Comprehensive analysis of lncRNA–miRNA–mRNA during proliferative phase of rat liver regeneration

Abstract This study aims to reveal the regulatory mechanism of lncRNAs–miRNAs–mRNAs network during the proliferative phase of liver regeneration (LR). High‐throughput sequencing technology was performed, and a total of 1,738 differentially expressed lncRNAs (DE lncRNAs), 167 known differentially expressed miRNAs (DE miRNAs), and 2,727 differentially expressed mRNAs were identified. Then, the target DE lncRNAs and DE mRNAs regulated by the same miRNAs were screened and a ceRNA regulatory network containing 32 miRNAs, 107 lncRNAs, and 270 mRNAs was constructed. Insulin signaling pathway, pyrimidine metabolism, axon guidance, carbohydrate digestion and absorption, and pyruvate metabolism were significantly enriched in the network. Through literature review and the regulatory relationship between lncRNAs and miRNAs, nine core lncRNAs were identified, which might play important roles during the proliferative phase of rat LR. This study analyzed lncRNA–miRNA–mRNA regulatory network for the first time during the proliferative phase of rat LR, providing clues for exploring the mechanism of LR and the treatment of liver diseases.


| INTRODUCTION
The liver is one of the important organs in human and animal, which is responsible for a variety of physiological functions. It has a strong ability of regeneration after liver loss or toxic injury. Liver regeneration is a highly organized tissue growth process of restoring the original framework structure and tissue specific function after liver injury (Jeon et al., 2013). It was usually divided into three phases including initiation, proliferation, and termination. In the proliferative phase, the main process is the proliferation of hepatocyte, which replicate once or twice under the synergistic action of various growth factors and inflammatory cytokines. Long noncoding RNA (lncRNA) is a class of RNA molecules with a length of more than 200 nucleotides (nt) and lacking an open reading frame that plays crucial roles in epigenetics, transcriptional regulation, and posttranscriptional regulation (Maruyama & Suzuki, 2012). Studies have shown that lncRNAs were not only involved in normal physiological activities but also related to the occurrence and development of various tumors (Chen et al., 2013). MicroRNA (miRNA) is a class of endogenous noncoding single-stranded RNA molecules with a length of approximately 22 nt. It plays important roles in regulating the expression of messenger RNAs (mRNAs) through specifically binding to the 3′-untranslated region (3′-UTR) of the encoding gene. There is evidence that miRNAs can regulate a variety of cell processes and developmental processes (Krol, Loedige, & Filipowicz, 2010).  proposed a competition endogenous RNA (ceRNA) hypothesis, which pointed that mRNA, transcriptional pseudogenes and long noncoding RNA could communicate to each other through their ability to compete for microRNA binding using microRNA response elements (MREs) (Wang, Zhang, He, & Gou, 2018;. Subsequently, increasing evidence indicated that lncRNAs, as ceRNA, were associated with a variety of cancers, including hepatocellular carcinoma (HCC) . Accordingly, it is necessary to explore the regulatory network of lncRNA-miRNA-mRNA during the proliferative phase of liver regeneration (LR). In present study, highthroughput sequencing technology was performed to obtain the miRNA, mRNA, and lncRNAs expression data during the proliferative phase of rat LR and lncRNAs-miRNAs-mRNAs regulatory network was established. Our findings might lay the foundation for further investigate the lncRNAs-miRNAs-mRNAs interaction network during LR and liver-associated diseases.

| Preparation of rat LR model after 2/3 hepatectomy
The healthy adult male Sprague-Dawley (SD) rats weighing 210-250 g were provided by Laboratory Animal Center of Zhengzhou University (Zhengzhou, China). These rats were raised in a controlled temperature room 19-23℃ with a relative humidity 50-70% and illumination time 12 hr/day (8:00-20:00), and permitted to freely have water and food. A total of 36 rats were randomly divided into six groups with six rats per group: Five partial hepatectomy (PH) groups and one normal group (CG). The rats in PH group were subjected to 2/3 PH in accordance with the method of Xu C. et al. (2010). They were anesthetized and killed at 0, 12, 24, 30, 36, and 72 hr after surgery. The right liver lobe was mixed each time point of six rats and restored in −80℃. All operations conformed to the Animal Protection Law of China and Animal Ethics.

| Sequencing of lncRNA and mRNA and identification of DE lncRNA and DE mRNA
The mirVana miRNA Isolation Kit (Ambion) was used to extract total RNA and the TruSeq Stranded Total RNA with Ribo-Zero Gold was used to construct the complementary DNA (cDNA) libraries. In brief, after total RNA extracted and ribosomal RNA digested, the RNA was broken into short fragments by the interrupt reagent. The first cDNA chain was synthesized using these short fragments as template and a random six-base as primer. Then the second cDNA chain was synthesized using the first cDNA as template and the dTTP was replaced with dUTP. After repairing the end, jointing adenylate 3′ ends and sequence adapters, the second cDNA chain was digested by UNG (Uracil-N-Glycosylase) enzyme, and the first cDNA with different joints was retained. Agarose gel electrophoresis was used to select the fragment size. Finally, polymerase chain reaction (PCR) amplification was performed. After the constructed library passed the quality inspection with Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara), Illumina sequencing platform (Hiseq X Ten) was used for sequencing.
Raw reads of fastq format were performed quality preprocessing by using Trimmomatic (0.36). After wiping off the adapter and lowquality reads, the clean reads were obtained. The hisat2 (2.2.1.0) was use to align the clean reads with the rat reference genome. For

| Sequencing of miRNA and identification of DE miRNA
The mirVana miRNA Isolation Kit (Ambion) and TruSeq Small RNA Sample Prep Kits were used to extract total RNA and construct the cDNA libraries. The whole process was carried out in strict accordance with the reagent instructions. First, T4 ligase was used to ligate a 5′ adapter and a 3′ adapter to the RNA molecules. Then a SuperScript II Reverse Transcription Kit (Invitrogen) was used to reverse-transcribed 5′ and 3′ adapter-ligated RNA to cDNA and PCR amplification was performed. Finally, the cDNA product was purified by RNA Gel Electrophoresis and gel recovery. The size and purity of the sample were determined using an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara). The Illumina sequencing platform was used for sequencing analysis.
Raw reads of fastq format were processed consisting of removing adapter and low-quality reads including sequences with quality score less than 20 and sequences with N base to obtain high-quality clean reads. The Bowtie2 was use to map the clean reads to mature miRNAs in miRBase 21.0 database. These consistent sequences were considered as the known miRNAs. The expression level of miRNAs was measured by TPM. The p value was calculated by Audic-Claverie statistic. The fold change ≥ 2 or fold change ≤ 0.5, and p value < 0.05 were used as the cut-off criteria.

| Function enrichment analysis
To analysis the biological function of lncRNAs, Gene Ontology (GO) enrichment analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were performed on the DE mRNAs and predicted target genes of DE miRNA and DE lncRNAs using GeneCodis3 bioinformatics resources (http://genecodis.cnb.csic.es) and DAVID Bioinformatics Resources 6.8 ((https://david.ncifcrf.gov/). GO enrichment analysis included biological process (BP), molecular function (MF), and cellular component (CC). The p values have been obtained through hypergeometric analysis corrected by the false discovery rate (FDR) method. Both GO terms and KEGG pathways were considered to be significantly enriched with FDR < 0.05.

| 18899
Of these mRNAs, 63 were upregulated, and 207 were downregulated. Each mRNA or lncRNA could be regulated by one or more miRNA and vice versa. 3.5 | The screening of core lncRNAs during the proliferative phase of rat LR The lncRNAs-miRNA-mRNA network was consisted with 107 lncRNAs, 32 miRNAs, and 270 mRNAs. Of these miRNAs, five were reported to play an important role during LR, including miR-21, miR-127, miR-34a, miR-378, and miR-125b, and they were regarded as the core miRNAs. The core lncRNAs were selected with a differently expression, and they were associated with the five miRNAs. Finally, nine core lncRNAs were correspondingly identified (Figure 3).

| DISCUSSION
Previous studies have shown that LR was regulated by a number of biological molecules including hormones, growth factors, and cytokines. However, most of these studies are limited to the protein-coding genes, and it is still largely unknown how these genes are regulated during LR. Therefore, it is necessary to find new regulators involved in LR for better understanding the mechanism.
Recent studies have shown that lncRNAs were important regulators of gene expression and associated with many important cellular physiological activities such as cell proliferation and differentiation (Ma et al., 2015;Zhu & Xu, 2013). LncRNAs could act as miRNA sponges to regulate the target mRNAs. The role of lncRNAs has been studied in a variety of cancer-related diseases including HCC. Staff et al. identified two miRNAs (miR-192 and miR204) could directly suppress lncRNA HOTTIP expression and interrupt GLS1-mediated glutaminolysis in HCC (Staff, 2016). Chen et al. indicated that lncRNA PTENP1 could modulate cell proliferation, migration, autophagy, and apoptosis by decoying miR-17, miR-19b, and miR-20a in HCC cell (C. L. Chen et al., 2015). However, the role of lncRNA-miRNA-mRNA network remains largely unknown during the proliferative phase of rat LR.
In this study, high-throughput sequencing was conducted to analyze the expression changes of lncRNAs, miRNAs, and mRNAs during the Some DE miRNAs have been reported to be associated with LR.
However, the function of most lncRNAs has not been studied. Then, an lncRNA-miRNA-mRNA interaction network was constructed during the proliferative phase of rat LR involving 107 lncRNAs, 32 miRNAs, and 270 mRNAs. To study the underlying role of lncRNAs during the proliferative phase of rat LR, GO enrichment analysis of the target mRNAs was performed. The result indicated that a large amount of significant GO terms were related to cellular metabolic process, cell adhesion, cellular response to stimulus, cell communication, and cell cycle, which had been reported as important physiological activities during the proliferative phase of rat LR (Erickson, Thompson, & Hixson, 2006;Kotsis et al., 2018;Loyer et al., 1994;Qin, Zhao, Chen, & Xu, 2006;Zheng, Weng, & Yu, 2009). KEGG pathway analysis identified nine signaling pathways including insulin signaling pathway and substance metabolism including pyrimidine metabolism, carbohydrate digestion and absorption, and pyruvate metabolism. Sasaki et al. indicated that insulin transmitted signal to intracellular regulators involved in hepatocyte growth through insulin receptor substrate 1 (IRS-1) during rat LR (Sasaki, Zhang, Nishiyama, Avruch, & Wands, 1993). The substance metabolism could supply energy and materials for the synthesis of DNA and proteins during the proliferative phase of rat LR (Yin, Chang, & Xu, 2017). In the lncRNA-miRNA-mRNA interaction network, five core miRNAs (miR-21-3p, miR-34a-5p, miR-127-3p, miR-378a-5p, and miR-125b-5p)  Some studies indicated that miR-21 was upregulated and played a significant role in modulating cell cycle progression and hepatocyte proliferation by targeting PTEN, FASLG, CCND1, BTG2, and PELI1 during LR (Castro et al., 2010;X. Chen et al., 2016;Li, Chan, Leung, Wang, & Xu, 2015;Marquez, Wendlandt, Galle, Keck, & McCaffrey, 2010;Ng, Song, Roll, Frandsen, & Willenbring, 2012;Song et al., 2010 which is known to promote DNA synthesis in hepatocytes after 2/3 PH (Song et al., 2010). So NONRATT007218.2, TCONS_00008697 and TCONS_00008701 might control hepatocyte proliferation during rat LR by interacting with miR-378a-5p. In this study, miR-127-3p was also predicted to target ENSRNOT00000079185 (ODC1), which was consisted with previous study. Hyun et al. showed that miR-125b could contribute to liver regeneration by mediating Hedgehog signaling (Hyun et al., 2015). It suggested that NONRATT001051.2 might be conducive to liver regeneration by targeting miR-125b-5p.
Many TFs have been reported during rat LR including E2F2, KLF2, STAT3, NFkappaB, AP-1, C/EBPbeta, and Nrf2. In this study, 21 transcription factors (CSRNP1, MYBL2, TEAD4, EGR2, ESRRA, ARID1B, ZBTB20, NFYC, SREBF2, HNF4A, PRKAG1, NFAT5, SREBF1, ERG, ZFP384, CLOCK, RORC, MAFB, GPBP1L1, ZBTB7C, and ADNP) were found to be involved in TF-miRNAs regulation network during the proliferative phase of rat LR. Zinc-finger protein ZBTB20, also named DPZF, HOF, and ZNF288, was a critical regulator of EGFR expression and hepatocyte proliferation in mouse liver regeneration . HNF4α, a member of the nuclear receptor family of transcription factors, could maintain hepatocyte differentiation in the adult healthy liver, and its loss may directly contribute to hepatocellular carcinoma development (Bonzo, Ferry, Matsubara, Kim, & Gonzalez, 2012). CLOCK, belonging to the bHLH-PAS family, located in the cell nucleus, played an important role in the regulation of liver gene expression (Malatesta, Baldelli, Marcheggiani, & Gazzanelli, 2003). The nuclear factor of activated T-cells (NFAT) transcription factors represented a family of gene transcription signaling intermediates that translate receptor-dependent signaling events into specific transcriptional responses using the Ras/Raf pathway, and NFAT4 played an important role in liver regeneration (Pierre et al., 2009). However, the function of most TFs was still unclear.
Some limitations were existed in this study. LncRNAs have a variety of functions. However, only the role of lncRNAs as miRNA sponges was analyzed through building the regulatory network of lncRNA-miRNA-mRNA and lncRNA-miRNA-TF. In addition, key lncRNAs predicted by bioinformatics analysis were not experimentally verified during the proliferative phase of rat LR.

| CONCLUSIONS
First, DE lncRNA, DE miRNA, and DE mRNA were analyzed by highthroughput sequencing technology, and then the lncRNA-miRNA-mRNA regulatory network was constructed according to the regulation mechanism of lncRNAs. Finally, through literature review and lncRNA-miRNA regulatory pairs, nine key lncRNAs, and five key miRNAs were screened out, which may play an important role during the proliferative phase of rat LR. This study provided clues for revealing the mechanism of LR and offered new ideas for the treatment of liver-associated diseases

ACKNOWLEDGMENTS
The study was financially supported by the National Natural Science Foundation of China (No. 31572270 and No. 31601038).