Congenital heart disease (CHD) is the most common congenital malformation, consisting of structural and functional abnormalities of the cardiovascular system that develop during the embryonic period and are present at birth. CHD affects 0.5–0.8% of all live births and is the leading cause of neonatal death [1‐3]. According to the changes in cardiac haemodynamics and pathophysiology, CHD can be divided into two main categories: cyanosis and nonmagnetic form. Tetralogy of Fallot (TOF) is the most common type of cyanotic congenital heart disease, accounting for 7–10% of all common CHDs [4, 5]. TOF consists of a ventricular septal defect (VSD), pulmonary stenosis, right ventricular hypertrophy, and aorta overriding the ventricular septum. With surgical correction progressing in recent years since the initial successful repair of TOF, the attention of the research community has shifted towards understanding causation [1, 6]. However, the exact pathogenesis of TOF remains elusive. Numerous studies have shown that miRNA expression should be disordered in TOF heart tissues [7]. To date, there have been few relevant reports on the comprehensive profiling of noncoding RNAs (ncRNAs) in TOF hearts.

Current knowledge and understanding have shown that temporal and spatial expression patterns of heart development-related genes are essential in the regulation of cardiomyogenesis, which means that both genetic and epigenetic factors play a crucial role throughout development [8, 9]. Abnormal functional connectivity in the regulation network leads to failure of cardiac cell lineage specification, commitment, and differentiation [10, 11]. Unfortunately, the underlying molecular mechanisms remain relatively poorly understood. Using next-generation sequencing technology, increasing evidence supports the deregulation of ncRNAs in the dysregulation of cardiomyogenesis [12, 13]. Different types of ncRNAs include microRNAs and a variety of long ncRNAs (lncRNAs), such as lncRNAs, antisense RNAs, pseudogenes, and circular RNAs (circRNAs). For circRNAs, using a high-throughput circRNA microarray is beneficial for detecting and studying circRNAs compared with circRNA-seq [14]. However, microarray can only explore known circRNAs. To identify novel circRNAs, next-generation sequencing technology is a suitably sensitive tool.

Recently, circRNAs, as hot topics and trends in scientific research, have provided further opportunities for better understanding the biological mechanisms of heart diseases [15, 16]. CircRNAs, as closed structures, are evolutionally conserved, tissue-specific, and relatively stable. In addition, accumulating integrative analyses have demonstrated that circRNAs are involved in regulating gene transcription and biological processes by acting as miRNA “sponges”, competing for endogenous RNAs (ceRNAs), including RNA transcripts, miRNAs, and circRNAs. All of the observed characteristic features give circRNAs obvious advantages in the exploration of new clinical diagnostic biomarkers and therapeutic targets [17]. However, there have been only a few reports exploring circRNAs in human TOF, even though an enrichment for various diseases/functions of these promising circRNA findings was identified.

In the present study, we performed microarray analysis and evaluated circRNA expression profiles in heart samples from healthy controls and foetuses with TOF. Moreover, by using bioinformatics analysis of these differentially expressed circRNAs, we screened key circRNAs and constructed a ceRNA network. Our results may contribute to a mechanistic understanding of the pathogenesis and the identification of potential therapeutic targets for TOF in the future.

It is well known that cardiac development is a dynamic and complex process. At the same time, cardiac development is finely regulated by the body itself, and minor developmental disorders (such as genetic or environmental factors) are sufficient to cause severe cardiac defects and subsequent embryonic or foetal death. TOF, as one of the most common cyanotic congenital heart diseases in the world, is estimated to account for 7%–10% of CHD cases [37] and is the result of the intersection of genetic, apparent, and environmental risk factors [27]. Studies have shown that rare extracardiac lesions in CNV and syndromic TOF may affect the reproductive suitability and propagation pattern of TOF [38], such as the 22q11.2 deletion and the associated 22q11.2 deletion syndrome [39]. Like other types of CHD, the exact cause of TOF is unclear. At present, with the help of new auxiliary treatment equipment, most children with TOF undergoing corrective surgical repair can survive to adulthood, but the long-term course of postoperative residual lesions in some children may still eventually lead to right ventricular dysfunction, ventricular arrhythmia and advanced sudden cardiac death [40, 41], which assumes a long-term economic burden on families and society. Studies have shown that TOF can occur in other noncardiac (syndromic) or isolated (nonsyndromic) environments. Among them, syndrome-type TOF accounts for approximately 20% of cases; it is mainly associated with 22q11.2 deletion syndrome (22q11del), and its clinical outcome has a worse prognosis [42, 43]. In a large study involving nine institutions, the most frequent harmful mutation in adults and infants with TOF was the NOTCH1 mutation, which is estimated to account for 4.5% of nonsyndromic TOF cases. The next most common harmful mutation site was FLT4, which accounted for 2.4% of cases. In addition, previously involved heart transcription factors (such as TBX1, NKX2.5, GATA4, HAND2, and GATA6, etc.) leading to pathogenic mutations exist in only 1.2% of the population [44], and the lack of efficient biomarkers and the unclear mechanisms underlying TOF are still challenges in the field of cardiology [5, 6, 27].

Previous research has demonstrated that genetic and epigenetic mechanisms, including mutation, histone modification, DNA/RNA methylation, ncRNA modifications and others, play vital roles in cardiac development [8, 12, 18]. Among these major events, ncRNAs may be potential biomarkers for better prognosis and diagnosis. However, research into circRNAs is just beginning. Studies have found that circRNAs are produced by reverse splicing of the precursor mRNA (pre mRNA) of exons of thousands of genes in eukaryotes, in which the downstream 50 splice site (Ss) is linked to the upstream 30 Ss, and the resulting RNA loop is linked by a 30–50 phosphodiester bond at the linkage site[46, 47]. As early as 25 years ago, researchers found only a few circRNAs, which were generally considered aberrant splicing by-products with little functional potential[45, 49]. circRNAs are more stable than linear ribonucleic acids because circRNAs lack a free end, and their cyclic structures prevent degradation by nucleic acid exonuclease[48].

Rapid advances in biochemical methods and the use of high-throughput sequencing technology in recent years have served to isolate and identify a wider range of circRNAs [51]. Genome-wide analysis showed that most of the circRNAs were abundant and conserved across species, displaying cell type-, tissue-, and developmental stage-specific expression patterns in eukaryotic cells, indicating that circRNAs played a vital role in the regulation of transcriptomics and biological processes, including heart-related diseases [29‐31].

Studies have shown that circRNAs are linked to the physiological and pathological development of many organisms, while some circRNAs are related to neuronal function, congenital immune response, cell proliferation, and pluripotency [53]. It has been found that circRNAs have various functions. At the molecular level, they participate in gene expression by isolating microRNA or protein, regulating RNA polymerase II (POL II) transcription, processing pre-interfering mRNA, and translating the resulting polypeptides [50]. Increasing evidence shows that circRNAs, as the binding sites for sponge RNAs and miRNAs competing for miRNAs, maintain RNA-binding protein and control the expression of alternative splicing and parental genes, indicating that cyclic RNA is becoming an important regulatory element at the transcriptional and posttranscriptional level [32, 52]. At present, competitive endogenous RNA (ceRNA) is a new field of RNA biology that implicates a large-scale regulatory network among multiple types of RNA (circRNAs, mRNAs, and miRNAs) at the transcription level [17]. A study has shown that circRNAs derived from PWWP2A can serve as endogenous miR-223 sponges to inhibit myocardial hypertrophy and heart failure [54]. The myocardial infarction-related circRNA (MICRA) level was found to be a robust predictor of LV dysfunction 3–4 months after myocardial infarction [55].

To date, this study is the first to provide systematic profiling of circRNA expression in TOF and to investigate further characterization of the role of the ceRNA network in the pathogenesis of TOF in the early life stage. Moreover, we identified and screened 214 significantly upregulated and 62 downregulated circRNAs through microarray analysis. Then, 19 circRNAs were chosen according to the established multistep screening strategy, and the circRNA-miRNA-mRNA regulatory module was constructed in an attempt to better explore TOF occurrence and development. We analysed the functions and pathways of key mRNAs participating in the ceRNA network using ChIP-X (ChEA) and the CTD database. These key mRNAs were mainly enriched in functions of “heart development” and “NOTCH signalling pathway”, which are closely associated with TOF [8]. In addition, we found that HIF1A is closely linked to human CHD, consistent with previous studies [33]. Studies have suggested that metabolic cardiac remodelling is associated with HIF1A deficiency and maternal diabetic exposure. The combination of diabetic pregnancy and HIF1A deficiency changes vascular homeostasis in the myocardium due to maternal diabetes and increases the risk of cardiovascular abnormalities in offspring [56]. In addition, reduced Nkx2-5 expression together with a sustained hypoxia-inducible factor 1α response led to embryo death [57].

This study has demonstrated that circRNAs can affect the occurrence and development of heart disease and cardiovascular disease, have potential as diagnostic or predictive biomarkers of disease, and offer new potential therapeutic targets. we rescreened the most promising ceRNA module by targeting HIF1A and performed correlation analysis to reveal that hsa_circ_0007798 has the positive correlation with HIF1A. Among the module, miR-199b-5p serves as a bridge between the circRNA and the mRNA. Coincidentally, miR-199b-5p was identified to participate in left ventricular remodeling, associated with pathologic cardiac hypertrophy [35-37]. These studies are in accordance with our finding showing hsa_circ_0007798/miR-199b-5p/HIF1A regulatory module is associated with the cardiac defects, including the TOF. However, compared with encoded RNAs and other noncoding RNAs (miRNAs and lncRNAs), our current understanding of circRNAs still has a significant gap, and the expression of circRNAs in human TOF myocardial tissue has not yet been reported. In this study, the Arraystar human circRNA chip was used to detect the expression profiles of circRNAs in TOF myocardial tissue and normal myocardial tissue, to analyse the differentially expressed circRNAs, to construct the ceRNA regulatory network of TOF using bioinformatics methods, to explore the possible pathogenesis of TOF without a genetic background, and to provide a new method for preventing the formation and development of TOF.