Circular RNAs in human cancer

CircRNAs can be sorted into three types, including exonic circRNAs, intronic circRNAs, and retained-intron circRNAs, which are produced from different cycling mechanisms. During the formation of circRNAs, specific sequences are conserved and sequences from upstream and downstream regions are linked together in a reverse direction, all of which contribute to mature circRNAs by back splicing [10]. Although the efficiency of back splicing is much lower than in linear RNAs [5], there is still a large intracellular abundance of circRNAs due to their stability and relatively long half-life period [11]. The earliest back splicing mechanism is dependent on complementary intron matches [12], which widely exist in the exonic circRNA formation process. CircRNAs structured from this mechanism have special repeated sequences, such as Alu and other short sequences, on their two sides [13, 14]. After transcription into RNA precursors, the exon sequences on the two sides of the cycling region will complementarily match with each other under the promotion of repetitive sequences, such as Alu, to form a circle shape. Following this, the spliceosome binds to the cycled RNA precursor under the effect of U6 and U2. Then, it selectively cuts out the exon in the cycling region by cooperating with the protein complex and the U5 core [15, 16]. At the same time, exon sequences connection are reversely cut, forming mature circRNA.

The mechanism of intronic circRNA formation is different from that of exonic circRNA. The formation of intronic circRNAs relies on GU-rich sequences near the 5′ splice site and C-rich sequences near the branch point. During the back splicing process, the two segments bind into a circle first. Then exonic sequences and intronic sequences in the binding part are cut out by the spliceosome. The remaining introns are eventually pieced together to form mature circRNA [17, 18]. Recently, Li et al. found another novel type of circRNA consisting of both exons and introns that overexpress themselves according to their complementary sequences [7]; however, the exact splicing mechanism has yet to be determined (Fig. 1a).

Similar to linear RNA, the biogenesis of circRNAs is regulated by various types of proteins, such as the RNA spliceosome and RNA binding protein (RBP). The RNA spliceosome and spicing process are vital research frontiers in biomedicine. The RNA spliceosome is a RNA-based enzyme with a U5 core consisting of 5 snRNAs and a number of proteins. Under the effect the U6 and U2 promoters, many proteins cooperatively process pre-RNAs [16]. Similar to mRNA formation, circRNAs can be spliced from linear RNA precursors by the spliceosome. Furthermore, one linear RNA molecule can be processed into different types of RNAs, including mRNA, circRNA, and lncRNA, through different splicing events. For example, HIPK3 pre-mRNA can be spliced into HIPK3 mRNA and its second exon can also form circRNA hsa_circ_0018082 [13]. There is also a competition between inverted splicing of circRNA, mRNA, and lncRNA during splicing [19], which may be a regulation mechanism of circRNAs. Another regulated protein is the Quaking (QKI) protein, a representative molecule. It can bind to specific downstream sites on the linear pre mRNA sequence and promote cycling and inverted splicing of circRNAs [20] (Fig. 1b).

With the application of next-generation sequencing and bioinformatics, many circRNAs are being discovered and integrated into the circRNA database [9] (Table 1). The Cirbase and Deepbase 2.0 databases contain over 150,000 circRNA genomes and mature RNA sequences of different species, including human, fruit fly, and nematode [21‐23]. The Cir2Traits database contains circRNAs relating to various diseases [24]. Circnet and circRNAbase contain alternatively spliced variants and their expression patterns in different tissues and possible miRNA regulated networks [21, 25]. The CircInteractome database contains corresponding genome sequences, primers for qRT-PCR, and siRNAs of mature circRNAs. It can also be used to predict RBP binding sites [26]. The CIRCpedia and TSCD databases contain RNA-seq data of several human cell lines and samples [27, 28].

Unlike linear RNA, 3′ and 5′ terminals are not exposed in circRNAs; thus, they are not sensitive to ribonucleases, such as exonuclease or RNase R and have a longer half-life period with a more prominent specificity in histology and pathology [29]. Meanwhile, circRNAs are easier to extract and detect with higher specificity compared with proteins; thus, circRNAs are an ideal molecular marker for cancer. With the application of RNA-seq [12, 30], changes in the expression of circRNAs are found to be prevalent in many cancers, including esophageal cancer [31], glioma [32], and hepatocellular carcinoma [33]. Further experiments and analysis on clinical files confirm that the expression levels of some circRNAs vary according to TNM stage, metastasis, and size the of tumor in esophageal cancer [34], hepatocellular carcinoma [35, 36], and colorectal cancer [37]. Free circRNA molecules are degraded in the extracellular environment, but studies show that circRNAs that are produced intracellularly can be secreted to the exterior of an exosome and be restored after an extended period of time. Li et al. found high expression of circ_0001649 in patients with liver cancer through RNA sequencing and qRT-PCR, and the expression level in the exosome is closely related to clinical characteristics, such as tumor size [33] (Table 2).

Because of their unique structure, circRNAs are more stable in tissues, serum, savory, or urine compared with circulating tumor DNAs (ctDNAs) or linear RNAs. There is also a greater possibility to screen for sensitive and specific molecular biomarkers from thousands of circRNAs with regard to the relatively small number of miRNAs. Testing circRNAs by RT-PCR and in-situ hybridization is also more specific and sensitive than detecting proteins by an antigen-antibody reaction. In conclusion, the use of circRNAs outweighs the use of current tumor biomarkers in many fields, and they could be promising molecular biomarkers and new targets for future drugs.

CircRNAs are widely involved in physiological and pathological processes, and studies have found that circRNAs can be involved in biological activities by acting as a miRNA sponge [6], binding RBPs [7], and translating peptides [8, 38]. Of these, the role of circRNAs as a miRNA sponge is the main mechanism of circRNAs in tumor cells. There are many miRNA binding sites on circRNAs. Thus, circRNAs can bind to miRNA as a RNA sponge and can regulate miRNA and downstream gene expression through a ceRNA mechanism and contribute to tumor progression. The following examples present current studies on circRNAs and their functions in human cancer.

CircRNAs are novel molecules that have been discovered in recent years. They are no longer recognized as products of transcription errors. On the contrary, they have become a popular research subject, after miRNAs and lncRNAs, owing to their important functions [52‐56]. The formation of a mature circRNA molecule has at least two mechanisms, which are different from selective splicing in linear RNAs; however, the detailed molecular mechanism requires further studies. There may be a third mechanism involving the formation of retained-intron circRNAs that contain both exons and introns, and this has also yet to be explored. Moreover, different types of proteins and miRNAs have been found to take part in the processing of pre-RNAs to circRNAs. The mechanism of circRNAs formation has yet to be discovered, and it remains an attractive mystery.

A tremendous amount of data indicate that circRNAs regulate other genes at the transcriptional and post-transcriptional levels. Some can even encode proteins directly and are thus involved in physiological and pathological processes, including influencing malignant tumors. Until now, the miRNA sponge has been found to be a possible mechanism for circRNA function. They can specifically bind to miRNAs and regulate gene expression by competing with competing endogenous RNA (ceRNA), including lncRNAs, mRNAs, and pseudogenes. The participation of circRNAs in the ceRNA network renders it more complete and complex. CircRNAs compete with mRNAs from the same preRNAs during the splicing process, and circRNAs were recently found to encode for proteins. This demonstrates their involvement in biological procedures, provides a more complete understanding of the RNA language, and enriches the central dogma [57‐59]. Knowledge of circRNAs still remains at a superficial level, and more functions remain to be discovered among over 30,000 molecules, as only a limited number of reports describe their functions. Exploration of their functions will bring humans closer to the essence of life.

CircRNAs are much more stable compared to linear RNAs owing to their unique structure. With regard to biological markers of cancer, circRNAs have demonstrated their wide perspective. Studies have already found differential expression of circRNAs in tissues and blood, and their expression differs from those of miRNA and lncRNA [60‐62]. It is possible to build a model for prognosis by combining them with various other biomarkers and thus improve the accuracy and specificity of diagnosis. The clinical applications of circRNAs continues to expand, with their use in samples, such as saliva and urine, and research on circRNAs is indeed a very promising aspect of the study of life sciences.

Previous studies have reported that circRNAs are one type of molecular marker in tumors and that they can influence apoptosis and infiltration. CircRNAs are potential targets for new drugs. For example, the fusion-circRNA found in leukemia is associated with chemotherapy resistance, and knocking out expression of f-circM9 can lead to apoptosis of leukemia cells and decrease its drug resistance to Ara-C [48]. On the other hand, circRNAs perform their functions by mediating miRNAs through a miRNA sponge. This process can also act as a treatment target for malignant tumors. It was also revealed that circRNAs can encode multiple peptides, which can also be used to treat tumors. However, more progress needs to be made between circ-RNA and tumor treatment since the process to effectively and efficiently transform circRNAs into targeted cells is still a major problem that needs to be solved. Although nanoparticles have been applied as transforming vehicles in gene therapy and the Synergistic Photodynamic/Photothermal/Chemotherapy of Cancer drug releasing system has also made some advancement, neither have been utilized in clinical practice. Further studies are still needed to reach a definite resolution.