4.1 Reagents and Rats
Primary antibodies were purchased from Abcam, including antibodies against collagen I (COL-1), collagen III (COL-3), connective tissue growth factor (CTGF), TGF-β, and β-actin (details in Table 1). Secondary goat anti-rabbit IgG and mouse anti-rabbit IgG antibodies were obtained from Solarbio Company (Beijing, China).
Thirty healthy male Sprague Dawley (SD) rats (6–8 weeks old), weighing 200–250 g, were purchased from Beijing Vital River Laboratory Animal Technology Co. Ltd (Beijing, China).
4.2 Construction of a CIH-Induced Rat Model
All animal studies were conducted in compliance with the guidelines established by the Animal Administration Committee of Tianjin Medical University (Approval number: TMUaMEC2016012). The rats were randomly divided into a control group and a CIH group, with 15 rats in each group. The rats were allowed free access to food and water and provided with a normal diet. The atrial fibrosis rat model was established by exposure to a CIH environment, using a previously described protocol . Briefly, the rats in the CIH group were placed in a chamber filled with alternating cycles of nitrogen and oxygen, whereas the control group rats were maintained under normal oxygen conditions. The intermittent hypoxia time was gradually increased from 4 hours each day to 8 hours each day during the first week (from 9:00 AM to 5:00 PM) and then fixed at 8 hours each day. After filling the chamber with nitrogen, the oxygen concentration decreased from the normal 21% to 8% and was maintained at 8% for 50 s. The oxygen level was then increased from 8% to 21% by aerating with oxygen and was maintained at 21% for 50 s. Each intermittent hypoxia cycle comprised 300 s, alternating between oxygen concentrations of 8% and 21%.
4.3 Assessment of Echocardiography (ECG)
Intermittent hypoxia exposure lasted for 8 weeks, after which all rats were weighed and anesthetized with ether to perform the ECG assay. Animals were fixed on a platform and assessed by a GE Vivid-7 Doppler ultrasonic machine (General Electric, USA). Interventricular septal thickness (IVS), left ventricular end-diastolic dimension (LVEDD), left ventricular end-systolic dimension (LVESD), left atrial diameter (LAD), and mean pulmonary artery pressure (mPAP) were monitored.
We performed a carotid artery intubation for the blood pressure and ECG monitoring of all rats in both groups. The right common carotid artery was separated from the vagus nerve without damage. Distal ligation was performed with a non-absorbable suture, and a single vascular clamp was fixed on the proximal carotid artery. After a wedge-shaped resection was performed in the middle area, the artery was inserted into a polystyrene catheter (1.5-mm-diameter, injecting heparin sodium) connected to a biological signal acquisition system. ECG data and arterial blood pressure variables were recorded in 10 cardiac cycles, including heart rate, RR interval, PR interval, QRS interval, QT interval, systolic blood pressure, diastolic blood pressure, mean blood pressure, and pulse pressure.
At the end of ECG monitoring, the hearts from both groups were collected, weighed, and washed with pre-cooled phosphate-buffered saline three times. The left atrial tissue of five rats in each group was used for hematoxylin-eosin (HE) staining, Masson’s trichrome staining, and immunohistochemistry (IHC); all other rat samples were frozen at −80 °C for molecular biological analysis.
4.4 HE Staining
The left atrial tissues were immersed in 4% paraformaldehyde for two weeks. After paraffin-embedding, these tissues were cut into 5-μm sections. The slices were dewaxed in xylene for 10 min, three times, and then moved through a 95%, 80%, and 70% alcohol gradient for gradual dehydration, for 5 min each. The specimens were then successively stained with hematoxylin for 3 min and eosin for 5 min, followed by reimmersion in alcohol and xylene. Morphological changes were observed under an Olympus CX-21FSI microscope system (Olympus, Japan), and Image-Pro Plus 7.0 software was applied for imaging analysis. Three fields of view were taken from each slide.
4.5 Masson’s Trichrome Staining
Masson’s trichrome staining was performed using a kit from Nanjing Jiancheng (Jiangsu, China). Briefly, the left atrial tissues were cut into 5-μm sections. The slides were dewaxed, rehydrated, and immersed in Bouin’s solution at 56 °C for 15 min. Then, the slides were dyed with Wiegert’s hematoxylin, Biebrich scarlet-acid fuchsin, and aniline blue, followed by fixation in 1% acetic acid. Finally, the slides were dehydrated and mounted. Three fields of view were taken from each slide.
The expression levels of fibrosis-related proteins, including COL-1, COL-3, CTGF, and TGF-β1, were detected using IHC assay. The baked slices were dewaxed with xylene and hydrated with an ethanol gradient (100%–70%). After successively incubating with citrate buffer and 3% H2O2 for 30 min, the slides were sealed with blocking solution and incubated with the primary antibodies overnight at 4°C. On the second day, the slides were rinsed and incubated with the corresponding secondary antibody (Zhongshan Jinqiao Biotechnology Company; Beijing, China) for 15 min, followed by 3,3′-diaminobenzidine (DAB) and HE staining. The slides were then examined and photographed (400×) using an Olympus BX53 fluorescence microscope (Tokyo, Japan). The DAB staining was analyzed by Image-Pro Plus 7.0 software (Media Cybernetics, Rockville, MD, USA). Three fields of view were taken for each slide. The mean optical density was calculated as the integral optical density/the total brown area.
4.7 Western Blotting
To further explore the expression of fibrosis-related proteins, we performed western blotting assay. Frozen artery tissue samples were homogenized in radioimmunoprecipitation assay lysis buffer and centrifuged (12,000 rpm for 15 min). Protein samples (10 µl) were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, transferred onto polyvinylidene difluoride membranes, and incubated with primary antibodies at 4°C overnight (Table 1). After washing with Tris-buffered saline containing Tween-20 3 times, the membranes were incubated with secondary antibody for 2 h. The signal was visualized using a chemiluminescence kit (Millipore, USA) and exposed using an autoradiographic system.
4.8 Identifying Differentially Expressed CircRNAs
High-throughput sequencing was performed by Shanghai OE Biotech Company. We used Trimmomatic  software to perform the original data preprocessing, including quality control, read number counts, and summarization. The CircBase database  contains numerous circRNA sequences from six species: humans, mice, nematodes, speartail, Drosophila melanogaster, and coelacanth. Here, we used CIRI software  to perform circRNA prediction. The prediction results were compared with the CircBase database to obtain known and novel circRNAs.
In addition, circRNA expression was quantified using the reads per million (RPM) method, and the RPM files for circRNAs in each sample were obtained, as were the expression boxplots.
The number of circular reads represents the read number compared with the circRNA back-spliced junction area, which is derived from the circRNA prediction software. The number of total reads (in millions) represents the read number for each sample (clean_data).
DESeq software was used to normalize the junction read counts of circRNA in each sample. A negative binomial distribution test was conducted to identify the differentially expressed circRNAs, based on the criteria of log2|fold change| > 1 and P-value < 0.05. Cluster analysis was performed for the differentially expressed circRNAs. Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were performed for the host genes of the differentially expressed circRNAs.
4.9 Prediction of CircRNA–miRNA Interactions
Miranda database was used to predict microRNAs (miRNAs) that interact with differentially expressed circRNAs. The miRNAs with greater connections to circRNAs were screened using a hypergeometric distribution test. The circRNA–miRNA target interaction network was visualized using R software.
4.10 qRT-PCR Analysis
Critical circRNA expression levels were further investigated by qRT-PCR analysis. TRIzol reagent (Invitrogen, USA) was used to isolate total RNA from atrial tissue samples. Total RNA was reverse transcribed into cDNA following the instruction of the reverse transcription kit (TIANGEN, Beijing, China). The obtained cDNAs were used for qRT-PCR analysis using a qRT-PCR kit (TransGen Biotech, Beijing, China) and a 7500 RT-PCR System (Thermo, USA). Primers were generated by Beijing BGI Company (Beijing, China), and the primer sequences are shown in Table 2. The cDNA levels were analyzed by the 2−ΔΔCt method, and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as an internal control.
4.11 Statistical Analysis
All experimental data are presented as the mean±standard deviation and were analyzed by two-tailed t-tests in SPSS 19.0 software (SPSS Inc., USA). P < 0.05 was considered a significant difference.