RETRACTED ARTICLE: Integrative analysis of the circRNA–miRNA regulatory network in atrial fibrillation

We aimed to investigate the circRNA–miRNA regulatory network in atrial fibrillation (AF) by using Cytoscape and HMDD v3.0. Finally, 120 differentially expressed circRNAs in peripheral blood monocytes of 4 AF patients were preliminarily screened by circRNA microarray. circRNA_4648, circRNA_4631, and circRNA_2875 were the first four circRNAs with the most binding nodes in the circRNA–miRNA network. The top three most frequent miRNAs for up-regulated circRNAs were hsa-miR-328 that interacted with 5 up-regulated circRNAs, hsa-miR-4685-5p with 4 up-regulated circRNAs, hsa-miR-3150a-3p, hsa-miR-4649-5p, hsa-miR-4783-3p, and hsa-miR-8073 with 3 up-regulated circRNAs,, while the top three most frequent miRNAs for down-regulated circRNAs were hsa-miR-328 that interacted with 14 down-regulated circRNAs, hsa-miR-4685-5p with 11 down-regulated circRNAs and hsa-miR-661 with 9 down-regulated circRNAs. According to HMDD v3.0, five up-regulated and eleven down-regulated circRNAs were found to interact with AF related miRNAs. These results indicated the possible regulatory network between circRNAs and miRNAs in the pathogenesis of AF.

Atrial fibrillation (AF), one of the most common arrhythmias in clinical practice, with a prevalence about 1-2% in the general population, is characterized with high relative risk of heart failure and embolic stroke. AF is also considered as a potential factor for high mortality and morbidity, especially in elderly individuals 1,2 . Recent growing reports indicate that structural remodeling and electrical remodeling are important pathophysiological contributors to onset and maintenance of AF 3,4 . However, exact mechanism of how AF occurs is still unknown.
To our knowledge, non-coding RNAs (ncRNAs), include a class of RNAs, such as long non-coding RNAs (lncRNAs), micro-RNAs (miRNAs) and circular RNAs (circRNAs), play crucial roles in regulating gene expression under pathological and physiological conditions [5][6][7] . circRNAs, a novel type of endogenous ncRNAs , have be reported as a key ncRNAs in gene regulation and the pathophysiology of cardiovascular diseases 8,9 . It has been well-known that dysregulated miRNAs can contribute to the prevalence of AF by deregulating transcription factor, regulating atrial excitability and increasing atrial arrhythmogenicity 10,11 . Accumulating studies indicate that circRNAs may interact with miRNAs by a sequence-driven sponging effect and the circRNA-miRNA-network has emerging roles in physiological and pathological processes of cardiovascular diseases 12,13 . However, to our knowledge, there are few studies pointing to the expression of circRNAs in AF, and circRNA-miRNA network in AF remains unclear.
In the present study, we analyzed and predicted the differentially expressed circRNAs in human monocytes from patients with AF and healthy controls using microarray, the potential regulatory network between circRNAs and miRNAs were explored by using Cytoscape and HMDD v3.0. We hypothesized that there were differentially expressed circRNAs in human monocytes and highly possible interaction between circRNAs and miRNAs, which would provide an important landmark for investigating the mechanism of AF.

Materials and methods
Study population and specimen collection. 10 patients with AF (AF group) and 10 matched healthy subjects (Control group) who excluded AF were enrolled (Table 1). 10 ml of peripheral blood was collected, monocytes were purified from peripheral blood and frozen for analysis. The diagnosis of AF was consistent with the criteria listed in the 2016 ESC Guidelines for the management of atrial fibrillation developed in collaboration with EACTS 14  The differentially expressed circRNAs of AF detected by microarray analysis. The total RNA in monocytes was extracted using Trizol reagent (Ambion, USA) and purified by QIAGEN RNeasy Mini Kit (QIGEN, Germany). Sample labeling and microarray hybridization were conducted by Outdo Bio-tech (Shanghai, P.R. China) with the same method as previously described 15 . Simply, the circRNAs were measured with the Agilent One-Color Microarray Based Gene Expression Analysis Low. The arrays were scanned by Axon microarray 4000B microarray scanner and extracted using Agilent Feature Extraction software (version 11.0.1.1). Quantile normalization and data processing were conducted by the Gene Spring GXv11.5.1 software package (Agilent, USA). The fold-change between AF patients and healthy controls was calculated. The statistical significance was calculated by t test and further filtered with fold change. circRNAs with foldchange > 2 and p < 0.05 were regarded as significant differential expression.
qRT-PCR validation of differentially expressed circRNAs. In order to confirm the results of microarray analysis, four upregulated circRNAs (circRNA_0031, circRNA_1837,circRNA_5901 and circRNA_7571) and four downregulated circRNAs (circRNA_2773, circRNA_5801, circRNA_7386 and circRNA_7577) were selected randomly for validation by qRT-PCR in all study population. Simply, 1 μl of cDNAs was added to 12.5 μl of SYBR-Green Gene Expression Master Mix (Applied Biosystems, Inc.), 10.5 μl of DEPC-treated water, and 0.5 μl of reverse and forward primers. The gene expression level of target circRNAs was normalized to the housekeeping gene GAPDH (Sangon Biotech, Shanghai, China) and calculated using the (2 −ΔΔCt ) method. The primer sequences for RT-PCR were shown in Table 2.
Construction of circRNA-miRNA regulatory networks. Acting as competing miRNA sponge, the sponging activity of differentially expressed circRNAs over corresponding miRNAs was calculated by the prediction of miRNA target binding sites using the miRanda software. Enrichment results of total differentially expressed circRNAs were sorted by p value, and the potential connections between circRNAs and miRNAs were further explored by using Cytoscape 3.4.0 (http://cytos cape.org/). Finally, the regulatory networks of circRNA-microRNA in AF patients were constructed.
Analyze the AF related circRNAs according to HMDD v3.0. In order to further explore the AF related circRNAs, we used the website of HMDD v3.0. HMDD v3.0, a database for experimentally supported human microRNA-disease associations, integrated many past publications about miRNA-disease associations, and offered evidence-stratified miRNA-disease data based on six categories of 20 evidence codes 16 . We used the keywords 'atrial fibrillation' to obtain AF related miRNAs from HMDD v3.0. If the differentially expressed circRNAs identified by microarray interacted with these reported AF related miRNAs, they were considered to be AF associated circRNAs.

Ethical approval. No treatment was tested in patients by the authors for this article. Informed consent was
obtained from all individual participants included in the study.

Discussion
In the present study, we provide two experimental findings on circRNAs involved in AF. On the one hand, there was significantly different expression profiles of circRNAs between AF patients and normal controls. On the other hand, five up-regulated (hsa_circRNA_7571, hsa_circRNA_3448, hsa_circRNA_1402, hsa_circRNA_4284 and hsa_circRNA_1415) and eleven down-regulated circRNAs (hsa_circRNA_2527, hsa_circRNA_4648, hsa_ circRNA_4624, hsa_circRNA_1496, hsa_circRNA_3138, hsa_circRNA_3138, hsa_circRNA_6086, hsa_cir-cRNA_2875, hsa_circRNA_3807, hsa_circRNA_4402, hsa_circRNA_4631 and hsa_circRNA_2773) were found to interact with AF related miRNAs and considered as the AF associated circRNAs by the construction of circRNA-miRNA network and the analysis using HMDD v3.0. www.nature.com/scientificreports/  www.nature.com/scientificreports/ Atrial electric remodeling associated with profound reduction of L-type Ca 2+ current and shortening of the action potential duration was the characteristic of both clinical and experimental AF. It was reported that miR-328, diminished L-type calcium current, shorted the atrial action potential duration, and increased AF vulnerability, would contribute to the atrial electric remodeling in AF and can be used as a diagnosis biomarker of AF 17,18 . Our findings indicated that hsa-miR-328 interacted with both up-regulated and downregulated circR-NAs, which was consistent with the reports and indicated that circRNA_7571, circRNA_3448, circRNA_1415, circRNA_4648, circRNA_4624, circRNA_4402, circRNA_2527 and circRNA_1496 colud be regarded as the diagnosis biomarkers of circRNAs for AF. miR-486 was related to the accumulation of superoxide anion, induction of DNA damage, reduction of cell proliferation and senescent phenotype in human fibroblasts 19 . Slagsvold et al. reported that hsa-miR-486 was upregulated in AF within left atria 20 . Another report from Wang et al. showed that hsa-miR-486 was found to be up-regulated in left atrial appendage in patients with AF 21 . Thus, hsa-miR-486 was considered as a AF related miRNA. At the same time, circRNA_1402, interacted with hsa-miR-486 in our findings could be considered as one of the AF related circRNAs.
A large number of studies have reported the relationships between the miRNAs (hsa-miR-133a, hsa-miR-574, hsa-miR-92a, hsa-miR-26b and hsa-miR-199a) and AF. For example, miR-133 has a cardioprotective role dependent on AKT serine/threonine kinase (AKT) signaling in control situation, inducing apoptosis in AF patients due to its down-regulation 22 . hsa-miR-26b increases IK1 current and membrane resting potential, the downregulation of hsa-miR-26b may reduce AF vulnerability 23 . hsa-miR-574 may promote electrical remodeling via Cav1.2 and contribute to cardiac arrhythmia pathogenesis of AF 24 .
hsa-miR-92a can attenuate cardiomyocyte apoptosis in AF patients induced by hypoxia/reoxygenation via the up-regulation of SMAD7 and down-regulation of nuclear NF-κB p65 25 . MiR-26b directly targeted KCNJ2. Both in vivo and in vitro inhibition of miR-26b increased IK1 and AF vulnerability, whereas overexpression of dampened AF vulnerability 26 . miR-199a down-regulation induces Sirtuin 1, a cardio-protective protein, as a compensatory mechanism to inhibit the process of oxidative stress which contributes to the pathogenesis of AF 27 . These miRNAs were considered as the potential biomarkers and therapeutic targets related to AF. Therefore, the differentially expressed circRNAs of circRNA_4284, circRNA_6086, circRNA_3138, circRNA_2773, circRNA_2875, circRNA_3807 and circRNA_4631 in the current study were more likely to be AF associated circRNAs.

Study limitations.
First, the small sample size does not provide sufficient power for such an analysis. Second, we just preliminarily investigated the circRNA-miRNA regulatory network in AF, the target gene or pathway analysis and functional assays of circRNA-miRNA regulatory network in the AF process should be further explored.

Conclusions
Our study showed that there were differentially expressed circRNAs in AF patients, five up-regulated and eleven down-regulated circRNAs were considered as the AF related circRNAs. The differentially expressed circRNAs had a possible regulatory network with miRNAs, which indicated the possible regulatory network between circRNAs and miRNAs in the pathogenesis of AF. www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/. © The Author(s) 2020