Comprehensive circRNA expression profile during ischemic postconditioning attenuating hepatic ischemia/reperfusion injury

Ischemic postconditioning (IPO) attenuates hepatic ischemia/reperfusion (I/R) injury. The aim of this study was to explore the role of circular RNAs (circRNAs) in the protective mechanism of IPO. In this study, microarray hybridization analysis was performed to determine the circRNA expression profile. Briefly, a total of 1599 dysregulated circRNAs were detected. The competitive endogenous RNA (ceRNA) network, including 6 circRNAs, 47 miRNAs and 90 mRNAs, indicated that the potential “housekeeping” function of circRNAs is dysregulated in hepatic I/R injury. Based on the validation results of selected circRNAs, miRNAs and mRNAs following qRT-PCR amplification, the mmu_circRNA_005186-miR-124-3p-Epha2 pathway was constructed. Dual-luciferase reporter analysis showed that miR-124-3p interacted directly with mmu_circRNA_005186 and Epha2 through the predicted binding sites, which suggested that mmu_circRNA_005186, serving as a miRNA sponge for miR-124-3p, regulated the expression of Epha2. Functionally, we explored the mechanism of mmu_circRNA_005186 in LPS-treated RAW264.7 cells which simulated the inflammation in hepatic I/R injury. We found that mmu_circRNA_005186 silencing attenuated the LPS-induced inflammation and was associated with miR-124-3p upregulation and Epha2 downregulation. Our study is the first to show that circRNAs are closely related to hepatic I/R injury and IPO and suggests that targeting mmu_circRNA_005186-miR-124-3p-Epha2 pathway might attenuate hepatic I/R injury.

SCIenTIfIC RepoRts | (2019) 9:264 | DOI: 10.1038/s41598-018-36443-8 kidney 11 , brain 12 and liver 13 . Multiple studies have found that minimizing cell necrosis and apoptosis is crucial for the benefits of IPO 14 . However, no studies have reported circRNA expression profiles in hepatic I/R injury, and whether IPO can attenuate liver I/R injury by mediating circRNA expression is unknown. Therefore, in this study, considering the significance of hepatic I/R injury, we explored circRNA expression patterns via microarray hybridization and predicted the functions of the host genes of differentially expressed (DE) circRNAs. We identified an important role of circRNAs in ischemic postconditioning attenuating hepatic I/R injury.

IPO attenuated liver I/R injury.
To determine whether IPO can attenuate hepatic I/R injury in this animal model, H&E staining was used to evaluate the histological injury grade. The H&E staining results showed that IPO alleviated the histological injury severity of the hepatic tissue ( Fig. 1).

CircRNA expression pattern during hepatic I/R injury assessed via microarray. Microarray
hybridization is an efficient method for detecting differential expression profiles and biological functions of cir-cRNAs. The results showed that a total of 1599 circRNAs were differentially expressed (p < 0.05 and fold change ≥2.0). Of these, 213 circRNAs were upregulated and 493 were down-regulated in the I/R group compared with the normal group. At the same time, 641 and 252 circRNAs were up-regulated and down-regulated, respectively, between the IPO and I/R groups (Fig. 2a,b). Hierarchical clustering analysis (heat map) was used to evaluate significantly expressed circRNAs, which were indicated by p < 0.05 and a fold change ≥2.0, among the normal, I/R and IPO groups (Fig. 2c). A Venn diagram analysis of the dysregulated circRNAs indicated that most of these RNAs (51.3% and 58.6%) were sequentially expressed (up-regulated in the I/R group and down-regulated in the IPO group) during IPO-attenuated I/R injury (Fig. 2d).
The predicted biological function of circRNA linear transcripts. To  to the enrichment score and are ranked by p-value; all of the top 10 GO terms for each domain were nearly identical (Fig. 3a). This result revealed that in the BP domain, the most meaningful, enriched GO terms were cellular metabolic process, primary metabolic process, metabolic process and cellular process. For the CC domain, the most meaningful, enriched GO terms were intracellular, intracellular part, cell part and cell. In addition, in MF the most meaningful, enriched GO terms were binding, protein binding and transferase activity. In addition, KEGG pathway analysis was performed, and the top 10 pathways in the compared groups were listed according to the enrichment score and were ranked by p-value (Fig. 3b). The results indicated that in the protective IPO mechanism, the role of circRNA is closely related to the mechanistic target of rapamycin (mTOR) signaling pathway and further validation are required to confirm these results.
Validation of selected circRNAs during hepatic I/R injury by qRT-PCR. Ten sequentially expressed circRNAs among the normal, I/R and IPO groups were screened according to p-value, fold change, raw intensity and type (Table 1) in validation experiments. The results confirmed that consistent with the microarray results, mmu_circRNA_005186, mmu_circRNA_011137, mmu_circRNA_013703, mmu_circRNA_29140, mmu_cir-cRNA_36837, and mmu_circRNA_43819 were significantly amplified by qRT-PCR (Fig. 4a).
mmu_circRNA_005186 acted as a miRNA Sponge for miR-124-3p and regulated the expression of Epha2. To determine whether mmu_circRNA_005186 directly targets miR-124-3p through the predicted binding sites (Fig. 6a), we performed a dual-luciferase reporter assay and found that miR-124-3p mimics significantly reduced the luciferase activity of mmu_circRNA_005186-WT (p < 0.01) but not the mutant one in 293T cells (Fig. 6b). Then, we applied the dual-luciferase reporter assay to validate whether miR-124-3p interacts with Epha2 by the predicted binding sites (Fig. 6c). The results indicated that luciferase activity was significantly lower in the miR-124-3p mimic (p < 0.01) than the negative control (NC), but this reduction was not observed with the mutant plasmids (Fig. 6d). Furthermore, by qRT-PCR, we found that the expression level of miR-124-3p was significantly upregulated after mmu_circRNA_005186 silencing in RAW264.7 cells (Fig. 6e,f). In addition, the expression level of Epha2 was significantly downregulated (Fig. 6g). Collectively, the above results demonstrated that mmu_circRNA_005186 enhanced Epha2 expression level by acting as a miRNA sponge for miR-124-3p in vitro.
Effect of mmu_circRNA_005186 on LPS-induced inflammation in RAW264.7 cell. Given that macrophages play a vital role in hepatic I/R injury, we sought to determine whether mmu_circRNA_005186 affects the LPS-induced inflammation in RAW264.7 cells through the mmu_circRNA_005186-miR-124-3p-Epha2 pathway. The RAW 264.7 cells were transfected with small interfering RNA against mmu_circRNA_005186 (si-005186) or si-circRNA-NC for 48 h and were stimulated by LPS. The supernatant concentrations of TNF-α and IL-1β released from RAW 264.7 cells were significantly lower in the si-005186 group than the negative control ( Fig. 7a,b). As expected, qRT-PCR identified that miR-124-3p was upregulated by 6-fold (p < 0.01) (Fig. 7c) while the expression level of Epha2 markedly decreased in the si-005186 group (Fig. 7d).

Discussion
Currently, IPO, an effective method for attenuating hepatic I/R injury 13 , has been used for several years; however, the beneficial mechanism of IPO is still undefined and vague. CircRNAs, a novel type of endogenous noncoding RNA that was recently rediscovered, has gradually been recognized as a molecular biomarker 15 and as a main mediator in the pathogenesis of tumor growth and metastasis 16 . For example, Hsa_circ_0001649 has been reported as a potential novel biomarker for hepatocellular carcinoma 17 . However, circRNA expression profiles in hepatic I/R injury have not been studied.
In the present study, we used microarray hybridization to detect differential expression profiles and to investigate the biological function of circRNA. A total of 1599 circRNAs were DE among the normal, I/R and IPO groups. A Venn diagram analysis indicated that the majority of the dysregulated circRNAs (51.3% and 58.6%) were sequentially expressed during IPO-induced I/R injury attenuation, which indicated that circRNA plays a vital role in I/R pathogenesis and IPO.
Increasing evidence indicates that circRNA, acts as a competing endogenous RNA or sponge, increases target gene expression by removing the inhibitory effect of miRNA 4,18 . For instance, miR-7a prevents myocardial ischemia-induced cell death by targeting inflammatory and apoptotic genes, however, the circRNA ciRS-7 sponges miR-7a, blocking the anti-apoptotic and cell-protective functions 19 . The heart-related circRNA HRCR is known to prevent cardiac hypertrophy and heart failure by sponging miR-223, which suppresses ARC expression 20 . In the present study, circRNA verification experiments revealed six dysregulated circRNAs following qRT-PCR amplification. To study the potential "housekeeping" functions of these identified circRNAs, a ceRNA network of circRNA-miRNA-mRNA interactions was constructed that includes 6 circRNAs, 47 miRNAs and 90 mRNAs. According to the validation results of selected miRNAs and mRNAs, several ceRNA pathways were constructed. Since the mechanism of circRNA has not been well elucidated, none of the potential circRNA-miRNA-mRNA pathways included in this study have been reported previously. The dual-luciferase analysis validated that miR-124-3p interacts directly with mmu_ circRNA_005186 and Epha2 through the predicted binding sites. The results suggest that mmu_circRNA_005186, serving as a miRNA sponge for miR-124-3p, regulated the expression of Epha2. Furthermore, we explored the mechanism of mmu_circRNA_005186 in LPS-induced RAW264.7 cells that simulated the inflammation in hepatic I/R injury. Through these functional experiments, we demonstrated that mmu_circRNA_005186 silencing attenuated the inflammation and was associated with miR-124-3p upregulation and Epha2 downregulation in vitro. Previous studies indicated that miR-124-3p commonly serves an integral roles in cell growth, apoptosis, invasion and metastasis in various cancers [21][22][23] . In addition, a recent study conducted by Shan Huang indicates that increased miR-124-3p inhibits neuronal inflammation and contributes to neurite outgrowth via their transfer into neurons in a mouse model with traumatic brain injury 24 , suggesting the potential anti-inflammatory role of miR-124-3p. Furthermore, several papers reported that miR-124-3p could be useful molecular marker in brain ischemic stroke 25,26 . Collectively these studies confirm that consistent with our present results, miR-124-3p may inhibit inflammation and attenuate hepatic I/R injury. Epha2 belongs to the membrane-bound tyrosine kinase family, which serves as an oncogene involved in promoting the formation and progression in different types of malignancies 27,28 . In the present study, Epha2 is the linear gene of mmu_circRNA_005186, and this linear gene has been reported to modulate HCC radiosensitivity via p38/ mitogen-activated protein kinase-mediated signaling pathways and miR-26b 29 . Ligand-stimulated EphA2 inhibits Akt in PTEN-deficient prostate cancer cells, glioma cells 30 , and other cell types 31,32 and the activation of mTOR/ Akt signaling pathway could protect hepatocytes from hepatic IR injury 33 , suggesting the potential regulating function of Epha2 in hepatic I/R injury. In addition, Saeki et al. reported that Epha2 promotes cell adhesion and spreading of monocyte and macrophage cell lines 34 . To the best of our knowledge, platelet and neutrophil adhesion explains the mechanism of hepatic I/R injury 35 . These studies provide novel insights and clues that Epha2 can be applied to study hepatic I/R injury. In summary, our results show that 1599 circRNAs are dysregulated in hepatic I/R and IPO, and that these circRNAs might play a crucial role in these processes with potential pathophysiological functions. Moreover, we reported for the first time that mmu_circRNA_005186 exerts its function by serving as a miRNA sponge for miR-124-3p, thereby promoting the function of Epha2. Combining the preceding evidence with our functional experiments, modulation of the mmu_circRNA_005186_miR-124-3p_Epha2 pathway may be one of the possible targets and strategy to attenuate hepatic I/R injury.

Methods and Materials
Ethics approval. This   RAW 264.7 cells were seeded at 2 × 10 6 cells in 1 ml of DMEM in 6-well plates and allowed to adhere for 12 h before stimulation by LPS (100 ng/ml, Sigma, USA). Cells were incubated with LPS for 12 h or 24 h.
Histological examination of the liver. Hematoxylin-eosin (H&E) staining was conducted to evaluate liver pathological injury. The grade of liver injury was evaluated via Suzuki's criterion 37 , which indicates the severity of sinusoidal congestion, cytoplasmic vacuolation, and parenchymal cell necrosis.
Total RNA isolation and quality assessment. Total RNA was extracted from nine frozen mouse liver tissues (three randomly selected samples from each group) and cells using TRIzol (Invitrogen, USA) according to the manufacturer's instructions. Subsequently, total RNA integrity and gDNA contamination was tested by qRT-PCR. The cDNA synthesis and PCR amplification were carried out by a GeneAmp PCR system 9700 (Applied Biosystems, USA) and ViiA 7 real-time PCR system (Applied Biosystems, USA), respectively. RNA expression was defined as the threshold cycle (Ct). GAPDH was amplified as the internal control for circRNAs and mRNAs and U6 was amplified as the internal control for miRNAs. The relative expression levels were calculated by the double-standard curve method. All primers (Supplementary Table S1) were designed and synthesized by Kangchen Bio-tech (Shanghai, China).
RNA labelling and circRNA microarray hybridization. RNA labelling and array hybridization were performed according to the manufacturer's protocol. The enriched circRNAs were amplified and transcribed into fluorescent cRNA. The labeled cRNAs were purified by a RNeasy Mini Kit (Qiagen). The concentration and specific activity of the labelled cRNAs (pmol Cy3/μg cRNA) were measured by a NanoDrop ND-1000 instrument. Each labelled cRNA (1 μg) was fragmented by the addition of 5 μl of 10× blocking agent and 1 μl of 25× fragmentation buffer; then, the mixture was heated at 60 °C for 30 min, and finally, 25 μl of 2× hybridization buffer was added to dilute the labeled cRNA. Hybridization solution (50 μl) was dispensed onto the gasket slide, and this slide was placed onto the circRNA expression microarray slide. The hybridized arrays were washed, fixed and scanned using an Agilent Scanner G2505C.
CircRNA expression profile data analysis. Agilent Feature Extraction software (version 11.0.1.1) was used to analyse acquired array images. When the data were extracted, quantile normalization and subsequent data processing were conducted using the R software Limma package. CircRNAs that were significantly DE between the two groups were identified through fold change and p-values (fold change ≥2.0 and p-value < 0.05). The log2 ratio was used for quantile normalization. Hierarchical clustering was conducted to show distinguishable circRNA expression profiles among the samples. Significantly DE circRNAs among the three groups were identified through volcano plot filtering.

Annotation of linear transcripts and GO and KEGG pathway analysis of host genes. Linear
transcripts overlapping with the chromosome location of the circRNA sequence were annotated. CircRNA distribution in the genome was determined by comparing the circRNA with genetic elements. GO analysis was used to investigate three domains of functionality: BP, CC and MF. Fisher's exact test in the Bioconductor topGO package was used to determine if more overlap existed between the DE list and the GO annotation list than would be expected by chance. The p-value produced by the topGO analysis denotes the significance of the GO term enrichment in the DE genes (p < 0.05). Pathway analysis was performed to functionally analyze and map genes to KEGG pathways. The p-value denotes the significance of the pathway correlated with the conditions (p < 0.05). Validation of selected circRNAs, miRNAs and mRNAs. Ten circRNAs were selected on the basis of a combination of p-value, fold change, raw intensity and type. Ten miRNAs and mRNAs were selected according to the ceRNA network. In addition, further qRT-PCR validation was performed by a SYBR Green PCR kit (Arraystar, USA) in triplicate for each sample.
Prediction of miRNAs and mRNAs related to circRNAs and construction of the circRNA-miRNA-mRNA Network. To identify predicted circRNA and miRNA interactions, Arraystar's in-house miRNA target prediction software was applied on the basis of TargetScan and miRanda algorithms. To improve prediction reliability, we set the context + score to less than −0.5. Then, the potential miRNA targets were predicted with miRanda and TargetScan databases, and the ceRNA network of circRNA-miRNA-mRNA was delineated using Cytoscape software.
Luciferase reporter assay. The sequence of mmu_circRNA_005186 or the 3′-UTR of the Epha2 mRNA, including the seed sequence of the miR-124-3p binding site and a mutated binding site, was cloned into the pmir-GLO dual-luciferase vector (Promega, USA).Then, 293T cells were co-transfected with miRNA-124-3p mimics and luciferase vectors using lipofectamine 2000 (Invitrogen, USA). After incubation for 48 h, firefly and Renilla luciferase activity was measured using the DualGlo luciferase assay system (Promega) according to the standard protocol.
ELISA assay for TNF-α and IL-1β. After treatment with LPS (100 ng/ml) for 24 h, the supernatant concentrations of TNF-α and IL-1β released from RAW 264.7 cells were measured by an automated microplate reader (Heales, China) using commercial ELISA kits according to the instructions of the kits (Abcam, USA).

Statistical analysis.
The microarray data are reported as the means ± SEM. All other statistical data were analyzed and visualized by GraphPad Prism 5.0 and Student's t-test. The p-values were used to evaluate statistical significance (p < 0.05).

Data Availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request or GEO database (GSE117524).