Expression and functional analysis of lncRNAs in the hippocampus of immature rats with status epilepticus

Abstract Long non‐coding RNAs (lncRNAs) have been implicated in the regulation of gene expression at various levels. However, to date, the expression profile of lncRNAs in status epilepticus (SE) was unclear. In our study, the expression profile of lncRNAs was investigated by high‐throughput sequencing based on a lithium/pilocarpine‐induced SE model in immature rats. Furthermore, weighted correlation network analysis (WGCNA), gene ontology (GO) analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis were performed to construct co‐expression networks and establish functions of the identified hub lncRNAs in SE. The functional role of a hub lncRNA (NONRATT010788.2) in SE was investigated in an in vitro model. Our results indicated that 7082 lncRNAs (3522 up‐regulated and 3560 down‐regulated), which are involved in cell proliferation, inflammatory responses, angiogenesis and autophagy, were dysregulated in the hippocampus of immature rats with SE. Additionally, WGCNA identified 667 up‐regulated hub lncRNAs in turquoise module that were involved in apoptosis, inflammatory responses and angiogenesis via regulation of HIF‐1, p53 and chemokine signalling pathways and via inflammatory mediator regulation of TRP channels. Knockdown of an identified hub lncRNA (NONRATT010788.2) inhibited neuronal apoptosis in vitro. Taken together, our study is the first to demonstrate the expression profile and potential function of lncRNAs in the hippocampus of immature rats with SE. The defined hub lncRNAs may participate in the pathogenesis of SE via lncRNA‐miRNA‐mRNA network.

Long non-coding RNAs (lncRNAs) are new members of the ncRNA family that are longer than 200 nucleotides in length. 7 LncRNAs have been implicated in the regulation of gene expression at the epigenetic, transcriptional or post-transcriptional level, even though they do not encode any protein products themselves. 7,8 Through binding to the specific miRNAs, lncRNA could function as a competing endogenous RNA (ceRNA) in regulating protein expression. 9,10 Various studies have revealed that lncRNAs are widely expressed in several human tissues and cells, 11,12 and participate in progression of multiple diseases, including tumorigenesis, 13,14 hepatitis C virus infection 15 and neurobiology of stress and depression. 16 In SE, H19-a lncRNA-was demonstrated to contribute to epileptogenesis by aggravating SE-induced neuronal loss, glial cell activation, mossy fibre sprouting and cognitive impairments in epileptic rats. 17 Whole-transcriptome screening revealed that H19 exhibited diverse functions related to epileptogenesis, including demyelination, immune and inflammatory responses, cell apoptosis and activation of MAPK. 18 UCA1, another lncRNA, may participate in the pathogenesis of epilepsy, which is evidenced by the dynamic change in the expressions of UCA1 and NF-κB during epilepsy. 19 However, the expression and potential function of lncRNAs in SE are still unclear.
Here, we determined the expression profile of lncRNAs by highthroughput sequencing based on a lithium/pilocarpine-induced SE model in immature rats. KEGG and GO analyses were performed to predict the potential function of dysregulated lncRNAs. The hub ln-cRNAs were determined by weighted correlation network analysis (WGCNA), and their potential function was predicted according to the lncRNA-miRNA-mRNA network. Furthermore, the functional role of a hub lncRNA (NONRATT010788.2) in SE was investigated in an in vitro model.

| SE model
The SE model was established as our previous study indicated. 20 Briefly speaking, female Sprague Dawley rats with mixed-sex litters were housed in a temperature-and light-controlled facility with food and water. The 25-days-old pubs were intraperitoneally (i.p.) injected with lithium chloride (125 mg/kg, Sigma) at 18 hours before pilocarpine injection (i.p. injection, 40 mg/kg, Sigma). Racine's scale was performed to determine the severity of convulsions of rats, and the animals with a score of 4-5 were used in the present study. Then, diazepam (10 mg/kg) was intraperitoneally injected to terminate the seizure attacks of the SE rats. The rats that were injected with the same amount of normal saline were defined as control. All the animals were purchased from the animal centre of Sichuan University, and related research was approved by the Sichuan University Committee on Animal Research. At 24 hours post-SE onset, the animals (n = 4 for SE group and control group each) were killed for hippocampus dissection. The hippocampus were preserved in RNAlater (−20°C, Qiagen) for further RNA extraction. HiSeqXten platform. Reads with more than 10% N (Unable to determine base information), with adapter sequence, or of low quality were removed from the raw reads to obtain clean reads. Finally, clean reads were compared with rat genome from NCBI using hisat 2. Differentially expressed genes were selected according to the threshold set for a fold change ≥ 2.0 and a unadjusted P value ≤ .05. P values were calculated with a t test. All of the raw data were supplied on line (BioProject: PRJNA532235 https ://datav iew.ncbi.nlm.

| LncRNA library construction and highthroughput sequencing
nih.gov/objec t/PR JNA 53223 5?revie wer=h0ao7 p9dhm maj16 jtik7 nas70u). Differential expression test was analysed performed with DESeq R packages according to the packages manual FDRs were controlled using the Benjamini-Hochberg method at an FDR of 5%.

| Weighted gene co-expression network analysis (WGCNA)
WGCNA is a gene co-expression network-based strategy for identifying key genes, which is a comprehensive collection of R functions. 21 WGCNA was based on the expression profiles of lncRNAs and mRNAs from SE and control group. The genes (lncRNAs and mRNAs) with similar expression trend were divided into one module.
The screening criteria for selecting genes of WGCNA analysis is P value < .05. Then, the GO, KEGG and co-expression network were analysed and conducted with external software package following the tutorials provided.

| Quantitative PCR analysis
The hippocampus was collected for total isolation by TRIzol reagent (Invitrogen) following the instructions of manufacturer. The reverse transcription was performed with the PrimeScript RT Reagent Kit (Perfect Real Time, Takara), following the manufacturer's instructions. Then, the cDNA samples were amplified for qPCR using TB Green Premix Ex TaqTM II (TliRNaseH Plus, Takara). The reaction of qPCR was set at 95°C (30 seconds) for pre-denaturation, then a total of 40 cycles (95°C for 5 seconds and 58°C for 30 seconds). The relative expression of RNAs was calculated based on the standard curve and C t value. The housekeeping gene, β-actin, was used as a loading control. cbi.pku.edu.cn/) was used for KEGG pathway analysis. Then, the significantly enriched GO or KEGG terms were analysed using hyper geometric test with P value ≤ .05.

| LncRNA-miRNA-mRNA network construction
To construct the lncRNA-miRNA-mRNA network, miRNA-mRNA, miRNA-lncRNA target relationships were predicted by target prediction database (http://www.targe tscan.org/vert_71/and http:// www.mirba se.org/). To assess the reliability of candidate ceRNA pairs, filtering strategy is used: (a) adjusted P value cut-off for each ceRNA pairs is set to .01 and (b) circRNA and gene should have the same direction in DEG analysis, because ceRNA pairs were reported to be a positive correlation inexpression. To calculate the probability that a circRNA is a target ceRNA, a Fisher exact test is executed for each pair (circRNA gene) separately. The correlation value cut-off was 0.90. Cytoscape version 3.6.1 was used for assembly and visualization of the network.

| Cell culture and treatment
The rat hippocampal neurons were purchased from JENNIO Bio.
Tec. and maintained in a humidified atmosphere (Thermo Fisher) containing 5%CO 2 at 37°C. The Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum (Gibco) was used for neuron cell culture. SiRNA targeting NONRATT010788.2 and siRNA-NC (RiboBio Tech.) were transfected with fiboFECT CP Transfection kit (RiboBio Tech.) following the manufacturer's protocol. At 48 hours after transfection, the cells were collected for total RNA extraction and related assay. All transfections were performed in triplicate.

| Statistical analysis
All of the numerical continuous data were presented as mean ± standard deviation. GraphPad Prism 5.0 was used for statistical analysis, and Student's t tests were used for comparing two groups. The P value < .05 was considered as statistically significant. All experiments were repeated four to six times.

| SE-induced changes in expression of lncRNAs in the hippocampus of immature rats
To investigate the expression profile of lncRNAs during SE development, lithium and pilocarpine were injected to establish the SE model in immature rats. After 24 hours, the hippocampus was collected for high-throughput sequencing. As shown in Figure 1A  and Abld8) were picked for qPCR determination. Our results indicated that the high-throughput sequencing results on the expression of selected genes corroborated with those of qPCR ( Figure 1E and F). Taken together, we identified the dysregulated lncRNAs and mRNAs in SE.

| WGCNA of dysregulated lncRNAS
To identify genes expressed together on a higher systems level, all dysregulated lncRNAs and mRNAs were clustered into nine gene modules based on expression trend ( Figure 3A). The brown module (r = −.97) and turquoise module (r = 1.0), which exhibited strongest correlation with the SE, were selected for further analysis. Hub gene analysis implied that most of the dysregulated genes in brown and turquoise modules were the hub genes and were highly correlated with the pathological phenotype of SE ( Figure 3B and C). In brown module, 104 lncRNAs and 1030 mRNAs were down-regulated in the hippocampus of SE rats ( Figure 3D). Meanwhile, in the turquoise module, 667 lncRNAs and 4482 mRNAs were up-regulated in the hippocampus of SE rats ( Figure 3E). Taken together, the above results indicated that the lncRNAs in the brown and turquoise modules regulated the pathogenesis of SE.
F I G U R E 1 Expression profile of lncRNAs in the hippocampus of immature rats following SE. A and B, The hippocampus of immature rats was collected for high-throughput sequencing at 24 h post-SE. Heatmap of differential expression of lncRNA (A) and coding RNAs (B). C and D, Distribution of differentially expressed lncRNAs (A) and coding RNAs. E and F, Comparison analysis of dysregulated lncRNAs (E) and coding RNAs (F) between high-throughput sequencing and qRT-PCR results

| Potential functional analysis of dysregulated lncRNAS in brown and turquoise modules
To further analyse the potential function of the dysregulated lncR-NAs in the brown and turquoise modules, the KEGG and GO-BP analyses were performed based on the predicted function of the dysregulated genes. Pathway analysis indicated that the brown module was involved in several pathways, including GABAergic synapse, cAMP signalling pathway, MAPK signalling pathway and Ras signalling pathway ( Figure 4A), while the apoptosis, Rap1 signalling pathway, cAMP signalling pathway, chemokine signalling pathway, inflammatory mediator regulation of TRP channels and PI3K-Akt signalling pathway were enriched in the turquoise module ( Figure 4B).
The highest enriched GO terms targeted by the dysregulated genes in brown module included ion transmembrane transport, brain development and axon guidance ( Figure 4C). The biological processes of angiogenesis, cell migration, cell proliferation, inflammatory response and apoptotic process were enriched in turquoise module ( Figure 4D). These results indicated that the hub genes in brown and turquoise modules may play a crucial role in the pathogenesis of SE.

| Dysregulated lncRNAS regulate SE development by acting as a ceRNA
To define the potential ceRNA mechanism under differentially expressed mRNAs and lncRNAs, a ceRNA network was constructed.
Thus, we constructed a coding lncRNA-miRNA-mRNA co-expression network based on the miRNA expression profile demonstrated by our previous study. 22 The network analysis showed that six pathways were regulated by the dysregulated lncRNAs from turquoise module ( Figure 5). As shown in Figure 5A Figure 5E). The PI3K-Akt signalling pathway was also regulated by 123 lncRNAs that correlated with 95 miRNAs ( Figure 5F). Collectively, these results suggested that dysregulated lncRNAs from turquoise module regulate SE development by acting as a ceRNA.

| Knockdown of NONRATT010788.2 inhibits neuronal apoptosis
To determine the function of the identified hub lncRNAs, an apoptosis-related lncRNA (NONRATT010788.2) in turquoise module was selected for further analysis. The high-throughput sequencing results indicated that NONRATT010788.2 was significantly up-regulated in the hippocampus of immature rats with SE ( Figure 6A). Further, qPCR determination confirmed the up-regulation of NONRATT010788.2 in the hippocampus of immature rats with SE ( Figure 6B). SiRNA targeting NONRATT010788.2 was used to transfect the rat hippocampal neurons, which were treated with Mg 2+ -free HEPES. Our results confirmed the downregulation of NONRATT010788.2 in rat hippocampal neurons transfected with siRNA targeting NONRATT010788.2 ( Figure 6C). TUNEL assay indicated that knockdown of NONRATT010788.2 efficiently inhibited the apoptosis of rat hippocampal neurons induced by Mg 2+ -free HEPES ( Figure 6D). Flow cytometry results also confirmed the inhibition of rat hippocampal neuron apoptosis by siRNA targeting NONRATT010788.2 ( Figure 6E). Thus, the identified hub lncRNA, NONRATT010788.2, promoted neuronal apoptosis. WGCNA is an R software package which can be used to search for clusters (modules) of highly correlated genes, for summarizing an intramodular hub gene, for weighted correlation network analysis, for example co-expression network analysis of gene expression data. 21 Based on the WGCNA, the module with a high correlation with the disease progression would be screened out. 29,30 The dysregulated genes in the module are usually defined as hub genes that are involved in the pathogenesis of disease. 29,30 In the present study, the differentially expressed lncRNAs and mRNAs were divided into nine modules. There was a strong correlation between SE and two modules (brown module and turquoise module). Further functional prediction and lncRNA-miRNA-mRNA co-expression network analysis indicated that the up-regulated hub lncRNAs in turquoise module were involved in SE progression through regulation of apoptosis, Rap1 signalling pathway, cAMP signalling pathway, chemokine signalling pathway, inflammatory mediator regulation of TRP channels and PI3K-Akt signalling pathway, which have been demonstrated to play an important role in the pathogenesis of SE. [31][32][33][34][35] Furthermore, the in vitro model was studied to confirm the functional role of an identified hub lncRNA (NONRATT010788.2) that was associated with apoptosis. The constructed co-expression network indicated that lncRNA, NONRATT010788.2, inversely correlated with miR-324-3p, which plays a negative role in apoptosis. 36,37 It is indicated that NONRATT010788.2 may bind to miR-324-3p and hence regulate neuronal apoptosis. These results confirmed the promotional role of NONRATT010788.2 in neuronal apoptosis, but further investigations are needed to demonstrate the underlying mechanism.
In conclusion, our study is the first to demonstrate the expression profile and potential function of lncRNAs in the hippocampus of immature rats with SE. The defined hub lncRNAs may participate in SE development through regulation of apoptosis, inflammatory responses and angiogenesis in the hippocampus of immature rats.
Thus, the dysregulated lncRNAs would be the potential therapy and diagnosis targets for SE. But, further experimental studies are needed to investigate the functions of these hub lncRNAs in SE, as well as the underlying mechanism.

ACK N OWLED G EM ENTS
Thanks to Yuanzhi Cheng from Chengdu Basebiotech Co., Ltd for providing assistances on bioinformatic analysis.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no potential conflicts of interest.

E TH I C A L A PPROVA L
All experimental procedures were approved by the Institutional Animal Care and Use Committees of Sichuan University.

CO N S E NT FO R PU B LI C ATI O N
Consent for publication is not applicable in this study. No individual person's data were used.

DATA AVA I L A B I L I T Y S TAT E M E N T
Not applicable.