The Emerging Role of Long Non-coding RNAs and Circular RNAs in Coronary Artery Disease

Coronary artery disease (CAD) is a common disorder caused by atherosclerotic processes in the coronary arteries. This condition results from abnormal interactions between numerous cell types in the artery walls. The main participating factors in this process are accumulation of lipid deposits, endothelial cell dysfunction, macrophage induction, and changes in smooth muscle cells. Several lines of evidence underscore participation of long non-coding RNAs (lncRNAs) and circular RNAs (circRNAs) in the pathogenesis of CAD. Several lncRNAs such as H19, ANRIL, MIAT, lnc-DC, IFNG-AS1, and LEF1-AS1 have been shown to be up-regulated in the biological materials obtained from CAD patients. On the other hand, Gas5, Chast, HULC, DICER1-AS1, and MEG3 have been down-regulated in CAD patients. Meanwhile, a number of circRNAs have been demonstrated to influence function of endothelial cells or vascular smooth muscle cells, thus contributing to the pathogenesis of CAD. In the current review, we summarize the function of lncRNAs and circRNAs in the development and progression of CAD.


INTRODUCTION
Coronary artery disease (CAD) is a common disorder caused by atherosclerotic processes in the coronary arteries. This condition can be asymptomatic or can result in fatal situations. In fact, CAD is the main cause of the mortality associated with coronary heart disorders (1). Atherosclerosis is regarded as a progressive inflammatory condition during which oxidative, hemodynamic, and biochemical factors destruct the function of endothelial cells (2). Subsequent alterations in the permeability of endothelial cells, accumulation of macrophages, production of inflammatory substances, and activation of smooth muscle cells are additional steps in the development of atherosclerosis (3,4). Two classes of regulatory non-coding RNAs, namely, long non-coding RNAs (lncRNAs) and circular RNAs (circRNAs), have been shown to affect the process of atherosclerosis and CAD development (5,6). Although both having regulatory effects on the expression of genes, they vary in terms of biogenesis and mechanism of action. LncRNAs have sizes of more than 200 nucleotides (7) and can function as signal, sequester, scaffold, guide, or enhancer RNAs to influence genomic organization or gene expression (8). They share several features with mRNAs such as the presence of RNA polymerase II binding sites, 3 ′ poly A tails and 5 ′ caps (9). On the other hand, circRNAs are single-stranded covalently enclosed molecules made via back-splicing of linear precursor transcripts (10). Both classes of transcript can influence function of endothelial cells or smooth muscle cells in the process of atherosclerosis. In the current review, we summarize the function of lncRNAs and circRNAs in the development and progression of CAD.

Role of LncRNAs in CAD
LncRNAs can affect CAD pathogenesis through regulation of immune responses, modulation of function of endothelial cells and vascular smooth muscles, and changing lipid metabolism. In some cases, a certain lncRNA can affect more than one route.

LncRNAs Regulate Immune Responses
H19 is a transcript encoded by a conserved imprinted gene cluster containing the insulin-like growth factor 2 gene (11). H19 has been shown to function as a molecular sponge for let-7. Expression of this lncRNA has been shown to be reduced in the muscle of patients with type-2 diabetes, as FIGURE 1 | Expression of LEF1-AS1 is increased in plasma and tissues of CAD patients. LEF1-AS1 inhibits miR-544a, thus decreasing PTEN. This is because miR-544a increases PTEN levels. PTEN diminishes AKT activity. LEF1-AS1 contributes in enhancement of reactive oxygen species (ROS) formation, elevation of inflammatory responses, and fibrosis through miR-544a/PTEN axis (28). On the other hand, expression of MEG3 is decreased in CAD tissues compared with normal tissues. Down-regulation of MEG3 is associated with up-regulation of miR-21. miR-21 binds with 3 ′ -UTR of RhoB and decreases its expression. RhoB has a role in activation of PTEN; therefore, MEG3 down-regulation is associated with the lower activity of PTEN (29). Atorvastatin has a regulatory role on PTEN through modulation of RhoB (30). well as animal model of this disorder. The consequent upregulation of let-7 decreases expression levels let-7 targets (12). This lncRNA has been overexpressed in patients with CAD despite its normal expression in other forms of cardiovascular disorders and therefore has been suggested as a marker for prediction of CAD. Notably, its expression levels have been associated with duration of CAD and serum concentrations of transforming growth factor β1 (TGF-β1). In vitro studies verified the effects of H19 overexpression in enhancement of TGF-β1 secretion (13). The lncRNA CoroMarker has been shown to be functionally clustered with genes, which are related with signal transduction, transmembrane transport, synaptic communication, and innate immune responses, while having negative correlation with inflammation-related genes. Small interfering RNA-mediated silencing of CoroMarker has reduced the production of proinflammatory cytokines (14). IFNG-AS1 is another overexpressed lncRNA in CAD patients whose expression has been considerably associated with Gensini score, as well as levels of inflammatory markers high sensitivity C-reactive protein (hs-CRP), tumor necrosis factor α (TNFα), and interleukin 6 (IL-6). On the other hand, IFNG-AS1 levels have been inversely associated with the levels of anti-inflammatory cytokine IL-10 level (15). Cho et al. (16) have demonstrated DQ485454 as the main ANRIL transcript in the endothelial cells. Expression of this transcript has been significantly higher in endothelial cells compared with THP-1 monocytes. Notably, they reported down-regulation of DQ485454 in CAD coronary arteries as compared with samples obtained from non-CAD arteries. Forced up-regulation of this transcript has attenuated cellular processes participating in CAD initiation as it decreased monocyte adhesion to endothelial cells, transendothelial monocyte migration, and endothelial cell migration. Moreover, expression of several CAD-related genes were altered after DQ485454 silencing (16). A microarraybased study has demonstrated down-regulation of NEXN-AS1 in human atherosclerotic plaques. This lncRNA interacts with the chromatin modifier BAZ1A and the 5 ′ part of the NEXN gene. Overexpression of NEXN-AS1 suppressed TLR4 oligomerization and nuclear factor κB (NF-κB) function, decreased endothelial production of adhesion proteins and inflammatory cytokines, and repressed adhesion of monocyte to endothelial cells (17).
Another study in CAD patients demonstrated down-regulation of Chast, HULC, and DICER1-AS1 in the peripheral blood samples (18). Expression of CASC11 has also been decreased in patients with CAD parallel with overexpression of TGF-β1. Functional studies revealed the impact of this lncRNA in the suppression of TGF-β1 expression in endothelial cells (19).

LncRNAs Alter Function of Endothelial Cells and Vascular Smooth Muscle Cells
ANRIL is another up-regulated lncRNA in CAD patients as well as animal model of this disorder. Overexpression of ANRIL decreases expression of miR-181b and is associated with risk of CAD in the subpopulations of elderly patients with history of smoking, hypertension, and hyperlipidemia. Overexpression of ANRIL in human coronary endothelial cells has downregulated miR-181b, increased p50/p65 expressions, enhanced viability of human coronary endothelial cells, and promoted release of inflammatory molecules and vascular-protecting proteins (20). Another study has revealed overexpression of  ANRIL in patients with acute coronary syndrome in association with levels of monocyte chemoattractant protein-1 and IL-10. Notably, these proinflammatory cytokines are produced in reaction to dysfunction of endothelial cells. ANRIL silencing has enhanced cell proliferation and tubule development and suppressed induction of inflammatory responses and apoptosis of endothelial cells. Such effects were linked to ANRIL-mediated suppression of let-7b and its impacts on the TGF-βR1/Smad signaling (21). Another study has demonstrated higher levels of ANRIL and MIAT in the atherosclerotic arteries when compared to the non-atherosclerotic ones (22). ANRIL can also regulate growth of vascular smooth muscle cells via modulation of CDKN2A/B locus, which has a direct effect on the pathobiology of atherosclerosis (23). H19 in addition to its role in the regulation of immune responses can influence vascular smooth muscle cells. Expression of this lncRNA has been increased in the injured neointima and in human atherosclerotic plaques but is scarcely deceted in normal vessels. H19 sequesters let-7 family microRNAs (miRNAs), which are known to shield vascular smooth muscle cells from oxidative damage (12,24). Expression of lnc-DC has been higher in patients with type 2 diabetes and CAD compared with diabetic patients without CAD. Such up-regulation has been accompanied by overexpression of STAT3. Yet, expression of these genes was not associated with the severity of CAD. Based on the observed correlations between expression of genes, authors have suggested the importance of JAK/STAT-related-lncRNAs in the pathogenesis of CAD (25). Expression of FAL1 has been shown to be elevated in CAD tissues and TNF-α-stimulated endothelial cells compared with normal and unstimulated cells. Up-regulation of FAL1 in endothelial cells has enhanced cell cycle progression, proliferation, and migration via modulation of PTEN/AKT pathway (26). In addition, HIF1a-AS1 partakes in the pathology of atherosclerosis via modulating apoptosis of vascular smooth muscle cells and endothelial cells (27). Expression of CASC11 has also been decreased in patients with CAD parallel with overexpression of TGF-β1. Functional studies revealed the impact of this lncRNA in the suppression of TGF-β1 expression in endothelial cells (19). Expression of LEF1-AS1 has been elevated in plasma and tissue samples of CAD patients, whereas expression of its target miRNA, i.e., miR-544a, has been decreased. LEF1-AS1 modulates proliferation and migration of smooth muscle cells via the miR-544a/PTEN route (28). Figure 1 shows the molecular cascade of contribution of LEF1-AS1 and MEG3 in CAD.

LncRNAs Regulate Lipid Metabolism
LincRNA-DYNLRB2-2 is an lncRNA whose expression is stimulated by Ox-LDL. This transcript enhances ABCA1associated cholesterol efflux and suppresses inflammatory responses via GPR119 in macrophage originated foam cells (31). CHROME is another up-regulated lncRNA in CAD patients whose expression is altered by nutritional and cellular cholesterol levels via the sterol-activated liver X receptor transcription factors. This lncRNA enhances cholesterol secretion and HDL synthesis through suppression of the activity of a number of miRNAs. CHROME silencing in human hepatocytes and macrophages enhances expressions of miR-27b, miR-33a, miR-33b, and miR-128, thus decreasing the levels of their shared target genes, particularly ABCA1, which controls de novo synthesis of HDL (Figure 2) (32). Expression of FAL1 has been shown to be  -- (44) elevated in CAD tissues and TNF-α-stimulated endothelial cells compared with normal and unstimulated cells. Up-regulation of FAL1 in endothelial cells has enhanced cell cycle progression, proliferation, and migration via modulation of PTEN/AKT pathway (26). GAS5 is another down-regulated lncRNA in CAD. Enforced up-regulation of GAS5 in animal models of CAD has improved hyperlipidemia, reduced myocardial damage, suppressed apoptosis of cardiac cells, and diminished oxidative stress, inflammatory damage, and aberrant induction of the Wnt/β-catenin pathway in cardiac tissue (33). The function of up-regulated and down-regulated lncRNAs in CAD is summarized in Tables 1, 2, respectively.

Prognostic Value of LncRNAs in CAD
LncRNAs can be used for evaluation of prognosis of CAD patients. For instance, a long-term follow-up study has demonstrated correlation between down-regulation of CASC11 and poor survival of patients with CAD (19). On the other hand, overexpression of LEF1-AS1 and ANRIL has been shown to be correlated with poor clinical outcome of patients with CAD (28,44). Kaplan-Meier analysis has also demonstrated association between ANRIL and LEF1-AS1 overexpression and short overall survival in CAD patients (28,44). Table 3 reviews the studies that appraised the prognostic role of lncRNAs in CAD.

Diagnostic Value of LncRNAs in CAD
Blood or serum levels of some lncRNAs can be used as diagnostic markers in CAD. The best diagnostic value has been reported for H19 where the receiver operating characteristic (ROC) curves showed the diagnostic power of 0.9367, signifying H19 as a suitable marker for CAD (13). Cai et al. (14) have profiled lncRNAs in circulating peripheral blood monocytes and plasma samples of CAD patients and healthy subjects. Their preliminary results demonstrated possible biomarker role for CoroMarker, BAT5, and IL21R-AS1 lncRNAs. The verification step in the larger cohort of CAD patients supported the biomarker role of CoroMarker. This lncRNA could differentiate CAD patients from healthy subjects with accuracy of 0.920 and in an independent manner from identified CAD risk factors and other cardiovascular disorders (14). Another study demonstrated the accuracy of 0.90 and 0.87 for HULC and DICER1-AS1, respectively, in differentiation between CAD patients and healthy individuals (18). In addition, up-regulation of IFNG-AS1 in CAD patients could be used to forecast the risk of CAD with accuracy of 0.755 (15). Table 4 summarizes the points regarding the diagnostic significance of lncRNAs in CAD.

LncRNAs Polymorphisms and CAD
A number of functional single-nucleotide polymorphisms (SNPs) in lncRNAs have been associated with susceptibility to CAD. ANRIL has been the mostly assessed lncRNA in this regard. For instance, rs1330049, rs2383206, rs10757278, and rs10757274 SNPs have been associated with risk of CAD in Asians (52).
On the other hand, rs2383207 and rs1333049 SNPs of ANRIL have not been associated with CAD risk in Han Chinese (53). In addition, rs1333040 and rs1004638 SNPs of ANRIL have not been associated with this disorder in Iranian population (54). H19 is another lncRNA whose association with risk of CAD has been assessed in some populations. Hu et al. (55) have reported an association between H19 rs2735971 and rs3024270 SNPs and susceptibility to CAD in a Chinese population, suggesting the significance of these SNPs as markers for prediction of risk of CAD in this population. Other SNPs within LINC00841, MALAT1, and lincRNA-p21 have been associated with risk of CAD in some ethnic groups ( Table 5).

Role of CircRNAs in CAD
A comprehensive circRNA profiling in CAD patients has shown up-regulation of 624 circRNAs and down-regulation of 171 circRNAs in these patients compared with healthy subjects. Subsequent validation in a larger cohort of patients supported upregulation of hsa_circ_0001879 and hsa_circ_0004104 in CAD patients. Remarkably, up-regulation of hsa_circ_0004104 has led to aberrant expression of atherosclerosis-associated genes in macrophages (63). Another high-throughput study of circRNAs expression in CAD patients has shown up-regulation of 18   circRNAs, whereas down-regulation has been shown in six circRNAs. Subsequently, authors have reported the role of nine circRNAs in the enhancement of TRPM3 expression through suppression of hsa-miR-130a-3p (64). Expression of circ-SATB2 has been shown to be increased in proliferative vascular smooth muscle cells in association with down-regulation of miR-939. Circ-SATB2 was able to augment expression of a target of miR-939, namely, STIM1. Up-regulation of circ-SATB2 decreases expression of SM22-α, a marker of contractile vascular smooth muscle cells. Functional studies verified the role of this circRNA in the regulation of differentiation, proliferation, apoptosis, and migration of vascular smooth muscle cells through enhancing STIM1 expression (65). Expression of circZNF609 has been shown to be reduced in peripheral blood leukocytes of CAD patients in correlation with levels of CRP and lymphocyte counts. Forced overexpression of circZNF609 has enhanced production of inflammatory cytokines IL-6 and TNF-α, while enhancing expression of IL-10. These effects are possibly mediated through sponging miRNAs (66). In an eminent study, Guo et al. (67) have assessed levels of hsa_circ_0002984, hsa_circ_0010283, and hsa_circ_0029589 in human peripheral blood mononuclear celloriginated macrophages from CAD patients and evaluated the consequences of overexpression or silencing of these circRNAs. Authors have reported down-regulation of hsa_circ_0029589 in macrophages, whereas up-regulation of the N6-methyladenosine levels of hsa_circ_0029589 in macrophages of patients with acute coronary syndrome has been shown. Notably, up-regulation of IRF-1 has diminished the expression of hsa_circ_0029589, but surged its m6A levels. Therefore, IRF-1 has been shown to enhance macrophage pyroptosis and inflammatory responses in acute coronary syndrome and atherosclerotic patients by obstructing circ_0029589 via increasing its m6A modifications (67). Tables 6, 7 show the list and function of up-regulated and down-regulated circRNAs in CAD, respectively.

Diagnostic Value of CircRNAs in CAD
ROC curve analysis has shown that hsa_circ_0001879 and hsa_circ_0004104 could differentiate CAD patients from healthy subjects with diagnostic values of 0.703 and 0.700, respectively. Notably, combination of expression levels of these circRNAs, in conjunction with CAD risk factors, could enhance the diagnostic power (63). Expression of circZNF609 has been shown to be reduced in peripheral blood leukocytes of CAD patients in correlation with levels of CRP and lymphocyte counts. In addition, down-regulation of circZNF609 has been associated with higher risks of CAD with accuracy of 0.761 (66). In a highthroughput circRNA profiling, Zhang et al. have demonstrated differential expression of 22 circRNAs between CAD patients and healthy subjects. Among these circRNAs, hsa_circ_0124644 has been shown to have the highest AUC value. Validation of their results in a larger cohort of CAD patients showed the diagnostic power of 0.769 for this circRNA (69). Combination of expression profile of certain circRNAs with conventional CAD risk factors has enhanced the diagnostic power ( Table 8).

CircRNA Polymorphisms and CAD
Finally, SNPs within circRNAs can alter the risk of CAD. Zhou et al. (73) have appraised the impact of two SNPs at the circFOXO3 flanking introns in the development of CAD in a Chinese population. They reported the association between the rs12196996 G allele and elevated susceptibility to CAD. Moreover, such association was more remarkable among younger individuals and non-smokers. The haplotype rs12196996G-rs9398171C has also been associated with susceptibility to CAD. Functionally, the rs12196996 GG genotype conferred lower amounts of circFOXO3 expression (73).

DISCUSSION
CAD is a pathogenic condition in which several cell types and molecules are involved. In fact, this condition results from abnormal interactions between numerous cell types in the artery walls. The main participating factors in this process are accumulation of lipid deposits, endothelial cell dysfunction, macrophage induction, and changes in smooth muscle cells (5). Non-coding RNAs can influence almost every aspect of this pathogenic process. Dysregulation of several lncRNAs and circRNAs has been noted in CAD. ANRIL has been among the most assessed lncRNAs in CAD. Whereas, most studies have demonstrated up-regulation of ANRIL in patients with CAD or animal models of CAD (20,22), a single study has reported down-regulation of certain transcript of ANRIL in CAD coronary arteries as compared with samples obtained from non-CAD arteries (16). Therefore, transcript variants of lncRNAs might   have different tissue specificity and diverse functional roles. Such detailed analysis of transcript variants has not been performed for other lncRNAs in the context of CAD. Dysregulated lncRNAs in CAD patients have functional interactions with Wnt/β-catenin, NF-κB, TGF-βR1/Smad, JAK/STAT, PTEN/AKT, and mTOR signaling pathways. Therefore, these signaling pathways are putative targets for therapeutic manipulations in CAD. Comprehensive studies are needed to explore the interactions between all mentioned lncRNAs and these pathways to find the most appropriate lncRNA for therapeutic interventions. The lncRNAs with the most robust interactions with higher numbers of these pathways are probably the most suitable targets. Moreover, CAD-related lncRNAs have interactions with a number of miRNAs such as miR-181b, let-7b, miR-544a, miR-34a, miR-33b, miR-26a, miR-27b, miR-21, miR-128, and miR-33a, suggesting the complicated interactions between diverse non-coding RNAs in the context of CAD. Identification of this multifaceted interaction network is a prerequisite for the development of anti-CAD strategies. Such network has a practical significance in the design of prognostic or diagnostic panels. Functional studies have confirmed the causal effects of lncRNAs/circRNAs in the pathogenesis of CAD, as the ectopic expression of these transcripts in human endothelial cells has led to dysregulation of proliferation, cell cycle transition, and migration in the favor of CAD development. Moreover, altered expression of these transcripts in immune cells has provoked immune responses and suppressed anti-inflammatory cytokines. A number of additional lncRNAs/circRNAs have functional impact on deposition of lipid in the vessels as demonstrated by in vivo experiments. The synergic effects of these transcripts in the pathobiology of CAD should also be evaluated in functional studies through establishment of double-knockout models.
Some SNPs within lncRNAs have been related with risk of CAD in certain populations. However, the results of these studies have not been replicated in other populations. Therefore, these data are not sufficient to propose these SNPs as markers for CAD in a pan-ethnic scale. Moreover, assessment of haplotypes of these SNPs and their association with risk of CAD would provide more reliable results in this regard.
In addition, circRNAs could partake in the development of CAD via modulation of proliferation, differentiation, or apoptosis in CAD-related cells such as vascular smooth muscle cells. Such roles may be mediated through modulation of expression of several target genes, particularly miRNAs. Similar to lncRNAs, circRNAs can serve as molecular sponges for miRNAs. Expression levels of some circRNAs can be used as diagnostic markers in CAD.
In brief, CAD has been associated with dysregulation of numerous lncRNAs and circRNAs. Moreover, these kinds of non-coding transcripts can be used as markers for prediction of risk of CAD and disease course. The potential of these transcripts as therapeutic targets should be appraised in upcoming investigations. The most important limitation of studies that assessed the functional role of lncRNAs/circRNAs in the pathogenesis of CAD is that they are mostly dependent on cell line studies or animal studies whose generalization to human subjects is not easy. Moreover, the consequences of the observed altered expression of lncRNAs/circRNAs in human subjects should be assessed in follow-up studies. Finally, the association between lncRNAs/circRNAs signature and response to the therapeutic option in CAD patients including coronary artery bypass graft, percutaneous coronary intervention, and medical therapies should be assessed in upcoming studies.