The emerging roles of non-coding competing endogenous RNA in hepatocellular carcinoma

LncRNAs, ncRNAs over 200 nt in length that originate from promoter-proximal, antisense and intergenic regions, carry out their regulatory function by interacting with genomic DNA, mRNAs and proteins as well as other ncRNAs [45, 46]. LncRNAs may act as sponges, scaffolds, decoys, signals and guides. LncRNAs can regulate gene expression at epigenetic, transcriptional, posttranscriptional and translational levels. Thus, these ncRNAs are considered to have pleiotropic impacts and considered “master regulators” of the genome [47, 48]. They exert regulatory effects in both physiological and pathological conditions, including in various cancers such as HCC [49‐51]. Emerging evidence has revealed that a large family of lncRNAs function as ceRNAs for modulation of the expression and biological features of miRNAs in HCC [48] (Table 2).

LncRNAs can exert their functions in the HCC pathophysiological process by participating in a certain pathway but through interactions with different miRNAs. The lncRNA metastasis-associated lung adenocarcinoma transcript-1 (MALAT1) promotes HCC migration and invasion by three pathways [57‐59]. First, MALAT1 competes with miR-30a-5p to regulate vimentin gene expression. Pan et al. [57] first demonstrated the sponge role of MALAT-1 for miR-30a-5p and described its functions in HCC as a ceRNA. The authors measured MALAT1 and vimentin expression levels in paired HCC and normal adjacent tissues by RT-PCR and predicted potential miR-30a-5p binding sites in MALAT1 and the target site of miR-30a-5p in vimentin using bioinformatic analyses. Co-transfection of plasmids and miR-30a-5p mimics into 293T cells was used to discover how miR-30a-5p regulates MALAT1 expression levels. Dual luciferase assay showed that miR-30a-5p significantly decreased vimentin expression level, and subsequent experiments showed that miR-30a-5p inhibits HCC cell migration and invasion. The role of MALAT1 in HCC cell growth was also tested using in vivo experiments, in which the authors injected HepG2 cells transfected with sh-MALAT1 into nude mice; the tumor volume and size in the shRNA knockdown group were smaller than that of the negative control group. These results provided in vivo evidence that MALAT1 promotes HCC growth. Second, MALAT1 competes with miR-204 to regulate SIRT1 gene expression. Hou et al. [58] used shRNA knockdown and loss-of-function strategies on MALAT1 in HepG2 cells to screen potential MALAT-1-interacting miRNA candidates, and identified miR-204 had shown the best correlation. The authors treated HepG2 cells with miR-204 inhibitors and mimics and found that MALAT-1 level was increased and the migration and invasion abilities of HepG2 cells were enhanced in miR-204 inhibitor group. The results demonstrated that MALAT1 derepresses SIRT1 by sponging and competitively binding to miR-204, causing enhanced migration, invasion and epithelial-mesenchymal transition (EMT) of HCC. Finally, MALAT1 also competes with miR-143-3p for the tumor suppressor ZEB1. Chen et al. [59] conducted a clinicopathological analysis in 56 paired HCC and non-tumor liver samples to explore the association between MALAT-1 expression level and clinical characteristics including age, sex, tumor size, tumor differentiation, TNM stage and distant metastasis. The authors also predicted possible interactions among MALAT1, miR-143-3p and ZEB1 based on a bioinformatics analysis. SiRNA knockdown of MALAT-1 was performed in several cell lines, and Matrigel invasion assays proved that si-MALAT-1 inhibited HCC cell invasion. Subsequent experiments confirmed the existence of the MALAT1/miR-143-3p/ZEB1 regulatory pathway and showed that high expression of MALAT1 in HCC led to increased ZEB1, thus promoting the proliferation and migration of HCC.

Several lncRNAs can influence HCC behavior by several distinct mechanisms through interactions with different miRNAs and their corresponding mRNAs. HULC is one of the most upregulated lncRNAs in HCC and was found to function in HCC by at least three mechanisms (Fig. 5a). Wang et al. [12] performed chromatin accessibility by real-time PCR (CHART-PCR) assays to evaluate the accessibility of genomic DNA to nuclease and analyze different regions of open chromatin around the HULC promoter. ChIP assays demonstrated that increased euchromatic histone modifications are correlated with transcription activation of HULC. Subsequent experiments revealed significantly increased miR-372 level after siRNA-mediated inhibition of HULC, indicating that HULC decreases miR-372 expression in HCC and enhances chromatin accessibility and transcription. Li et al. [53] observed that a higher expression level of HULC in HCC tissues is associated with many enhanced EMT features, including histological morphology, physiological behavior and EMT markers such as E-cadherin, N-cadherin, ZO-1, vimentin, β-catenin, Snail and ZEB1. The authors proposed that HULC exhibits a negative regulatory effect on miR-200a-3p and upregulates ZEB1, resulting in enhanced EMT and promoted growth and metastasis of HCC. A HULC/USP22/Sirt1 protective autophagy pathway was further identified that attenuates HCC cell sensitivity to chemotherapeutic agents [52]. In the study, autophagy-related proteins were markedly increased upon HULC overexpression, but were significantly reduced upon Sirt-1 silencing. These results revealed a HULC-Sirt1-autophagy pathway, which was further discovered to be able to weaken the chemosensitivity of HCC cells toward oxaliplatin. As another example, the lncRNA X inactive-specific transcript (XIST) inhibits the proliferation of HCC cells through the miR-497-5p/PDCD4 or miR-92b/Smad7 pathway, but promotes HCC progression by silencing miR-194-5p and derepressing MAPK1 [92‐94].

Together, these studies on the associations between lncRNAs and HCC show that lncRNAs function in a multidimensional network to influence HCC, not only by interacting with a large group of miRNAs and affecting their correlated pathways, but also by participating in a number of distinct mechanisms. Notably, some lncRNAs show both tumor suppressive and oncogenic functions, which calls for additional studies to determine how these seemingly contrary functions are balanced and regulated.

CircRNAs are relatively more resistant to degradation compared with linear RNA because the 3′ and 5′ ends of circRNAs are covalently linked [101]. Apart from higher stability compared with linear RNAs, circRNAs are highly homologous to their linear counterparts and contain more MREs. Thus, circRNAs can act as effective miRNA sponges to disrupt miRNA-mediated target repression and regulate genes at both transcriptional and post-transcriptional levels as robust ceRNAs [102]. Emerging evidence has demonstrated that circRNAs may counteract miRNA-mediated repression of linear mRNAs by functioning as miRNA sponges (Table 3).

Similar to other ceRNAs, circRNAs competitively inhibit miRNAs via their MREs. CircRNAs function in tumor signaling pathways through their regulation of cell proliferation, apoptosis, tumor invasion and migration. Qin et al. [142] demonstrated significantly decreased expression of hsa_circ_0001649 expression in HCC tissues (P = 0.0014), which was related to tumor embolus (P = 0.017) as well as tumor size (P = 0.045). Yao et al. [143] reported the significantly decreased expression of circZKSCAN1 (hsa_circ_0001727) in HCC (P < 0.05), which was associated with vascular invasion (P = 0.002) and cirrhosis (P = 0.031) as well as tumor number (P < 0.01). Hu et al. reported increased levels of circASAP1 in HCC patients with postoperative metastasis or recurrence [103]. Transplanted HCC mouse models verified that circASAP1 promoted proliferation as well as invasion of HCC through miR-326/miR-532-5p-MAPK signaling (Fig. 5b). Xenografts of tumors with shRNA-mediated knockdown of circASAP1 showed decreased numbers of metastatic pulmonary nodules, while the xenograft model derived from circASAP1-overexpressing cells had a higher rate of lung metastasis. In addition, circASAP1 enhanced the infiltration of tumor associated macrophages in xenografts, but no alterations of infiltrating neutrophils or fibroblasts were observed. Zhang et al. observed that overexpression of circSMAD2 attenuated migration, invasion and EMT of HCC by sponging miR-629 [139]. Taken together, these results provide evidence that circRNAs have a dual role in regulating tumor signaling pathways by competing with miRNAs with contrary functions.

CircRNAs are resistant to exonucleases, resulting in relatively stable expression and intracellular accumulation. Exosomal excretion is thus an important route of in vivo elimination of circRNAs. In a study by Su et al., SMMC-7721 and HepG2 cell-derived exosomal circRNA Cdr1 was transfected into 293T cells [104]. Consistent with its intracellular function, exosomal Cdr1as increased the proliferation and migration of 293T cells, indicating that circRNAs function in the cytoplasm and exosomes. Zhang et al. reported a similar finding in the circ-DB/miR-34a/USP7/Cyclin A2 pathway [114].

The activating receptor natural-killer group 2 member D (NKG2D) is a tumor-associated immune response marker, and its expression indicates immune response mediated by CD8 + T cells and γδ + T cells as well as NK cells in tumors. Zhang et al. proposed the circTRIM22-12/miR-191/TET1 pathway and validated it in circTRIM33-12-overexpressing nude mice xenograft models. circTRIM33-12 expression level was positively correlated with NKG2D-positive cell numbers in HCC, suggesting that circTRIM33-12 sponged miR-191/TET1 and exerted an anti-tumor function by not only participating in the tumor proliferation–related signaling pathway, but also contributing to immune surveillance and T cell activation [136].

Considering the high stability of circRNAs in body fluids, including blood and saliva, circRNAs are potential superior biomarkers for HCC diagnosis. The miR-328-3p/HIF1AN-NOTCH2 and miR-892a/HDGF-NCL-PI3K-AKT pathways regulated by circCDYL facilitate Survivin and Myc expression in HCC, suggesting that a combination of circCDYL, HIF1AN and HDGF can be used for the diagnosis of Barcelona Clinic Liver Cancer stage 0 and A of HCC [144]. CircRNA_104075, another potential diagnostic biomarker, was discovered to upregulate the expression of YAP by inhibiting miR-582-3p. Zhang et al. used the CRISPR/Cas9 strategy to establish a knockdown mouse model of HNF4a. HNF4a was formerly proved to be an HCC-promoting transcription promoter. HNF4a knockout mice were smaller in size, with a low liver weight, and had low expression of circ_104075, which indicates that circ_104075 transcription is stimulated by HNF4a. Intriguingly, an m6A modification in the 3′-untranslated region of YAP mRNA facilitates the interaction between YAP, miR-582-3p and circRNA_104075, demonstrating how circRNA function can be influenced by methylation or other modifications and the role of epigenetics in circRNA function [108].

Similar to lncRNAs, circRNAs exert their diverse functions in HCC through sponging miRNAs. CircRNAs exhibit multiple roles in different pathways and participate in HCC pathogenesis through various pathophysiologic processes. In addition, circRNAs show enhanced stability, making them promising candidates for novel cancer biomarkers.

Pseudogenes were previously described as non-functional, genomic junk DNAs sharing homology with the parental encoding sequence. Pseudogenes were characterized by a loss of protein-coding regions due to truncation or mutation and suggested to have no function. Transcriptomic and proteomic analyses have not only confirmed the ability of pseudogenes to produce transcripts and proteins, but also revealed novel miRNA/gene/pseudogene regulatory networks in cancer biology [145, 146]. Studies on the influence of pseudogenes in different types of tumors have revealed their roles in HCC (Table 4).

OCT4-pg4 is a pseudogene related to OCT4, a transcription factor that is involved in proliferation, pluripotency and self-renewal ability in embryonic stem cells and germ cells [149]. The expression of OCT4-pg4 is positively associated with OCT4, and both are increased in HCC. Wang et al. found that OCT4-pg4 serves as a molecular decoy towards miR-145 and prevents OCT4 inhibition, thus promoting HCC growth and tumorigenicity [10]. Survival analysis of 54 cases of HCC suggested a significant correlation between increased OCT4-pg4 expression levels and shortened overall survival as well as disease-free survival in HCC patients.

RACGAP, a member of the GTPase activation family, is an oncoprotein that enhances the proliferation and migration of HCC by activating the RhoA/ERK signaling pathway. The RACGAP gene is also the first gene reported as an independent biomarker for HCC recurrence. Wang et al. discovered the significant up-regulation of the RACGAP1P pseudogene in HCC, which was related to shortened survival, larger tumor size, elevated AFP level and advanced clinical stage [147]. Luciferase assays and in vivo assays demonstrated that miR-15-5p was sequestered from its endogenous target RACGAP by RACGAP1P, causing increased RACGAP expression and contributing to the RACGAP oncogenic network (Fig. 5c).

In addition to oncogenic pseudogenes, tumor suppressive pseudogenes have also been identified. The INTS6 pseudogene inhibits the tumor growth in several types of human cancers through G1 cell cycle arrest. Peng et al. showed that INTS6 inhibited HCC cell growth, migration and survival [148]. Furthermore, INTS6P1 facilitated tumor suppression by competing with oncogenic miR-17-5p. In vitro and in vivo assays also showed that both INTS6 and INTS6P inhibited HCC cell growth, migration and survival.

These studies indicate that pseudogenes serve as ceRNAs and affect tumorigenesis through ceRNETs. With respect to their role in HCC carcinogenesis, pseudogenes may be used as prognostic indicators, stratification factors or therapeutic targets, and these functions may lead to the development of precise and individualized therapeutic strategies for HCC. It is likely that HCC-associated pseudogenes acting as ceRNAs are rare. Nevertheless, with the widespread application of whole-genome high-throughput sequencing technology, we believe that more HCC-associated pseudogenes will be uncovered and applied in disease management.

Several ncRNAs that act as ceRNAs have been identified in the circulation and other body fluids in HCC and other cancers [108, 156‐158]. These results suggested the potential that ncRNAs that act as ceRNAs and their associated ceRNETs may be useful for clinical applications as diagnostic biomarkers of HCC (Table 5).

Ming et al. demonstrated that the lncRNA BCYRN1 functions as a ceRNA to play a vital role in HCC [159]. The authors analyzed BCYRN1 expression in plasma from 124 HCC patients, 79 cirrhosis patients, 68 hepatitis B patients and 74 healthy controls. Plasma BCYRN1 expression of HCC patients was significantly higher than that of healthy controls (p < 0.001) or hepatitis B patients (p < 0.01) [159]. Furthermore, combined detection of BCYRN1 and AFP improved the diagnosis of HCC in their cohort. Their data also indicated that BCYRN1 in plasma performed well in HCC diagnosis.

Zhang et al. found that circ_104075 acts as ceRNA to sponge miR-582-3p to stimulate tumorigenesis via YAP [108]. Circ_104075 exhibited a sensitivity of 96.0% and a specificity of 98.3% in HCC diagnosis. Moreover, some recent studies revealed that panels consisting of ncRNAs and their related ceRNETs can also be used as diagnostic biomarkers for HCC. Asmaa et al. [156] constructed an lncRNA-associated ceRNET (lncRNA-RP11-513I15.6-miR1262-RAB11A mRNA) according to an in silico analysis, followed by validation in serum specimens and additional clinical and experimental assays. Receiver operating characteristic curve analysis revealed that RAB11A mRNA, lncRNA-RP11-513I15.6 and miR-1262 in the ceRNET were effective biomarkers to differentiate HCC subjects from healthy controls (area under curve: 0.963, 0.847 and 0.822, respectively). Moreover, the combination of serum exosomal miR-1262, lncRNARP11-513I15.6 and AFP measurements improved the diagnostic accuracy to detect HCC at the early stage to nearly 100% sensitivity and 76.7% accuracy [156]. With the development of specific algorithms for ncRNA detection and quantification, we believe that more and more ncRNAs that function as ceRNAs will be identified as diagnostic biomarkers for HCC.

Several studies have explored the clinical application of ncRNAs that function as ceRNA as prognostic biomarkers and potential therapeutic targets for HCC (Table 5). Cao et al. found that the lncRNA SNHG6-003 functions as a ceRNA to promote cell proliferation and induce drug resistance in HCC [85]. High expression of SNHG6-003 closely correlated with tumor progression and poor survival in HCC patients. The results suggested that targeting the ceRNA network involving SNHG6-003 may be a treatment strategy against HCC. Besides the tremendous diagnostic value in HCC, the prognostic value of BCYRN1 was also proven in a recent study. Ding et al. demonstrated that overexpression of BCYRN1 significantly expedited HCC cell growth, clone formation and movement abilities, while downregulation of BCYRN1 had the opposite effects [160]. The authors also found that BCYRN1 was overexpressed in HCC samples, which was associated with unfavorable prognosis in patients with HCC. The study findings indicated that BCYRN1, miR-490-3p, and POU3F2 formed a ceRNA mechanism to modulate the occurrence and progression of HCC, suggesting these can be served as potential target molecules for the management of HCC [160]. Recently, Lin et al. established a circRNA-miRNA-mRNA regulatory network by integrating the analysis of differentially expressed circRNAs, miRNAs and mRNAs in HCC [9]. The functional enrichment analysis of differentially expressed circRNA-related mRNAs, screened from the constructed HCC-associated ceRNETs, revealed that differentially expressed cirRNAs were significantly associated with cell proliferation, cell adhesion and cell migration. The differentially expressed circRNAs are involved in various processes including cell cycle, peroxisome proliferator-activated receptor signaling, chemical carcinogenesis and p53 signaling, playing vital roles in HCC progression. The refined circRNA-miRNA-mRNA regulatory modules associated with HCC carcinogenesis further uncovered three key circRNAs (hsa_circ_0004913, hsa_circ_0007456 and hsa_circ_0078279) that may play important roles in carcinogenesis and progression of HCC [9]. These circRNAs may competitively bind to miR-182 and miR-346 to affect their respective regulatory networks. Functionalization of previously uncharacterized transcripts could be achieved, partially by identifying the ceRNA interactors, to present a framework for predicting and validating ceRNA interactions; this strategy may be universally applied to any transcript. In a scale-free network, links between nodes follow a power-law distribution, which suggests that most nodes have only a few links, while a few nodes can possess a large number of links [164]. These post-transcriptional ceRNETs can be considered as scale-free networks, because their structure is constructed upon smaller interconnected subnetworks, in which the nodes (ceRNAs) are linked by a large number of connections (miRNAs) [11]. The few highly connected nodes, also known as target hubs, were found to be critically involved in biological processes [164]. Therefore, these critical nodes (three circRNAs and two miRNAs), identified by the constructed circRNA-associated ceRNETs, may represent ideal therapeutic targets for HCC.

As promising biomarkers in the screening, diagnosis and prognosis for HCC, ncRNAs as ceRNAs are also potential therapeutic targets. However, there are still significant gaps in our current understanding of ncRNA functions as ceRNAs and we are still far from being able to incorporate ncRNAs into clinical practice. Large-scale studies and clinical trials are required to validate the role of ncRNAs as ceRNAs in the clinical application for HCC.