Differentially expressed circRNA and functional pathways in the hippocampus of epileptic mice based on next‐generation sequencing

Abstract

Epilepsy is a clinical syndrome caused by the highly synchronized abnormal discharge of brain neurons. It has the characteristics of paroxysmal, transient, repetitive, and stereotyped. Circular RNAs (circRNAs) are a recently discovered type of noncoding RNA with diverse cellular functions related to their excellent stability; additionally, some circRNAs can bind and regulate microRNAs (miRNAs). The present study was designed to screen the differentially expressed circRNA in an acute seizure model of epilepsy in mice, analyze the related miRNA and mRNA, and study their participating functions and enrichment pathways. In order to obtain the differential expression of circRNA in epilepsy and infer their function, we used next‐generation sequencing and found significantly different transcripts. CIRI (circRNA identifier) software was used to predict circRNA from the hippocampus cDNA, EdgeR was applied for the differential circRNA analysis between samples, and Cytoscape 3.7.2 software was used to draw the network diagram. A total of 10,388 differentially expressed circRNAs were identified, of which 34 were upregulated and 66 were downregulated. Among them, mm9_circ_008777 and mm9_circ_004424 were the key upregulated genes, and their expression in the epilepsy group was verified using Quantitative real‐time PCR (QPCR). The analysis indicated that the extracted gene ontology terms and Kyoto Encyclopedia of Genes and Genomes pathways were closely related to several epilepsy‐associated processes. This study determined that mm9_circ_008777 and mm9_circ_004424 are potential biomarkers of epilepsy, which play important roles in epilepsy‐related pathways. These results could help improve the understanding of the biological mechanisms of circRNAs and epilepsy treatments.

1. INTRODUCTION

Epilepsy is a destructive neurological disease characterized by repeated spontaneous seizures, which have serious health and economic burdens, and greatly reduce the quality of life for those affected. ¹ , ² , ³ , ⁴ This common chronic neurological disease affects almost 50 million people worldwide. ⁵ Regardless of the type of primary injury, common pathological reactions occur around the damaged brain tissue during the incubation period, forming an abnormal neural network and resulting in neuronal synchronization overexcitation. This reaction is often spontaneous and recurrent. Genes related to epilepsy are strictly regulated by various epigenetic mechanisms, and their chronic changes in the hippocampus may play a major role in epilepsy. ⁶ , ⁷ MiRNAs are noncoding RNAs of 20–24 bp length, which can regulate the expression of hundreds of target genes, and have been extensively studied. ⁸ , ⁹ However, recent evidence suggests that another noncoding RNA called circRNA, can regulate gene expression with a higher precision. ¹⁰ , ¹¹ , ¹² , ¹³ Certain properties of circRNAs indicate that they may have a special role in the pathogenesis of central nervous system diseases. ¹⁴ , ¹⁵ CircRNAs are characterized by a covalent closure, mainly produced by reverse splicing and lack of 5′‐3′ polarity and a poly(A) tail, which is the process of reverse splicing of downstream exons to upstream exons. CircRNAs are major role players of the epigenetic regulatory mechanism in the pathogenesis of various diseases; they can function as miRNA sponges, thereby reducing their ability to target mRNAs. ¹⁶ , ¹⁷ Their base pairs directly target mRNA and can trigger the cleavage of mRNA, depending on the degree of complementarity. The current research showed that circRNA is involved in a variety of biological processes (BP), such as gene regulation. ¹⁸ Therefore, in order to determine the potential role of circRNA in epilepsy, we screened the differentially expressed circRNA in an acute seizure model of epilepsy in mice, analyzed the related miRNA and mRNA, and studied their participating functions and enrichment pathways.

2. MATERIALS AND METHODS

2.1. Animal experiment

Twelve healthy adult male Kunming mice weighing 20–22 g and aged 8–10 weeks were used in this study. They were randomly divided into two groups, the epilepsy group and the control group, with six mice in each group. Both groups were subjected to behavioral observation and video EEG monitoring. In the epilepsy group, lithium chloride (127 mg/kg) was injected intraperitoneally, atropine (1 mg/kg) was injected 24 h later, and pilocarpine hydrochloride (10 mg/kg) was injected 30 min later to induce acute epilepsy. After drug administration, the behavior of the animal was observed based on the standard of Racine in 1972. ¹⁹ If no epileptic seizures occurred, pilocarpine was injected every 30 min, and the mice were continually observed. The injections were stopped once status epilepticus appeared. After the seizures continued for 60 min, 10% chloral hydrate (300 mg/kg) was injected intraperitoneally to terminate the seizures. The control group was given 0.9% normal saline 20 mL/kg by intragastric administration every day. The above operations were performed under video monitoring with the video EEG device NATION7128W (Shanghai Nuocheng Electric Co, Ltd). The video monitoring required the insertion of an electronic probing into the hippocampus, which might cause trauma and potentially significant changes in the RNA expression profile of the hippocampus. ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ Therefore, in this study, electrodes were inserted into the left and right temporal cortex. After, the mice were euthanized by cervical dislocation and the brain was immediately removed. The hippocampus was dissected and stored at −80°C. All animals were managed in accordance with the standard procedures approved by the Institutional Animal Care and Use Committee of The First Affiliated Hospital of Guangxi University of Chinese Medicine (license: SYXK Gui 2009‐0001).

2.2. Sample processing and circRNA collection

Total RNA was extracted using RNAiso Plus Total RNA extraction reagent (9109, TAKARA) following the manufacturer's instructions. RNA integrity was checked using an Agilent Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA, US). RNAClean XP Kit (A63987, Beckman Coulter, Inc, CA, USA) and RNase‐Free DNase Set (79254, QIAGEN, GmBH, Germany) were further used to purify the qualified total RNA. Ribo‐Zero rRNA removal kit (Epicentre, MRZMB126) was used to digest total RNA to remove linear RNA and fragment the RNA. The first strand cDNA was synthesized immediately after fragmentation. Following this, 25 μL of the thawed second‐strand master mix was added to 25 μL of the cDNA first‐strand synthesis product and mixed well to synthesize the second‐strand cDNA. Next, end repair and adenylate 3′ ends were performed, and the adapters were ligated. Lastly, PCR was used to amplify and purify the enriched cDNA templates, which produced the sequencing library to be tested.

2.3. Library quality control

Qubit® 2.0 Fluorometer (Invitrogen, Q32866) and Qubit™ dsDNA HS Kit (Invitrogen, Q32854) were used to detect the library concentration degree. The FlashGel Dock system combined with FlashGel DNA Cassette detected the size and purity of the library. The peak size of the library was generally around 350 bp (according to the size of the library insert).

2.4. Next‐generation sequencing

Next‐generation sequencing was performed with a high‐throughput sequencer (Illumina Hiseq 2000/2500, Miseq) to sequence the cDNAs. Since the raw reads obtained by sequencing might contain unqualified reads with low overall quality, which might impact the analysis quality, they were filtered out using Seqtk (the genome version is GRCm38.p4 [mm10]; ftp://ftp.ensembl.org/pub/release-83/fasta/mus_musculus/dna/Mus_musculus.GRCm38.dna.primary_assembly.fa.gz). Finally, the CIRI software established by Gao et al. ²⁶ was used to predict the circRNA sequence data.

2.5. Quality control of circRNA sequencing results and analysis of circRNA differential expression

We selected five circRNAs with the most significant differences for qPCR verification, and the relative mRNA expression was calculated using 2^−ΔΔct method with normal saline group as control. The quality control standard was as follows: the amount of data was about 10 G/sample, and the proportion of base quality in each direction greater than 20 (Q20) was not less than 90%. All samples in the sequencing volume met the requirements, and the sequencing results conformed to the standards, which could be used for data analysis. The results are shown in Table 1. EdgeR was used to analyze the different circRNA between samples; the p‐value was obtained, multiple hypothesis testing correction was performed, and the false discovery rate was calculated to determine the p‐value threshold. The corrected p‐value is the q‐value. At the same time, we calculated the multiple of differential expression based on the SRPBM value (fold‐change). ²⁷ In addition, a volcano map was constructed based on p value and log2 (p < 0.05 and | log2 [fold change] | > 1).

TABLE 1.

All samples met the requirements and sequencing results met the quality standard

Sample name	Seq. type	Orientation	Raw bases (G)	Q20 ratio
shenliyanshuizu‐B4	mRNA	Forward/Reverse	13.14	97.19%
dianxianzu‐A5	mRNA	Forward/Reverse	10.91	97.37%
shenliyanshuizu‐B5	mRNA	Forward/Reverse	13.07	96.59%
dianxianzu‐A2	mRNA	Forward/Reverse	11.05	97.18%
dianxianzu‐A7	mRNA	Forward/Reverse	12.99	97.06%
dianxianzu‐A3	mRNA	Forward/Reverse	13.59	97.14%
shenliyanshuizu‐B2	mRNA	Forward/Reverse	12.46	97.29%
dianxianzu‐A4	mRNA	Forward/Reverse	11.41	97.01%
shenliyanshuizu‐B1	mRNA	Forward/Reverse	12.58	97.19%
shenliyanshuizu‐B6	mRNA	Forward/Reverse	13.44	96.89%
dianxianzu‐A8	mRNA	Forward/Reverse	10.38	97.32%
shenliyanshuizu‐B7	mRNA	Forward/Reverse	12.76	97.15%

Abbreviation: Q20, bases of Q ≥ 20/all bases of sequencing.

2.6. Prediction of circRNA that regulate miRNA and related target genes

Based on the differential expression of circRNAs, their official names were obtained from circBase (http://www.circbase.org/). The Regrna2.0 database (http://regrna2.mbc.nctu.edu.tw/detection.html) was used to predict miRNA regulation by circRNA. mRNA was predicted through the TargetScan database (http://www.targetscan.org/vert_72/). The default parameters psRobot were used to obtain miRNA to predict the target mRNA.

In addition, we established the competing endogenous RNA (ceRNA) network relationship through the abovementioned differentially expressed circRNA, miRNA, and mRNA, and used Cytoscape 3.7.2 software for visual data analysis.

2.7. Functional enrichment of differentially expressed genes (gene ontology [GO] and Kyoto encyclopedia of genes and genomes [KEGG])

According to the location information of circRNA, the protein coding genes corresponding to circRNA, the parental genes, were obtained. We performed GO and KEGG functional pathway enrichment analysis on the parental genes corresponding to the differential circRNA. The threshold was set at q‐value ≤0.05, which was defined as GO terms that were significantly enriched in differentially expressed genes.

2.8. Quantitative real‐time polymerase chain reaction (qPCR) detected circRNA expression

Quantitative PCR (PCR machine, Beijing Donglin Prosperity Technology Co, Ltd, DL9700) was used to verify the expression of differential circRNAs. First, the sample tissue was ground and homogenized prior to the addition of TRIzol (Tiangen Biochemical Technology [Beijing] Co, Ltd). Standard manufacturer's protocol was followed for the TRIzol‐based RNA extraction. After the addition of DEPC H₂O (Sigma, 40718) to fully dissolve the RNA, the RNA was stored at −80°C (Thermo Fisher Instruments Co, Ltd, ULTS1368). The ultramicro nucleic acid analyzer (Hangzhou Aosheng Instrument Co, Ltd, Nano‐200) was used to detect RNA concentration.

2.9. Statistical analysis

SPSS 17.0 software was used for statistical analysis. When conforming to the normal distribution, one‐way analysis of variance was used for comparison between groups, and the LSD test was used for pairwise comparison. The rank sum test was used for comparison between non‐normally distributed groups.

3. RESULTS

3.1. Differential expression of circRNA

A total of 10,388 circRNAs were identified (FC >2 or <0.5, p < 0.0.1). Among these circRNAs, 34 were upregulated and 66 were downregulated (p ≤ 0.05, fold‐change ≥2). Table 2 lists the top 15 differentially expressed and currently known circRNAs. Figure 1(A) contains the genome coverage distribution, which shows the chromosome coverage of linear RNA and circRNA reads in each sample group. The outermost circle in the figure is the genome. Each circle represents the chromosome read coverage of the sample, and the red represents the circRNA linker read coverage of the sample. The Perl script was used to classify and count the predicted circRNA, and the result is shown in Figure 1(B). The circRNA expression profile is shown in the heat map (Figure 1C), and the hierarchical clustering shows the difference in circRNA expression: there are significant differences between the hippocampal tissue of epileptic mice and the control mice. The scatter plot of expression correlation between samples and the volcano plot of differential genes are shown in Figure 1(D,E).

TABLE 2.

The top 15 differentially expressed and currently known circRNAs based on P‐value

gene_id	log2FC	p‐value	Regulation
mm9_circ_000233	1.17214296	0.000515965	DOWN
mm9_circ_005164	1.54298014	0.000661442	DOWN
mm9_circ_016270	1.38021686	0.001621806	UP
mm9_circ_015216	2.72107335	0.001708148	DOWN
mm9_circ_008777	3.37497860	0.002235524	UP
mm9_circ_005327	2.17124059	0.002811562	DOWN
mm9_circ_004424	1.89741591	0.004170345	UP
mm9_circ_003739	3.09149267	0.004548419	DOWN
mm9_circ_009955	1.30805614	0.004950204	DOWN
mm9_circ_018127	2.81692055	0.008826780	DOWN
mm9_circ_003677	1.25053930	0.012102298	DOWN
mm9_circ_000453	2.12683680	0.023333050	DOWN
mm9_circ_001254	1.16967134	0.031013153	UP
mm9_circ_008922	3.35538010	0.031189069	DOWN
mm9_circ_006079	3.35704374	0.032701765	UP

FIGURE 1.

Genomic coverage distribution of each sample group (A). Histogram of circRNA classification statistics of each sample group (B). Heat map (C), scatter map (D), and volcano map (E) of circRNA gene distribution in the epilepsy group and the control group; the ordinate is ‐log10 (P value) and the abscissa is log2 (Fold‐change). In the scatter plot and volcano plot, red indicates upregulation of circRNA and blue indicates downregulation of circRNA

3.2. Functional and pathway enrichment analysis

According to the position information of the differential circRNA, the protein‐coding gene corresponding to the genomic position of the circRNA, the parental gene, was obtained. Following this, we performed GO function analysis on the parental genes corresponding to the differential circRNA. First, the number of genes corresponding to the three levels of BP, cell composition, and molecular function (MF), were calculated, as shown in Figure 2(A). The BP were mainly enriched in biological regulation, cellular processes, metabolic processes, regulatory BP, monomer processes, and so on. Cellular component was mainly concentrated in cells, cytoplasm, cell membrane, and organelles. MF was mainly manifested in binding and catalytic activity. For the results of GO enrichment analysis, the first 30 GO analyses are shown in Figure 2(C). In that picture, rich factor refers to the ratio of the number of genes in the GO entry of the differential circRNA parental genes to the total number of genes. The larger the rich factor, the greater the degree of enrichment. The q‐value is the p‐value after correction for multiple hypothesis testing. The smaller the value, the more significant the enrichment. In addition, the same principle as GO enrichment was also used to perform KEGG pathway statistics on differential genes (Figure 2B) and enrichment analysis (Figure 2D). The most significantly enriched KEGG pathways were the ErbB signaling pathway, VEGF signaling pathway, regulation of actin cytoskeleton, GnRH signaling pathway, and T‐cell receptor signaling pathway.

FIGURE 2.

GO and KEGG pathway analysis of differentially expressed genes. (A) GO functional classification statistics chart of differential circRNA parental gene, statistics on the number of genes corresponding to the three levels of biological process, cellular component, and molecular function. (B) KEGG pathway classification statistics of differential circRNA parental genes; (C) and (D) are bubble charts of the first thirty genes that were significantly enriched in GO enrichment analysis and KEGG pathway enrichment analysis, respectively. The enrichment factor represents the ratio between differentially expressed genes and all annotated genes enriched in this pathway. The bubble scale represents the number of different genes; the depth of the bubble color represents the p‐value

3.3. circRNA–miRNA–mRNA ceRNA regulatory network

The related miRNA and target gene mRNA of differentially expressed circRNA were predicted. This included 21 miRNAs and 45 target gene mRNAs, and then the circRNA–miRNA–mRNA regulatory network was drawn, as shown in Figure 3. The node size was adjusted according to the degree value; the greater the degree value, the larger the node.

FIGURE 3.

circRNA is red and diamond‐shaped; miRNA is gray and inverted triangle; mRNA is blue and oval

3.4. QPCR verification of differential circRNA

Firstly, the selected circular RNA sequence were queried through the circBase database (http://www.circbase.org/), and the circular name was converted into an official ID, which are shown in Table 3. qPCR results showed that relative to the control group, the expression of mmu_circ_000023, mmu_circ_0000086, and mmu_circ_0000872 were upregulated in the epilepsy group (p < 0.05); the expression of mmu_circ_0001663 and mmu_circ_0000705 had no significant difference (p > 0.05). The result is shown in Figure 4.

TABLE 3.

CircRNA name and ID comparison

Gene type	Name	ID
Circular RNA	mm9_circ_008777	mmu_circ_0000230
Circular RNA	mm9_circ_006079	mmu_circ_0000086
Circular RNA	mm9_circ_004459	mmu_circ_0001663
Circular RNA	mm9_circ_004424	mmu_circ_0000872
Circular RNA	mm9_circ_016270	mmu_circ_0000705

FIGURE 4.

The expression of target gene in epilepsy group and control group

4. DISCUSSION

CircRNA, a new member of the RNA family, is a small noncoding RNA with highly conservative and specific tissue expression. CircRNAs have become a hot spot in transcriptomics research; since they lack the poly(A) tails at the 5′ and 3′ ends and covalently form a loop structure, they are not easily degraded by the exonuclease RNaseR. ²⁸ CircRNAs play multifunctional roles as microRNA (miRNA) sponges, regulators of transcription and post‐transcription, parental gene expression, and translation of proteins in various diseased conditions. ²⁹ In order to study the role of circRNA in epilepsy, we used next‐generation sequencing technology to study the expression profile of circRNA in the entire genome of mice with acute epilepsy (n = 6) versus normal control mice. We found several circRNAs, of which 34 were upregulated and 66 were downregulated. Additionally, we found predicted miRNAs and target gene mRNAs related to differential circRNAs. After the study was completed, a circRNA–miRNA–mRNA regulatory network was constructed; mm9_circ_004424 and mm9_circ_008777 were selected as key genes and qPCR verified that they were highly expressed in the epilepsy group. However, unfortunately, whether the change of circRNA is the cause of epilepsy or the result of epilepsy, the experiment has not yet reached a conclusion. Further discussion will be needed in subsequent cell and animal experiments. Presently, a few circRNAs related to the occurrence and development of epilepsy have been identified. Gong et al. ³⁰ found that the expression of circRNA‐0067835 in the tissues and plasma of patients with refractory epilepsy was significantly reduced and confirmed that circRNA‐0067835 regulates the expression of FoXo3 by acting on the corpus cavernosum of miR‐155, thereby regulating the refractory progress in the expression of sexual epilepsy. Lee et al. ³¹ considered that circRNA dysregulation might regulate a variety of disease‐related mRNAs through circRNA–miRNA–mRNA interactions, thus playing a pathophysiological role in epilepsy. At present, no research report on the functions of mm9_circ_004424 and mm9_circ_008777 has been retrieved. This highlights the significance of this research, which provides references and clues for future epilepsy‐related circRNA research and provides ideas and evidence for the study of circRNA regulatory mechanism of epilepsy.

Through GO function and KEGG pathway analysis, the BP and signaling pathways involved in differential genes include biological regulation, cellular processes, metabolic processes, binding, catalytic activity, ErbB signaling pathway, VEGF signaling pathway, chemokine signaling pathway, GnRH signaling pathway, and T‐cell receptor signaling pathway have been determined. Zhu et al. ³² used samples from symptomatic epilepsy patients with cavernous hemangioma (CA), and after excluding the influence of CA, found that the levels of NRG1 and ErbB4 proteins in the temporal cortex of symptomatic epilepsy patients significantly increased. This indicated that the NRG1/ErbB4 signal pathway might be involved in the pathogenesis of primary epilepsy by regulating the excitability of neurons. Vascular endothelial growth factor (VEGF) is an attractive target for regulating brain function at the neurovascular interface. It has both advantages and disadvantages in epileptic seizures. Han et al. ³³ found that activating the VEGF/VEGFR2 signaling pathway could actively promote the initial stage of neurogenesis and reduce cognitive dysfunction after epileptic seizures. At present, chemokine pathways have attracted an increasing amount of attention. Chemokines are small cytokines secreted by blood cells, which can be used as signals for leukocyte migration, including CCL2, CCL3, CCL5, CX3CL1, and its receptors. ³⁴ Chemokines can directly affect the excitatory receptors of neurons, and they are expressed both pre‐ and post‐synaptically. Recently, numerous studies have shown that they are closely related to the occurrence and development of epilepsy. Chemokines are not only produced and released due to seizures, but they may also cause seizures. ³⁵ , ³⁶ Research has shown that the GnRH signaling pathway is related to epilepsy. ³⁷ Most of these pathways have been confirmed, and our research results are consistent with the current research results, indicating that the results of this research have a certain theoretical basis. Therefore, we infer that mm9_circ_004424 and mm9_circ_008777 may play GO functions through biological regulation, cellular processes, metabolic processes and catalytic activity, and participate in the occurrence of epilepsy through signal pathways such as the ErbB signaling pathway, VEGF signaling pathway, chemokine signaling pathway, and GnRH signaling pathway.

In summary, this study used next‐generation sequencing technology to analyze the circRNA expression profile data in acute epileptic seizure mice versus normal mice and screened the differential expression of circRNA between the epileptic hippocampus and normal hippocampus. Preliminary explorations of functions and signal pathways were carried out to provide a meaningful reference for the basic research of epilepsy regulating‐related circRNA. The results showed that MM9_circ_008777 and MM9_circ_004424 can be used as potential diagnostic indicators in the diagnosis of epilepsy, as well as the possibility of clinical intervention for their expression to achieve treatment or remission of epilepsy.

CONFLICT OF INTEREST

The authors declared no potential conflicts of interest.

Liao X‐Q, Yu H‐C, Diao L‐M, et al. Differentially expressed circRNA and functional pathways in the hippocampus of epileptic mice based on next‐generation sequencing. Kaohsiung J Med Sci. 2021;37:803–811. 10.1002/kjm2.12404