EIF4A3-Mediated CircRNA_100290 Promotes GC Cell Proliferation, Invasion and EMT via miR-29b-3p/ITGA11 Axis

Background Circular RNA (circRNA) has been reported as an important regulator in the development and progression of various carcinomas. However, the role of circRNA_100290 in gastric cancer (GC) is still unclear. This study aimed to investigate the role of circRNA_100290 in GC invasion and metastasis and its possible mechanism. Methods The expression of circRNA_100290 in GC cells and tissues were examined using quantitative real-time polymerase chain reaction (qRT-PCR). The role of circRNA_100290 in cell proliferation, migration, and invasion was evaluated on AGS and HGC-27 cell lines in vitro. Bioinformatics tools, dual-luciferase reporter assay, Western blot assay and qRT-PCR were used to explore the downstream pathways of circRNA_100290. The mechanism underlying the regulation of the expression of circRNA_100290 was explored using RNA immunoprecipitation, qRT-PCR, and Western blot assays. The expression of circRNA_100290 was found signicantly upregulated in GC cells and 102 GC tissues, high expression of circRNA_100290 in GC was closely related to Borrmann’s types, lymph node metastasis and tumor-node-metastasis staging. In vitro, knockdown of circRNA_100290 in AGS and HGC27 cells signicantly inhibited cell proliferation, migration, and invasion. Mechanistically, dual-luciferase reporter assay conrmed a direct binding between circRNA_100290 and miR-29b-3p, which targets ITGA11, an oncogene which is closely related to epithelial–mesenchymal transition (EMT). In addition, EIF4A3, one of RNA binding proteins (RBPs), could inhibit the formation of circRNA_100290 via enriching anking sites of circRNA_100290. Low expression of EIF4A3 in GC was related to a worse prognosis. and a tumor suppressor in GC. CircRNA_100290might serve as a potential biomarker and an effective target for GC diagnosis and therapy.


Results
The expression of circRNA_100290 was found signi cantly upregulated in GC cells and 102 GC tissues, high expression of circRNA_100290 in GC was closely related to Borrmann's types, lymph node metastasis and tumor-node-metastasis staging. In vitro, knockdown of circRNA_100290 in AGS and HGC-27 cells signi cantly inhibited cell proliferation, migration, and invasion. Mechanistically, dual-luciferase reporter assay con rmed a direct binding between circRNA_100290 and miR-29b-3p, which targets ITGA11, an oncogene which is closely related to epithelial-mesenchymal transition (EMT). In addition, EIF4A3, one of RNA binding proteins (RBPs), could inhibit the formation of circRNA_100290 via enriching anking sites of circRNA_100290. Low expression of EIF4A3 in GC was related to a worse prognosis.
Conclusions Elevated circRNA_100290 in GC promotes cell proliferation, invasion and EMT via miR-29b-3p/ITGA11 axi and might be regulated by EIF4A3. CircRNA_100290 might be a promising biomarker and target for GC therapy.

Background
Gastric cancer (GC) is the fth most frequently diagnosed malignancy and the third leading cause of cancer death worldwide 1 . According to Chinese cancer statistics, in 2014, the new cases of GC in China were approximately 410,000 and the deaths were approximately 294,000, which was second only to lung cancer 2 . Although abundant advances have been made in diagnostics and new therapeutic approaches of GC, a large number of patients with GC end up with poor prognosis. Therefore, appropriate molecular biomarkers of early diagnosis and potential treatment targets for GC need to be developed.
CircRNAs are a unique category of RNA molecules formed via back-splicing. They have neither 5′-3′ polarities nor polyadenylated tails, which were rst identi ed in plant viruses in the 1970s and now exist widely in eukaryotes 3,4 . Recently, various circRNAs have been found to participate in tumorigenesis and progression 5,6 . CircRNAs are suitable biomarkers of diseases usually because they own covalently closed-loop structures, are more stable than corresponding linear RNAs, and are insusceptible to the degradation of RNase R. In addition, circRNAs often exhibit speci c expression in various diseases and tissues 4,7,8 . More convincing evidence demonstrated that circRNAs were dysregulated in GC.
CircRNA_001569 and circPDSS1 were reported to be overexpressed in GC and accelerated GC progression by sponging miR-145 and miR-186-5p, respectively 9,10 . Circ_0027599/PHDLA1, circLARP4, and circPVRL3 functioned as tumor suppressors and inhibited the growth and metastasis of GC cells [11][12][13] . Previous studies reported that circRNA_100290 was upregulated and functioned as a miRNA sponge in oral squamous cell carcinoma and colorectal cancer 14,15 . However, the role of circRNA_100290 in GC is still unclear. In addition, splicing factors and RNA-binding proteins (RBPs) might regulate the formation of circRNAs via back-splicing 16,17 , but the regulatory mechanism of circRNA_100290 in GC is still unknown.
The present study was novel in demonstrating that circRNA_100290 was overexpressed in GC samples and cell lines. In vitro, silencing the expression of circRNA_100290 suppressed GC cell proliferation, induced G0/G1 phase arrest, and impeded migration, invasion, and EMT via miR-29b-3p/ITGA11 axis. The present study provides a promising biomarker and an effective target for GC treatment. Cell culture and transfection GC cell lines AGS, BGC-823, SGC-7901, and HGC-27 and human immortalized normal gastric epithelial cells GES-1 were provided and identi ed by Genechem Co., Ltd (Shanghai, China). Hsa_circ_100290 and EIF4A3 siRNA and paired negative control sequence were designed and synthesized by GenePharma (China). The si-CircRNA_100290 sequence of sense strand was 5′-CUCAUGCUUAGGCUUGAUUdTdT-3′; the sequence of the antisense strand was 3′-dTdTGAGUACGAAUCCGAACUAA-5′. The si-EIF4A3 sequence of sense strand was 5′-CGAGCAAUCAAGCAGAUCAdTdT-3′, and the sequence of the antisense strand was 3′-dTdTGCUCGUUAGUUCGUCUAGUdTdT-5′. AGS and HGC-27 cells were prepared for si-circRNA_100290 and si-EIF4A3 transfection using Lipofectamine reagent (GenePharma, China) according to the manufacturer's protocol. The knockdown e ciency was detected by quantitative real-time polymerase chain reaction (qRT-PCR) assay 48 h after transfection.

Plasmid transfection and luciferase assays
Luciferase reporter gene plasmid containing the 3′-UTR region or mutated 3′-UTR region of circRNA_100290 and miR-29-3p overexpression plasmid were constructed by GeneChem Co., Ltd (Shanghai, China). Then, 20 ng reporter construct and 80 ng miRNA expression plasmid, along with 4 ng Renilla luciferase plasmid, were co-transfected into H293T cells in a 96-well plate using jetPRIME transfection reagent (PolyPlus, France) as described by the manufacturer. The transfection e ciency was evaluated by uorescence microscopy. The luciferase activity was measured 48-h after transfection using the dual-luciferase reporter assay system as described by the manufacturer (Promega, USA).

Western blot
Western blot assay was conducted followed by previous report 18

RNA immunoprecipitation assay
The RNA-Binding Protein Immunoprecipitation (Millipore, USA) was used to perform a RIP assay.
According to the manufacturer's protocol, 1 × 10 7 cells were harvested and lysed in complete RIPA lysis buffer. RNA magnetic beads were conjugated to anti-EIF4A3 (Abcam, USA) or control anti-IgG (Millipore, USA). The Ct value of circRNA_100290 was detected by qRT-PCR.

Colony formation assay
Two hundred GC cells were seeded in six-well plates and incubated for 11 days. After being washed, xed, and stained with 0.01% crystal violet, the cell clones were imaged and evaluated using Quantity One software.
Cell counting kit (CCK-8) assay Two thousand GC cells were plated in 96-well plates. 10 µl CCK-8 (Solarbio, Beijing, China) reagent were seeded and incubated for 3 h at 37 °C atmosphere. Then, the absorbance at 450 nm was detected and recorded for ve consecutive days.
Flow cytometry analysis for cell cycle The cell xation was conducted using 70% ethanol for 24 h. Then, the cells were stained using PI and RNase reagent and incubated for 30 min at 37 °C. The cell cycle distribution analysis was performed using ow cytometry device (BD, USA).
Wound healing assay About 3 × 10 5 cells were plated into a 6-well dish. A linear scratch was made using a sterile 10-µL pipette tip when the cell convergence degree reached 90%. The cell scratch wound was washed with PBS and treated with RPMI1640 supplemented with 3% fetal bovine serum. The cell scratch wound was imaged under a microscope after 0, 24, and 48 h. Then, the wound healing rate was analyzed using ImageJ software.

Transwell migration and invasion assay
Transwell inserts (Corning, USA) were used to perform cell migration and invasion assays. The membrane was not coated with Matrigel (BD, USA) for cell migration assay. GC cells were added into the upper chamber and cultured in an incubator at 37 °C for 24 h. Then migrated cells were xed, stained with crystal violet and recorded. For the cell invasion assay, the membrane was coated with Matrigel and serum-free medium mixture (BD, USA), and the culture was incubated at 37 °C for 48 h.

Statistical analysis
GraphPad Prism 8.0.2 and SPSS 25.0 software was used for statistical analysis. Data were presented as mean ± standard error of the mean. The χ2 test, Student t test, and one-way analysis of variance were used for comparisons. Pearson correlation coe cient was calculated to measure correlation between factors. TA P value less than 0.05 was considered statistically signi cant.

Results
Expression of CircRNA_100290 was upregulated in GC tissues qRT-PCR was performed to examine the expression of circRNA_100290 in 102 paired GC tissues. The results showed that the expression of circRNA_100290 in GC was signi cantly higher than that in paired adjacent noncancerous tissues (Fig. 1A). Moreover, analysis of clinicopathological characteristics revealed that the expression of circRNA_100290 in GC tissues was closely related to Borrmann's types, lymph node metastasis and TNM stage (Table 1). Expression of miR-29b-3p decreased in GC and could be sponged by CircRNA_100290 CircRNAs customarily function as miRNA sponges to bind functional miRNAs. Hence, RNA22 V2.0 was used to predict miRNAs having potential binding sites with circRNA_100290 (Fig. 1B). Then, miR-29b-3p was chosen for further research. The qRT-PCR assay was conducted to examine the expression of miR-29b-3p in 31 matched GC tissues. The results showed a diminished expression of miR-29b-3p in GC tissues (Fig. 1C). Subsequently, the expression of miR-29b-3p in GC was validated by analyzing data downloaded from the EBI database which contained 184 GC tissues and 168 normal gastric epitheliums, and a same expression trend was observed (Fig. 1D). The correlation analysis results demonstrated a negative correlation between the expression of circRNA_100290 and miR-29b-3p in GC tissues (r = − 0.3656, P = 0.047; Fig. 1E). Additionally, luciferase reporter plasmids were transfected into H293T cells to further explore the relationship between circRNA_100290 and miR-29b-3p. The dual-luciferase reporter assay revealed that miR-29b-3p mimics reduced the luciferase activity in the wild-type group, indicating miR-29b-3p as a target for circRNA_100290 (Fig. 1F&G).These data suggested that miR-29b-3p might act as a tumor suppressor and circRNA_100290 could serve as a molecular sponge for miR-29b-3p in GC.

Expression of ITGA11 increased in GC and negatively correlated with the expression of miR-29b-3p in GC tissues
Starbase was used to predict the possible target genes of miR-29b-3p. ITGA11, one potential target gene, might have binding sites in 3′-UTR with miR-29b-3p ( Fig. 2A). The expression and function of ITGA11 in GC were not completely clear. qRT-PCR was conducted to examine the expression of ITGA11 in 31 GC tissues. The results showed that the expression level of ITGA11 in GC tissues was higher than that in paired noncancerous tissues (Fig. 2B). Data were downloaded from the TCGA database to further explore the role ITGA11 plays in GC. Clinicopathological characteristic analysis showed ITGA11 expression was closely related to Lauren's types, invasion depth and TNM stage (Table 2). Furthermore, the correlation analysis showed a negative correlation between the expression of miR-29b-3p and ITGA11 in GC (r = − 0.3168, P = 0.009; Fig. 2C). Ln. Lymph node Subsequently, the prognostic value of the expression of ITGA11 in 354 GC patients from the TCGA database was evaluated. ITGA11 high-expression group displayed a lower 5-year survival rates (Fig. 2D). High expression of ITGA11 predicted poor prognosis in GC. In addition, the protein-protein interaction (PPI) analysis on ITGA11 was conducted using the STRING software. Potential proteins interacting with ITGA11 included integrin family members, collagen family members, transforming growth factor family members, and so on. Gene ontology (GO) analysis revealed that the aforementioned proteins were mainly involved in migration-related biological processes, such as extracellular matrix organization, cell adhesion and migration, and so on (Fig. 2E).
CircRNA_100290 promoted proliferation, colony formation, and cell cycle distribution of GC via miR-29b-3p/ITGA11 axis The expression levels of circRNA_100290 and miR-29b-3p were assessed in four GC cell lines and GES-1.
The expression of circRNA_100290 was upregulated in AGS, BGC-823, and HGC-27 compared with GES-1, while miR-29b-3p was downregulated correspondingly (Fig. 3A&B). AGS and HGC-27 cell lines were chosen for further knocking down of circRNA_100290 as a higher circRNA_100290 expression. After knocking down circRNA_100290, the expression of miR-29b-3p and ITGA11 presented increasing and decreasing trends, respectively in both cells (Fig. 3C&D).
The CCK8 assay was performed to compare the cell viability of GC. For ve consecutive days, the absorbance value in si-circRNA_100290 AGS and HGC-27 cells was found to be obviously lower compared with that in the control groups, indicating that knocking down circRNA_100290 inhibited GC cell proliferation (Fig. 3E&F). The results of cell colony formation assay demonstrated that the colony numbers in si-circRNA_100290 AGS and HGC-27 cells were less than those in control groups (Fig. 3G-3J). Reduced circRNA_100290 suppressed the colony formation ability of GC cells. The cell cycle distribution was detected by ow cytometry. Decreased circRNA_100290 induced the increased percentage of the G0/G1 phase and downregulated the ratio of S and G2/M phases in si-circRNA_100290 AGS and HGC-27 cells (Fig. 3K-3N). The results suggested that knocking down circRNA_100290 might induce G0/G1 arrest and inhibit GC cell proliferation CircRNA_100290 accelerated GC cell migration and invasion through regulating EMT The cell wound healing and Transwell migration assays were performed to measure the effect of circRNA_100290 on cell migration and invasion abilities. Cell wound healing assay demonstrated that the wound healing rate was signi cantly lower in si-circRNA_100290 AGS and HGC-27 groups compared with the control groups (Fig. 4A). The statistical diagrams of AGS and HGC-27 groups showed that the most signi cant difference between the experimental and control groups was observed after 48 h (Fig. 4B&C).
Transwell migration assay showed that the impaired migration ability was observed in si-circRNA_100290 AGS and HGC-27 cells. The number of GC cells migrating into the lower chamber was less in si-circRNA_100290 groups compared with the control (Fig. 4D-4G). The results of Transwell invasion assay also revealed that the reduced expression of circRNA_100290 damaged the cell invasion ability. The number of AGS and HGC-27 cells invading into the lower chamber remarkably reduced after knocking down circRNA_100290 (Fig. 4H-4K). Moreover, Western Blot assay demonstrated that knocking down of circRNA_100290 induced the increased E-cadherin and reduced expression of N-cadherin and Vimentin (Fig. 4L).

EIF4A3 could bind anking region of circRNA_100290 and inhibited circRNA_100290 expression in GC
To nd out the molecular mechanism underlying the regulation of the expression of circRNA_100290, Circinteractome, one of the bioinformatics tools, was used to predict circRNA_100290 related RBPs.
EIF4A3 had the most potential binding sites matched with circRNA_100290 and its anking regions compared with other RBPs (Fig. 5A). Therefore, EIF4A3 was chosen for further study. RIP assay was performed to determine the binding between EIF4A3 and anking regions of circRNA_100290 in AGS and HGC-27 cells. The RIP-qPCR results showed 8.18-and 4.31-fold enrichment of the anking site of circRNA_100290 in AGS and HGC-27 cells, respectively (Fig. 5B). In addition, after the knocking down of EIF4A3, an increased level of circRNA_100290 was observed in both cells (Fig. 5C).
Expression of EIF4A3 was decreased in GC, and low-expressed EIF4A3 predicted poor prognosis of patients with GC qRT-PCR and Western blot assays were performed to assess the expression of EIF4A3 in 31 matched GC tissues. The results showed that the expression of EIF4A3 at both mRNA and protein levels was signi cantly downregulated in GC (Fig. 5D&E). Besides, the prognostic value of the expression of EIF4A3 in 876 patients with GC from the GEO database was evaluated by using the Kaplan-Meier plotter website. The results showed that the low expression of EIF4A3 was associated with worse overall survival (OS) and rst progression (FP). The median OS in the low-EIF4A3 expression group was 26.27 months, shorter than that in the high-EIF4A3 expression group (30.9 months). Comparably, the median FP in the low-EIF4A3 expression group was 14.1 months shorter than that in the high-EIF4A3 expression group (21.73 months) (Fig. 5F&G). Subsequently, clinicopathological analysis was conducted to explore the role EIF4A3 plays in GC by using datasets GSE62254 27 containing 300 GC patients from GEO database. Results revealed that the expression of EIF4A3 in GC was closely related to sex, age, Lauren's type, invasion depth, TNM staging, and ACRG genotyping ( Table 3). The data above suggested that EIF4A3 might serve as a suppressor in GC, and reduced expression of EIF4A3 predicted worse prognosis of patients with GC.

Discussion
CircRNAs, a new class of noncoding RNAs (ncRNAs), has gradually gained attention. It has been reported that exonic-originated circRNAs located mainly in the cytoplasm usually function as miRNA 'sponges' 6 .
CircRNAs also participate in transcriptional or post-transcriptional regulation 28,29 . Moreover, several circRNAs containing internal ribosome entry sites could be translated into peptides 30 . However, the acknowledged circRNAs and their regulatory mechanism in GC have not been thoroughly elaborated.
CircRNA_100290, one of circRNAs discovered recently, is located on chromosome 1, and its parental gene is SLC30A7. A recent study reported that circRNA_100290 was abnormally highly expressed in colorectal cancer and promoted the proliferation of colorectal cells 15 . In the present study, circRNA_100290 was signi cantly upregulated in GC, and was closely related to invasion depth, lymph node metastasis, and TNM stage. Functionally, silencing circRNA_100290 in AGS and HGC-27 cells signi cantly inhibited cell viability, colony formation, migration, and invasion ability, and induced the G0/G1 phase arrest in vitro. These results suggested that circRNA_100290 might serve as an oncogene in GC.
CircRNAs might function as miRNA sponges via binding miRNAs and regulate downstream target genes, also known as competing endogenous RNA regulatory mechanism. In our study, the expression of miR-29b-3p was found to be decreased in GC and this trend was validated by analyzing data from the EBI database. MiR-29b-3p is a member of miR-29 family, decreased expression of miR-29 family members has been reported in various tumors, including lung cancer, esophageal cancer, hepatocellular carcinoma, and so on [31][32][33] . The bioinformatic prediction and correlation analysis indicated miR-29b-3p might share the complementary binding sites with circRNA_100290, which was con rmed by the dual-luciferase reporter assay, suggesting that circRNA_100290 could function as a sponge for miR-29b-3p. Taken together, our results suggested that miR-29b-3p acted as a tumor suppressor and interacted with circRNA_100290 by sponging in GC.
ITGA11, a candidate target of miR-29b-3p, was further studied. ITGA11 is a member of the integrin family.
However, the role of ITGA11 in GC has not been reported so far. The present study reported an elevated expression of ITGA11 in GC, which was reversed to the expression of miR-29b-3p. Pathological factors analysis and survival analysis indicated that high expression of ITGA11 predicted a worse prognosis of GC patients. ITGA11 might serve as an oncogene in GC. Furthermore, silencing circRNA_100290 in AGS and HGC-27 cells led to the increased miR-29b-3p and diminished ITGA11, suggesting circRNA_100290 could sponge miR-29b-3p to increase the expression of ITGA11, thereby promoting GC cell proliferation, migration and invasion.
Growing evidence demonstrates that circRNAs usually regulate tumor progression and metastasis by affecting EMT 37,38 . In our study, knocking down of circRNA_100290 induced altered expression of several EMT markers, accompanied by the release of miR-29b-3p and blocking of ITGA11. Previous studies have reported the involvement of miR-29b family members in EMT 39 . In addition, there is report reveals that miR-29b could inhibit EMT and metastasis by targeting a network of pro-metastatic motivators involved in angiogenesis, collagen remodeling, and proteolysis 40 . Shin et al. reported that exogenous miR-29b mediated an anticancer effect by impeding the activation of ITGA11 36 . In our study, by using PPI and GO analyses, ITGA11 was found to have a close connection with EMT-related proteins and could be involved in EMT. These results re ected that circRNA_100290 promoted EMT mainly via the miR-29b-3p/ITGA11 axis.
CircRNAs are formatted via back-splicing regulated by RNA splicing factors or RBPs 41,42 . In this study, EIF4A3 was found to have the most predicted binding sites among other RBPs with both anking regions and circRNA_100290 itself, and RIP assay con rmed the direct interaction between anking regions of circRNA_100290 and EIF4A3. As a member of the DEAD-box protein family, EIF4A3 is located mainly in the nucleus, and is a part of the exon junction complex necessary for nonsense-mediated mRNA decay 43 .
In addition, EIF4A3 is also involved in various biological processes, including mRNA translation initiation and RNA splicing 44,45 . It was inferred that EIF4A3 was probably involved in the transcriptional regulation of circRNA_100290. Wang

Conclusions
In conclusion, our study reveals that the expression of circRNA_100290 was upregulated in GC. Further, the knockdown of circRNA_100290 signi cantly inhibited GC cell proliferation, migration, and invasion in vitro. The potential mechanism could be that circRNA_100290 regulated EMT via targeting miR-29b-3p/ITGA11. In addition, EIF4A3 could serve as a negative regulator of circRNA_100290 and a tumor suppressor in GC. CircRNA_100290might serve as a potential biomarker and an effective target for GC diagnosis and therapy.

Declarations
Ethics approval and consent to participate The studies involving human participants were reviewed and approved by the ethics committee of The First A liated Hospital of China Medical University. The patients/participants provided their written informed consent to participate in this study.

Consent for publication
Not applicable.

Availability of data and materials
The data that support the ndings of this study are available from the corresponding author upon reasonable request.

Competing interests
The authors declare that they have no competing interests. Authors' contributions GW and DS were responsible for performing the experiments and drafted the manuscript. GW and Wh L were responsible for acquisition and analysis of data. GW and DS provided and collected the clinical data. YX were responsible for designing the experiments and supervising the study. All authors read and approved the nal manuscript.