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

Gastric cancer (GC) is an aggressive malignancy with a high mortality rate and poor prognosis, primarily caused by metastatic lesions. Improved understanding of GC metastasis at the molecular level yields meaningful insights into potential biomarkers and therapeutic targets. Covalently closed circular RNAs (circRNAs) have emerged as crucial regulators in diverse human cancers including GC. Furthermore, accumulating evidence has demonstrated that circRNAs exhibit the dysregulated patterns in GC and have emerged as crucial regulators in GC invasion and metastasis. However, systematic knowledge regarding the involvement of circRNAs in metastatic GC remains obscure. In this review, we outline the functional circRNAs related to GC metastasis and drug resistance and discuss their underlying mechanisms, providing a comprehensive delineation of circRNA functions on metastatic GC and shedding new light on future therapeutic interventions for GC metastases.

Background

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].

Fig. 1

Biogenesis and function of circRNAs. A. The flanking introns of circularized exons contain reverse complementary sequences, which form a circular structure through direct base pairing, or RBP binding sites, which generate a circular structure via RBPs dimerization. The introns are removed or retained to form exonic circRNA (ecircRNA) or exon-intron circRNA (EIciRNA). Intronic circRNA (ciRNA) is derived from an intron by preventing intron debranching after splicing. Mitochondria encoded circRNAs (mecciRNAs) probably circularize via a splicing-independent mechanism. B. CircRNAs serving as miRNA sponges. C. CircRNAs promoting the transcription of targets through interacting with proteins or complexes such as U1 snRNP. D. CircRNAs behaving as protein decoys. E. CircRNAs working via generating polypeptides

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.

CircRNAs participate in EMT

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).

Table 1

A list of circRNAs related to GC metastasis

CircRNA	CircBase ID	Expression	Property in metastasis	Molecular mechanism	Refs
circTHBS1	hsa_circ_0034536	Up	Enhancer	Modulate the miR-204-5p/INHBA axis and interact with the RBP, HuR	[48]
circCCDC66	hsa_circ_0001313	Up	Enhancer	Activate c-Myc/TGF-β signaling pathway	[49]
circ_0001829	hsa_circ_0001829	Up	Enhancer	Sponge miR-155-5p to upregulate SMAD	[50]
circOXCT1	hsa_circ_0004873	Down	Repressor	Sponge miR-136 to upregulate SMAD4	[51]
circAXIN1	hsa_circ_0005838	Up	Enhancer	Encode a novel protein, AXIN1-295aa	[52]
circFGD4	hsa_circ_0000390	Down	Repressor	Sponge miR-532-3p to upregulate APC	[53]
circREPS2	hsa_circ_0139996	Down	Repressor	Sponge miR-558 to upregulate RUNX3	[54]
circAKT3	hsa_circ_0000199	Up	Enhancer	Sponge miR-198 to upregulate PIK3R1	[55]
circ_0023409	hsa_circ_0023409	Up	Enhancer	Sponge miR-542-3p to upregulate IRS4	[56]
ciRS-7	hsa_circ_0001946	Up	Enhancer	Sponge miR-7 to upregulate PTEN	[57]
circTNPO3	hsa_circ_0001741	Down	Repressor	Interact with the RBP, IGF2BP3	[58]
circFNDC3B	hsa_circ_0006156	Up	Enhancer	Interact with the RBP, IGF2BP3	[59]
circ_100876	hsa_circ_0023404	Up	Enhancer	Sponge miR-665 to upregulate YAP1	[60]
circPRRX1	hsa_circ_0004370	Up	Enhancer	Sponge miR-665 to upregulate YWHAZ	[61]
circRanGAP1	hsa_circ_0063526	Up	Enhancer	Regulate the miR-877-3p/VEGFA axis	[62]
circ_0044366	hsa_circ_0044366	Up	Enhancer	Sponge miR-29a to upregulate VEGF	[63]
circURI1	hsa_circ_0000921	Up	Repressor	Interact with the splicing factor hnRNPM	[64]
ebv-circLMP2A	-	Up	Enhancer	Form a positive feedback loop with HIF1α	[65]
circNRIP1	hsa_circ_0004771	Up	Enhancer	Sponge miR-149-5p to upregulate AKT1	[66]
circNEK9	hsa_circ_0032683	Up	Enhancer	Sponge miR-409-3p to upregulate MAP7	[67]
circRELL1	hsa_circ_0001400	Down	Repressor	Sponge miR-637 to upregulate EPHB3	[68]
circSHKBP1	hsa_circ_0000936	Up	Enhancer	Modulate the miR-582-3p/HuR/VEGF axis and interact with HSP90	[69]
circMRPS35	hsa_circ_0000384	Down	Repressor	Recruit the histone modifier, KAT7	[70]
circMAPK1	hsa_circ_0004872	Down	Repressor	Encode a MAPK1-109aa protein	[71]
circRPL15	hsa_circ_0064574	Up	Enhancer	Sponge miR-502-3p to upregulate OLFM4	[72]
circUBE2Q2	hsa_circ_0005151	Up	Enhancer	Modulate the miR-370-3p/STAT3 axis	[73]
circAGO2	hsa_circ_0135889	Up	Enhancer	Interact with the RBP, HuR	[74]
circHuR	hsa_circ_0049027	Down	Repressor	Transcriptionally repression in cis	[75]

TGF-β/SMAD signaling pathway

The TGF-β/SMAD signaling is a classic pathway in cancer metastasis [47]. The circRNA circTHBS1, which is highly expressed in GC and associated with poor prognosis, is reported to promote the malignant behaviors and EMT of GC cells by triggering the INHBA/TGF-β pathway [48]. Mechanistically, circTHBS1 behaves as a miR-204-5p sponge to enhance the INHBA expression, and it also stabilizes the INHBA mRNA mediated by HuR, consequently activating the TGF-β pathway (Fig. 2AI) [48]. The circCCDC66 expression is elevated in GC and related to tumor stage and lymphatic metastasis [49]. Gain- and loss-of-function studies have revealed that circCCDC66 promotes GC metastasis by activating c-Myc and the TGF-β signaling pathways [49]. In another case, hsa_circ_0001829 promotes GC cell migration and invasion in vitro and GC metastasis in vivo via modulating the miR-155-5p/SMAD axis [50]. A similar ceRNA mechanism also applies to circOXCT1, which interacts with miR-136 to relieve the repressive effect on its target SMAD4, inhibiting GC EMT and metastasis [51].

Fig. 2

Molecular mechanisms of circRNAs related to GC metastasis. A. Roles of circRNAs in signaling pathways associated with EMT. I. The circRNA circTHBS1 increases the INHBA level via adsorbing miR-204-5p in a sponge form, and stabilizes the INHBA mRNA via sequestering HuR protein, leading to the activation of the TGF-β pathway. II. CircAXIN1-encoded a novel protein, AXIN1-295aa interacts with APC to activate the canonical Wnt/β-catenin signaling pathway. III. The circRNA circAKT3 activates the PI3K/AKT signaling by serving as a ceRNA against miR-198 to upregulate PIK3R1. IV. CircTNPO3 competitively binds to IGF2BP3, leading to the destabilization of the MYC mRNA, consequently repressing the expressions of MYC and its target SNAIL. V. The circFNDC3B level is significantly increased in GC and circFNDC3B interacts with IGF2BP3 protein and CD44 mRNA to form a ternary complex, resulting in the upregulation of CD44, which facilitates EMT in GC. VI. The circRNA circ_100876 interacts with miR-665 to relieve the repressive effect on its target YAP1, which is involved in the transcriptional activation of EMT-related genes. B. Roles of circRNAs engaged in angiogenesis. I. CircRanGAP1 is validated to stimulate angiogenesis via modulating the miR-877-3/VEGFA axis. II. CircURI1, a highly expressed circRNA in GC, sequesters the splicing factor, hnRNPM protein in a sequence-dependent manner to modulate alternative splicing of a subset of migration-related genes, such as VEGFA, consequently inhibiting GC metastasis. III. Ebv-circLMP2A promotes angiogenesis through forming a positive feedback loop with HIF1α to improve the VEGFA expression. Under hypoxia, HIF1α up-regulates ebv-circLMP2A, and ebv-circLMP2A interacts with KHSRP to destabilize the VHL mRNA, resulting in VHL down-regulation and HIF1α accumulation. C. Exosomal circRNA in GC. The circRNA circSHKBP1 promotes GC progression via the miR-582-3p/HuR/VEGF axis, and sequestering HSP90 to suppress STUB1-mediated HSP90 ubiquitination. Additionally, increased exosomal circSHKBP1 could facilitate co-cultured cell growth. D. Other pivotal pathways or targets involved in GC metastasis. I. The circRNA circMRPS35 inhibits GC tumorigenesis through the recruitment of histone acetyltransferase KAT7 to the promoters of FOXO1/3a genes, activating the FOXO1/3a transcription, consequently triggering the FOXO1/3a pathway. II. The circRNA circMAPK1 exerted an anti-tumor effect on GC invasion via generating a 109aa protein forming as a molecular sponge for MEK1, thus inhibiting the phosphorylation of MAPK1 and eventually leading to the inactivation of the MAPK pathway. III. The ciRNA circAGO2 interacts with HuR protein to promote its activation and enrichment on the 3’ UTR of HuR targets, resulting in repressing the AGO2/miRNA-mediated gene silencing involved in cancer progression. IV. The circRNA circHuR sequesters CNBP from the HuR’s promoter, leading to the repressions of HuR and GC progression

Wnt/β-catenin signaling pathway

The Wnt/β-catenin signaling pathway is indispensable among the pathways regulated by circRNAs in EMT [47, 52–54]. The circAXIN1 expression is significantly up-regulated in GC compared to the corresponding non-tumor gastric tissues [52]. Silencing of circAXIN1 suppresses GC cell proliferation, migration, and invasion, whereas the ectopic expression of circAXIN1 promotes GC malignancy in vitro and in vivo [52]. Mechanistically, a novel protein AXIN1-295aa encoded by circAXIN1 competes with parental AXIN1 protein to bind APC and release β-catenin, consequently activating the canonical Wnt/β-catenin signaling pathway to facilitate GC progression (Fig. 2AII) [52]. Additionally, Dai et al. have proposed that the circFGD4 expression is markedly attenuated in GC tissues and negatively correlated with lymphatic metastasis and the short prognosis of GC patients [53]. Furthermore, circFGD4 shows its anti-tumor effect on GC tumorigenesis and metastasis by modulating the miR-532-3p/APC/β-catenin axis [53]. Similarly, circREPS2 exhibits a decreased level in GC and inhibits GC migration and invasion via repression of the RUNX3/β-catenin pathway by sequestering miR-558 [54].

PI3K/AKT signaling pathway

The PI3K/AKT signaling pathway is frequently activated in EMT during metastasis and a series of dysregulated circRNAs have been found to interfere with this pathway [47, 55–57]. For example, GC-specific circAKT3 activates the PI3K/AKT signaling by repressing miR-198-mediated inhibition of PIK3R1, a regulatory subunit of PI3K (Fig. 2AIII) [55]. The circRNA hsa_circ_0023409 is highly expressed in GC tissues and markedly correlated with tumor size, histological grade, and TNM staging, nominating it as a potential prognostic marker for GC [56]. Functionally, hsa_circ_0023409 exerts the oncogenic effects on GC progression and metastasis by competitively sponging miR-542-3p to enhance the expression of IRS4, which contributes to activating the PI3K/AKT pathway [56]. A well-characterized circRNA, CDR1as (ciRS-7), is markedly up-regulated in GC and linked to poor survival in an independent validation cohort, and promotes GC cell migration and metastasis via antagonizing the miR-7-mediated expression of PTEN, which is broadly regarded as a negative regulator of the PI3K/AKT signaling pathway [57, 76].

Other pathways

Several additional circRNAs have been gradually characterized to engage in other EMT signaling pathways [58-61]. For example, circTNPO3 is significantly downregulated in GC compared with matched noncancerous tissues and plasma circTNPO3 owns the ability to serve as a potential diagnostic biomarker [58]. In vitro and in vivo observations reveal that circTNPO3 suppresses GC proliferation and metastasis [58]. Mechanistically, circTNPO3 competitively interacts with IGF2BP3 and subsequently destabilizes the MYC mRNA, ultimately inhibiting MYC and its target SNAIL, a primary and key inducer of EMT (Fig. 2AIV) [58]. The circRNA circFNDC3B appears to be increased in GC significantly and facilitates cell migration, invasion and EMT of GC cells by forming a ternary complex of circFNDC3B-IGF2BP3-CD44 mRNA (Fig. 2AV) [59]. In addition, circ_100876, a significantly up-regulated circRNA in GC, contributes to GC migration and invasion by serving as a molecular sponge for miR-665 to regulate the expression of YAP1, which activates a transcriptional program involved in EMT (Fig. 2AVI) [60].

Collectively, these findings strongly indicate that circRNAs can modify several critical biological pathways relevant to GC metastasis.

CircRNAs regulate angiogenesis

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 (VEGFAe7IN) [64]. CircURI1-induced VEGFAe7IN 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].

Exosomal circRNAs and GC metastasis

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.

Other metastasis-related pivotal pathways or targets

FOXO1/3a pathway

The FOXO1/3a pathway stimulates the expressions of the downstream targets, including p21, p27, Twist1, and E-cadherin [70, 90]. The circRNA circMRPS35 is identified from circRNA profiles of three paired GC and the corresponding non-tumor tissues, whose level is associated with clinicopathological characteristics and prognosis in GC patients [70]. Biologically, in vivo observations and in vitro experiments reveal that circMRPS35 inhibits GC cell proliferation and invasion [70]. Furthermore, mechanistic studies reveal that circMRPS35 combats GC tumorigenesis by recruiting histone acetyltransferase KAT7 to transcriptionally activate the FOXO1/3a genes, consequently triggering the FOXO1/3a pathway (Fig. 2DI) [70].

MEK-MAPK pathway

The MEK-MAPK signaling pathway is mainly involved in GC proliferation and metastasis [71, 91]. The circRNA circMAPK1 exhibits a decreased level in GC compared to the corresponding adjacent non-tumor tissues and is inversely correlated with GC tumor size, lymphatic invasion, TNM stage, and poor OS [71]. Functional investigations implicate that circMAPK1 suppresses GC proliferation and invasion in vitro and in vivo [71]. Mechanistically, circMAPK1 exerts the anti-tumor effect through encoding a MAPK1-109aa protein as a molecular sponge for MEK1, thus suppressing the phosphorylation of MAPK1 and eventually resulting in the inactivation of the MAPK pathway (Fig. 2DII) [71].

STAT3 pathway

Signal transducer and activator of transcription 3 (STAT3) is a widely-characterized oncogene in diverse human cancers [92, 93]. The circRNA circRPL15, up-regulated in GC tissues and correlated with short survival, enhances GC cell migration and invasion, and inhibits apoptosis by sequestering miR-502-3p from the OLFM4 mRNA to activate the STAT3 pathway [72]. A similar ceRNA mechanism also applies to circUBE2Q2, which interacts with miR-370-3p to relieve the inhibitory effect on its target STAT3 in GC, promoting proliferation, glycolysis, and metastasis [73].

Human antigen R

Human antigen R (HuR), a classic RBP, is frequently up-regulated in multiple human cancers including GC and plays a vital role in cancer progression and metastasis [94]. An intronic circRNA circAGO2 generated from the first intron of AGO2 is increased in GC and boosts GC metastasis in vitro and in vivo [74]. Mechanistic studies reveal that circAGO2 physically interacts with HuR protein to facilitate its activation and enrichment on the 3’ UTR of HuR targets, inhibiting AGO2/miRNA-mediated gene silencing associated with cancer progression (Fig. 2DIII) [74]. In another case, circHuR, predominantly localized in the nucleus, is downregulated in GC tissues and suppresses GC cell growth, invasion, and metastasis [75]. Mechanistically, circHuR interacts with CNBP and subsequently restrains its binding to the promoter of HuR, leading to the repressions of HuR and GC progression (Fig. 2DIV) [75].

Interplay between circRNAs and drug resistance in GC

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].

Table 2

CircRNAs involved in drug resistance in GC

CircRNA	CircBase ID	Drug	Expression	Drug resistance	Targets	Refs
circVAPA	hsa_circ_0006990	Cisplatin	Up	Enhance	miR-125b-3p, STAT3	[97]
circAKT3	hsa_circ_0000199	Cisplatin	Up	Enhance	miR-198, PIK3R1	[55]
circARVCF	hsa_circ_0092330	Cisplatin	Up	Enhance	miR-1205, FGFR1	[98]
circCCDC6	hsa_circ_0001313	Cisplatin	Up	Enhance	miR-618, BCL-2	[99]
circFN1	hsa_circ_0058147	Cisplatin	Up	Enhance	miR-182-5p	[100]
circPVT1	-	Cisplatin	Up	Enhance	miR-30a-5p, YAP1	[101]
circ_0000260	hsa_circ_0000260	Cisplatin	Up	Enhance	miR-129-5p, MMP11	[102]
circ_0032821	hsa_circ_0032821	Oxaliplatin	Up	Enhance	miR-515-5p, SOX9	[103]
circPVT1	-	Paclitaxel	Up	Enhance	miR-124-3p, ZEB1	[104]
circNRIP1	hsa_circ_0004771	5-fluorouracil	Up	Enhance	miR-138-5p, HIF-1α	[105]
circDLG1	hsa_circ_0008583	anti-PD-1	Up	Enhance	miR-141-3p, CXCL12	[106]
circCUL2	hsa_circ_0000234	Cisplatin	Down	Suppress	miR-142-3p, ROCK2	[107]
circMCTP2	hsa_circ_0000657	Cisplatin	Down	Suppress	miR-99a-5p, MTMR3	[108]
circ_0000144	hsa_circ_0000144	Oxaliplatin	Down	Suppress	miR-502-5p, ADAM9	[109]
circ_0000376	hsa_circ_0000376	Bupivacaine	Down	Suppress	miR-145-5p	[110]
circ_0000520	hsa_circ_0000520	Herceptin	Down	Suppress	PI3K-AKT pathway	[111]

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.

Clinical significance of circRNAs in GC

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.

Table 3

Clinical significance of circRNAs in GC (Cases more than 50)

CircRNA	CircBase ID	Sample	Expression	Clinicopathologic Features	Prognosis	Refs
circTHBS1	hsa_circ_0034536	Tissue	Up	Size, stage, grade, LNM	OS	[48]
circCCDC66	hsa_circ_0001313	Tissue	Up	Stage, LNM	-	[49]
circOXCT1	hsa_circ_0004873	Tissue	Down	Stage, LNM	OS	[51]
circAXIN1	hsa_circ_0005838	Tissue	Up	Stage, grade, LNM	-	[52]
circFGD4	hsa_circ_0000390	Tissue	Down	Grade, LNM	OS	[53]
circREPS2	hsa_circ_0139996	Tissue	Down	Size, stage, grade	-	[54]
circAKT3	hsa_circ_0000199	Tissue	Up	Size, stage, grade, chemoresistance	OS	[55]
circ_0023409	hsa_circ_0023409	Tissue	Up	Size, stage, grade	OS	[56]
ciRS-7	hsa_circ_0001946	Tissue	Up	Stage, LNM	OS	[57]
circTNPO3	hsa_circ_0001741	Tissue, plasma	Down	Differentiation	-	[58]
circ_100876	hsa_circ_0023404	Tissue	Up	Stage, LNM, BVI, LVI	DFS	[60]
circRanGAP1	hsa_circ_0063526	Tissue, plasma	Up	Size, stage, LNM	OS	[62]
circURI1	hsa_circ_0000921	Tissue	Up	Stage, tumor metastasis	-	[64]
ebv-circLMP2A	-	Tissue	Up	Stage, LNM, tumor metastasis	OS, DFS	[65]
circNRIP1	hsa_circ_0004771	Tissue	Up	Size, LNM	OS, DFS	[66]
circRELL1	hsa_circ_0001400	Tissue, plasma	Down	Stage, LNM, differentiation	OS, DFS	[68]
circSHKBP1	hsa_circ_0000936	Tissue	Up	Size, stage, vascular invasion	OS	[69]
circMRPS35	hsa_circ_0000384	Tissue	Down	Size, stage, LNM	OS	[70]
circMAPK1	hsa_circ_0004872	Tissue	Down	Size, stage, LNM	OS	[71]
circUBE2Q2	hsa_circ_0005151	Tissue, plasma	Up	Size, lymphatic invasion	-	[73]
circAGO2	hsa_circ_0135889	Tissue	Up	-	OS	[74]
circHuR	hsa_circ_0049027	Tissue	Down	Stage, tumor metastasis	OS	[75]
circVAPA	hsa_circ_0006990	Tissue	Up	-	-	[97]
circFN1	hsa_circ_0058147	Tissue	Up	Stage, grade, chemoresistance	-	[100]
circCUL2	hsa_circ_0000234	Tissue	Down	Stage, LNM, differentiation	OS	[107]
circMCTP2	hsa_circ_0000657	Tissue	Down	Size, stage, grade, chemoresistance	OS, DFS	[108]
circ_0110389	hsa_circ_0110389	Tissue	Up	Stage, differentiation	OS, DFS	[123]
circOSBPL10	hsa_circ_0008549	Tissue	Up	Stage, grade	OS, DFS	[124]
circPVT1	-	Tissue	Up	Stage, nervous invasion	OS, DFS	[125]

Conclusions and future perspectives

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.

Conclusions

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.