Bioinformatics-Based Identification of CircRNA-MicroRNA-mRNA Network for Calcific Aortic Valve Disease

Background Calcific aortic valve disease (CAVD) is the most common native valve disease. Valvular interstitial cell (VIC) osteogenic differentiation and valvular endothelial cell (VEC) dysfunction are key steps in CAVD progression. Circular RNA (circRNAs) is involved in regulating osteogenic differentiation with mesenchymal cells and is associated with multiple disease progression, but the function of circRNAs in CAVD remains unknown. Here, we aimed to investigate the effect and potential significance of circRNA-miRNA-mRNA networks in CAVD. Methods Two mRNA datasets, one miRNA dataset, and one circRNA dataset of CAVD downloaded from GEO were used to identify DE-circRNAs, DE-miRNAs, and DE-mRNAs. Based on the online website prediction function, the common mRNAs (FmRNAs) for constructing circRNA-miRNA-mRNA networks were identified. GO and KEGG enrichment analyses were performed on FmRNAs. In addition, hub genes were identified by PPI networks. Based on the expression of each data set, the circRNA-miRNA-hub gene network was constructed by Cytoscape (version 3.6.1). Results 32 DE-circRNAs, 206 DE-miRNAs, and 2170 DE-mRNAs were identified. Fifty-nine FmRNAs were obtained by intersection. The KEGG pathway analysis of FmRNAs was enriched in pathways in cancer, JAK-STAT signaling pathway, cell cycle, and MAPK signaling pathway. Meanwhile, transcription, nucleolus, and protein homodimerization activity were significantly enriched in GO analysis. Eight hub genes were identified based on the PPI network. Three possible regulatory networks in CAVD disease were obtained based on the biological functions of circRNAs including: hsa_circ_0026817-hsa-miR-211-5p-CACNA1C, hsa_circ_0007215-hsa-miR-1252-5p-MECP2, and hsa_circ_0007215-hsa-miR-1343-3p- RBL1. Conclusion The present bionformatics analysis suggests the functional effect for the circRNA-miRNA-mRNA network in CAVD pathogenesis and provides new targets for therapeutics.


Introduction
Calcifc aortic valve disease (CAVD) is a condition caused by calcifcation of the aortic valve or aortic annulus, resulting in a hemodynamic manifestation of aortic valve stenosis or regurgitation [1]. It is a chronic progressive disease that increases in prevalence with age [2,3], leading to an increasing proportion of acquired valvular heart disease. CAVD has become a signifcant factor of disease burden in the elderly [4], which has the characteristics of high morbidity and mortality [1]. Treatment of CAVD mainly relies on surgery [5], which includes surgical valve replacement and percutaneous valve prosthetic implantation. However, not all patients are eligible for surgical treatment. In terms of drug therapy, the antihypertensive drugs and statins are efective in treating atherosclerosis but do not reverse or slow the process of CAVD [6]. In conclusion, there is an urgent need to explore the key regulatory molecules in the pathogenesis of CAVD to provide new targets for its pharmacological treatment.
About 90% of the mammalian genome is transcribed into noncoding RNAs (ncRNAs), whose functions have not been fully studied [7]. With the development of deep RNA sequencing (RNA-seq) technology and novel bioinformatics methods, a wide variety of circular RNAs (circRNAs) types have been discovered and identifed [8]. As a series of novel noncoding RNAs, circRNAs are characterized by a covalent closed loop structures lacking a 5′ cap or a 3′ Poly A tail [9]. Te high abundance, relative stability, and evolutionary conservation of circRNAs distinguish it from traditional linear RNAs. CircRNAs has signifcant advantages in developing applications as a novel clinical diagnostic marker because it is able to better adsorb miRNAs from organisms than linear mRNAs and lncRNAs. However, the function of circRNAs is still unclear. Tere is growing evidence that belongs to competing endogenous RNAs (ceRNAs), which contain microRNA response elements (MREs). Te specifc RNAs with MREs can impair miRNA activity by sequestration, resulting in upregulation of miRNA target gene expression, which is known as ceRNA hypothesis [10]. Te cirRNAs has been found to exert an important biological response in cardiovascular disease [11][12][13], but studies on CAVD are still limited. Wang et al. found that circRIC3, as a miR-204-5p sponge, positively regulates the expression of the calcifcation-promoting gene dipeptidyl peptidase-4 (DPP4), leading to CAVD [1]. Yu et al. reported that circRNA TGFBR2 positively regulates TWIST1 through sponge phagocytosis of miR-25-3p by inhibiting osteoblast diferentiation and preventing valve calcifcation in human VICs [14]. Tese results reconfrm that circRNAs are critical in the development of CAVD. However, studies on circRNA-associated ceRNA networks in CAVD remain scarce. Terefore, studying of circRNA-miRNA-mRNA networks complements the lack of ncRNA in the exploration of CAVD pathogenesis. It is promising to fnd markers that can be used as diagnostic predictors of the disease and provide new insights for the treatment of CAVD.
To investigate how the circRNA-miRNA-mRNA network regulates CAVD pathophysiological processes, we gain insight into the signaling regulation within the tissues leading to involvement in CAVD progression and discover relevant therapeutic targets. We screened CAVD-related circRNA, miRNA, and mRNA datasets in the Gene Expression Omnibus (GEO) database ( Figure 1). By performing diferential expression analysis, DE-circRNAs and DE-miRNAs targets were predicted. Following that, we constructed circRNA-miRNA-mRNA networks in CAVD. Te common mRNAs (FmRNAs) were analyzed by gene ontology (GO) function enrichment analysis, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis, and protein-protein interaction (PPI) network construction.
We identifed 8 hub genes from PPI for circRNA-miR-NA-hub gene network visualization, and this regulatory network will be a potential therapeutic target for treating CAVD diseases. Te schema of the bioinformatics analysis is shown in Figure 1.

Methods
2.1. RNA Array. Te CAVD datasets from the National Center for Biotechnology Information (NCBI) GEO database (https://www.ncbi.nlm.nih.gov/geo/) were evaluated and screened. Te GSE155119 dataset of circRNA expression was found in GPL26192 platform. Te GSE87885 dataset of miRNA expression was found at GPL22555 platform. On the GPL10558 platform and GPL570 platform, we found the GSE83453 and GSE51472 datasets for mRNA expression. Te calcifed aortic valve contains 20 samples, including 3 circRNA samples, 2 miRNA samples, and 15 mRNA samples. Te noncalcifed aortic valve contains 19 samples, with 3 from the circRNA dataset, 3 from the miRNA dataset, and 13 from the mRNA dataset.

Prediction of DE-CircRNAs and DE-miRNAs Targets.
Te DE-circRNAs target miRNAs were predicted by using the online software, the Encyclopedia of RNA Interactomes (ENCORI). DE-miRNAs target genes were predicted by online websites, miRTarBase, TargetScan, and miRDB (Table 1), respectively. Te mRNAs recognized by miRTarBase, TargetScan, and miRDB websites were considered candidate targets. Te information of websites is shown in Table 1.

GO Function Analysis.
Te database for annotation, visualization, and integrated discovery (DAVID, https:// david.ncifcrf.gov/) is an online web-based bioinformatics resource, which can provide tools for analyzing the function of large lists of genes/proteins [19]. Te FmRNAs were input into DAVID online software to annotate the target genes with GO function. Te results are visualized by R software.

KEGG Pathway Analysis. KEGG Orthology-Based
Annotation System (KOBAS) is one of the most widely used web servers for gene/protein functional annotation and gene set enrichment. Te KOBAS website was used to map FmRNAs to the KEGG pathway. KEGG pathway analysis visualization was performed using R software.
2.7. PPI Network. STRING (https://string-db.org/) database is the software for predicting protein-protein interactions. Te PPI network of FmRNAs was established using the STRING database (version 11.0). A combined score of >0.4 was considered the cutof to indicate a signifcant PPI pair.

CircRNA-miRNA-Hub Gene Network.
Eight FmRNAs with a high degree of PPI were selected to fnd the relevant noncoding RNAs regulating them from the total DE-circRNAs and DE-miRNAs. Tese relevant data were imported into Cytoscape software (version 3.6.1) for analysis and visual graphing.

PPI Network.
Target gene data predicted by FmRNAs were uploaded to the STRING database for the construction of the PPI network ( Figure 6). Te MYC, ITPR1, EGFR, CACNA1C, RASGRP1, MECP2, RBL1, and WEE1 were the   Genetics Research hub genes with high degree values. Te hub genes expression was shown in Figure 7.

Discussion
CAVD, the most common valvular disorder, is the leading cause of aortic stenosis. Te most efective treatment is surgery or interventional valve replacement [20], which has complications and does not guarantee long-term success [21]. Tere is an absence of approved pharmacological treatments to stop the progression or treat CAVD [22]. Te ceRNA hypothesis has been proposed as a model for regulating gene expression during disease progression in recent years [10]. Tere is growing experimental evidence that multiple noncoding RNAs, including circRNAs, small noncoding RNAs, pseudogenes, and lncRNAs may have ceRNA activity [23]. More importantly, circRNAs, which are highly resistant to nucleases, maintain high abundance in the cytoplasm and better regulate miRNAs. However, it was only circRIC3 and circRNA TGFBR2 that were studied in CAVD disease [1,14]. Construction of circRNA-miRNA-mRNA regulatory networks is essential to understand the pathophysiological progression of CAVD as the basis for developing novel therapeutics. We have constructed circRNA-miRNA-mRNA regulatory networks based on the sponge activity of circRNA. Most circRNAs in the coexpression network remain unknown.

Genetics Research
Bioinformatic analysis of the DE-circRNAs target genes in network showed that transcription was the most important BP identifed by GO analysis. Vadana et al. found that expression of SMAD and Runt transcription factors increased calcium deposition in CAVD [24]. Te KEGG pathway of the target gene is signifcantly enriched in the cell cycle, MAPK and TGF-β pathway. It has been previously shown that MAP2K1 mutations activate p-ERK-dependent cell cycle progression and autophagy, exhibiting arterial valve stenosis [25]. Inhibiting the p38-MAPK signaling pathway can reduce ALP activity and calcifcation deposition to ameliorate aortic valve calcifcation [26]. We successfully established 3 circRNA-miRNA-hub gene networks relevant to CAVD, which include hsa_circ_0026817hsa-miR-211-5p-CACNA1C, hsa_circ_0007215-hsa-miR-1343-3p-RBL1, and hsa_circ_0007215-hsa-miR-1252-5p-MECP2. Normal aortic valves are composed of valve endothelial cells (VECs) and valve interstitial cells (VICs), which play an important role in maintaining valve morphology and function [27]. Dysfunction of VICs and VECs is the key to the progression of CAVD. Upregulated hsa_circ_0026817 in CAVD may target hsa-miR-211-5p to regulate CACNA1C. It has been shown that miR-211-5p overexpression inhibits cell cycle by decreasing cyclin D1 levels [28]. Inhibition of cyclin D1 essentially abolishes fbrotic responses which are associated with VICs proliferation [29,30]. Downregulation of miR-211-5p in CAVD leads to aortic valve fbrosis via the regulation of cyclin D1 in VICs. CACNA1C is the gene encoding the L-type voltage-gated Ca 2+ channel [31]. Te activation of cytosolic L-type Ca 2+ channel leads to the entry of small amounts of Ca 2+ into the cytoplasm and triggers Ca 2+ release from the sarcoplasmic reticulum by activating ryanodine receptor 2 (RyR2). RyR2 was predominantly expressed in VICs, and inhibition of RyR2 prevents valvular calcifcation [32]. Matsui et al. identifed high expression of CNCNA1C in calcifed valves and verifed the involvement of CACNA1C in CAVD progression by afecting valve calcifcation in VIC cells [33].
Down-regulated hsa_circ_0007215 in CAVD may regulate both hsa-miR-1343-3p/RBL1 and hsa-miR-1252-5p/ MECP2. Upregulated miR-1343-3p in CAVD might directly infuence valve endothelial cells (VECs) growth through the TGF-β signaling pathway. Te surface of the heart valves is covered with VECs [34], which forms a barrier between the blood and the internal valve tissue [35]. In the aortic valve, TGF-β1 is predominantly localized to VEC and found to decrease the phosphorylation of RBL1 at the G1/S boundary, thereby inhibiting the development of cells into S phase [36,37]. hsa-miR-1343-3p/RBL1 pathway was involved in CAVD by regulating the VECs cycle. Overexpression of miR-1252-5p might take part in CAVD by promoting MAPK signaling pathway [38], which has been shown to be involved in regulating Ca 2+ entry into cells and mediating osteogenic diferentiation of VIC in CAVD [39,40]. MECP2, a target gene of miR-1252-5p, is an important regulator for the maintenance of normal cardiac development and myocardial structure [41]. Te shorter e2 splice isoform of MECP2 can activate the MAPK pathway [42], which is involved in determining the structure of healthy heart [43]. Our fndings suggest that 3 circRNA-miRNA-mRNA networks could be contributing factors for CAVD.
In conclusion, the pathogenic efects of the ceRNA network in CAVD may be associated with the regulation of VICs and VECs. Te identifed 3 circRNA-miRNA-hub gene axes may constitute the underlying pathophysiology of CAVD ( Figure 9). Tis ofers new insights into pharmacological interventions for CAVD. In considering the multiple factors that are responsible for CAVD disease, including collagen accumulation and resident cytopathic  remodeling [44], these circRNA-miRNA-mRNA axes could also be involved in CAVD formation. We addressed this issue through further analysis; hsa_circ_0026817-hsa-miR-211-5p-CACNA1C, hsa_circ_0007215-hsa-miR-1343-3p-RBL1, and hsa_circ_0007215-hsa-miR-1252-5p-MECP2 may be a new efective and potential target for the treatment of CAVD.

Conclusion
Te establishment of CAVD is a result of the contribution of multiple regulatory factors. We constructed the circRNA-miRNA-mRNA regulatory network by microarray data mining and comprehensive bioinformatics analysis. It reveals that hsa_circ_0026817-hsa-miR-211-5p-CACNA1C, hsa_circ_0007215-hsa-miR-1252-5p-MECP2, and hsa_-circ_0007215-hsa-miR-1343-3p-RBL1 axes may play a crucial part in CAVD and may provide new insights into the pathogenesis and therapeutic targeting of CAVD.

Data Availability
Te data used in this study are publicly available and allow unrestricted reuse through open licenses. All datasets in this study were downloaded from the GEO database. Tese datasets were taken from the following public domain resources: https://www.ncbi.nlm.nih.gov/geo/. Te GEO public database allows researchers to download and analyze public datasets for scientifc purposes.

Disclosure
Linghong Song, Yubing Wang and Yufei Feng were the cofrst authors.

Conflicts of Interest
Te authors declare that they have no conficts of interest.

Authors' Contributions
LS, YW, and YF were responsible for the reliability of the submitted data and drafted the article. HP, SJ, CW, and JD performed the statistical analysis and interpretation of the data. XS, YQ, and WG were responsible for the evaluation and guidance of the full text. LP provided fnal approval of the submitted version. All authors read and approved the fnal manuscript. Linghong Song, Yubing Wang, and Yufei Feng contributed equally. Cell growth Figure 9: Schematic representation of the role of three circRNA-miRNA-mRNA regulatory networks in promoting CAVD. Has_-circ_0026817-hsa-miR-211-5p-CACNA1C encodes enhanced L-type calcium channels that promote calcium infux into VICs, leading to calcifcation. In addition, hsa_circ_0026817 targets hsa-miR-211-5p through sponge activity, regulates cyclin D1, and promotes proliferation of VICs. Hsa_circ_0007215-hsa-miR-1252-5p-MECP2 regulates calcium channels through the MAPK pathway and is involved in calcifcation of VICs. Hsa_circ_0007215-hsa-miR-1343-3p-RBL1 is regulated by TGF-β signaling. RBL1 is inhibited in VEC, blocking G1-S phase cell cycle progression and regulating VECs proliferation. In hsa_circ_0007215-hsa-miR-1343-3p-RBL1, TGF-β signaling is able to regulate RBL1 to block cell cycle progression in G1-S phase and regulate proliferation of VECs.