Circles reshaping the RNA world: from waste to treasure

CircRNAs are a group of non-coding RNAs that are ancient yet newly recognized regulating molecules. After the earliest discover in plant-based virus in 1976 [9], they just come into scene in recent years and are found widely spread in eukaryotic cells from flies to mammals and humans thanks to the technological breakthroughs in high-throughput deep sequencing [10‐12]. But still, circRNA was regarded as the functionless junk-RNA developed from mRNA splicing thus did not get enough attention as they should be. Until 2013, a famous and classic paper published by Hansen TB et al. on the journal of nature has totally changed the fate of circRNA and turn it into a focal point of scientific research [3]. The subsequent findings indicate that circRNAs have great potential to perform special regulating roles in biological development. Thousands of circRNAs were also suggested to present in plant cells and might act as a negative regulator of its parental gene through genome-wide identification and potential function analysis of circRNAs in rice [13]. Similar with lncRNAs, the expression profile of circRNAs is specific among different cell types indicating its possible regulatory function. More and more studies have suggested that they contribute disease initiation and progression, possess the potential of turning into new clinical diagnostic and prognostic markers, and serve new insights into the therapy of cancers.

Although more and more human circRNAs were identified and uploaded online using molecular biology strategies combined with new bioinformatic approaches, the underlying mechanisms of circRNA generation are still not clear yet. There is an assumption that at least four distinct paths involved in the formation of circRNAs [2, 10, 14, 15].

Due to the confusion of nomenclatural method in different databases and the poor understanding in biogenesis and characteristics of circRNAs, generating a systematic circRNAs classification is not an easy task. Up to now, there are two separate ways to divide circRNAs based on different principles.

Large-scale RNAseq has been widely employed for many purposes in scientific research and substantially changed the transcriptomics field, especially the circRNAs, as it enables an unprecedented amount of high resolution data to be acquired. Except for the established softwares which are powerful enough to analyze the sequence data, several current online databases containing circRNAs collected from GenBank annotations or published articles enable researchers to organize the circRNAs conveniently.

As a “bridge” between DNA and protein, the complex regulatory role of RNA has long been underestimated. In eukaryotic cells, protein-coding RNA (mRNA) only occupies about 2% of the genome, the rest massive number of transcripts were classified to non-protein coding RNAs (ncRNAs). CircRNA was found stable and abundant in cells across species, involving in complex regulatory networks of ncRNAs and playing a indispensable role in the development of non-neoplastic diseases and tumors [43, 44].

Just several years ago, circRNA was still considered the waste of error splice, while up to now, they were proved gradually to possess better stability, escape the degradation of RNAase and widely spread in eukaryotic cells [1, 49, 50]. These deeper understanding of circRNAs’ characteristics paves the way for the reveal of their four crucial functions (Fig. 4).

The enormous abundance of circRNAs in brain compared to other normal or cancerous human tissues announces their preferred derivation from neural genes and intensive accumulation in neural tissues [68]. But this dose not cover the whole story, circRNAS was revealed increasing substantially relative to their linear isoforms during CNS aging. This age-related accumulation of circular RNA might have implications for its potential of working as a promising aging marker [37]. Besides the apparent aging indicators like skin winkles and gray hair, circRNA might be able to tip the aging of brain and internal organs in advance, alarming human to take some preventive measures before it is too late. In addition, many brain circRNAs changed their abundance abruptly at a time corresponding to synaptogenesis, they also exhibit a substantial up- or downregulation when the homeostasis of neuronal activity changes [19]. Brain circRNAs are suggested to regulate synaptic function and go hand in hand with the development of nervous system, this important discovery adds further dimension to our understanding of the molecular pervasiveness and importance of circular RNAs.

CiRS-7 was firstly found in large amounts in human brain tissue and was also termed CDR1 (cerebellum degeneration-related antigen 1, CDR1,also known as CDR34) because of its transcription in the antisense orientation with respect to the CDR1 gene, which has been implicated to be crucial in autoimmune neurologic disorders like Alzheimer’s disease [69‐71]. CircRNA CANRIL was identified to be relevant to the risk of human atherosclerosis [2]. Also, the over-expressed HRCR (heart-related circRNA) attenuates the development of cardiac hypertrophy and heart failure in mice by directly binding to miR-223 and acting as an endogenous super sponge to inhibit miR-223 activity, which results in the increasing expression of ARC, a target for miR-223, in heart cells [6].

Some circRNAs are even able to possess the function of regulating cell cycle progression. Circ-Foxo3 is a highly expressed circular RNA in non-cancer cells, silencing endogenous circ-Foxo3 accelerates cell proliferation while ectopic expression of circFoxo3 hampers cell cycle progression [72]. Circ-Foxo3 was shown to bind with the cell cycle proteins CDK2 and p21 to form a ternary complex. Normally, CDK2 (cyclin-dependent kinase 2) is a well-known cell cycle facilitator interacting with cyclinA and cyclinE [73], while p21 (cyclin-dependent kinase inhibitor 1) restrains these activities and arrest cell cycle progression [74]. The formation of circ-Foxo3-p21-CDK2 ternary complex blocks cell cycle progression by arresting the function of CDK2 [72]. May be the dysregulation of circ-Foxo3 is relevant to the neuromuscular degenerative diseases or even the germination of cancer.

Some researchers provide evidences that circ_001569 acts mechanically as a miRNA sponge to inhibit miR-145 activity, and subsequently up-regulates miR-145 targets E2F5, BAG4 and FMNL2 to promote cell proliferation and invasion in CRC (colorectal cancer) [7]. Cir-ITCH was also shown to inhibit the development and progression of CRC and ESCC (esophageal squamous cell carcinoma) [75, 76]. Nowadays, a larger number of circRNAs was found to significantly express differentially between ovarian epithelial carcinoma and normal adjacent tissue than mRNA. Furthermore, the expression profile of circRNA in primary ovarian epithelial carcinoma and metastatic carcinoma is also different [8]. Hsa_circ_002059 has been proved to be down-expressed in gastric tumor samples compared to the normal gastric mucosa and may be used as a potential biomarker in the early diagnosis of gastric cancer [77]. The hsa_circRNA_104912 level was significantly lower in LSCC (laryngeal squamous cell carcinoma) than in corresponding adjacent non-neoplastic tissues, playing an important role in the tumorigenesis of LSCC [78]. We can assume that perhaps the different expression of circRNA is associated with tumor metastasis and invasion. In brief, circRNAs are gradually found to play a potential and necessary role in the biogenesis of all kinds of tumors. Further, circRNAs present a more robust expression pattern across patients than their linear counterparts indicating their suitability as biomarkers in highly heterogeneous cancer transcriptomes for diagnostic and therapeutic implications.

Chromosomal translocations give rise to oncogenic fusion gene thus encode fusion proteins which have been shown to contribute to tumorigenesis [79, 80]. PML/RARa and BCR-ABL are the most famous and recurrent fusion gene in patients with APL (acute promyelocytic leukemia) and CML (chronic granulocytic leukemia) respectively [81]. These two well-established fusion gene has become the effective therapeutic targets and reliable prognosis assessment indicators. Then whether such genomic alterations also affect noncoding RNAs, particularly circRNAs? Jlenia Guarnerio found that cancer-associated chromosomal translocations did give rise to f-circRNA (fusion circRNAs), these newly recognized cirRNAs can be generated by transcribed exons of distinct genes in translocation process [82]. Patients carrying the PML/RARa or MLL/AF9 translocation (confirmed by the use of convergent primers spanning the break point) was confirmed to display the expression of several f-circRNAs composed by different exons. Furthermore, the formation of f-circRNAs is not an exclusive feature of leukemic cells, it is also universal in tumors of different histological origins. The EWSR1/FLI1 [83] and EML4/ALK1 [84] translocation common in solid tumors such as Ewing Sarcoma and lung cancer, could also generate associated f-circRNAs. F-circRNAs promote cellular transformation, contribute cell survival upon therapy and confer resistance to treatment in tumor cells. But f-circM9 (fusion circRNA MLL/AF9) was proved probably not sufficient to trigger tumorigenesis on its own, it plays an active role in favoring leukemia progression in vivo when coupled with other oncogenic stimuli (e.g., the presence of the oncogenic fusion protein) [82]. This paper expands our current knowledge concerning molecular mechanisms of cancer onset and progression, indicating potential diagnostic and therapeutic implications.

There is research shows that circular and linear expression exhibits an inverse trend in many cancer related pathways. Signaling pathways like NF-kB, ILK and TGF-β, which are essential for epithelial to mesenchymal transition and typically over expressed for mRNA in metastatic tumors, are suggested down regulated for circRNA [8]. Similarly, proliferative signaling pathways (PI3k/AKT, JAK/STAT), angiogenesis pathways (VEGF, ILK) and HGF signaling were activated for linear mRNA transcripts and down regulated for circular RNA in metastases. Conversely, PTEN and PPAR signaling pathways, whose inactivation leads to carcinogenesis, were observed a downregulation in mRNA transcripts yet a corresponding upregulation of circRNA in metastases [8]. As we know, ITCH is an anti-cancer protein targeting p63, p73, and Notch1 gene and usually involve in tumor formation and chemosensitivity [85]. Cir-ITCH was shown to play an inhibitory role in the development and progression of colorectal cancer and esophageal squamous cell carcinoma and it may interact with the miRNAs (miR-7, miR-17, miR-214) thus increasing the level of ITCH, which works as inhibitor of Wnt /β-catenin signaling pathway, exerting an anti-neoplastic effect [75, 76].

Majority of current studies all lend a strong support for the notion that circRNAs play fundamental roles in various disease, especially tumors. However, the function of large amounts of circRNAs which have been detected inside animal cells still remains largely unknown and awaits further detailed study. Thankfully, the results of existing researches have several interesting implications concerning circRNAs biology and toward disclosing the authentic involvement of circRNAs in various disease and cellular process.

As already stated in our previous section, the expression levels of circRNA was observed an increase after ADAR knockdown compared with the host mRNA. This finding is actually very interesting. You can imagine that if the level of ADAR decreases transiently in physiological state, it is perhaps the perfect time when circRNAs start to play a role. Prior studies have shown that stem cells can differentiate into neurons if ADAR gene expression in embryonic stem cells is reduced. Then, whether circRNA is involved in stem cell differentiation? What role dose circRNA plays in abnormal differentiation of hematopoietic stem cell associated hematological malignancy? Whether the high stability of circRNA has a relationship with the memory retention of human brain and the memory-mediated killing effect of T-lymphocytes? Maybe circRNA can help us fully understand and effectively treat the brain degenerative diseases and haematological malignancies.

We all know that circRNAs highly exist in permanent cells (neuron and cardiomyocyte) and accumulate as time goes on, then how to explain this phenomenon? Is the process of aging gives enough time for circRNAs to gather or the accumulation effect of circRNA promotes the occurrence of aging? What about the target miRNAs? Do they still bind with circRNAs sponge and increase synchronously during the aging process? What will happen to miRNAs after their release in a inactive state from the cleaved circRNAs? What contributes do the interaction between circRNA, miRNA and their parental gene serves in the process of aging and functional organ degeneration? These questions are all worthy of further study. Maybe 1 day human can unveil the mystery of aging by researching these promising RNAs.

As a footnote to distinguish varied circRNAs, the naming of circRNAs today is still based on the RNA sequencing tools, which differs significantly from each other. This chaos existing in naming method causes the difficulty of circRNA classification. A normative and widely accepted naming system is necessary. With the gradual understanding of circRNA’s biogenesis and functions, there is a hope that majority of the circRNAs can be named according to the well-studied miRNA they bind or the star gene they regulate. The most famous example is ciRS-7 and CDR1as. Considering the many-to-one relationship between miRNA, cirRNA and their mother-gene, an internal numbering system is required. With this complete naming system, we can assume a circRNA’s source and potential function after the first sight of its name, isn’t great?