Circular RNA: From Biogenesis to Functional Paradox

ABSTRACT

Circular RNA (circRNA) is a new class of RNA molecules which, unlike linear RNA, form a covalently closed continuous loop without 5′-cap and 3′-tail. New research has revealed thousands of endogenous circRNAs in mammalian cells. CircRNAs are generated mainly from exonic or intronic sequences, and in their biogenesis, reverse complementary sequences and RNA binding proteins (RBP) play an important role. A majority of them are stable, conserved and resistant to RNase R, and often show development type / tissue-specific expression. Many roles of circRNAs are coming to the forefront such as microRNA (miRNA) sponges, transcriptional regulators and parental gene expression modifiers. Growing evidence shows that circRNA can play important roles in regulating various diseases, such as atherosclerotic vascular disease, neurological disorders and cancer, and potentially act as diagnostic or predictive biomarkers for certain diseases. Similar to long noncoding RNAs (lncRNAs) and miRNAs, circRNAs are also becoming a new research hotspot in RNA biology and may be widely involved in life processes. In this review we examine the biogenesis and the properties of circRNAs, their functions and potential significance in various diseases.

INTRODUCTION:

 

Circular RNA has emerged as a new class of endogenous RNA that forms a closed loop without 5′-3 ‘polarities or polyadenylated tails [1], making them much more stable than linear RNA and resistant to RNA degradation by exonuclease or RNase R.  CircRNA was first identified from transcript of a candidate tumor suppressor gene [2]. Very few circRNAs were discovered in recent years [3, 4, 5, 6, 7, 8]. These molecules were long considered as molecular flukes- specific for certain pathogens, such as the hepatitis δ virus [9] and some plant viruses [10]. However, this definition of “artifact” has recently been modified in the light of several recent studies revealing a significant amount of circRNAs present in different cell types of organisms extending from archaea to mammals. With recent advances in deep RNA sequencing technologies and bioinformatics, a large number of circRNAs were discovered and their diverse properties revealed. Like linear mRNAs, circRNAs were found to be specifically expressed in cells and tissues. For example, hsa_circRNA_2149 has been found to be present in CD19 + leukocytes but not CD34 + leukocytes, neutrophils or HEK293 cells [11]. Some nematode circRNAs appeared to be expressed in oocytes but were found absent in 1 or 2-celled embryos [12]. CircRNAs form >14% of actively transcribed genes in human fibroblasts and in some cases the abundance of circular molecules exceeded that of associated linear mRNA by >10-fold [13]. They are mainly cytoplasmic, but some appear to be enriched in the nucleus as well [14]. Its presence has also been validated in extracellular body fluids [cell-free saliva] of healthy individuals [15]. Majority of the circRNAs are endogenous non-coding RNAs and only a small portion of exogenous circRNAs have been identified, such as hepatitis δ virus (HDV) and engineered circRNAs with internal ribosome entry sites (IRES) [16, 17]. Primarily they arise from exons, while some come from introns and have miRNA response elements (MRE) [13]. CircRNAs show least polymorphisms at the predicted miRNA target sites [18]. In addition to some circRNA, most sequences have been highly conserved evolutionarily between species [19]. These properties indicate that circRNAs are potentially involved in transcriptional and post-transcriptional processes and can become ideal biomarkers in the diagnosis of diseases like Cancer, Neurological disorders and cardiovascular diseases.

BIOGENESIS:

CircRNAs are derived from pre-mRNAs and generally catalysed by either the spliceosomal machinery or by groups I and II ribozymes [20]. The inhibition of canonic spliceosome using isoginkgetin, a pre-mRNA splicing inhibitor, reduces the levels of both circRNAs and spliced linear transcripts, thus providing evidence of the role of spliceosome in the biogenesis of circRNA [21]. CircRNAs have several origins with majority of them deriving from exons of the coding regions and the rest from 3′-UTR, 5′-UTR, introns, intergenic regions and antisense RNA. CircRNA can be derived from both canonical and non-canonical cleavage processes. Different from the orthodox splicing of linear RNA, a single gene locus can produce varied circRNAs through a selection of alternative retro-splicing sites [22]. So far, three types of circRNAs have been identified: exonic circRNAs (ecircRNAs), intronic RNAs (ciRNAs) [23] and exon-intron circRNAs (ElciRNAs) [24]. A study revealed 83% of circRNAs overlapping with protein coding regions and exonic circRNAs account for the largest class found in animals and plants [25]. The length of ecircRNA varies from hundreds to thousands of nucleotides (nts) with an average estimated length of approximately 547 nts and extends about less than five exons. Some ecircRNAs can interact with microRNAs and / or RBPs, and many of them surround the other exon that contains canonical translation start codon [26].

Lariat-driven circularization or exon skipping model (Fig. 1a):

Exon skipping (ES) is a common type of alternative splicing with a well-known effect on mRNA formation [27]. A new study, deciphered the important role that this process may have in ecircRNA biogenesis as well. During exon skipping a large lariat containing the exon(s) is formed which subsequently undergoes internal cleavage to remove the intron and generate ecircRNA or EIciRNA [28]. The analysis of RNA Seq data sets has shown that the majority of skipped exon(s) can produce ecircRNA in human umbilical vein endothelial cells stimulated by TGF-β or TNF-α [29]. In addition, a recent study has shown that a lariat-containing exon production [presumably from exon skipping] is a very common step in the ecircRNA biogenesis of Schizosaccharomyces pombe [30]. More studies are required to test whether circularization proceeds only with inherent properties of the lariat-containing exon(s) or other factors such as RBP.

Intron-pairing driven circularization or Direct Back Splicing Model (Fig. 1b):

 

In direct splitting, exons are split in non-canonical order, a border point first attacks the downstream 5′-splice site (splice donor) at its 2′-hydroxyl group and then attacks the resulting 3′-hydroxyl end upstream of the 3′- splice site (splice acceptor) which forms a circRNA [22]. It has been verified and suggested that the intron-pairing-driven circulation and cis-acting factors, such as reverse complementary sequences including IRAlus, play role in circRNA biogenesis [28, 31]. A genome wide analysis in Drosophila has revealed that the lack of nucleotide motif for intron pairing in many gene loci produces abundant circRNA and the length of flanking introns appears to be a critical factor for back-splicing [32, 33]. CircRNAs in human cells were detected using a minigene expression vector in which constructed flanking introns are significantly shortened (21). CircRNA contains some repeated transposable sequences, such as Alu repeats and other short sequences, on their two flanking introns that form stable base pairs [34]. After transcription into RNA precursors, the exon sequences on both sides of the cycling portion will hold together with each other under the influence of repetitive sequences, such as Alu, to form a circular shape. After this, the spliceosome binds to the cyclic molecule under the effect of U6 and U2 and selectively cut the exons in the cycle region by interacting with the protein complex and the U5 nucleus [35,36]. At the same time, exonic connections are reversed, forming mature circRNA. In addition, many circRNAs do not appear to be strong intron base pairs (62 and 91% in C. elegans and humans, respectively) [37].

The above given mechanisms occur in-vivo and correlate with the canonical spliceosome, but some evidence has shown that intron-paired circularization may occur more often than lariat-driven circularization [38].

Types of CircRNAs:

Intronic circular RNAs (ciRNAs)

A new type of circRNA derived from introns was reported in human cells and was called circular intronic RNA (ciRNA). They represent a small part of circRNA, of which only 19.2% exist in humans [39] and a very small fraction in plants. The mechanism of ciRNA formation differs from that of ecircRNA. Unlike ecircRNAs, the ciRNA has 2′-5′ head-tail joint (Fig 3) and different properties in terms of stability, subcellular localization, abundance, conservation and functions. Intronic circRNA shows nuclear localization and less conserved sequence than exonic circular RNA, and former being more exclusively found in human cells. CiRNA synthesis depends on a consensus motif containing a 7-nt GU-rich element near the 5’splice site and an 11-nt C-rich element near branch point site

During the back splicing event, the two segments first bind to form a circular structure followed by the incision of exonic and intronic sequences in the binding portion of the spliceosome, and finally the residual introns are stitched together to form mature circRNA [40]. (Fig.4).

Exon– intron circRNAs (ElciRNAs)

A group of scientists found a novel type of circRNA where exons circulate together with introns, where the latter are “retained” between their exons and these are termed as exon-intron circRNA (ElciRNA).

This characteristic of introns sandwiched between exons makes this subclass unique even if they share certain functions with both ecircRNAs and ciRNAs. Similar to ecircRNA, ElciRNA’s reverse complementary sequences are in their long flanking introns, indicating that there is a common mechanism for their biogenesis. Like ciRNAs, EIciRNA is predominantly in the nucleus and has been shown to be bound to RNA Pol II to promote the transcription of their parental genes in cis by interaction with U1 small nuclear ribonucleoprotein (snRNP). Direct back-splicing includes two pathways: First, intron pair-driven circularization and RBP pair-driven circularization that depend on sequence-specific RNA-binding proteins and second exon- skipping that is essential to ensure proper production of circRNA. RBPs form a bridge between the flanking introns leading the splice donor and splice acceptor sites close to each other, thereby promoting biogenesis for both ciRNAs as well as EIciRNAs [41]. (Fig. 2)

In fact, as the specific mechanism of the EIciRNA generation is unknown. Many ecircRNAs or ElciRNAs may arise from the same gene locus through alternative circularization. Zhang et al. suggested a model of “alternative circularization” similar to alternative splicing (Fig.2) and found that competition in RNA pairing through complementary sequences (either repetitive or non-repetitive) over or within individual flanking introns can significantly affect splicing and exon circularization. Complementary sequences within individual flanking introns may be sufficient to promote linear mRNA generation. Conversely, complementary sequences across flanking introns may favour exon circularization. The competition between inverse complementary sequences can result in multiple circRNA transcripts being processed from a single gene (Fig 2). Alternative circularization may be species specific depending on the different distributions of complementary sequences across different species. The presence of complementary sequences is necessary but not sufficient for exon-circularization, suggesting that the mechanism is very complicated and possibly governed by other factors, such as RBP [34].

Gao et al. explored internal components including alternate splitting events (AS), such as exon skipping, intron retention (IR) and alternate 5 ‘or 3’ cleavage sites (A5SS and A3SS) within the circRNA and showed that the frequency of AS events varies in different cell types, suggesting their potential roles in gene regulation [42].

 

REGULATION:

The regulation of circRNA biogenesis basically depends on the cis regulatory elements and the trans acting factors that govern splicing. Both the cis elements and the trans factors promote circRNA biogenesis by getting downstream donor and upstream acceptor sites in the vicinity. In the case of cis elements, it has been found that inverted Alu repeats are enriched in introns flanking the circularized exons and non-repetitive complementary sequences can also serve to promote circularization both in vivo and in vitro [13, 27, 37]. So far, at least three RBPs have been confirmed to serve a regulatory role in the circRNA biogenesis. Primarily, a well-known splitting factor Muscleblind (MBL) binds to circMbl flanking introns to facilitate the circularization of its own second exon and provoke the formation of circRNA that functions as RBP to bridge two flanking introns together. Similarly, another RBP, Quaking (QKI), belongs to STAR (signal transduction and activation of the RNA) family of RBP known to play a role in pre-mRNA splicing, shown to improve the formation of many circRNA by binding at recognition elements in upstream and downstream introns, and may facilitate circRNA biogenesis during epithelial to mesenchymal transition (EMT) [43]. QKI affects pre-mRNA together with mRNA turnover and translation, and has been involved in a number of diseases including cancer [44]. QKI knockdown and MBL overexpression can in turn decrease or increase circRNA output, respectively. For the generation of circRNA, there is a requirement for a specific (MBL / QKI) binding site within flanking introns. Another regulatory enzyme is adenosine deaminase acting on RNA (ADAR-1), a double-stranded RNA editing enzyme, which negatively regulates the circRNA biogenesis and counteracts their expression by melting the stem structure. It has been proposed that A-to-I editing in the double-strand RNA region mediated by ADAR-1 reduces the RNA pairing potential across the flanking introns, thereby suppressing the back-splicing required for circRNA formation [45]. Therefore, RBPs can act as activators or inhibitors for the formation of circRNA. Similar to linear RNA, the biogenesis of circRNA is also regulated by the spliceosomal RNA machinery. The RNA spliceosome is RNA-based enzyme with a U5 core consisting of 5 snRNA and several proteins. Under the effect of U6 and U2 promoters, the multiple proteins interact during pre-RNA process [36]. In addition, a linear RNA molecule can be processed into different types of RNA, including mRNA, lncRNA and circRNA, through various splicing events. For example, HIPK3 pre-mRNA can be divided into HIPK3 mRNA and its other exon can also form circRNA hsa_circ_0018082 [34]. Taken together, these evidences suggest that the biogenesis of the circRNA may depend on different factors that are likely to work together to regulate the results of back-splicing. In addition, different circRNAs can be regulated with different mechanisms and their production in cells seems to be more complicated than previously believed.

CIRCULAR RNA TRANSPORTATION AND STABILITY:

Circular RNAs have long half-lives because they are naturally resistant to RNA exonuclease degradation due to their unique covalent bond between the 5 ‘and 3’ ends. This high stability suggests that the progressive accumulation of circRNA is dominated by their slow turnover instead of the production, resulting in accumulation in the cells if their levels are not adequately controlled by cellular mechanisms. Two mechanisms were suggested for their transport; First, using the exon junction complex to assist in export from the core [46] and another is the secretion of circRNA from cells via extracellular vesicles (EVs). EVs are membrane-bound vessels released from cells and may contain cellular components, including proteins, lipids, and RNAs. Different EV types, including exosomes, have been characterized on the basis of their biogenesis or release pathways. RNA Seq data derived from patients with colorectal cancer (CRC) showed that in exosomes, the expression level of circ KLDHC10 significantly increased in serum in CRC patients compared with normal healthy controls. Also the accumulation of circRNA during neuronal differentiation, synaptic development and foetus development has been observed [47]. Therefore, the excess of circRNA in body fluids and blood due to high RNase resistance allows them to serve as potential biomarkers in various diseases.

Nevertheless, cellular levels of circRNAs are highly controlled, endonucleases provide access points for the exonucleases, probably by facilitating the disintegration of circRNAs [48]. The main RNA endonucleases in eukaryotic cells include Ago-2 (which works in RNA silencing), angiogenin (which cleaves tRNA during stress), CPSF73 (which works in the formation of mRNA 3 end), IRE1 (which works in ER stress), RNase L (which is involved in native immunity) and SMG6 (which is important for nonsense-mediated breakdown). CircRNA detected in EVs including ciRS-7/CDR1asindicates that cells may use EVs to transport circRNA for their communication with other cells, possibly as part of the ceRNA network. The CDR1as / ciRS-7 contains an almost perfect miR-671 target site that can be cleaved by Ago-2 to trigger the degradation of the transcript. Possibly, the packaging of circulating RNAs in EVs is either to eliminate excess circular RNA or involvement in cell-to-cell communications.

 

FUNCTIONS:

CircRNAs as miRNA sponges or competing endogenous RNAs

 

Accumulated evidence indicates the role of circRNAs as potent competing endogenous RNA (ceRNA) molecules or miRNA sponges, but since the circRNA has no free ends, these are predicted to avoid microRNA mediated deadenylation [49]. CeRNA molecules such as mRNA, lncRNA and pseudogenes, contain shared microRNA response elements that allows competition for miRNA binding [50] suggesting its role in miRNA functioning and regulating gene expression. The strongest evidence of the sponging activity of the circRNA is derived from a study of exonic circRNA ciRS-7/ CDR1ase, that sponges mR-7 or CDR1 antisense and murine sex-determining region Y (Sry). Both ciRS-7 and Sry are known to bind miRNAs without degradation, making them powerful ceRNAs. Murine Sex-determining region Y (Sry) gene responsible for mammalian sex determination can produce a testis-specific single exon circular transcript. Transfection studies in HEK293 cells have shown that once co-transfected with circRNA Sry, expression vector and pJEBB-138, this single-exon circRNA can bind at 16 binding sites with miR-138 and can be co-precipitated with Argonaute 2 (AGO2) [26]. These studies indicate that Sry acts as a miR-138 sponge. Natural circular antisense transcript termed as cerebellar degeneration-related protein 1 transcript (CDR1as) is the translational product of the cerebellar degeneration-related protein 1 (CDR1) gene. Also it was shown that CDR1as can bind with miRNAs and can be degraded by miR-671 [51].

The binding site of miR-671 shows a little variation across species and possesses near-perfect complementarity. Research studies revealed that CDR1as has atleast 70 conserved seed matches for miR-7 and is bound by miRNA binding proteins (e.g. Argonaute proteins). The altered expression of CDR1as and its inhibitor miR-671 decreases the expression of miR-7 target genes such as alpha synuclein gene (SNCA), epidermal growth factor receptor (EGFR) gene and Insulin receptor substrate 2 (IRS2) gene [52, 53, 54]. CDR1as is highly expressed in nervous tissue and the overexpression of CDR1as in zebrafish embryos, that lacks the cdr1 locus, significantly reduces the mid-brain size and mimics the phenotype of miR-7 loss-of-function causing morphological defects in the region.

Most of the ecircRNAs are known to contain lesser number of putative miRNA binding sites. However, bioinformatics analysis of mammalian ecirRNA data generated by circRNA-Seq experiments revealed very few circRNAs with more than 10 miRNA binding sites. For example, circRNA cir-ITCH is known to span several exonic binding sites of the E3 ubiquitin protein ligase (ITCH) gene and acts as a miRNA sponge of miR-7, miR-17 and miR-214. In contrast to circRNAs in mammals, D. melanogaster circRNAs possess at least one thousand well-conserved miRNA seed matches [32], but fly circRNAs functioning as miRNA sponges is yet to be elucidated. Overall, whether sponging activities of circRNAs is a general phenomenon or not and how balance between networks of circRNAs, miRNAs and ceRNAs is maintained to regulate cellular homeostasis is still a mystery.

CircRNAs as regulators of alternative splicing or transcription, mRNA trap and transcriptional regulations

CircRNAs are known to regulate alternative splicing and transcription. It was reported that circRNA CircMbl, generated by the second exon of a general splicing factor MBL/MBNL1 competes with canonical pre-mRNA splicing process. CircMbl and its flanking introns have conserved sites that strongly and specifically bind MBL modulating the MBL levels that significantly affects circMbl formation [55]. Therefore, MBL, may regulate alternative splicing by modulating the balance between canonical splicing and circMbl biogenesis. CircRNAs could act as mRNA traps and these mRNA traps act to regulate protein expression by sequestering the translation start site. The importance of mouse formin (Fmn) gene for limb development that can produce exonic circRNA via backsplicing of the Fmn coding region is known. Knockout mice lacking the splice acceptor upstream of Fmn coding sequence show normal limb development and non-measurable expres­sion of the exonic circRNAs, but they show an incompletely penetrant renal agenesis phenotype.

Moreover, the exonic circRNA functions as a ‘mRNA trap’, generating a noncoding linear transcript, thereby decreasing the expression level of the Formin protein [56]. It was further uncovered in human fibroblasts that many of the circRNAs containing single exons contain a translation start site and thus can function as mRNA traps [38]. Exonic circRNAs derived from the second exons of the HIPK2 and HIPK3 loci contain the canonical ATG (start codon) in human and mouse cells. In case of HIPK3 it was found that the exonic circRNA is considerably more abundant than the linear protein-coding transcript [13]. Their functioning as mRNA traps is of great importance as they can regulate the phenotypic effect of any target gene. For example, in dystrophinopathy patients, the mRNA trapping by dystrophin exonic circular RNAs might enhance the disease phenotype leading to inactive DMD transcripts, further reducing the pool of translatable mRNAs [57]. However, emerging ther­apies for these dystrophinopathies like antisense oligonucleotides against certain exons that regulate splicing are currently in clinical trials [58].

CircRNAs regulate the parental gene expression

 

As demonstrated in Fig 4 circRNAs regulate the expression of their parental genes. CircRNA biogenesis is mainly dependent on key flanking RNA elements essential for the intronic lariat to escape from debranching because they have less affinity for microRNA target sites, indicating their distinctive functionality [59]. It has been found that some circRNAs present in the nucleus interact to a great extent with RNA polymerase II (Pol II) machines, thereby modulating host-transcriptional activity in a cis-acting manner [60]. A special class of circRNAs e.g. CircEIF3J and CircPAIP2 of EIciRNA have been associated with RNA Pol II in human cells and are predominantly localized in the nucleus. They interact with the U1 subunit of small nuclear penteconucleotides (snRNPs) thereby acting in cis mode to improve the transcription of their parent genes [24]. Likewise, Li and others revealed that both cir-ITCH and its 3′-UTR share some miRNA binding sites for miR-7, miR-17 and miR-214, that increases the expression of ITCH (61). Thus it is speculated that intrinsic circRNAs such as ciRNA and EIciRNA, as opposed to ecircRNA can function effectively in cytoplasmic regulatory processes that seem to predominantly regulating the transcription process in the nucleus.

Translation potential of CircRNAs

 

Lack of essential structures critical to efficient translation initiation regarded circRNAs with no protein-coding ability, but researchers were propelled to search for the translational potential of circRNAs due to its origin from protein-coding sequences, presence of open reading frames (ORFs) and cap-independent translation of certain linear mRNAs. Naturally occurring circRNA in mammalian cells with protein coding ability was primarily “hepatitis δ agent”; a satellite virus of the hepatitis B [9]. Early in 1995, in vitro study revealed mammalian translation apparatus could initiate translation of engineered circRNAs containing internal ribosome entry site (IRES) elements [17]. The circular mRNA containing a simple green fluorescent protein (GFP) ORF could direct GFP expression in E. coli due to the presence of prokaryotic ribosome binding sites [16]. When IRES gets inserted into a green fluorescent protein (GFP), the resulting circRNA serves as mRNA directing robust GFP protein synthesis [33]. Subsequently, it was discovered a circRNA of a virusoid (220 nt in length) associated with rice yellow mottle virus generates a 16-kD protein [62]. A circRNA database named circRNADb, containing 32,914 human exonic circRNAs was established recently [63] that provides detailed information to predict the translatability of certain circRNAs, the genomic sequences, IRE sites and ORFs. There are no experimental evidences to prove that ecircRNAs serve as mRNAs. Also no naturally occurring ecircRNA that undergoes translation (i.e., bound to polysomes) has been discovered so far [64]. Unlike linear products of genes that were significantly abundant in the ribosome bound fractions, no circular species of these genes were bound to monosome or polysomes. Ribosome footprinting data for human bone osteosarcoma epithelial cells (U20S) also does not support the natural existence of circRNAs that undergo translation (12). It has been recently reported that the most common base modification of RNA [N6-methyladenosine (m6A)] promoted efficient protein translational initiation from circRNAs in human cells. Additionally, it was discovered that consensus m6A motifs enriched in circRNAs require a single m6A site for driving translational initiation. Initiation factor eIF4G2 is required for initiation and the process is further enhanced by methyltransferase METTL3/1 and inhibited by demethylase FTO. Polysome profiling analysis with mass spectrometry and computational prediction revealed widespread translational potential of hundreds of endogenous circRNAs.  This study expanded the coding landscape of human transcriptome, and suggested a role of circRNA-derived proteins in various cellular responses (65).

CircRNAs AND CANCER:

The evolutionary conservation of circularization conceals some important functions, and it was proposed that the circRNA has important roles to play in a series of cellular processes and the initiation and development of various diseases such as cancer. Dysregulation of several circRNAs in various cancers have been found to be linked with maintenance of various cancer phenotypes.

For instance, the expression of cir-ITCH in esophageal squamous cell carcinoma (ESCC) is downregulated. Cir-ITCH facilitates ubiquitin-mediated Dvl2 degradation and decreases the expression of the oncogene c-myc thereby inhibiting canonical Wnt signaling pathway and shows antitumor activity (66). Moreover, RNA-Seq data from 12 patients with normal colon mucosa and colorectal cancer tissues, confirmed global reduction of circRNAs in CRC cases [67]. Many circRNAs were also found in serum exosomes of CRC patients and could distinguish CRC patients from healthy controls. CircRNA hsa_circ_002059 is less expressed in gastric cancer tissues in comparison to normal tissues, hence can be used as a novel biomarker for gastric cancer diagnosis. Significant difference between plasma levels of circRNA hsa_circ_002059 between postoperative and preoperative gastric cancer patients was elucidated [68]. In laryngeal squamous cell cancer (LSCC) tissues significant differential expression of circRNAs were found through microarray analysis data [69].  In vivo progression of leukemia is favoured when fusion circRNA (fcircRNA) is coupled with other oncogenic stimuli.  Fusion circRNAs generated from transcribed exons of distinct genes give rise to cancer associated chromosomal translocations [70]. ZKSCAN1 gene and its circular RNA circZKSCAN1 through different signaling pathways are known to inhibit hepatocellular carcinoma cell growth, migration, and invasion [71]. As research of circRNAs continues, they may be found to play roles in other tumors as well.

CircRNAs as miRNA sponges in Cancer

 

CircRNAs reduce the expression of miRNA-mediated gene regulation in a variety of cancers. This interaction with tumor related miRNAs indicates a great significance in tumor biology [72].  Gene ontology (GO) enrichment analysis of human disease associated miRNA data collected from 174 diseases including cancer, found that circRNAs interact with these disease-associated miRNAs associated with particular biological processes. A breakthrough study reported circRNAs to function as miRNA sponges which are known to naturally sequester and inhibit their target miRNA activity [73]. The recognition of target miRNA is mainly dependent on complementary sequences between the seed region of miRNA (2–7 nts in the mature miRNA sequence) and its target sites on circRNAs or ceRNAs. Mutations in miRNAs seed regions and target sites have a high impact on the miRNA–ceRNA interactions [74]. The dysregulation of crosstalk between miRNAs and ceRNAs significantly affect cancer pathogenesis, suggesting correlation with miRNAs as well as involvement of circRNAs in malignant tumours. For example, the ciRS-7/miR-7 axis is involved in cancer via down regulating gene expression of oncogenes like XIAP and EGFR. Tumor suppressor genes such as KLF4 are also inhibited by the same axis, which promotes the initiation and development of cancer, such as gastric cancer, cervical cancer, hepatocellular carcinoma, schwannoma tumor, tongue cancer, lung neoplasm and colorectal cancer by sequestering and inhibiting miR-7 activity [75, 76, 77]. Furthermore, miR-7 upregulates E cadherin which reduces epithelial to mesenchyme transition (EMT) thereby participating in tumorigenesis and cancer progression [78]. Dysregulation of Wnt signaling pathway is considered as a widespread theme of cancer biology. For example, cir-ITCH regulates Wnt/β-catenin signaling pathway via ubiquitination and degradation of phosphorylated Dvl2 and is therefore involved in tumor formation and chemosensitivity [79].

Additionally, E6/E7 a viral oncogene is known to overexpress miR-7 activity in HPV-positive human HeLa cells [80]. Also Cir-ITCH has been reported sponging various miRNAs such as miR-20a and miR-7 in colorectal cancer and miR-7, miR-17, and miR-214 in esophageal squamous cell carcinoma. Some circRNAs like circ-ZEB1.19, circZEB-1.17, circZEB1.5 and circZEB1.33 are implicated in the suppression of the lung cancer progression by acting as miR-200 sponge.

CircRNAs as potential biomarkers in Cancer  

 

CircRNAs are abundant and conserved over different species and has tissue and developmental stage specific properties. In essence, they are strongly expressed in blood, saliva and circulating exosomes, serving as potential biomarkers for diagnosis, prognosis and therapy for cancer patients. Upregulation of circRNA_100855 and downregulation of circRNA_104912 in laryngeal squamous cell cancer tissues (LSCC) is significantly associated with tumor stage and metastasis, which suggests an important role in the LSCC tumor genes. Similarly, downregulation of circ_002059 in gastric cancer is correlated with TNM staging, distal metastasis, sex and age that represent its potential for gastric carcinoma diagnosis [68]. Many circRNAs are known to express themselves in various human cancers including hepatocellular carcinoma and colorectal carcinoma [81, 82, 83] indicating the critical role of circRNA in physiological and pathological processes of cancer. In summary, these findings suggest significant biological roles of the circRNA in the development of various cancers and the ability to act as potential new biomarkers.

ROLE OF CIRCRNAS IN NEUROLOGICAL DISEASES

 

CircRNAs are known to be highly enriched in mammalian neuronal tissues exhibiting differential expression in different brain areas, including hippocampus, striatum, prefrontal cortex, olfactory cortex and cerebellum. A large number of circRNAs have been found to be upregulated during neuronal differentiation and play important roles in neuronal functions. [45, 55]. Sequencing studies of dissected brain tissues and differentiated neuronal cell lines revealed thousands of circRNAs enriched in synaptic fractions of neurons unlike their linear counterparts. Brain circRNAs are highly conserved between rodents and humans, and are developmentally regulated. High-resolution RNA in situ hybridization technique revealed that prevalent fraction of circRNAs from genes encode for synaptic proteins in cultured hippocampal neurons and hippocampal slices and confirmed unequivocal localization of circRNAs in both the cell body and dendrites [84]. Brain-specific genes carry more sequence features that promote circRNA formation. Furthermore, on average more circRNAs in brain are produced from the linear transcripts that correlate with the brain-biased expression of RBPs and various splicing factors for circRNA biogenesis. In addition to biogenesis, high stability of circRNAs also contributes to their accumulation in brain tissues like post-mitotic neurons.

As circRNAs are naturally present in mammalian cells, their expression is highly speculated under normal as well as diseased conditions. In diseased conditions circRNAs are primarily expressed in mid brain region and play significant roles in various nervous system disorders such as epilepsy, Parkinson’s disease (PD), Alzheimer’s disease (AD), multiple sclerosis, and schizophrenia [85] (Table 1). Silencing of miR-7 in cortical neuronal progenitors give rise to microcephaly like brain defects [86]. Likewise, it has been shown that midbrain development in humans and Zebrafish is impaired due to ectopic expression of ciRS-7 and miR-7 [87]. Also various circRNAs are derived from the genes that show genetic links with abnormal neuro developmental phenotypes, such as FBXW7, DOPEY2, and RMST [88].

One of most common and dreadful neurological disorder is Alzheimer Disease (AD) which is associated with downregulation of miRNAs like let-7i, miR-9, miR-15, miR-146b, miR-181c, miR-210, miR-338, and miR-45 [89, 90, 91]. Also, deficiency in the levels of Ubiquitin Conjugating Enzyme (UBE2A), an autophagic, phagocytic protein that is essential for the clearance of AD-amyloid peptides in Alzheimer’s disease is linked to deficits in a natural circular miRNA-7 sponge, ciRS-7. miR-7 is a well conserved, inducible miRNA abundant in the brain of the humans and murine central nervous system (CNS) that displays restricted spatiotemporal expression during development, maturity, and disease, and the manipulation of its neurobiology has considerable diagnostic, prognostic and therapeutic potential in various neurological disorders [92]. Genome-wide association studies (GWAS) have shown strong links between circRNAs with single-nucleotide polymorphisms (SNPs) and various neurological diseases. In AD, ciRS-7 targets miRNA-7 trafficking which not only promotes deficits in the expression of the UBE2A but EGFR as well [93]. Upregulation of miR-7 due to functional deficiency of CDR1as (inhibitor of ciRS-7) may lead to the downregulation of AD-relevant targets [94] suggesting that CDR1as may participate in AD pathogenesis. NF-kB-regulated miRNAs like, miRNA-9, miRNA-125b, miRNA-146b, and miR-155 are progressively upregulated in AD and results in downregulation of their target mRNAs including complement factor H (CFH), negative regulator of the inflammatory response [95]. Also, circRNA from SRY gene regulates acyl protein thioesterase 1 (APT1) and acts as a natural miRNA sponge to repress miR-138 activity which impacts memory and learning abilities [96, 97]. Alzheimer’s disease is characterized by abnormal accumulation of β-amyloid peptide [Aβ] in the brain so enhanced cleavage of β-amyloid precursor protein [APP] by β-site APP-cleaving enzyme 1 (BACE1) has an imperative implication in AD. Recently it was reported that ciRS-7 rather than affecting gene expression levels has an essential role in regulating and reducing BACE1 and APP protein levels by promoting their degradation via the lysosome and proteasome. Consequently, ciRS-7 overexpression reduces the generation of Aβ, indicating a potential neuroprotective role of ciRS-7.

In Parkinson’s disease (PD) alpha synuclein is the main protein involved which is the component of Lewy bodies in nigral neuronal cells of PD brain. Aggregation of alpha-synuclein has been reported to be directly under the control of miR-7 which induces alpha synuclein downregulation and thus can protect cells against oxidative stress [98].  MiR-7 targets the nuclear factor (NF)-κB signaling pathway and protects against 1-methyl-4-phenylpyridinium-induced cell death [99]. CiRS-7/CDR1as as an inhibitor of miR-7 [100] may point towards the involvement of ciRS-7 inhibitor in PD. Other miRNAs like miR-153, let-7, and miR-34a/b are also downregulated in PD [101, 102, 103].

The cytoplasmic aggregation of intron lariats attracts TAR DNA- binding protein (TDP43), a major component of ubiquitinated protein aggregates found in sporadic amyotrophic lateral sclerosis (ALS). Moreover, knockdown of debranching enzyme 1 activity is effective for suppressing TDP-43 toxicity in the human neuronal cell line and primary rat neurons, suggesting the possibility of using circRNAs as a potential therapeutic means for ALS [104].

Some circRNAs have specific virus miRNA binding sites, therefore, can affect the immune responses. CircRNA hsa 2149 was detected in CD19+ leukocytes but not in CD341 leukocytes, neutrophils, or HEK293 cells. CircRNA 100783 was found to be involved in chronic cell aging CD8⁺T cells, so could be a novel biomarker for inflammatory conditions. With the suppression of runt-related transcription factor 3 (RUNX3), miR-138 could balance expressions of T helper 1 (Th1) and T helper 2 (Th2) cells [105], deciphering the participation of circRNAs in inflammatory neuropathy.

CircRNAs play a vital role in development of nervous system neoplasms as well. Expression analysis of various tumor cell lines showed widespread expression of CDR1as in neuroblastomas and astrocytoma cell lines. Also, that miR-7 expression was down-regulated in astrocytoma and neuroblastoma as compared to normal brain tissue. A study on glioblastoma cell line demonstrated that miR-7 could repress EGFR expression and downregulates IRS-1 and IRS-2 expression by inhibiting the protein kinase B activity [106]. However, in future, further study on structure and functions of circRNAs will broaden our insight regarding the disease pathogenesis that would be beneficial for exploiting novel technologies for diagnostic and treatment purposes.

 

ROLES OF CIRCRNA IN CARDIOVASCULAR DISEASES

Recent studies have deciphered essential roles of circRNAs in the initiation and development of cardiovascular diseases (Table 2). MiR-223 is a well-known inducer of cardiac hypertrophy, heart failure [107] and hypertrophy in cardiomyocytes [108]. Its down-stream target is ARC (apoptosis repressor with CARD domain) [109] which has a defensive role in cardiomyocyte hypertrophy and apoptosis [110, 111, 112]. Studies have confirmed that the heart-related circRNA (HRCR) leads to ARC upregulation through inhibiting the activity of miR-223 by sponging directly. In this manner HRCR upregulation indirectly prevents the development of cardiac hypertrophy and heart failure. This potential of HRCR could be targeted for drug development for such disorders. Another example is Cdr1as. It induces Myocardial Infarction (MI) which is the leading causes of deaths and disabilities around the world. Prolonged myocardial ischemia leads to myocardial cell apoptosis [113]. As Cdr1as acts as miR-7 sponge, it plays a protective role in myocardial cells through regulating PARP [poly ADP-ribose polymerase] negatively and decreasing apoptosis [114]. SP1 and PARP are miR-7 target genes and can inhibit miR-7a-induced decrease of cell apoptosis during hypoxia (115). They also play proapoptotic roles during MI development [116, 117]. Therefore, Cdr1as can promote MI injuries by inducing the expression of miR-7a targets like PARP and SP1, which indicates the key role of Cdr1as/miR-7a axle in MI-induced myocardial apoptosis. Another important circRNA Circ-Foxo3 is derived from a member of the fork head family of transcription factors which is called Foxo3 that demonstrates a positive relation between circ-Foxo3 and senescence, thereby establishing its pro senescence function [118]. It was found that the expression of the circ-Foxo3 represses cell proliferation and cell cycle progression [119] and hinders the entry of various antisenescence transcription factors like ID1, E2F1, FAK, and HIF1into cell nucleus. These findings provide new insights into the inhibition of cardiac senescence and myocardial protection.

Several GWAS have unveiled the correlation between SNPs on chromosome 9p21 near the INK4/ARF locus and Arteriosclerotic Vascular Disease [ASVD] [120, 121] suggesting that chromosome 9p21.3 is correlated to the susceptibility of ASVD. cANRIL is an antisense transcript generated from the INK4A/ARF locus and SNPs near INK4/ARF locus regulate its transcription through modulating ANRIL splicing and cANRIL production indicating that the structure and expression of cANRIL species are formed via alternative splicing [122]. Modified cANRIL structure is hypothesized to cause changes in PcG-mediated INK4/ARF silencing and atherosclerosis susceptibility and concluded that cANRIL expression may be a useful susceptibility marker which is of pathogenic relevance to ASVD [123]. Jacobs and colleagues first unravelled the inhibitory effects of Polycomb group (PcG) complexes on INK4/ARF locus [124]. Later on it was revealed that cANRIL influences the PcG-mediated INK4/ARF silencing through recruiting PcG complexes [125, 126]. A novel circular RNA identified as circRNA_010567 has been found promoting myocardial fibrosis via sponging miR-141 and targeting TGF-b1. CircRNA_010567 was also found to be up-regulated in diabetic mice cardiac fibroblasts (CFs) and myocardium treated with angiotensin II. It is predicted to sponge miR-141, as silencing of circRNA_010567 upregulates miR-141 and downregulated TGF- β1 and reveals its direct influence on TGF-β1 expression, thereby suppressing fibrosis-associated protein resection in CFs, including Col I, α-SMA and Col III CircRNA_010567/miR-141/TGF-β1 axis providing a novel insight in cardiovascular diseases as it has an essential regulatory role in the mice model of diabetic myocardial fibrosis [127]. A study of peripheral blood circRNAs of coronary artery disease (CAD) patients and control individuals revealed that 22 circRNAs were differentially expressed. Then, after following stricter screening criteria and verification, hsa_circ_0124644 was established as a diagnostic biomarker of CAD [128].

CircRNA AND STRESS

Dysregulation of a circRNA FOXO3 (circ-FOXO3) results in sequestering proteins involved in various cellular stress pathways. Its expression was enhanced in both ROS exposed mouse embryonic fibroblasts as well as doxorubicin treated mice. Cardiac senescence induced circ-FOXO3 on upregulation via interacting with several proteins of stress response pathways including E2F1, ID-1, HIF-1a and FAK. In the same study, the association of circ-FOXO3 with above mentioned proteins was studied by enriching circ-FOXO3 using protein specific antibodies, and detection of these proteins via pull down using a biotin-labelled probe against circ-FOXO3. Its induction reduces the nuclear translocation of E2F1, ID-1, and HIF1a as well as mitochondrial localization of FAK. While as, its knockdown lead to the accumulation of these proteins in nucleus and mitochondria. Taken together, these studies suggest that one potential function of circRNAs is to reduce the subcellular localization of proteins via sequestration. Differential expression of several circRNAs was found in peripheral blood mononuclear cells (PMBCs) of patients with major depressive disorder (MDD). Hsa_circRNA_103636 downregulated in MDD patients compared to healthy controls, is significantly altered after 2 months of antidepressant regimens [129], suggesting its novel potential as a biomarker for the diagnosis and treatment of MDD. CircRNAs produced by Homeodomain-interacting protein kinase-2 and 3 (HIPK-2 and 3) enzymes are mainly activated during genotoxic stress [130] and targeting such circRNAs may be fruitful for patients with stress as it is one of the causative factors for the occurrence of neurological diseases.

While many stresses are a continuous battle to regulate homeostasis, other stresses such as stress due to microbial infection, are acute in nature and require a rapid and efficient response. E.g., LPS induces the NF-κB signaling pathway, hence promotes transcription of factors involved in the various antimicrobial response pathways. Upon LPS treatment a circRNA circ-RasGEF1B is upregulated via the NF-κB pathway and regulates the expression of intracellular adhesion molecule-1 (ICAM-1) which is involved in immune response signaling. Its knockdown decreased the ICAM-1 expression at transcriptional level. Interestingly, it was found that its knockdown resulted in a small but significant decrease in the stability of mature mRNAs but not that of pre-mRNA. Bioinformatics analysis confirmed the absence of any shared miRNA-binding site between circ-RasGEF1B and the ICAM-1 transcript, ruling out the possibility that circ-RasGEF1B acts directly via miRNA-based regulation. But still it remains unclear, how circ-RasGEF1B regulates ICAM-1 mRNA stability [131].

CONCLUSION

With the advent in high-throughput sequencing technologies, ncRNAs have come under the limelight and have become an attraction for many RNA biologists in the field of research. Circular RNAs are novel RNA molecules that form a covalent loop that is a closed continuous structure. They can bind to miRNAs and sponge their activity, regulate transcription of various genes and affect parental gene expression. Moreover, circRNAs may be involved in onset of various diseases, such as cancer, cardiovascular diseases and many more. CircRNA formation depends upon canonical splicing mechanism, sequence-intrinsic properties such as complementary repeats, length of intron flanking, circularized exon and some proteins including ADAR, hnRNPs, MBL, QKI, and SR. Understanding the biogenesis of circRNAs shall pave the way for introduction of circRNAs in vivo. As far as functions of the circRNAs are concerned, a few of them have been revealed. The binding sites in circRNAs are crucial for their activity, functioning of various target proteins and miRNAs can be modulated by modifying their binding sites. Clearly this therapy is wholly and solely based on the fact that miRNAs that are overexpressed in various diseases act as wrongdoers and circRNAs can sponge their activity thereby stopping their deleterious effects. Taken together due to their higher stability, circRNAs affect many life processes, have characteristics to serve as predictive or diagnostic biomarkers of numerous life-threatening diseases and have potential to serve as new therapeutic targets. Nevertheless, as compared to coding RNA, lncRNA and miRNA, there are still gaps in our understanding of circRNA biology. This precious world of circRNAs has bestowed upon us with new insights into the “dark matter” of the genome. Their molecular mechanisms involved in development of various diseases are yet to be fully elucidated. But with the advent of research and technology, future studies would decipher the functions of circRNAs in terms of pathological and physiological processes. Specific, sensitive and reliable biomarkers would help in the early disease diagnosis, high risk population identification, assess response and finally to develop targeted therapies. MicroRNAs have been proved to be efficient disease markers exploiting them for replacement therapy using miRNA mimics, which is currently in early clinical trials [132]. In the future circRNA mimics may be used in the same way to treat various diseases. Though research in circRNAs is still in its infancy, it may provide the next generation of personalized medicine, due to their tissue specificity, stability and suitability in “liquid biopsies.”