METTL14-mediated m6A modification of circORC5 suppresses gastric cancer progression by regulating miR-30c-2-3p/AKT1S1 axis

N6-methyladenosine (m6A) RNA methylation and circular RNAs (circRNAs) have been shown to act vital roles in multiple malignancies including gastric cancer (GC). However, there is little knowledge about how m6A modification of circRNAs contributes to GC progression. The association of METTL14 expression with the clinicopathological characteristics and prognosis in patients with GC was assessed by Western blot, Immunohistochemistry and public datasets. In vitro and vivo function experiments were conducted to investigate the role of METTL14 in GC. Furthermore, m6A-circRNA epitranscriptomic microarray was utilized to identify METTL14-mediated m6A modification of circRNAs, which were validated by methylated RNA immunoprecipitation (Me-RIP), RT-qPCR and rescue experiments in GC cells. The sponge of circORC5 with miR-30c-2-3p was confirmed by luciferase gene report and RNA immunoprecipitation assays. The expression, localization and prognosis of circORC5 in GC were evaluated by fluorescence in situ hybridization. The effects of METTL14 and (or) circORC5 on miR-30c-2-3p-mediated AKT1S1 and EIF4B were estimated by RT-qPCR and Western blot analyses. We found that METTL14 was downregulated in GC tissue samples and its low expression acted as a prognostic factor of poor survival in patients with GC. Ectopic expression of METTL14 markedly repressed growth and invasion of GC cells in vitro and in vivo, whereas knockdown of METTL14 harbored the opposite effects. Mechanically, m6A-circRNA epitranscriptomic microarray and Me-RIP identified circORC5 as the downstream target of METTL14. Silencing of METTL14 reduced the m6A level of circORC5, but increased circORC5 expression. Moreover, circORC5 could sponge miR-30c-2-3p, and reverse METTL14-caused upregulation of miR-30c-2-3p and downregulation of AKT1S1 and EIF4B. In addition, circORC5 possessed a negative correlation with miR-30c-2-3p and indicated a poor survival in GC. Our findings demonstrate that METTL14-mediated m6A modification of circORC5 suppresses gastric cancer progression by regulating miR-30c-2-3p/AKT1S1 axis.

in China [2]. Recent decades have witnessed the great progress in the treatment of GC including endoscopic resection, targeted therapy, and immunotherapy [3]. But, the prognosis of advanced patients is still poor owing to tumor invasion and metastasis. Thus, comprehending the molecular mechanisms of tumorigenesis is of importance to the early diagnosis and treatment of GC.

Clinical data
Clinical and pathological data for 385 cases of GC patients and 33 paired tumor tissue samples were downloaded from The Cancer Genome Atlas database (http:// xena. ucsc. edu/ getti ng-start ed/). Clinicopathological characteristics including age, sex, stage, pathological stage, Tumor-Node-Metastasis (TNM) stage, survival and recurrence were also collected. 10 paired GC tissues were stored in liquid nitrogen and frozen at − 80 °C. A human tissue microarray containing 90 paired tumor tissues from GC patients (Lot No. XT14-008) was purchased from the Shanghai Outdo Biotech Company (Shanghai, China). Our study protocol was approved by the Ethics Committee of Shanghai Sixth People's Hospital.

Immunohistochemical analysis
The tissue microarray was deparaffinized, rehydrated, and microwaved-heated in sodium citrate buffer (10 mmol/L, pH6.0) for antigen retrieval. Then, the slides were incubated with Mouse anti-METTL14 monoclonal antibody (1:100, Lot No. CABT-B9471, NY, USA) by SABC (mouse IgG)-FITC immunohistochemical Kit (Lot. No. A0130, Wuhan, China). The protein expression of METTL14 (H score) was assessed by two independent pathologists. H-score = ΣPi(i + 1), where "Pi" represents the percentage of positive cells in all cells in the section, and "i" stands for coloring intensity.

RNA extraction and real-time quantitative PCR (RT-qPCR)
Total RNA was extracted using a RNA extraction kit (77,064, QIAGEN) and cDNA synthesis was performed using a reverse transcription kit (Promega, Madison, USA) according to the manufacturer's instructions. PCR was conducted using the SYBR Green Master Mix (Q111-02, VAZYME). After the reactions were completed, relative gene expression level was calculated using the 2 −ΔΔCt . The transfection efficiency of the vectors was defined by the extent of downregulation or upregulation of the target gene. The primer sequences used were indicated in Supplementary Table S1.

Human m 6 A-circRNA epitranscriptomic microarray
Total RNA from each sample was quantified using the NanoDrop ND-1000. The sample preparation and microarray hybridization were performed based on the Arraystar's standard protocols. Briefly, the total RNAs were immunoprecipitated with anti-m 6 A antibody (Nano materials intestinal injury). The modified RNAs were eluted from the immunoprecipitated magnetic beads as the "IP". The unmodified RNAs were recovered from the supernatant as "Sup". The "IP" and "Sup" RNAs were treated with RNase R, and then labeled with Cy5 and Cy3 respectively as cRNAs in separate reactions using Arraystar Super RNA Labeling Kit. The cRNAs were combined together and hybridized onto Arraystar Human circRNA Epi-transcriptomic Microarray. After washing the slides, the arrays were scanned in two-color channels by an Agilent Scanner G2505C.

In vivo tumor growth assay
Male nude mice (6 weeks old) were purchased from the Shanghai Laboratory Animal Central (SLAC, Shanghai, China). SGC-7901 cells (1 × 10 7 ) transfected with the METTL14 overexpression vector or no-load lentivirus vector were resuspended in 200μL of sterile PBS and injected subcutaneously into the right flanks of mice. After 4 weeks, the mice were sacrificed, and the xenografted tumors were collected for hematoxylin-eosin (HE) staining and IHC analysis. Tumor volume was calculated using formula: volume = (length x width 2 )/2. The animal experiments were approved by the Ethics Committee of Shanghai Sixth People's Hospital.

Statistical analysis
Statistical analysis was performed with GraphPad Prism 7 (La Jolla, CA, USA). In brief, the values are expressed as the mean ± standard deviation (SD). Student's test and analysis of variance were used for comparisons between groups. Kaplan-Meier analysis was used to assess the association of circORC5 or miR-30c-2-3p with GC prognosis. A Cox proportional hazard model was used to assess the risk of circORC5 or miR-30c-2-3p in GC. Pearson Correlation Analysis was used to analyze the correlation of circORC5 with miR-30c-2-3p. The categorical data were analyzed by chi-square of Fisher's exact tests. P < 0.05 was considered statistically significant.

Downregulation of METTL14 predicted poor prognosis in patients with GC
To investigate the potential role of METTL14 in GC, we first analyzed the RNA sequencing data from TCGA, which showed decreased METTL14 expression in human GC tissue samples relative to normal tissue (Fig. 1A, P < 0.05). We further detected the protein levels of METTL14 in 10 paired GC tissue samples by Western blot, which indicated that METTL14 was remarkably downregulated in GC compared to paired normal samples (Fig. 1B, C, P < 0.0001). Tissue microarray containing 90 pairs of cancerous and matched normal tissue was analyzed by IHC staining, which further validated the downregulation of METTL14 in GC tissues (Fig. 1D, Then, we found that downregulation of METTL14 showed no association with clinicopathological characteristics including pathological stage and TNM stage in patients with GC (Supplementary Table S2), however, Kaplan-Meier analysis uncovered that the patients with low-METTL14 expression indicated poor overall survival as compared with those with high-METTL14 expression (Fig. 1E, P = 0.042). Moreover, independent two cohorts were established by extracting TCGA and clinical data of GC patients from GSE22377. In coinciding with our results, low expression of METTL14 harbored worse survival in GC (Fig. 1E, P < 0.01). In addition, multivariate analysis revealed that METTL14 low expression as well as age and pathological stage was a risk factor for poor survival in patients with GC from TCGA cohort  Table S3, P = 0.005). These results suggested that METTL14 was a reliable prognostic factor in patients with GC.

Knockdown of METTL14 facilitated proliferation and invasion of GC
Given the fact that METTL14 expression is decreased in GC tissues, we speculated that METTL14 may function as a tumor suppressor in GC. We detected the mRNA and protein levels of METTL14 in normal gastric epithelial cell line (GES1) and GC cell lines (SGC-7901, MKN-28, BGC-823, AGS and MGC-803) by RT-qPCR and Western blot, and found that METTL14 possessed high expression in AGS and MGC-803, but low expression in SGC-7901 ( Fig. 2A). Therefore, to confirm the roles of METTL14 in GC, we established METTL14-kockdown cell model in AGC and MGC-803 cells with siRNA, and METTL14 overexpression cell model in SGC-7901 cells with plasmids. The transfection efficiency was determined by RT-qPCR and Western blot (Fig. 2B). The cell viability and proliferative capabilities of GC cells were then assessed by MTT and colony formation assay. As expected, knockdown of METTL14 remarkably promoted cell viability (Fig. 2C) and proliferative potential (Fig. 2D). In addition, Transwell invasion assay indicated that downregulation of METTL14 dramatically facilitated invasion capabilities of GC cells (Fig. 2E). Likewise, overexpressed METTL14 exhibited inhibitory effects on cell viability (Fig. 2C), colony formation (Fig. 2D) and invasion capabilities (Fig. 2E) relative to the control group.

METTL14 acted by m6A-dependent modification of circORC5
To elucidate the underlying mechanisms of METTL14 in GC progression, we first detected the effects of METTL14 on the m 6 A level in MGC-803 and AGS cells by MeRIP, which showed that the m 6 A level was markedly decreased in METTL14-knockdown MGC-803 and AGS cells (Fig. 3A). Then, m 6 A-circRNA epitranscriptomic microarray revealed that 444 circRNAs increased in m 6 A levels but 454 decreased in m 6 A levels in METTL14-knockdown MGC-803 cells compared with the control group (Fig. 3B), among which, hsa_ circ_0030632, hsa_circ_0047481 and hsa_circ_007612 were the top three downregulated circRNAs in m 6 A levels in METTL14-knockdown cells (Fig. 3C). MeRIP-PCR further validated that the m 6 A level of hsa_ circ_0007612 (circORC5) rather than hsa_circ_0030632 and hsa_circ_0047481 was decreased by knockdown of METTL14 in MGC-803 and AGS cells (Fig. 3D), whereas RT-qPCR analysis showed that circORC5 mRNA level was increased by METTL14 knockdown in MGC-803 and AGS cells (Fig. 3E).
The transfection efficiency of si-circORC5 in MGC-803 and AGS cells was determined by RT-qPCR analysis ( Supplementary Fig. S1). Functionally, we found that circORC5 depletion reduced colony formation and invasion capabilities, and counteracted si-METTL14 induced colony formation and invasion potential in MGC-803 and AGS cells (Fig. 3F, G). These results suggested that METTL14 acted by m 6 A-dependent modification of circORC5.

Identification and characteristics of circORC5 in GC
According to the circRNA annotation in Circular RNA interactome (https:// circinteractome.nia.nih.gov/index. html), hsa_circ_0007612 is derived from the linear gene origin recognition complex subunit 5 (ORC5) and termed as circORC5 (Fig. 4A). Compared with linear ORC5, cir-cORC5 harbored higher stability after treatment with RNase R exonuclease (Fig. 4B). Cytoplasmic and nuclear RNA analysis showed that circORC5 was preferentially localized in the cytoplasm in MGC-803 and AGS cells (Fig. 4C). FISH further validated that green fluorescent distribution of circORC5 was mainly in the cytoplasm of GC (Fig. 4D), and circORC5 had increased expression in GC tissues compared with the adjacent normal (Fig. 4E). Kaplan-Meier unveiled that GC patients with circORC5 high expression harbored poorer survival as compared with those with circORC5 low expression (Fig. 4F).

CircORC5 had a negative correlation with miR-30c-3p in GC
According to m 6 A-circRNA profiling and miRbase, cirORC5 was identified to have the potential to bind with five miRNAs (Fig. 5A). We analyzed the expression levels of these five miRNAs in GC and found that miR-30c-2-3p possessed most significant decrease in 387 unpaired and 41 paired GC tissues (Fig. 5B, P < 0.0001). FISH analysis further validated that miR-30c-2-3p expression levels were markedly downregulated and negatively correlated with circORC5 in GC tissues (Fig. 5C, D). FISH analysis indicated that circORC5 was also localized in the cytoplasm of GC tissue (Fig. 5E).

METTl14 inhibited in vivo tumor growth
To elucidate whether METTl14 suppressed in vivo tumor growth, we established a METTl14 or NC stably transfected SGC-7901 cell line, which was then subcutaneously injected into the flank of nude mice. After an observation for 33 days, we found that, the volumes of xenograft tumors induced by METTl14 transfected SGC-7901 cells were smaller than those by NC transfected cells (Fig. 8A). The growth curve demonstrated that, the tumors in METTl14 transfected group presented a reduction in a time dependent manner (Fig. 8B), and the tumor volume and weight were alleviated in METTl14 transfected group compared with the NC group (Fig. 8C). HE and IHC staining indicated that the tumor proliferation marker Ki-67 was downregulated in METTl14 transfected group compared with the NC group (Fig. 8D).

Discussion
Accumulating data indicate that METTL14 as a m 6 A "writer" shows a decreased expression in CRC [7,26], bladder cancer [8] and breast cancer [9], but an increased level in thyroid and pancreatic cancers [10,11]. Downregulation of METTL14 predicts a poor prognosis in patients with CRC and breast cancer [7,9,26]. METTL14 has been reported to be downregulated in GC [28], but its clinical implication in patients with GC is unknown. We herein found that, in coinciding with previous studies, METTL14 levels were lowered in GC and low expression of METTL14 was a prognostic factor of poor survival in patients with GC. According to the previous studies, the loss of METTL14 in tumor-associated macrophages promotes tumor growth [29] and SUMOylation of the m 6 A-RNA methyltransferase influences its function [30]. These reports may explain why METTL14 is downregulated in cancer but need be further studied in GC. Functionally, METTL14 has dual roles in cancer. It can repress growth, invasion and metastasis in multiple malignancies [6][7][8][9]26], but accelerate tumor progression [10,11]. A previous study showed that METTL14 could suppress the proliferation and invasion by inhibiting the PI3K/AKT/mTOR signaling [28]. Herein, we also found Mechanically, METTL14 participates in tumorigenesis not only by m 6 A-dependent modification of mRNAs [7,8,11], but also by modification of miR-375 or lncRNA XIST [6,26]. Previous report indicated that circGFRα1 mediated by METTL14 facilitates self-renewal of female germline stem cells [31]. Herein, m 6 A-circRNA profiling and Me-RIP were utilized to identify that METTL14 could exhibit m 6 A-dependent modification of circORC5 in GC cells. Downregulation of METTL14 decreased the m 6 A level of circORC5, but increased circORC5 expression. Knockdown of circORC5 repressed the colony formation and invasive capabilities, and attenuated METTL14 loss-induced tumor promoting effects in GC cells. Our findings unveiled that METTL14 might act by m 6 A-dependent modification of circORC5. It is known that circRNAs act as the sponges of miR-NAs implicated in GC progression [17][18][19][20][21][22]. Our previous studies indicated that circLARP4 or circDLST can act as a sponge of miR-424-5p or miR-502-5p to impact growth and invasion of GC [23,25]. Herein, we found that cir-cORC5 was upregulated in GC tissue samples and associated with poor survival in patents with GC. CircORC5 could bind with miR-30c-2-3p and decrease its expression in GC cells. It has been reported that miR-30c-2-3p is downregulated in pancreatic ductal adenocarcinoma [32] and breast cancer [33], and represses cell proliferation and cycle progression [33,34]. We also found that miR-30c-2-3p was downregulated in GC tissue samples, possessed a negative correlation with circORC5 expression and reversed circORC5-induced cell proliferation in GC cells. METTL14 deficiency decreased miR-30c-2-3p expression while downregulation of circORC5 increased its expression. LINC00346 and hsa_circ_0072995 can also sponge miR-30c-2-3p to promote tumor growth [35,36]. Our findings uncovered that circORC5 mediated by METTL14 promoted GC growth by sponging miR-30c-2-3p.
It has been shown that RAB31, member RAS oncogene family as a target of miR-30c-2-3p favors GC Fig. 7 METTL14 mediated circORC5 to regulate miR-30c-2-3p/AKT1S1 axis. A RT-qPCR analysis of the effects of transfection with si-METTL14 and (or) si-circORC5 on miR-30c-2-3p expression in MGC-803 and AGS cells. B The luciferase activity of the WT luc-AKT1S1 or luc-AKT1S1 after transfection with miR-30c-2-3p mimics in MGC-803 and AGS cells. C The luciferase activity of the WT luc-EIF4B or luc-EIF4B after co-transfection with miR-30c-2-3p mimics in MGC-803 and AGS cells. D RT-qPCR E Western blot analysis of the effects of transfection with miR-30c-2-3p mimic on AKT1S1 and EIF4B expression in MGC-803 and AGS cells. F Western blot analysis of the effects of transfection with si-METTL14 and (or) si-circORC5 on AKT1S1 and EIF4B expression in MGC-803 and AGS cells. Data are the means ± SEM of three experiments. *P < 0.05; **P < 0.01; ***P < 0.001 cell proliferation [37]. We also identified AKT1S1 and EIF4B as the direct targets of miR-30c-2-3p in GC cells. AKT1S1 is upregulated in HCC, associates with poor prognosis in patients with HCC and promotes HCC growth [38]. ElF4B-PI3K-AKT signaling promotes proliferation and inhibits apoptosis in nasopharyngeal carcinoma [39]. Herein, we found that miR-30c-2-3p or knockdown of circORC5 decreased the expression of ElF4B and AKT1S1 and circORC5 deficiency reversed the promoting effects of METTL14 knockdown on ElF4B and AKT1S1 expression in GC cells. Our findings indicated that, METTL14 suppressed GC growth by increasing the m 6 A level of circORC5 and decreasing its expression, and circORC5 sponged miR-30c-2-3p and upregulated ElF4B and AKT1S1, contributing to GC tumorigenesis (Fig. 9).