Circular RNAs in vascular diseases

Vascular diseases are the leading cause of morbidity and mortality worldwide and are urgently in need of diagnostic biomarkers and therapeutic strategies. Circular RNAs (circRNAs) represent a unique class of RNAs characterized by a circular loop configuration and have recently been identified to possess a wide variety of biological functions. CircRNAs exhibit exceptional stability, tissue specificity, and are detectable in body fluids, thus holding promise as potential biomarkers. Their encoding function and stable gene expression also position circRNAs as an excellent alternative to gene therapy. Here, we briefly review the biogenesis, degradation, and functions of circRNAs. We summarize circRNAs discovered in major vascular diseases such as atherosclerosis and aneurysms, with a particular focus on molecular mechanisms of circRNAs identified in vascular endothelial cells and smooth muscle cells, in the hope to reveal new directions for mechanism, prognosis and therapeutic targets of vascular diseases.


Introduction
The vasculature is the most extensive organ system in the body that delivers oxygen and nutrition to all tissues. Vascular diseases are the leading cause of morbidity and mortality worldwide. Therefore, there is an urgent need to identify novel diagnostic biomarkers and therapeutic strategies for vascular diseases.
Circular RNAs (circRNAs) are a type of single-stranded RNA molecules characterized by their covalently closed loop configuration, making them highly resistant to ribonuclease (RNase) digestion. Initially considered as junk RNAs produced by alternative splicing error (1), circRNAs have recently been shown to possess a wide variety of biological functions, including acting as miRNA/RNA binding protein (RBP) sponges, regulating parental genes, and even encoding proteins. This protein-encoding function suggests a potential role of circRNAs in gene therapy. Moreover, circRNAs are also promising biomarker candidates for diseases due to their stable expression, cell type specificity, and relatively high levels in body fluid (2).
Here, we briefly review circRNA biogenesis, degradation, and functions. We summarize circRNAs discovered in major vascular diseases such as atherosclerosis and aneurysm, with a focus on elucidating the molecular mechanisms of circRNAs identified in vascular endothelial cells (VECs) and vascular smooth muscle cells (VSMCs) in the hope to reveal new directions to identify mechanisms, prognosis method and therapeutic targets for vascular diseases.

Identification of circRNAs
In 1976, plant viroids were first identified as circRNAs (3). Later, circRNAs were observed in the cytoplasm of eukaryotic cell lines through electron microscope (4), which was suspected to be viral RNA genome. In the subsequent years, only a handful of mammalian genes were found to transcribe into circRNAs without clearly defined functions. Over a long period of time, circRNAs are considered as error-spliced junk RNA byproducts without critical biological functions (1).
Compared with mRNAs and lncRNAs, circRNAs possess several crucial and unique properties. Firstly, they are highly stable compared with linear RNAs. They can resist RNase digestion due to their covalent loop structure (12), thereby escaping from canonical linear RNA degradation. Secondly, the sequences of most circRNAs are highly conserved among species (10, 13). Thirdly, circRNAs expression is tissue specific (14,15) during development (16). They are highly enriched in the mammalian brain (17) and human platelets (18). Moreover, they are also present in body fluids, including blood, saliva, and urine (19,20).
Fourthly, the expression of circRNAs changes during the transition from physiological to pathological conditions (21,22).
In the lariat-driven circularization model, the downstream 3' splice donor site ligates to the upstream 5' splice acceptor site to  form a 3' → 5' phosphodiester bond and generating a lariat consisting of the skipping exons. This lariat then undergoes canonical splicing, releasing the intron and producing an ecircRNA (10) ( Figure 1A).
In intron-paring-driven circulation model, the pairing of flanking introns brings the splice sites into close proximity, facilitating backsplicing to create ElciRNAs. Once backsplicing is completed, ElciRNAs can eliminate the introns through canonical splicing, ultimately producing ecircRNAs (10) ( Figure 1B). RNA paring can occur at either the repetitive complementary sequence (such as Alu elements) or nonrepetitive complementary sequences (27). These sequences are typically located at the upstream and downstream introns.
In protein-driven circulation model, RBPs can interact with the specific donor and acceptor splice sites, bringing them closer to promote backsplicing, resulting in the formation of EIciRNA and ecircRNA (28) ( Figure 1C). For example, RBP Quaking (QKI) has been demonstrated to bind to sites flanking circRNAforming exons to promote circRNA formation. Insertion of QKI binding sites into linear RNA can induce exon circularization (29). In intron cyclization model, a consensus motif containing a 7 nt GU-rich element near the 5' splice site and an 11 nt C-rich element close to the branchpoint site can promote the branching enzyme to escape and a debranching failure, ultimately leading to ciRNA formation (25) (Figure 2).

Functions of circRNAs
CircRNAs act as miRNA and RBP sponges CircRNAs, based on the specific sequence motif they contain, can bind to and function as molecular sponges for miRNA and RBP. Through binding and quenching of miRNA/RBP, circRNAs reduce the availability of these molecules to their natural targets, indirectly modulating the expression of the targeted genes.
Subsequently, it was revealed that certain circRNAs can act as sponges for multiple miRNAs. For instance, circCHIPK3 can sponge nine different miRNAs through 18 potential binding sites, thereby promoting cell proliferation (22). These miRNAsponging circRNAs are primarily located in the cytoplasm and can selectively bind to specific miRNAs or groups of miRNAs with their miRNA binding sequences. These binding quenches miRNAs and reduces their binding to target gene transcript, and indirectly enhances the expression of miRNA-targeted genes (16,30).
CircRNAs can also function as sponges of RBPs and indirectly regulate gene expression. For example, the second exon of RBP muscleblind (MBL/MBNL1) can circularize to form circMbl in flies and humans. This circMbl and its flanking introns could bind strongly and specifically to MBL proteins to decrease Mbl mRNA production (31). Another example is circPABPN1, which shares HuR binding sequences with linear PABPN1 mRNA. CircPABPN1 competes with linear PABPN1 mRNA for HuR binding, an essential RBP recognizing AU-rich elements in the 3' UTR of its target mRNAs. This competition enhances mRNA stability and translation (32).

CircRNAs regulate parental gene transcription
CircRNAs can regulate their parental gene expression and these circRNAs are primarily located in the nucleus. It was previously reported that intron retention would functionally interfere with the transcriptomes (33), but the exact mechanism remained unclear. So far, both ciRNAs and EIciRNAs, which contain intron sequences, have been reported to locate in the nucleus and regulate their parental gene expression.
CiRNAs, which derive from lariat introns, are abundant in the nucleus and could regulate their parental gene expression through interaction with Pol Ⅱ (25). For instance, the depletion of ci-ankrd52 and ci-sirt7, which are ciRNAs derived from ANKRD52 and SIRT7 introns, leads to a reduction in the Frontiers in Cardiovascular Medicine transcription efficiency of ANKRD52 and SIRT7 pre-mRNA, respectively (25), suggesting that these ciRNAs exert influence on parental gene transcription by interacting with Pol Ⅱ. Similarly, EIciRNAs, arising from exon-intron sequences, also possess this regulatory function. EIciRNAs can interact with U1 small nuclear ribonucleoprotein (snRNP), further recruit Pol Ⅱ to its gene promoter region, and enhance parental gene expression (26). For example, knocking down ElciEIF3J and ElciPAIP2 decreased their binding with U1 snRNP and Pol Ⅱ, eventually decreasing EIF3J and PAIP2 transcription (26).

CircRNAs encode peptides
CircRNAs lack linear mRNA's 5' cap structure and thus cannot use the highly efficient cap-dependent translation mechanism (34) to produce protein. As a result, the majority of circRNAs do not encode peptides. However, they can employ cap-independent mechanisms for translation. Studies have shown that naturally occurring circRNAs can utilize internal ribosome entry site (IRES) elements (35) or m 6 A-dependent translation mechanism (36) to initiate protein synthesis.
In 1986, it was discovered that the hepatitis delta virus possessed a circular RNA genome and could encode proteins in mammalian cells (5). Later, in 1995, the construction of virus IRES into circRNA could initiate the protein coding of ORF through eukaryotic translational mechanism (37). Some naturally occurring circRNAs even contain endogenous IRES sequences that can be recognized by eIF4G2 to initiate translation (35).
Furthermore, the RNA base modification N-methyladenosine (m 6 A) also promotes circRNA translation. The m 6 A reader YTH domain family protein 3 (YTHDF3) can recognize m 6 A modifications and recruit eIF4G2 to initiate translation, a process further enhanced by the methyltransferases METTL3/14 (36). The m 6 A modification can enhance the translation efficiency of circRNAs (38,39).

CircRNA degradation
CircRNAs, unlike mRNA with a linear open end, cannot be directly degraded by exoribonuclease. Thus, circRNAs are stable and have a longer half-life in cells. However, the mechanisms underlying circRNA degradation are not fully elucidated. It was initially shown that circRNA rpsT could be degraded by RNase E (40,41). A subunit, Rrp44, of yeast exosome has been shown to harbor the endonuclease activity and can cleave circRNAs; however, the degradation efficiency of Rrp44 for circRNAs is lower compared to linear RNAs (42).
Furthermore, in response to viral infection and activation, the widely expressed cytoplasmic endoribonuclease RNase L can globally degrade circRNAs, thereby contributing to the activation of the double-stranded RNA (dsRNA)-activated protein kinase (PKR) and autoimmunity (43). CircRNAs that carry m 6 A modification can be endoribonucleolytic cleaved by YTHDF2 (m 6 A reader protein)-HRSP12 (adaptor protein)-RNase P/MRP (endoribonucleases) axis (44).

CircRNAs in vascular diseases
Vasculature systems harbor the heavy duty to deliver oxygen and nutrients to all the tissues and organs in the body. Any disruptions in this intricate system can result in vascular diseases. Two critical cellular components of blood vessels are vascular ECs and SMCs. ECs form the innermost layer of blood vessels and are responsible for secreting various vasoactive substances, such as nitric oxide and angiogenic factors. ECs thus play a pivotal role in regulating vascular tone, maintaining vascular homeostasis, and promoting angiogenesis (45). Vascular SMCs are the main component of the vascular medial layer and are crucial for maintaining vascular structure. The constriction and dilation of VSMC layers regulates vascular tone, blood flow, and blood pressure (46). It has been established that endothelial dysfunction and the phenotypic switch of SMCs contribute to the development and progression of numerous vascular diseases.
With the rapid advancements in next-generation sequencing technologies and bioinformatic tools, researchers have identified dysregulated circRNAs in various vascular diseases, including conditions like atherosclerosis and aneurysms. These circRNAs are appealing due to their stability, presence in body fluids and plasma, and specific tissue expression patterns. Consequently, there is growing anticipation that circRNAs hold the potential to serve as valuable biomarkers for predicting, diagnosing, assessing treatment effectiveness, and determining prognosis in vascular diseases (2). While circRNAs have multiple functions, most of those identified in the context of vascular diseases have primarily been demonstrated to function as miRNA sponges. Functions such as parental gene regulation and peptide encoding have not been reported.

CircRNAs in atherosclerosis
Atherosclerosis (AS) is a chronic inflammatory disorder characterized by lipid deposition and fibrous cap formation in the arterial wall (47). The initiation of AS is closely associated with low-density lipoprotein (LDL), a particle containing the apolipoprotein B component (48,49). Endothelial dysfunction and SMC phenotypic switch play critical roles in atherosclerosis initiation and progression (50).
Oxidized LDL (oxLDL) is widely used in in vitro cell culture systems to replicate the pathological processes involved in AS initiation. Numerous studies have confirmed that oxLDL contributes to endothelial dysfunction. OxLDL exposure can lead to reduced viability, suppressed migration, apoptosis, inflammatory responses, and oxidative stress in ECs (50). Furthermore, oxLDL can induce a transition in VSMCs from a quiescent, contractile state to a proliferative and synthetic state, both of which play pivotal roles in the formation of atherosclerotic lesions (51).
In recent years, numerous circRNAs have been reported to be dysregulated during the development of AS, with their roles in endothelial dysfunction and SMC phenotypic switching explored. These findings offer a novel direction for potential therapeutic strategies in the treatment of AS, as summarized in Table 1. While many immune cells, such as macrophages and lymphocytes, also play critical roles in AS, they are not discussed here.

EC-related circRNAs
In a study by Wen et al. (52), a comparison of circRNA profiles was conducted in serum exosomes obtained from patients with stable plaque atherosclerosis (SA) and those with unstable/ vulnerable plaque atherosclerosis (UA). Their findings revealed a positive correlation between circKIAA1429 (circRNA-0006896) levels in serum exosomes and triglyceride, LDL cholesterol, and C-reactive protein levels in UA patients. Moreover, it was observed that serum exosomes derived from UA patients could upregulate circKIAA1429 expression in HUVECs, reduce microRNA-1264 levels, elevate DNMT1 levels and STAT3 phosphorylation to reduce SOCS3 expression, and notably enhance HUVEC proliferation and migration when compared to both mock and SA groups.
Besides miRNA sponging, circRNA has also been reported to bind with proteins in SMCs. CircANRIL impaired exonucleasemediated pro-rRNA processing and ribosome biogenesis in VSMCs and macrophages by binding to 60S-ribosome assembly factor PES (83). CircANRIL induced nuclear stress and p53 activation, further induced apoptosis, and inhibited proliferation of VSMCs (83).

EC and SMC-related circRNAs
CircRNAs have been reported to regulate both EC and SMC functions. Overexpression of circGRN (circ_0044073) promoted the proliferation of HUVECs and HUVSMCs by interacting with miR-107 and activating the JAK/STAT (84) signaling pathway (85). CircGRN enhanced oxLDL-treated VSMC dysfunction by serving as miR-377-3p sponge to increase AURKA expression (86).

CircRNAs in aneurysm
Aneurysm refers to the local or diffuse dilation or bulge of a blood vessel (87). The thin and weakened vessel wall is more susceptible to dissection or rupture. Most aneurysms are asymptomatic, but the rupture of aortic aneurysms (AA) or intracranial aneurysms (IA) can be life-threatening (88, 89). Although the molecular mechanism of aneurysm formation is not completely understood, dysregulation of VSMCs and aorta wall matrix degradation are critical pathological changes in AA development (90). This often involves the transition of VSMCs from a contractile phenotype to a proliferative and inflammatory one. Furthermore, the destruction of the extracellular matrix (ECM) is a significant feature of AA, mediated by an imbalance between metalloproteinases (MMPs) and tissue inhibitors of metalloproteinases (TIMPs).

AA-related circRNAs
CircRNAs have been reported to regulate SMC proliferation and apoptosis, mainly through their regulation of miRNA expression or miRNA sponge functions ( Table 2). CircBTBD7 (hsa-circ-000595) was increased in AA tissue as well as in hypoxic aortic SMCs. Knockdown of circBTBD7 could increase miR-19a expression and reduce hypoxia-induced apoptosis of VSMCs (91). CircCCDC66 was upregulated in AA. Depletion of circCCDC66 enhanced VSMC proliferation and inhibited apoptosis via acting as a miR-342-3p sponge to promote CCDC66 transcription (92). CircChordc1 promoted the VSMCs contractive phenotype and enhanced their growth by vimentin degradation and GSK3β/β-catenin signaling activation, extenuating vascular wall remodeling, and reversing aneurysm progression (93). CircCDR1as served as an inhibitor of miR-7, leading to increased expression of the miR-7 target CKAP4. This promotes the proliferation and reduces the apoptosis of VSMCs (94). CircCBFB functioned as a sponge for miR-28-5p, releasing GRIA4 and LYPD3 from miR-28-5p suppression to promote VSMC growth (95). CircEIF2S2 (hsa_circ_0092291) reduced angiotensin II-induced damage in HAVSMCs by serving as a miR-626 sponge and upregulated COL4A1 expression (96). CircTMEM189-UBE2V1 (hsa_circ_0002168) interacted with miR-545-3p to upregulate CKAP4 levels, facilitating proliferation and restraining apoptosis in VSMCs (97).
CircRNAs have also been reported to regulate ECM degradation ( Table 2), suggesting their critical involvement in AA. Wang et al. identified 65 differentially expressed circRNA in abdominal AA tissues and found that circRBM33 was upregulated in AA samples and angiotensin II stimulated VSMCs. CircRBM33 overexpression increased MMP2 expression and reduced TIMP2 expression, leading to ECM degradation. CircRBM33 acting as miR-4268 sponges to upregulate EPHB2 and inhibit TIMP-1 expression to mediate ECM degradation
CircRNAs implicated in endothelial dysfunction also play a role in IA. Zhang et al. analyzed circRNA microarray of ECs isolated from RIA and UIA and identified that circIGF2BP3 (circ_0079586) and circRanGAP1 expression were upregulated in RIA.
In a study by Huang et al., 216 IA patients and 186 healthy volunteers were selected to assess the expression of circITGAL (hsa_circ_0000690) in their peripheral blood. They discovered that circITGAL expression was lower in individuals with multiple IAs than in healthy volunteers, indicating its role as a potential biomarker for IA diagnosis and is closely related to the volume of hemorrhage (108). In another study, Wu et al. analyzed circRNA expression profiles in peripheral blood using circRNA microarrays to compare healthy volunteers with patients harboring saccular aneurysm wall enhancement (AWE). This research revealed that circPGAP3 (hsa_circ_0007990) expression upregulated in UIA patients compared with healthy people. And circPGAP3 expression was significantly higher in UIA patients with AWE than those without. Thus, circPGAP3 could be a novel biomarker for UIA (109). Chen et al. analyzed differentially expressed circRNAs between normal superficial temporal arteries and IA samples and found that circDUS2 was upregulated in IA tissues, suggesting its potential role as biomarker (110).

Conclusion and future perspective
Over the past decade, our understanding of the functions of circRNAs has begun to emerge. Increasing evidence supports the notion that circRNAs, far from being regarded as "junk RNA", play pivotal roles in a wide array of biological processes. In the context of vascular diseases like atherosclerosis and aneurysms, numerous circRNAs have been identified within EC and SMC, and their involvement in the pathological progression of these vascular diseases has been indicated. Furthermore, some circRNAs have exhibited differential expression levels in the serum or exosomes of patients with atherosclerosis or aneurysms compared to the general population, suggesting the potential utility of circRNAs as biomarkers for these vascular diseases and for predicting disease prognosis.
However, despite these significant strides, the biological functions and molecular mechanisms of circRNAs in vascular diseases remain incompletely understood. To date, the majority of circRNAs reported to play a role in EC and SMC function in atherosclerosis and aneurysms primarily function as miRNA sponges or regulators of miRNAs. In-depth studies using in vivo animal models are scarce, and the roles of circRNAs in functions such as binding to RBPs, regulating parental genes, or encoding peptides have yet to be identified. Further research in these areas is needed to unravel circRNA functions in vascular diseases and to exploit their potential therapeutic and diagnostic applications.
Studying circRNA functions at the DNA level, both in terms of gain and loss of function, is historically challenging, as these approaches can inadvertently affect the levels of their linear RNA counterparts. However, recent advancements in RNA circularization techniques, such as the improved PIE (permuted intron exon) methods (113), have demonstrated high efficiency and the ability to synthesize bulk circRNAs in vitro. This has opened up new avenues for circRNA research, particularly in overexpression studies, as it allows for manipulating circRNA levels without altering the linear RNA transcript from their parental genes.
Furthermore, due to the inherent stability and longer half-life of circRNAs, in vitro synthesized circRNAs containing specific gene open reading frames hold great promise for gene therapy in the treatment of vascular diseases. These synthesized circRNAs can serve as valuable tools for exploring novel therapeutic strategies.
By integrating bioinformatic methodologies, extensive circRNA profiling, and cutting-edge circRNA synthesis techniques, a more comprehensive understanding of the roles played by circRNAs in vascular diseases can be achieved. This enhanced comprehension Liu et al. 10.3389/fcvm.2023.1247434 Frontiers in Cardiovascular Medicine of circRNA mechanisms will pave the way for the development of innovative therapies for vascular diseases.

Author contributions
All authors listed have made a substantial, direct, and intellectual contribution to the work and approved it for publication.

Funding
This work is supported by grants to SM (National Natural Science Foundation of China Grant 82170512 and Startup Grant from Guangzhou Laboratory) and to TZ (Guangzhou Science and Technology Plan Project 202201011255).

Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Publisher's note All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.