Emerging roles of circular RNAs in gastric cancer metastasis and drug resistance

Gastric cancer (GC) is an aggressive and heterogeneous malignancy [1, 2]. With a median overall survival (OS) of 16 months among all patients, GC remains the fourth leading cause of cancer-related mortality worldwide [1‐3]. Metastasis is a crucial process characterized by increased invasion and the ability of cancer to spread from its site of origin to other regions of the body, accounting for 90% of cancer-related deaths [4, 5]. Most GC patients are diagnosed at advanced stages and are frequently accompanied by invasion and metastasis, such as lymph node and peritoneum metastases [6, 7]. In metastatic (late) GC patients, the clinical outcomes are extremely poor, while the 5-year overall survival rate of early GC patients can reach over 90% [4, 7]. In addition, metastatic GC has long been considered less effective for surgical treatment and more resistant to drug therapy [8, 9]. Up to date, no effective methods or approaches are applied to treat metastatic GC [8, 9]. Recently, significant advances have been made in clarifying GC metastasis [5, 10]; however, the overall delineation of the molecular mechanisms is limited and ambiguous. Therefore, an in-depth understanding of GC metastasis at the molecular and cellular levels is imperative to identify potential biomarkers for diagnosis and therapeutic targets for intervention.

Covalently closed circular RNAs (circRNAs) are single-stranded endogenous RNA molecules with loop structures and are resistant to exonuclease activity [11‐13]. The biogenesis of circRNAs is widely acknowledged via a back-splicing event from precursor RNA (pre-RNA), which is facilitated by the flanking reverse complementary sequences, such as Alu elements, and is regulated by some RNA binding proteins (RBPs), including QKI, DHX9, FUS, Sam68, hnRNP L, hnRNPM and ADARs (Fig. 1A) [14‐23].

Thousands of circRNAs across species have been identified and characterized through high-throughput sequencing combined with bioinformatic analyses in the past decade [24, 25]. Most circRNAs are chiefly derived from known protein-coding genes, consist of a single or multiple exon(s) (exonic circRNAs, ecircRNAs), and generally localize to the cytoplasm [13]. The most prominent function of cytoplasmic ecircRNAs is to serve as competing endogenous RNAs (ceRNAs) or miRNA sponges to lift the inhibitory effects of miRNAs on their downstream targets (Fig. 1B) [24, 26‐28]. Interestingly, the intronic sequences between the circularized exons may be retained, forming exon-intron circRNAs (EIciRNAs) [29]. EIciRNAs are proved to enhance their parental gene expressions in cis via binding to the U1 small nuclear ribonucleoprotein (snRNP) complex in the nucleus (Fig. 1C) [29]. Intronic lariat precursors escaping from debranching produce intronic circRNAs (ciRNAs), which could regulate RNA polymerase II (Pol II)-mediated transcription in the nucleus [30, 31]. Besides, circRNAs directly interact with RBPs to regulate key targets as protein scaffolds or antagonists in various biological processes as well (Fig. 1D) [32‐34]. In addition, a small fraction of ecircRNAs undergoes cap-independent translation to encode small peptides through the internal ribosome entry site (IRES)-driven mechanisms, although the vast majority of circRNAs are thought to be non-coding RNAs (Fig. 1E) [35‐37]. Recently, a novel class of circRNAs encoded by mitochondria (mecciRNAs) has been reported to facilitate the mitochondrial entry of nuclear-encoded proteins by serving as molecular chaperones [38].

Accumulating evidence has pointed out the aberrant expression patterns of circRNAs and their regulatory roles in cancer progression and metastasis [39‐44]. Systematic and comprehensive knowledge regarding circRNAs related to GC metastasis expands our understanding of the underlying mechanisms of metastatic GC. In the present review, we overview the current research status of circRNAs in GC metastasis, including modulating epithelial-mesenchymal transition (EMT), regulating angiogenesis, exosomal circRNAs, and drug resistance, and discuss the potential clinical application value of circRNAs in GC. We hope to provide insights into circRNAs-mediated GC metastasis and their potential as putative biomarkers or therapeutic targets of GC in the future.

EMT, a highly complex and dynamic process, is recognized as a vital step driving the early phase of cancer metastasis [45, 46]. Recently, several circRNAs have been reported to participate in EMT by modulating various signaling pathways, such as TGF-β/SMAD, Wnt/β-catenin, and PI3K/AKT pathways [47]; thereby, we summarized up-to-date information on circRNAs engaged in these signaling pathways in GC metastasis (Table 1).

Angiogenesis, defined as the formation of new blood vessels sprouting from preexisting vessels, is well-regarded as an important initial step in cancer metastasis [77‐79]. Several signaling pathways, including VEGFA and HIF1α signaling, can continuously induce angiogenesis, aggravating cancer progression [80, 81]. Recently, several circRNAs have been reported to participate in GC metastasis by regulating VEGFA- or HIF1α-mediated angiogenesis [62‐65].

The circRNA circRanGAP1 is validated to sponge miR-877-3p to increase the VEGFA expression, stimulate angiogenesis and promote GC metastasis (Fig. 2BI) [62]. A similar ceRNA mechanism also applies to circ_0044366, which binds to miR-29a to derepress the VEGF expression and thus facilitates angiogenesis and migration in GC [63]. The circRNA circURI1 back-spliced from exons 3–4 of URI1 has been identified from circRNA profiling of 5 paired GC and adjacent non-cancerous (paraGC) specimens [64]. CircURI1 exhibits a remarkably higher expression in GC than paraGC tissues and is negatively associated with metastasis in GC patients [64]. Functional studies perform that circURI1 inhibits GC metastasis in vitro and in vivo. Mechanistically, circURI1 behaved as a decoy of hnRNPM in a sequence-dependent manner to modulate alternative splicing of a subset of genes related to cell migration, thus suppressing GC metastasis (Fig. 2BII) [64]. VEGFA is a direct and functional target of circURI1, and circURI1 can promote exon 7 inclusion of VEGFA (VEGFA^e7IN) [64]. CircURI1-induced VEGFA^e7IN possesses a greater ability to prevent the circURI1-silencing-mediated promoting effect on GC cell invasion than exon 7 exclusion of VEGFA [64, 82]. This study firstly reported the engagement of circRNA-modulated alternative splicing in cancer metastasis [64].

Additionally, virus-encoded circRNA has also been found to engage in angiogenesis in GC [65, 83]. Epstein-Barr virus (EBV)-derived circRNA LMP2A (ebv-circLMP2A) is correlated with distant metastasis and poor prognosis in EBV-associated GC (EBVaGC) [65]. Furthermore, the ebv-circLMP2A expression is positively correlated with the expressions of HIF1α and VEGF in clinical samples of EBVaGC and a mouse model [65]. Ectopic expression of ebv-circLMP2A promotes angiogenesis and GC cell migration under hypoxia, while ebv-circLMP2A knockdown reverses these effects [65]. Mechanistic studies reveal that HIF1α and ebv-circLMP2A form a positive feedback loop, which promotes angiogenesis in EBVaGC [65]. Briefly, under hypoxia, HIF1α induces the ebv-circLMP2A expression, and ebv-circLMP2A interacts with KHSRP to enhance the VHL mRNA decoy mediated by KHSRP, resulting in HIF1α accumulation (Fig. 2BIII) [65].

Exosomes are small extracellular vesicles with an average diameter of ~100 nanometers, containing an abundant cargo of proteins and different RNA species, including circRNAs, which can enhance substance exchange between cells and improve signal transduction [84, 85]. Accumulating evidence has demonstrated that exosomes play emerging roles in regulating cancer metastasis and treatment through the transfer and exchange of molecules during cell-cell communications [86, 87]. Recently, circRNAs have been shown to be abundant in exosomes and exosomal circRNAs might be regarded as circulating biomarkers for metastatic disease in GC patients [88, 89].

Multiple exosomal circRNAs from the plasmas of GC patients are involved in GC invasion and metastasis [66‐69]. CircNRIP1 possesses a significantly higher expression level in exosomes from GC plasma than in normal tissues and engages in exosomal crosstalk between GC cells [66]. GC cells co-cultured with exosomes derived from circNRIP1-overexpressed cells exhibit higher metastatic potential than control cells via the tail vein metastasis model [66]. Simultaneously, exosomal circNRIP1 promotes GC metastasis in vivo and regulates EMT by activating the AKT1/mTOR signaling pathway via sponging miR-149-5p [66]. Similarly, circNEK9, an up-regulated circRNA in GC tissues, accelerates GC proliferation by serving as a ceRNA against miR-409-3p to target MAP7 [67]. Additionally, the exosome-mediated transfer of circNEK9 performs promotive effects on GC cell migration and invasion [67]. Sang et al. have uncovered that exosomal circRELL1 is down-regulated in GC, and its delivery mediated by GC cells-derived exosomes stimulate autophagy by modulating the miR-637/EPHB3 axis in GC progression [68]. In another case, circSHKBP1 is remarkably upregulated in both GC tissues and serum and is significantly associated with advanced TNM stage and poor survival [69]. Mechanistically, exosomal circSHKBP1 promotes GC cell migration and invasion via modulating the miR-582-3p/HuR/VEGF axis, and inhibiting HSP90 ubiquitination through sequestering HSP90 to obstruct its interaction with STUB1 (Fig. 2C) [69]. These promising results provide novel insights into therapy and the predictions of GC prognosis.

Although chemo- and radio-therapy are recognized as the most effective and extensive treatment methods for GC patients after surgery during the past few decades, the clinical applications are still limited owing to the intrinsic and acquired resistance, resulting in the occurrence of distant metastasis in GC patients [1, 3, 95]. Additionally, targeted therapy and immunotherapy with immune checkpoint inhibitors for GC have emerged [96]. Convincing evidence has confirmed that diverse circRNAs influence drug resistance in GC therapeutic responses (Table 2) [55, 112].

Cisplatin (CDDP) is one of the most effective chemotherapeutic agents for patients with GC, especially those in advanced stages [113, 114]. The circVAPA expression is elevated in CDDP-resistant GC cells, and circVAPA facilitates GC cell migration, invasion, and CDDP resistance [97]. Further mechanistic investigations indicate that circVAPA exerts its oncogenic activity through sponging with miR-125b-5p to increase the STAT3 expression [97]. Similarly, several other circRNAs such as circAKT3, circPVT1, circFN1, and circ_0000260, also enhance CDDP resistance and malignant progression in GC [55, 98‐102]. Oxaliplatin (OXA) is a widely used anti-cancer medicine [115]. The circRNA circ_0032821 is significantly increased in OXA-resistant GC cells and their derived exosomes, and contributes to OXA resistance, GC cell migration and invasion through derepressing SOX9 via sequestering miR-515-5p [103]. Paclitaxel (PTX) is an effective first-line chemotherapy drug in GC treatment, and circPVT1 contributes to PTX resistance and GC cell invasion via serving as a ceRNA against miR-124-3p to target ZEB1, a crucial transcriptional inhibitor of E-cadherin [104]. 5-fluorouracil (5-FU) is currently a first-line agent for the clinical treatment of GC, and circNRIP1 promotes hypoxia-induced 5-FU resistance via modulating the miR-138-5p/HIF-1α axis in GC [105]. Anti-programmed cell death protein 1 (PD-1) monoclonal antibody is a commonly used immune-checkpoint blockade agent for GC immunotherapy [116]. The circRNA circDLG1 facilitates GC progression and anti-PD-1 resistance via miR-141-3p-mediated the regulation of CXCL12 [106].

On the other hand, various circRNAs reverse drug resistance in GC treatment [107‐109]. Peng et al. have unveiled that circCUL2 displays a decreased level in GC tissues and possesses a repressively regulatory function in CDDP resistance, GC cell migration, and invasion via miR-142-3p/ROCK2-mediated autophagy activation [107]. Another circRNA circMCTP2 is reported to inhibit CDDP resistance of GC cells via the miR-99a-5p/MTMR3 axis [108]. The circRNA hsa_circ_0000144 exerts inhibitory effects on OXA resistance, GC cell proliferation, and metastasis through up-regulating ADAM9 mediated by miR-502-5p [109]. Bupivacaine, a local anesthetic commonly used in the resection operation of GC patients, reduces the circ_0000376 level in GC cells, and circ_0000376 partially reverses bupivacaine-mediated repressive effects on GC cell viability and metastasis via sponging miR-145-5p [110]. Herceptin, a targeted therapy drug, is a humanized monoclonal antibody specifically binding to HER2 and acts as an antitumor role in GC [117]. The circRNA hsa_circ_0000520 is significantly reduced in GC and reverses the Herceptin resistance of GC cells by inhibiting the PI3K/AKT signaling pathway [111].

Taken together, these studies provide the possibility that a combination of circRNAs-based therapy with chemotherapy, targeted therapy or immunotherapy may be a valuable approach to overcome drug resistance and prevent metastasis in GC in the future.

CircRNAs have multiple remarkable characteristics which provide tremendous potential for serving as biomarkers and therapeutic targets owing to the covalently closed-loop structure, disease-specific and dynamic expression pattern and high conservation across species [118‐122]. For example, according to a study by Liang and colleagues, hsa_circ_0110389 has been identified as a diagnostic/prognostic biomarker and therapeutic target for GC [123]. Similarly, circOSBPL10 might serve as a novel proliferation factor and prognostic marker of the OS and disease-free survival (DFS) of GC patients [124]. In another case, Chen et al. have displayed that the circPVT1 level is an independent prognostic biomarker for OS and DFS in GC patients [125].

Since exosomes can be detected in various body fluids, including plasma, saliva, urine, and cerebrospinal fluid, exosomal circRNAs might be ideal noninvasive biomarkers for the diagnosis and/or prognosis of gastric cancer [88, 126]. For instance, the circSHKBP1 expression is significantly increased in GC serum and positively correlated with advanced TNM stage and poor survival [69]. Furthermore, GC cell exosomes enhance co-cultured cell growth by delivering circSHKBP1 [69]. These findings indicate that circSHKBP1 is a promising circulating biomarker for GC diagnosis and prognosis [69]. Additionally, the circRNA circRanGAP1 exhibits a significantly higher expression in plasma exosomes derived from GC patients than the healthy controls. It promotes GC cell migration and invasion, indicating that plasma exosomal circRanGAP1 might serve as a promising biomarker for GC patients [62]. The circRNAs that show potential as biomarkers in GC are summarized in Table 3.

Current active research in circRNAs has brought us a range of exciting findings implying that circRNAs are of great importance in various diseases [11, 118, 127‐129]. A tremendous amount of evidence has demonstrated the abnormal expression pattern of circRNAs in GC and the involvement of circRNAs in GC metastasis and drug resistance [11, 64, 126]. We have systematically described a series of dysregulated circRNAs in GC and elucidated their underlying molecular mechanisms in GC metastasis and drug resistance (Tables 1 and 2).

To date, various circRNA candidates have been validated and engaged in GC metastasis based on a series of molecular and cellular experiments [64, 66‐69, 124, 125]. However, a global and comprehensive understanding of circRNAs related to GC metastasis is still scarce. To gain better and deeper insight into the aberrant expression pattern of circRNAs involved in GC metastasis, genome-wide circRNA profiling with high throughput sequencing from metastatic and non-metastatic GC tissues is a powerful approach to address this issue.

Four subclasses of circRNAs have been identified, including ecircRNAs, EIciRNAs, ciRNAs and mitochondria-encoded circRNAs (mecciRNAs) [11, 38, 130]. Current literature about circRNAs in GC metastasis generally includes ecircRNAs and ciRNAs, their functions and the molecular mechanisms [72‐75, 94, 126]. Nevertheless, two other kinds of circRNAs and their functions have not been evaluated, which presents an exciting field to explore further.

The well-characterized mechanism of circRNAs is to sequester miRNAs to regulate the expressions of targeted genes [11‐13]. A single circRNA could function as a scaffold for several different miRNAs [123]. Conversely, a miRNA can target multiple circRNAs as well [60, 61]. Identification and construction of the circRNA-miRNA regulatory network will help to systematically decipher the roles of circRNAs in GC metastasis in the future. In addition to the ceRNA mechanism, circRNAs have various molecular modes of action, including participating in epigenetic regulations, modulating alternative splicing, and generating protein [64, 71, 75]. We expect a burst of circRNA studies to elucidate some novel mechanisms of action in GC metastasis in the upcoming years.

Considering that circRNAs possess unique features such as tissue- and developmental stage-specific patterns, structural resistance to exonucleases and longer half-lives, and specific circRNAs play essential roles in GC metastasis and drug resistance, manipulating circRNA abundance appears to be a promising therapeutic strategy for the advanced GC treatment [126‐128, 131, 132]. Furthermore, combining circRNAs-based therapeutic interventions with traditional chemotherapy or targeted therapy offers a unique opportunity to conquer drug resistance in advanced GC patients [97‐111, 113‐117]. However, choosing crucial target circRNAs of interest is still a problem. Furthermore, precisely and effectively delivering circRNAs into targeted cells for tumor treatment is also a significant issue that needs to be solved.

In summary, the advances in circRNAs research will be essential to unravel their potential significance in GC. Furthermore, a better understanding of the association between circRNAs and GC would make circRNAs promising candidates as valuable biomarkers or potential targets in GC treatment.