids Res. 2014:gku1003.

36. Crespel A, Rigau V, Coubes P, Rousset MC, De Bock F, Okano H, et al. Increased number of neural progenitors in human temporal lobe epilepsy. Neurobiol Dis. 2005;19(3):436–50. pmid:16023586

37. Fahrner A, Kann G, Flubacher A, Heinrich C, Freiman TM, Zentner J, et al. Granule cell dispersion is not accompanied by enhanced neurogenesis in temporal lobe epilepsy patients. Exp Neurol. 2007;203(2):320–32. pmid:17049346

38. Sharma J, Mulherkar S, Mukherjee D, Jana NR. Malin regulates Wnt signaling pathway through degradation of dishevelled2. J Biol Chem. 2012;287(9):6830–9. pmid:22223637

39. Busceti CL, Biagioni F, Aronica E, Riozzi B, Storto M, Battaglia G, et al. Induction of the Wnt Inhibitor, Dickkopf‐1, Is Associated with Neurodegeneration Related to Temporal Lobe Epilepsy. Epilepsia. 2007;48(4):694–705. pmid:17437412

40. Li L, Chin L-S, Shupliakov O, Brodin L, Sihra TS, Hvalby O, et al. Impairment of synaptic vesicle clustering and of synaptic transmission, and increased seizure propensity, in synapsin I-deficient mice. Proc Natl Acad Sci USA. 1995;92(20):9235–9. pmid:7568108

41. Choi YS, Cho HY, Hoyt KR, Naegele JR, Obrietan K. IGF‐1 receptor‐mediated ERK/MAPK signaling couples status epilepticus to progenitor cell proliferation in the subgranular layer of the dentate gyrus. Glia. 2008;56(7):791–800. pmid:18338791

42. Berdichevsky Y, Dryer AM, Saponjian Y, Mahoney MM, Pimentel CA, Lucini CA, et al. PI3K-Akt signaling activates mTOR-mediated epileptogenesis in organotypic hippocampal culture model of post-traumatic epilepsy. J Neurosci. 2013;33(21):9056–67. pmid:23699517

43. Kumar V, Zhang M-X, Swank MW, Kunz J, Wu G-Y. Regulation of dendritic morphogenesis by Ras–PI3K–Akt–mTOR and Ras–MAPK signaling pathways. J Neurosci. 2005;25(49):11288–99. pmid:16339024

44. Downward J. Targeting RAS signalling pathways in cancer therapy. Nat Rev Cancer. 2003;3(1):11–22. pmid:12509763

45. Mikkonen M, Soininen H, Kälviäinen R, Tapiola T, Ylinen A, Vapalahti M, et al. Remodeling of neuronal circuitries in human temporal lobe epilepsy: increased expression of highly polysialylated neural cell adhesion molecule in the hippocampus and the entorhinal cortex. Ann Neurol. 1998;44(6):923–34. pmid:9851437

46. Perosa S, Porcionatto M, Cukiert A, Martins J, Passeroti C, Amado D, et al. Glycosaminoglycan levels and proteoglycan expression are altered in the hippocampus of patients with mesial temporal lobe epilepsy. Brain Res Bull. 2002;58(5):509–16. pmid:12242104

47. Kole R, Krainer AR, Altman S. RNA therapeutics: beyond RNA interference and antisense oligonucleotides. Nat Rev Drug Discov. 2012;11(2):125–40. pmid:22262036

48. Muona M, Berkovic SF, Dibbens LM, Oliver KL, Maljevic S, Bayly MA, et al. A recurrent de novo mutation in KCNC1 causes progressive myoclonus epilepsy. Nat Gen. 2015;47(1):39–46.

49. Jeong JH, Jin JS, Kim HN, Kang SM, Liu JC, Lengner CJ, et al. Expression of Runx2 transcription factor in non‐skeletal tissues, sperm and brain. J Cell Physiol. 2008;217(2):511–7. pmid:18636555

50. Hinoi E, Takarada T, Yoneda Y. Glutamate signaling system in bone. J Pharma Sci. 2004;94(3):215–20.

51. Heinrich C, Lähteinen S, Suzuki F, Anne-Marie L, Huber S, Häussler U, et al. Increase in BDNF-mediated TrkB signaling promotes epileptogenesis in a mouse model of mesial temporal lobe epilepsy. Neurobiol Dis. 2011;42(1):35–47. pmid:21220014

52. Yoshihara Y, Onodera H, Iinuma K. Abnormal kainic acid receptor density and reduced seizure susceptibility in dystrophin-deficient mdx mice. Neuroscience. 2003;117(2):391–5. pmid:12614679

53. Toninelli GF, Bernardi C, Quarto M, Lozza G, Memo M, Grilli M. Long‐lasting induction of Notch2 in the hippocampus of kainate‐treated adult mice. NeuroReport. 2003;14(7):917–21. pmid:12802175

54. Baccarini AB, Sachidanandam R, Brown B. Knocking Down the Circular RNA ciRS-E2 Blocks Cancer Cell Proliferation Demonstrating Circular RNAs as a New Therapeutic Target. Mol Ther. 2016;24:S30–S1.