Research Focus

Introduction Circular RNA, a type of regulatory RNAs, recently caught the attention of scientists. In general, RNA transcribed inside the nucleus requires to undergo splicing which introns will be removed and exons will be linked in a linear order. Under certain circumstances which are not fully understood, the splice donor of downstream exon will join the splice acceptor of upstream exon and form a circular configuration (Figure 1A, left).


Figure 1 The biogenesis of circular RNAs (A) The lengths of the introns spanning the exons of circular RNA are generally longer. These intron can promote backsplicing through complementary sequence pairing or the protein-mediated intron interaction. (B) The side product of exon skipping can produce circular RNA via secondary splicing. (images adapted from (4))

In spite of the fact that circular RNA is commonly found in virus, whether it can produce functional molecules in multicellular organisms bothered the scientific community for a long time. In 1990s, scientists had discovered circRNAs in both human and murine cells, but without knowing the functions of these molecules, circRNAs are thought as the side products or the results of mis-splicing. Due to the development of next-generation sequencing and the advance of computational power in the last decade, scientists now can have a global view and re-examine the functions of circular RNA. In 2012 – the year for re-discovering circRNA, three independent teams published and provide the evidence that circRNA are commonly expressed in diverse tissues in higher eukaryotic cells (1-3).

The molecular mechanism of circRNA biogenesis Most of the circular RNAs are generated from protein-coding genes through canonical splicing without U2 or U12 preference. Currently, there are two models proposed for the circRNA biogenesis. In the first model, the formation of circRNA is facilitated by the interaction between the spanning introns. The results of the genome-wide analysis showed that the flanking introns for the circRNAs are longer, and thus favor the interactions between the spanning introns which can be enhanced through either the complementary base pairing or RNA-binding proteins-assisted association. In the second model, the circRNAs come from the secondary splicing in the lariat molecule (intra-lariat splicing) (Figure 1B).

Molecular characteristics and functions of Circular RNAs Circular RNAs are distinct to the lariat molecules, although both have enclosed configuration. While lariat molecules are enclosed by a 2’-5’ phosphodiester bond, circular RNAs are by 3’-5’ phosphodiester bond. As in a circular configuration, circRNAs are lack of 5’-cap and 3’-polyA tail, and thought to be more stable in cells. The circular RNAs harbor diverse molecular functions such as a molecular sponge, transcription regulator, splicing modulator, scaffold, and temples for encoding proteins. Sponging miRNAs is one of the most studied functions. Thus circRNAs can regulate mRNA stability through binding with miRNAs. The circRNAs sponging miRNA typically reside in the cytoplasm, and associate with ‘RNA-induced silencing complex’ (RISC). This subcellular localization is consistent with its function regulating mRNA stability. Taking ciRS-7 as an example, ciRS-7 has more than 70 binding sites of miR-7, and is capable to regulate the levels of miR-7-targeting mRNAs. Similar to the concept of miRNA sponge, the circRNA can work as a molecular sponge for RNA-binding proteins. In addition to the function as a molecular sponge, a subclass of circRNAs reside in nuclei due to retained introns or unidentified mechanisms. These nuclear circRNAs can interact with gene promoters and regulate transcriptional activation.

The backsplicing of circular RNAs shares the same splice signals with its parental transcript and uses the same spliceosome. Therefore, circRNA is considered as a potential competitor for the linear splicing of the parental gene. Nevertheless, circRNAs promote the interaction between p21 and CDK, serving as a molecular scaffold (4). In contrast to promoting protein interaction, it has been reported that cricRNA also can suppress protein interaction. Furthermore, few examples of circRNAs can be translated. The actual biological functions of these molecules remains elusive and are worth for further investigation.


Figure 2 The characterization of circRNAs in colorectal cancer (A) Distribution of exonic circRNAs based on their location in gene body. (B) RNA from HCT116 and HT-29 were treated without or with RNase R (-/+) prior to reverse transcription.  –RT: reverse transcription without reverse transcriptase. NC: negative control for PCR.(C) The distribution of circCCDC66 in subcelluar compartments. RNU6B: as a marker for nuclear fraction; rRNA: 18S rRNA as cytoplasmic marker. (D) The levels of circCCDC66 in clinical speciemsns (N: nontumor; P: polys; T: tumor) (E) Kaplan-Meier survival curves for patients with low and high expression levels of circCCDC66. (Part of text and images were adapted from (5))

Applications of circRNAs in cancer biology In the early stage of circRNA study, although it was reported that the tumor tissues express more diverse circRNAs, there were not much information known in human cancers. In addition, the correlation analysis indicated that tumor tissues express lower levels of circRNA without knowing the exact functional circRNAs in cancer. Thus, we decided to investigate whether circRNAs play oncogenic roles during the development of cancer (5). To address this question, we performed RNA-seq on paired tumor and adjacent non-tumor colorectal tissues. From the analyses, we identified 74 circRNA candidates. The majority of circRNA candidates come from the coding region of genes, and a small fraction comes from 5'- or 3'-untranslated regions (Figure 2A). To further validate the results of bioinformatic analyses, reverse transcription-quantitative PCR was used to check if these circRNA candidates were expressed in the human tissues. The results showed that circRNAs from CCDC66, CCNB1 and CDK13 were all expressed in the colorectal cancer cell-lines, and were resistant to the treatment with RNase R, an exoribonuclease. Thus, circRNAs which has no 5'- or 3'-end wouldn't be a substrate of RNase R. To further dissect the functions of circRNAs, we found most of the circRNAs reside in the cytoplasm (Figure 2C), implying they may play roles as miRNA sponge. After validating the presence of circRNAs, we characterized the expression levels of circRNAs in nontumor, polyps and tumor tissues, and found that the level of circCCDC66 was readily upregulated in pre-tumorous polyp tissues and even higher in the tumor tissues (Figure 2D). Nevertheless, if we classified the patients according to the levels of circCCDC66, the group with higher levels has worse prognosis (lower survival rate, Figure 2E). Taken together, these results implied the oncogenic roles of circCCDC66 in colorectal cancer.

To further investigate how circCCDC66 promotes the development of colorectal cancer, small interfering RNA targeting backsplice junction was used to knock down the level of circCCDC66. The results showed that the colorectal cancer cell-lines with circCCDC66 knockdown had lower capacities of invasion and migration. In contrast, the colorectal cancer cells with circCCDC66 overexpression had better ability in proliferation, suggesting that circCCDC66 plays an oncogenic role during the progression of colorectal cancer. Next, we want to know what the molecular mechanisms underlying circCCDC66-induced oncogenesis. Our bioinformatic analyses indicated that circCCDC66 harbors 102 binding sites for various miRNAs unlike ciRS-7 which is more specific to single miRNA. These binding sites belong to 101 miRNAs. To validate whether circCCDC66 exerts its oncogenic effect through these miRNA binding sites, we identified the target genes of these miRNAs, and found that relatively more upregulated oncogenes were the targets of circCCDC66 (37.5%), while only 20% of tumor suppressor was the targets of circCCDC66, suggesting that circCCDC66 promotes cancer progression through preferential protection on oncogenes. To further evaluate this hypothesis, we assayed the levels of circCCDC66-regulated oncogenes in colorectal cancer cell-line with circCCDC66 overexpression, and found that circCCDC66-regulated oncogenes such as DNMT3B, EZH2, MYC and YAP1 were upregulated by circCCDC66 overexpression (Figure 3A). To further characterize the upregulation of these oncogenes were mediated by the antagonization of miRNAs, we cloned 3'-untranslated region (UTR) to the downstream of a renilla luciferase (Figure 3B), and discovered that the downregulated MYC 3'-UTR reporter by circCCDC66 knockdown can be rescued by the miRNA inhibitors (Figure 3C).


Figure 3 CircCCDC66 promotes the progression of colorectal cancer through preferential protection of oncogenes (A) Results of RT-qPCR for circCCDC66-regulated oncogenes, DNMT3B, EZH2, MYC, YAP1 and non-circCCDC66 regulated genes. (B) The illustration shows the construct of MYC 3’-UTR reporter and the alignment between MYC 3’-UTR and showed miRNAs. (C) MYC 3’-UTR reporter activity from HCT116 with/without circCCDC66 overexpression. (D) Growth curves of xenografted tumors derived from HCT116 with or without circCCDC66 knockdown. (E) The developed tumors from HCT116 transfected with control oligonucleotides (siCON) or oligonucleotides against circCCDC66 (siJCT) in the mice cecum. The gross outlines of tumors were indicated in yellow (upper panel). The representative images for haematoxylin and eosin staining for tumors derived from HCT116 transfected with control oligonucleotides (siCON) or oligonucleotides against circCCDC66 (siJCT) (bottom panel). T: tumor. Yellow arrowheads indicate tumor masses in the layers of smooth muscle; yellow dashed lines: the intact boundary between tumor mass and layers of smooth muscle. (F) Gross view of tumor nodules established in the liver. Yellow arrowheads indicate tumor nodules in the liver. (Part of text and images were adapted from (5))

Finally, the results of the xenografted model in mice showed that the growth of tumors-derived from HCT116 with circCCDC66 was hindered compared to the cells treated with control siRNA oligonucleotides (Figure 3D). To more closely mimic the growth of tumors in vivo, HCT116 cells were orthotopically injected into the cecum of mice. As the results showed, the local invasion of the tumor into the smooth muscle layer was suppressed by the circCCDC66 knockdown while the cells transfected with control siRNA oligonucleotides readily disrupted the boundary between the epithelial layer and smooth muscle layer. Similar to colorectal cancer in human, the orthotopic model of colorectal cancer in the cecum of mice has a tendency to metastasize to the liver. The results showed that tumors derived from cells with circCCDC66 knockdown established fewer tumor nodules in the liver, suggesting that the expression of circCCDC66 is critical for the development of tumors at remote sites.  Taken together, our data helped to identify the first oncogenic circRNA with multiple lines of supporting evidence.

Colorectal cancer currently is the cancer with the highest prevalence rate and ranked third in the cause of cancer-related death. There were more than 10 thousands of newly diagnosed cases. It is critical for the patients to be diagnosed and treated at early stages. Although it has been widely promoted that people should do a colonoscopy and fecal occult blood test on a regular basis, the development of molecular diagnosis using circCCDC66 may be a more feasible and convenient tool. On the side of therapy, the treatments of colorectal cancer are based on the stages including surgical resection, irradiation, chemotherapy and/or combinations of these treatments. However, once the tumors develop resistance and relapse, finally leading to the death of patients. Thus it is important to investigate new pathological pathways and identify novel targets for developing the new strategy to cope with colorectal cancer. Our preliminary study has demonstrated that aberrantly expressed circRNA is a new pathological pathway and paved the path to the development of new therapeutic strategy.


1. Hansen TB, et al. Natural RNA circles function as efficient microRNA sponges. Nature 2013; 495:384-388

2. Wu Q, et al. Homology-independent discovery of replicating pathogenic circular RNAs by deep sequencing and a new computational algorithm. Proc Natl Acad Sci U S A 2012; 109:3938-3943

3. Salzman J, et al. Circular RNAs are the predominant transcript isoform from hundreds of human genes in diverse cell types. PLoS One 2012; 7:e30733

4. Hsiao KY, et al. Circular RNA - New member of noncoding RNA with novel functions. Exp Biol Med (Maywood) 2017; 242:1136-1141

5. Hsiao KY, et al. Noncoding Effects of Circular RNA CCDC66 Promote Colon Cancer Growth and Metastasis. Cancer Res 2017; 77:2339-2350

(The English version of introduction was drafted by Doris Tai during her summer internship in 2020)