Heart failure (HF) is typically characterized by pathological remodeling with a decrease in cardiac output to fulfill the metabolic demands of the body and is often promoted by diseases that contribute to pressure or volume overload, such as hypertension, myocardial infarction (MI), valvular heart disease and cardiomyopathy^[1]. HF affects a considerable number of individuals worldwide, with estimates surpassing 64 million people^[2]. Therefore, attempts to decrease economic burden of HF have become a major global public health issue.

Cardiac remodeling initially protects the heart as an adaptive response, which refers to a series of compensatory changes in the size, shape, and function of the myocardium in response to a variety of stimuli^[3], but when these changes are prolonged, progressive cardiac dysfunction occurs. MI, ischemia/reperfusion (I/R), injury following coronary revascularization and on-pump cardiac surgery, pressure and volume overload, and neuroendocrine activation are high-risk stimuli that trigger cardiac remodeling^[4]. Histopathological hallmarks, including cell death, defective autophagy, cardiac hypertrophy, fibrosis, inflammation and metabolic and mitochondrial dysfunction, are involved in cardiac remodeling^[5].

RNA-sequencing (RNA-Seq) technology has been applied to identify novel transcripts, such as noncoding RNAs (ncRNAs) and alternative splicing isoforms of each transcript^[6]. Accumulating evidence has revealed that ncRNAs, such as microRNAs (miRNAs), PIWI-interacting RNAs (piRNAs), circular RNAs (circRNAs), and long noncoding RNAs (lncRNAs), are pivotal regulators of gene expression involved in cardiac remodeling^[7-9]. CircRNAs, with a covalently closed loop structure, confer greater stability than linear RNAs. Most circRNAs are present at relatively low levels in cells, and some of them represent significant or even predominant transcript isoforms^[10]. CircRNAs exert their biological functions through specifically combining miRNAs, binding to RNA-binding proteins (RBPs), translating proteins, and modulating gene transcription^[11,12].

The present review focuses on the functional roles of circRNAs in cardiac remodeling leading to HF. CircRNA biogenesis and its mechanisms of action are briefly overviewed. Moreover, a landscape of the newest findings is described with respect to the role of circRNAs in cardiac remodeling and HF, with a focus on their regulation of cardiac cell death, cardiac hypertrophy, fibrosis and mitochondrial metabolism dysregulation. Furthermore, the current knowledge of the potential of circRNAs for diagnostic and therapeutic purposes is reviewed, and directions for future work are proposed to explore the pharmacological potential of circRNAs in HF.

CircRNAs are covalently closed RNAs that are resistant to the 3′-exonuclease RNase R^[13]. CircRNAs are formed from precursor mRNAs (pre-mRNAs) through the back-splicing. CircRNA molecules are primarily generated via the following two modes of back-splicing: (1) Intron complementary sequence-driven circularization. Reverse complementary repeat sequences (such as Alu elements) exist within the flanking introns. These sequences can form secondary structures such as “stem loops” through base pairing, thereby bringing the splice sites into proximity to initiate circularization. (2) RNA-binding protein (RBP)-driven circularization. Specific RBPs (e.g., QKI and FUS) bind to specific motifs within introns. Through protein dimerization or conformational changes, these RBPs bridge two distal splice sites, inducing circularization.

These two circularization modes can result in three subclasses of circRNAs: intronic circRNAs (IcircRNAs), exonic circRNAs (EcircRNAs), and exon-intron circRNAs (EicircRNAs). IcircRNAs are generated via a lariat-driven mechanism, whereas EcircRNAs and EicircRNAs are produced through intron circularization^[14].

An increasing number of studies have revealed different roles of circRNAs, ranging from the regulation of transcription to the sponging of RNA/protein and translation of protein. The biological functions of circRNAs are diverse and primarily involve four major mechanisms of action: (1) microRNA sponges: The “competing endogenous RNA (ceRNA)” hypothesis posits that circRNAs contain binding sites for specific miRNAs. Some cytoplasmic circRNAs contain multiple miRNA binding sites, allowing a single circRNA to bind a number of miRNAs to block the functions of target miRNAs^[15]. (2) Protein interactions: The interactions between circRNAs and proteins can be categorized into three scenarios: a circRNA binds to two proteins, thereby cementing their interaction; a circRNA binds to one protein and enhances or disrupts its interaction with another protein; and a circRNA binds to two proteins that interact with each other, thereby abolishing the existing interaction^[16]. (3) Translating proteins or peptides: Due to the lack of a 5' cap structure, circRNAs employ cap-independent translation. The internal ribosome entry site (IRES) contained in circRNA can recruit ribosomes to initiate protein translation. Nucleotide sites modified by m⁶A can be recognized by m⁶A reader proteins, thereby recruiting translation initiation factors to start translation of the downstream open reading frame (ORF). Some circRNAs with a continuous ORF can undergo multiple rounds of the rolling circle translation along with the circular template by ribosomes^[17]. (4) Regulating gene transcription: IcircRNAs or EicircRNAs in the nucleus can interact with U1 small nuclear ribonucleoprotein (U1 snRNP) to form a complex that acts on RNA polymerase II, thereby initiating gene transcription. Alternatively, IcircRNAs or EicircRNAs may interact with RNA Pol II to modulate gene expression^[18] [Figure 1].

The increasing evidence has shown that circRNAs play important roles in the pathophysiological mechanisms of cardiac remodeling and HF. Hereafter, studies assessing the functional roles of circRNAs in myocardial remodeling and HF using both in vitro and in vivo models are presented [Tables 1 and 2, Figure 2].

Increasing evidence has shown that the aberrant expression of ncRNAs is associated with the pathophysiology of different conditions^[76]. Therefore, ncRNA-based biomarkers, such as miRNAs and lncRNAs, have been proposed as innovative tools for the diagnosis and prognostication of cardiovascular diseases, including HF^[77].

CircRNAs have specific properties that are useful in the field of clinical biomarkers; they are protected from degradation by exoribonucleases because of their covalently closed circular structure. Moreover, the diagnosis of cardiovascular diseases could be improved by combining the levels of circulating circRNAs with clinical information in an artificial intelligence prediction model^[78].

Several circRNAs possess excellent biomarker potential in cardiac remodeling [Table 3]. The levels of circ_0112085, circ_0053919, circ_0062960 and circ_0014010 were markedly increased in the plasma of HF patients compared with those in the plasma of healthy controls. In addition, the level of circ_0062960 was positively associated with the serum level of B-type natriuretic peptide^[79]. Particularly, Cdr1as was upregulated in the circulation of patients with chronic HF^[80], and circ_0118464 was markedly increased in the epicardial adipose tissue of patients with HFpEF^[81].

Four upregulated circRNAs were identified in the plasma of patients with dilated cardiomyopathy: circ_0003258, circ_0051239 and circ_0051238 in lamin A/C-related dilated cardiomyopathy and circ_0089762 in ischemic dilated cardiomyopathy^[82]. Han et al. reported that circ_0097435 expression was significantly elevated in 40 HF patients. Circ_0097435 expression is increased in doxorubicin-treated cardiomyocytes, promoting cardiomyocyte apoptosis, indicating that circ_0097435 is a potential plasma marker of myocardial injury^[83].

Circulating circRNAs are usually contained in extracellular vesicles (EVs), such as exosomes, microvesicles and apoptotic bodies, which can protect circRNA cargos from degradation^[84-87]. RNA-seq of cardiac EVs from mouse hearts after I/R injury revealed that approximately 185 circRNAs were differentially expressed between these mice and Sham control mice^[88]. The plasma exosomal RNA of 15 chronic coronary syndrome (CCS) patients and 15 noncardiac chest pain patients (NCCP, control group) was subjected to an RNA-seq assay to screen for dysregulated circRNAs. A total of 276 exosomal circRNAs were dysregulated in CCS patients. Among the 103 upregulated circRNAs, hsa_circ_0075269 and hsa_circ_0000284 are suggested to be potential biomarkers for CCS^[89].

Moreover, the levels of circ_0097435 were higher not only in the plasma of patients with HF than in that of healthy subjects but also in exosomes isolated from these patients^[82]. RNA sequencing revealed 893 dysregulated exosomal circRNAs in patients with AMI. Exosomal circ_0001558, circ_0001535, and circ_0000972 were significantly upregulated in AMI patients. In addition, the expression of circ_0001558 increased 4.45-fold in a second cohort of AMI patients, indicating that it is a promising diagnostic indicator for AMI^[90].

These findings motivate further characterization of the potential of circulating circRNAs encapsulated in EVs as indicators of pathological cardiac remodeling and HF. Accurate and sensitive quantification of exosomal circRNAs is critical for their application as clinical biomarkers. In addition to reverse transcription quantitative PCR (RT-qPCR), digital PCR (dPCR) technology has recently emerged as a new-generation platform for exosomal circRNA quantification. However, the use of exosomal circRNAs as diagnostic markers for cardiac remodeling and HF has several limitations, such as the isolation and characterization of exosomes and methods for the quantification of circRNAs. Currently, commercial kits are used to extract exosomes from blood, but the purity of these exosomes does not meet the minimal information for diagnostic evaluation^[91]. The number of clinical samples should be expanded to validate the diagnostic efficacy of the potential exosomal circRNA biomarkers for cardiac remodeling and HF. Moreover, further investigations are necessary to evaluate biomarker performance characteristics, such as sensitivity, specificity and accuracy, and compare biomarker values with existing indicators. In the future, dPCR combined with machine learning will be an important strategy for the analysis of exosomal circRNAs for cardiac remodeling and HF diagnosis.

To date, circRNA-based approaches have been investigated in preclinical studies. Small interfering RNA (siRNA) targeting the junction site sequence of circRNA is a commonly used strategy to knock down circRNA expression. However, enhancing the stability of siRNA, optimizing the targeted delivery of siRNA to cardiac cells and minimizing off-target effects of the designed siRNA are fundamental. Compared with an RNAi-based approach, the CRISPR/Cas13 system could suppress circRNAs more efficiently and with lower off-target effects, but its effectiveness in vivo warrants further investigation^[92]. Moreover, the delivery system of CRISPR/Cas13 and the side effects of Cas13 expression in vivo are elusive and required further demonstration.

Lentiviruses containing instructions for generating circRNAs are typically used for circRNA overexpression in vivo^[93]. Synthetic circRNAs can also be used to administer circRNA. The engineering of artificial circRNAs affecting interactions with detrimental miRNAs and RBPs is a potential future therapeutic approach for HF. CircRNA sponges (circmiRs) with bulged binding sites for prohypertrophic miRNAs, such as miR-132-3p and -212-3p, were expressed via AAV9 vectors in the myocardium of TAC mice, and myocardial hypertrophy was attenuated, and cardiac function was preserved^[94]. Similarly, artificial circSP199a includes complements of the seed sequence of miR-199a-5p, and -3p was designed for functional studies. Exogenous administration of circSP199a, specifically sponges miR-199a-5p and -3p, results in the amelioration of myocardial hypertrophy and fibrosis in TAC mice^[95]. Moreover, artificial circRNAs can be designed to carry IRES and ORF elements for protein translation, which could be a viable option for therapeutic protein expression^[96]. However, artificial circRNAs are not widely used to mimic endogenous circRNAs. Currently, the technology of circRNA chemical synthesis and the enhancement of their protein compilation capability require further improvement and development.

Despite the therapeutic potential of circRNAs, their stability, half-life, immunogenicity and delivery routes to ensure cardiac uptake and additional pharmacological issues remain to be determined. Synthetic circRNAs are homogeneous and do not elicit an immune response in vivo. Moreover, a suitable delivery system for exogenous circRNA can transport circRNAs to cells and enable them to evade surveillance by the autoimmune system.

Smaller liposomes are known to have greater abilities to escape phagocyte uptake, penetrate targeted cells and ameliorate the risk of systemic toxicity^[97]. Stabilized lipid-based nanoparticles (LNPs) are a new kind of liposome that include solid lipid nanoparticles, nanostructured lipid carriers, and cationic lipid amphiphiles and possess improved internal physical stability^[98,99]. LNPs, as promising vehicles, have emerged across the biomedical industry to deliver therapeutic circRNA drugs. However, the distribution and expression kinetics of circRNAs depend on the optimization of lipid components, LNP adjuvants and routes of administration^[100,101].

CircRNAs are recognized as specific types of lncRNAs that participate in the pathogenesis of cardiac remodeling and HF. CircRNAs play critical roles in regulating gene expression at the transcriptional, posttranscriptional, and translational levels. The regulatory function of circRNAs in cardiac remodeling and HF mechanisms, including cardiomyocyte death, cardiac hypertrophy, fibrosis, EC dysfunction and functional disorders of mitochondria, has received increasing attention and in-depth development in recent years.

CircRNAs possess unique properties and are promising candidates for translational medicine research in cardiac remodeling and HF. CircRNAs as potential HF biomarkers require the persistent optimization of standard operating techniques and a reduction in costs for detection in clinics. From the perspective of therapeutic drugs, effective strategies focused on the precise molecular design and heart-targeted delivery of circRNAs need to be developed, as well as eliminating off-target effects and undesired immune responses. Further investigations involving mechanistic research, technological innovation, and translational studies of circRNA drugs will improve the therapeutic potential of these drugs in combating cardiac remodeling and HF.