1. circRNAs as microRNA Sponges
MicroRNAs are small, evolutionarily conserved n
on-c
RNAoding RNA (ncRNA) molecules whose genes are generously distributed across the human genome; approximately 2600 microRNAs have been identified
[1]. They participate in post-transcriptional gene regulation, affecting such crucial cellular processes as proliferation, migration, metabolism, and differentiation. The microRNA genes are transcribed into primary microRNAs by RNA polymerase II and then processed by the DROSHA complex into pre-microRNAs exported to the cytoplasm and processed by DICER to produce mature microRNAs. They interact in the cytoplasm with the effector RNA-induced silencing complex (RISC), which binds their target messenger RNAs (mRNAs) with a sequence complementary to the microRNA usually embedded within ‘mRNA’s 3′-UTR. They thus downregulate gene expression by either prompting mRNA degradation or blocking its translation at the ribosome
[2].
Numerous circular RNAs (circRNAs) containing multiple complementary target sequences act as a sponge for specific microRNAs, thus effectively modulating the activity of entire signaling pathways. Unsurprisingly, many can modulate cellular protein functions by forming complex network interactions.
For example, circ
EPB41 (
hsa_circ_0000042) promotes the stemness of non-small cell lung cancer cells in vivo and in vitro. Thorough analysis showed that
miR-486-3p is a direct target of circ
EPB41, and its expression was downregulated in tumorous tissues of non-small cell lung cancer patients. High-throughput sequencing and bioinformatics analysis showed that
miR-486-3p targets the 3′UTR of eIF5A—a translation initiation factor that regulates tumor stem cell differentiation. Moreover, the quelling of circ
EPB41 suppressed self-renewal capacity and decreased the expression of stemness markers. Conversely, downregulation of
miR-486-3p or overexpression of eIF5A restores cell proliferation and invasiveness after circ
EPB41 silencing
[3].
Evidence suggests that microRNA-sponging circRNAs are also involved in regulating breast cancer CSCs’ function and invasion capacity. Liu et al. identified circ
NOLC1 as a competing endogenous RNA for
miR-365a-3p, thus enhancing the expression of STAT3. Downregulation of circ
NOLC1 expression inhibited the ability of pleural effusion-derived breast cancer cells to form mammospheres and led to the suppression of stemness-related genes. Notably, the overexpression of STAT3 rescues circ
NOLC1 depletion-attenuated proliferation and CSC activity in breast cancer. The authors also indicated that the overexpression of circ
NOLC1 rescues propofol-attenuated proliferation of breast cancer CSCs
[4].
Epstein-Barr virus (EBV)-associated gastric cancer is a distinct subtype of this malignancy with unique molecular characteristics
[5], yet, until recently, it was unclear whether CSCs exist and play any role in its physiopathology. By long-term treatment of the EBV-associated gastric cell line with a chemotherapeutic agent in mice, a highly aggressive cell line was obtained and consisted mainly of stem-like CD44+/CD24-subpopulation. These cells expressed an EBV-encoded circRNA, ebv-circ
LMP2A, that induces and maintains stemness phenotypes. To this effect, ebv-circ
LMP2A sponges
miR-3908 and enhances the TRIM59/p53 pathway, and its high expression is significantly associated with metastasis and a poor prognosis in patients with this type of cancer. These findings have thus provided evidence for the existence of CSCs in EBV-associated gastric cancer, shedding light on the pathogenic mechanism of ebv-circ
LMP2A [6].
Thus, circRNA-mediated microRNA sponging emerges as an emblematic competing endogenous RNA mechanism that enables forces significant enough to shape cancer cell phenotypes.
2. circRNAs as Components of Protein Complexes
The patterns of circRNA interaction with proteins are seemingly more complex than circRNA-microRNA interactions, as several different modes of action have been described. It has been confirmed that binding between circRNAs and proteins affects the subcellular localization of protein complexes and can alter the levels and longevity of both RNA and protein players
[7][8][9].
The DICER protein complex is a critical factor in the biogenesis of most small regulatory RNAs. This enzyme belongs to the ribonuclease III family, which cleaves long double-stranded RNA molecules into short molecules, including microRNAs and small interfering RNAs (siRNAs). DICER malfunction is associated with the global loss of mature microRNAome in both in vitro and in vivo models, supporting pro-oncogenic cellular transformation
[10]. MicroRNAome, an assortment of all microRNAs being expressed in a given cell at a given time, is an essential readout of cellular homeostasis. Faulty expression or localization of microRNA processing machinery components leads to impaired microRNAome composition and is thought to be associated with cancer formation
[7]. However, the mechanism of such deregulation remained unclear. A comparison of DICER distribution in the cellular compartments of glioblastoma CSCs and non-malignant neural progenitor cells (NPCs) indicated that while DICER was mainly cytosolic in NPCs, it was decidedly nuclear in various glioblastoma CSC subtypes. Detailed studies discovered a nuclear RNA/protein complex consisting of the DICER, strictly nuclear RBM3 protein, and circ
MALAT1 (
hsa_circ_0002082) derived from the long non-coding
MALAT1 oncogene. Knockdown of circ
MALAT1 restored cytosolic DICER localization, thus reestablishing microRNAome homeostasis in diverse subpopulations of glioblastoma CSCs. The significance of circ
MALAT1 was apparent upon its knockdown in glioblastoma CSCs, resulting in lessened clonality and tumorigenicity and prolonged survival of circ
MALAT1 knockdown glioblastoma CSC tumor-bearing animals
[7].
Due to their ability to fine-tune scores of genes, microRNAs have been recognized as master guardians of final differentiation. Most microRNAs dictate terminal differentiation programs; therefore, the molecular signature of mass inhibition of microRNAs allows cancer cells to avoid these programs
[11]. The discovery of sweeping deregulation of the microRNAome in glioblastoma CSCs (via circ
MALAT1 action) may indicate a potential therapeutic target for unleashing differentiation programs in these cells. Restoring the pre-malignant composition of the microRNAome promotes the differentiation of cancer stem-like cells, making them more susceptible to treatment. Such an approach would thus allow the eradication of a subpopulation vital for tumor recurrence post-therapy.
CircRNAs can also modulate the activity of the transcription factors. The circRNA circ
RPPH1 (
hsa_circ_0000512) is elevated in glioblastoma CSCs, correlating with poor patient survival. It binds with UPF1 (RNA helicase), which maintains the stability of a complex, and ATF3 (transcription factor), which increases
UPF1 transcription and activates TGF-β signaling. Significantly, the characterized feedback loop contributes to the constant expression of the stem cell marker—Nestin, maintaining the oncogenic features of glioblastoma CSCs. The silencing of circ
RPPH1 significantly inhibited the proliferation and clonogenicity of glioblastoma CSCs both in vitro and in vivo, while its overexpression enhanced their self-renewal
[12].
CircRNAs were also implied in liver carcinoma CSCs, where circRNAs’ role has so far remained elusive. Recently, circ
IPO11 was shown to be significantly increased in these CSCs, where it recruited topoisomerase 1 to the
GLI1 gene promoter, leading to the activation of Hedgehog signaling that plays a critical role in the self-renewal of liver cancer CSCs’ tumorigenicity
[13].
In summary, circRNAs can modulate the expression/activity of specific protein complexes by affecting the sub-cellular localization of essential components, disrupting complex assembly, turnover, composition, and activity, thus affecting entire pathways and complex molecular readouts.
3. circRNAs Encode Proteins
Linear transcripts that are the matrix for protein synthesis contain certain structural elements, e.g., the 5′ cap structure, required for translation. CircRNAs, as molecules lacking these structures, were thus regarded as true non-protein-coding RNAs. However, studies have shown that specific and arguably rare circRNAs can serve as a template for protein synthesis. Some circRNAs, especially exonic ones, contain open reading frames (ORFs) that can be actively translated
[14].
Due to their unique structure, circRNAs are thought to be translated primarily through cap-independent mechanisms such as through an internal ribosome entry site (IRES)
[15] and N6-methyladenosine (m6A) modification
[16], although the mechanisms of translation of many coding circRNAs are still unknown.
Emerging evidence suggests that circRNA-derived proteins play a significant biological role in the cellular stress response and are involved in cancer progression. A representative example of this mechanism of action is recently described as
hsa_circ_0006401, whose expression in metastatic colorectal cancer was significantly increased compared with non-metastatic one
[17]. This particular circRNA contains an ORF spanning a splicing junction that encodes (Encyclopedia of DNA Elements) a 198 amino acid-long peptide expressed in human colon cancer and adenocarcinoma tissue samples, promoting an aggressive phenotype in colorectal cancer cells
[17].
CircRNAs, which encode proteins, have also been described in CSCs. A recent study indicated that the circular E-cadherin, which encodes a previously unknown peptide variant of the secretory E-cadherin protein, promotes the tumorigenicity of glioblastoma CSCs. E-cadherin protein activates EGFR independently of EGF, while its inhibition significantly suppresses tumorigenicity
[18]. In another case, a protein encoded by circ
EZH2—EZH2-92aa, overexpressed in glioblastoma cells, induced the evasion of glioblastoma CSCs’ responses to NK cells. EZH2-92aa inhibits the transcription of Major Histocompatibility Complex class I polypeptide-related sequence A/B and indirectly inhibits the transcription of UL16-binding protein by stabilizing EZH2. A functional approach showed that stable knockdown of EZH2-92aa enhances NK cell-mediated glioblastoma CSCs eradication in vitro and in vivo, synergizing with anti-PD1 therapy. Thus, the EZH2-92aa peptide encoded by circ
EZH2 is a decisive immunosuppressive factor
[19].
The translation potential Is a previously underappreciated and exciting line of research in the context of circular RNAs. The discovery of their coding capabilities shed new light on their function and unexpectedly shattered their reputation as true non-coding molecules. In determining the function of proteins derived from circRNAs, it should be noted that some have functions that mirror their host genes’ products, while some may have roles that are at odds with the parental transcripts’ products. Many questions remain open despite discovering potential mechanisms of circRNAs translation and identifying many functional peptides encoded by these non-coding RNAs. It is thus vital to delineate the biological significance of circRNAs translation and determine the translation mechanisms that enable protein synthesis from circRNAs.
4. circRNAs Regulate Gene Expression
Some circRNAs regulate gene expression at the post-transcriptional stages, e.g., by acting as molecular sponges that bind to and block microRNAs. Others can regulate the expression of their parental genes via attracting transcription factors and chromatin modifiers. Recently, a circRNA called circRNA activating
MAFF (cia-
MAF) was identified. It is highly expressed in liver cancer and its CSCs. Cia-
MAF binds to the
MAFF promoter, recruits the TIP60 chromatin-modifying complex, and ultimately promotes
MAFF expression. As a result,
MAFF promotes the expression of
CD44, a crucial CSCs’ marker that upholds their self-renewal. Loss of cia-
MAF function weakens the link between the TIP60 complex and the MAFF promoter
[20]. In another study, downregulation of circ
REEP3 (
hsa_circRNA_400564) inhibits the tumorigenicity of colorectal cancer and its metastatic potential while impairing their stemness. Mechanistically, circ
REEP3 recruits the chromatin-remodeling protein CHD7 to the promoter of the renowned oncogene FKBP10, activating its transcription. Additionally, circ
REEP3 enhanced the interaction between RIG-1 and RNF125 to promote ubiquitination-dependent degradation of RIG-1, leading to the suppression of antitumor immunity
[21].
Thus, circRNAs can orchestrate the interaction of transcription factors with promoters by recruiting proteins or entire complexes, consequently shaping the gene expression landscape in CSCs.
5. circRNAs Control Protein Lifespan and Turnover
CircRNAs can regulate the pool of proteins present in a cell at any given time by controlling their lifespan and possible degradation. One example of such a circRNA is circ
FNDC3B (
hsa_circ_0006156). m6A-modified circ
FNDC3B plays a tumor suppressive role in colon cancer CSCs by increasing
RNF41 mRNA stability and expression and thus promoting ASB6 degradation via RNF41-mediated ubiquitination. RNF41 silencing abrogated circ
FNDC3B-suppressed stemness and metastatic potential of colorectal CSCs. In vivo experiments showed that overexpression of circ
FNDC3B or RNF41 alone suppressed tumor growth, stemness, and liver metastasis through modulation of ASB6
[22].
Although many studies have covered the detailed mechanistic and signaling contexts of the ferroptosis pathway, the role of non-coding RNAs (ncRNAs), especially circRNAs, in the process is still unclear. One such molecule is circ
LRFN5 (
hsa_circ_0031751), which modulates ferroptosis in glioblastoma CSCs. This circRNA is downregulated in glioblastoma compared to normal brain tissues. It has been shown that circ
LRFN5 can inhibit glioblastoma CSCs’ viability, neurosphere formation, stemness, and tumor formation in vivo by binding to the PRRX2 protein, thus promoting its degradation via a proteasomal pathway mediated by ubiquitin. As PRRX2 maintains the expression of GTP cyclohydrolase I (GCH1), its degradation disables GCH1 activity, leading to lipid and ROS accumulation as well as glutathione depletion, thereby inducing ferroptosis and reducing the carcinogenicity of CSCs
[23].
In conclusion, circRNAs regulate protein stability in CSCs by inhibiting proteins’ activity, stability, and turnover, thus participating in a dense network of connections, e.g., regulating cell death.
6. circRNAs Regulate mRNA Stability
CircRNAs can also manage the pool of mRNAs present in the cell and their biological activity. Chen et al. identified an rt-circRNA, rt-circ
E2F, which is highly expressed in liver cancer and CSCs and plays an essential role in their self-renewal and activity. Rt-circ
E2F interacts with
E2F6 and
E2F3 mRNAs, attenuating their turnover, thus, increasing
E2F6/E2F3 activity. Moreover, this circRNA promotes the association of
E2F6/E2F3 mRNAs with IGF2BP2 and inhibits their association with m6A reader—YTHDF2, thereby inhibiting
E2F6/E2F3 mRNA decay. Both
E2F6 and
E2F3 are required for the self-renewal of liver CSCs and activation of the Wnt/β-catenin pathway. Thus, inhibiting these pathways is a promising strategy for preventing liver tumorigenesis and metastasis
[24].
Although the number of reports on such functions of circRNAs in CSCs is limited, recent studies confirm that circRNAs can affect RNA turnover in this type of cells.