Human GC specimens and adjacent non-tumor tissue samples were obtained between October 2014 and March 2016 from surgical specimens of patients with GC undergoing surgery at The Affiliated Huaian No.1 People’s Hospital of Nanjing Medical University. The samples were snap−frozen in liquid nitrogen and stored at -80˚C for the next experiments. We obtained written informed consent from all of these patients. This study was approved by the Research Ethics Committee of The Affiliated Huaian No.1 People’s Hospital of Nanjing Medical University.

Human GC cells, including BGC-823(CBP60477), HGC-27(CBP60480), MKN-45(CBP60432), AGS(CBP60476), SGC-7901(CBP60500), HGC-27(CBP60480) and immortalized normal gastric epithelial cell line GES-1(CBP60512), were grown in RPMI 1640 medium (#2381, Haihu Technology, Haidian, Beijing, China) (with 10% fatal bovine serum(FBS) and 1% antibiotics) and obtained from Nanjing COBIOER Biotechnology Co., Ltd. (Nanjing, Jiangsu, China). Regarding cell transfection, LipoFectMax reagent kits (HYY Medicine, Guangzhou, Guangdong, China) were adopted to transfect miR-1224 mimics (MOI 0.01), miR-1224 inhibitors (MOI 0.01), negative control (NC) inhibitors (MOI 0.01), or negative control (NC) mimics (MOI 0.01) into SGC-25 and SGC-7901 cells. For the analysis of viability 200 cells per concentration weach concentration used were scored. The miRNA mimics and inhibitors used in this study were synthesized by BioTend Biotechnology Co., Ltd. (Huangpu, Shanghai, China).

Total RNA was extracted from specimens and cells utilizing the TRIzol reagent (PuliKe Biotech, Shapingba, Chongqing, China). Subsequently, qRT-PCR analysis for mRNA (SATB homeobox 1 (SATB1), c-myc, cyclin D1, β-catenin) and miR-1224 examination (Primers were summarized in Table 1 ) were conducted using Cells Direct one step qRT-PCR kit (Thermo Fisher, Pudong, Shanghai, China) and RiBoBio miDETECT miRNA qRT-PCR Starter kit (RiBoBio, Guangzhou, Guangdong, China), respectively, based on an ABI QuantStudio Dx real-time PCR system (Daan Gene, Guangzhou, Guangdong, China) (30). Data were normalized to GAPDH or U6 expression and calculated with the 2^−△△Ct methods.

CCK-8 assays kits were obtained from ExcellBio Technology Co., Ltd. (Taicang, Jiangsu, China). Briefly, a total of 5 × 10³ SGC-25 or SGC-7901 cells (per well) were placed into 96-well plates after the transfection of miR-1224 mimics for 24 h. Afterward, the CCK-8 reagent (10 μl) was added at the indicated time. Sequentially, a DNM-9606 microplate reader instrument (PerLong Medical, Shunyi, Beijing, China) was employed to detect the OD value at the absorbance of 450 nm.

An EdU Apollo Imaging kit (RiboBio, Guangzhou, Guangdong, China) was employed to conduct the EdU assays (31). Specifically, the miR-1224 mimics or negative control (NC) mimics-transfected SGC-25 or SGC-7901 cells (about 4000 cells per well) were planted into 48-well plates. Twenty-four hours later, EdU solution (50 μM final concentration) was added to each well. The treated cells were maintained for an additional 2 hours and added with DAPI solution (32). Finally, the cells were fixed with formaldehyde (4%) and observed using fluorescence microscopy.

BGC-823 and SGC-7901 cells transfected with miR-1224 mimics or NC mimics were placed into 6-well plates (500 cells per well) and maintained in RPMI-1640 medium (10% FBS) for about 2 weeks. Subsequently, PBS was used to wash the cell colonies, and crystal violet solution (0.2%; SNOTA Biotech, Shenyang, Liaoning, China) was utilized to stain the colonies. Finally, the stained colonies were counted by a microscope (YDF-90; YuanRen, Pudong, Shanghai, China).

First, protein lysates, which were extracted from corresponding miRNA mimics-transfected cells using RIPA buffer, were separated by 8-12% SDS-PAGE. Then, the isolated proteins were transferred to Millipore polyvinylidene fluoride membranes. Thereafter, the primary antibodies targeting proteins were prepared and incubated with the membranes. Next, the examination of these proteins was performed using an ECL assay kit (TellGen, Pudong, Shanghai, China) after the corresponding secondary antibodies were probed with the membranes. The primary antibodies targeting Vimentin (1: 1000, ab92547, Cambridge, UK), N-cadherin (1:500, BD Biosciences, 610,921), β-Catenin(1: 1000, ab68183, Abcam, Shenzhen, China), cyclin D1(1: 1000, ab16663, Abcam, Shenzhen, China) and c-myc(1: 1000, ab32072, Abcam, Shenzhen, China) and GAPDH(1: 1000, ab8245, Abcam, Shenzhen, China) for this experiment were all purchased from NeoBioscience Biotechnology Co., Ltd. (Shenzhen, Guangdong, China).

Collected cells were planted into 6-well plates and subsequently grown to 70-80% confluence, followed by transfecting miR-1224 mimics or NC mimics. After culturing for 24 h, the monolayer GC cells were scraped by a sterile 200 µl pipette tip (Xinlong Biology, Xuhui, Shanghai, China). The cells were washed, and pictures of the wounded areas at 0 h and 48 h were taken using a microscope (YDF-90; YuanRen, Pudong, Shanghai, China).

The invasiveness of cells after treatment was conducted using BD Biosciences 8μm-pore-size Invasion Chambers (MingYang Biotech, Chaoyang, Beijing, China) pre-treated with Matrigel. Cells (1 × 10⁴) in RPMI-1640 medium were placed into the upper chamber while 700 μl 20% FBS was added into the lower chamber. Then, cells invading through the insert membranes were dyed with crystal violet (0.2%; SNOTA Biotech, Shenyang, Liaoning, China). The images were taken using a microscope (YDF-90; YuanRen, Pudong, Shanghai, China).

TargetScan was adopted to predict the putative target genes of miR-1224. Luciferase reporter assays were performed following the instructions. The 3’UTR sequence of special AT-rich binding protein 1 (SATB1) predicted to interact with miR-1224 or the mutated sequence was synthesized by Gema Technology (Pudong, Shanghai, China) and then inserted into pGL3-promoter Luciferase vector. Besides, successful clones were determined with sequencing technology. The 293T cells were seeded into 6-well plates for 1 day and co-transfected with 20 ng of the above-synthesized vectors and 20 nM of miR-1224 mimics or control mimics applying Lipofectamine 2000 (Invitrogen, Beijing, Haidian, China). Finally, the luciferase activity 48 hours after transfection was determined using the dual-luciferase reporter assay system (Promega, Hangzhou, Zhejiang, China).

In nude mice of Balb/c (male, 4-week-old) (n = 6), the tumor growth of SGC7901 cells in vivo was explored. About 5 × 10⁶ cells were transfected with miR-1224-overexpression. After five days of the above injection, tumor growth was examined every five days. Our group slaughtered the mice on the thirtieth day and then extracted samples from two groups. The width and length of the tumor were measured at the indicated time, and the bodyweight of mice was recorded after the samples were collected. The animal-related protocol was approved by the Animal Ethics Committee of The Affiliated Huaian No.1 People’s Hospital of Nanjing Medical University.

The gene ontology (GO) analyses (including BP: biological process, CC: cellular components and MF: molecular functions) and biological pathways analyses were conducted with FunRich software(http://www.funrich.org/). The interaction network of DE miRNAs-potential target genes was generated by the miRNet algorithm (33). The potential miRNAs target genes were predicted by starbase, TargetScan, and miRDB online algorithms (34–36). The potential transcription factors (TFs) for miR-1224 were predicted through FunRich software(http://www.funrich.org/). Additionally, the potential-TFs-miR-1224 interaction network was generated by Cytoscape software(https://cytoscape.org/).

All the statistical analysis was conducted using SPSS19.0 for Windows (SPSS, Inc., Chicago, IL, USA). The experiment was repeated three times, and each experiment contained three replicates. Furthermore, a student’s t-test was performed to analyze the differences between the two groups. Data from more than two groups were compared using one-way ANOVA. Kaplan-Meier methods were employed for the overall survival curve, and the differences were tested by the log-rank test. Besides, survival data were assessed using a Cox proportional hazards model. p < 0.05 indicated statistical significance.

RT-PCR was performed to explore whether miR-1224 was dysregulated in GC. It was observed that miR-1224 expression in GC specimens dramatically decreased compared with matched normal specimens ( Figure 1A ). Besides, the decreased expressions of miR-1224 were detected in five GC cells ( Figure 1B ). These findings verified that miR-1224 expressions were reduced in GC tumors.

Then, whether miR-1224 was associated with the clinical progression of GC patients was further explored. Based on the relative miR-1224 expressions in 128 samples, 128 patients were classified into two groups: high-group (n = 63) and low-group (n = 65). The associations between miR-1224 expressions and clinicopathological parameters in GC were summarized in Table 2 . It can be found that low miR-1224 expression was significantly associated with the TNM stage (p = 0.007). Moreover, the Kaplan-Meier survival assay suggested that the low expressions of miR-1224 were positively correlated with poorer overall survival (p = 0.0086) and disease-free survival (p = 0.0035) of GC patients ( Figures 1C, D ). Finally, the independence of miR−1224 was assessed using multivariate analyses. miR−1224 was a potential independent prognosis factor in GC (p < 0.05, Table 3 ). Overall, miR−1224 levels were associated with the clinical progress of GC patients and would be a tumor suppresser and new biomarker.

The remarkable reduction of miR-1224 expression in GC implied possible biological function in tumorigenesis. Because the expression of miR-1224 was relatively higher in SGC-7901 and BGC-823, we chose them for further functional assays. Hence, two GC cell lines (SGC-7901 and BGC-823) were transfected with miR-1224 mimics or NC mimics to determine the effects of miR-1224 on cellular growths. Subsequently, qPCR assays were conducted to determine the miR-1224 expression. The data indicated that the levels of miR-1224 were distinctly overexpressed in the transfection of miR-1224 mimics ( Figure 2A ). According to the CCK-8 assays, enforced expression of miR-1224 notably reduced the cellular growth of GC tumor cells ( Figures 2B, C ). Besides, cell proliferation was evaluated through EdU assays. The results suggested that the enhanced expression of miR-1224 resulted in a remarkably decreased cellular proliferation of GC cancer cells, consistent with the results of CCK-8 assays ( Figures 2D, E ). Moreover, the number of tumor cell colonies with miR-1224 overexpression significantly decreased compared to the controls, as demonstrated by colony formation assays ( Figure 2F ). Then, the possible roles of miR-1224 during the growth of SGC7901 cells in vivo were studied. Particularly, the enforced miR-1224 expression distinctly alleviated the tumor size and tumor weight ( Figures 3A, B ). As exhibited in Figure 3C , the positivity of Ki-67 was reduced via miR-1224 promotion. Furthermore, miR-1224 exhibited a higher level in samples from miR mimic group while SATB1 displayed a decreased level ( Figure 3D ). These data suggested that miR-1224 contributed to the inhibitory effects of GC cellular growth.

Next, the dynamic changes of migratory and invasive ability of tumor cells after the transfection of miR-1224 mimics were monitored to investigate the functional impact of miR-1224 overexpression on GC cell metastatic potentials. Wound healing assays demonstrated that a remarkable reduction in wounded areas was observed in cancer cells transfected with miR-1224 mimics, reflecting that the enhanced expression of miR-1224 markedly abrogated the migration of GC cells ( Figure 4A ). According to the results of transwell assays, the cells transfected with miR-1224 mimics displayed much fewer invasive cells compared with NC mimics-transfected cells. Thus, ectopic expression of miR-1224 contributed to the inhibition of GC cell migration ( Figure 4B ). Considering that epithelial-mesenchymal transition (EMT) was involved in regulating cancer metastasis, western blot assays were performed to detect the levels of EMT-related markers. The marked decline of N-cadherin and vimentin expression was observed in GC cells when they exhibited an overexpressed miR-1224 ( Figure 4C ). These results confirmed that the enhanced expression of miR-1224 attenuated the invasion and migration of GC cells.

Three online databases (starbase, miRDB, and TargetScan) were explored for their possible targets that displayed oncogenic properties to reveal the mechanisms involved in the tumor-suppressive roles caused by miR-1224 overexpression. The Venn diagram illustrated that there were 55 commonly potential target genes ( Figure 5A ). Next, whether the 55 common genes were associated with cancer development was investigated. Therefore, the gene ontology (GO) analysis was conducted using FunRich software. Biological process analysis revealed that the 55 common genes were correlated with energy pathways, metabolism, and signal transduction ( Figure 5B ). The cellular components and molecular functions for the 55 common genes were also analyzed ( Figures 5C, D ). Besides, biological pathways demonstrated that the 55 common genes were related to multiple cancer-correlated pathways, such as TRAIL signaling, VEGF and VEGFR signaling, S1P pathway ( Figure 5E ). Among the 55 commonly potential target genes, SATB1, a previously reported oncogene, attracted our attention. This was because the SATB1-gene interaction network generated by FunRich software suggested that SATB1 was closely associated with other cancer-correlated genes, such as GATA3, HDAC1, and MTA2 ( Figure 5F ). Afterward, the above prediction was confirmed. The binding site between miR-1224 and the 3’-UTR of SATB1 mRNA was presented in Figure 5G . We performed qPCR and found that SATB1 levels were distinctly upregulated in GC tumor specimens compared to normal tissues ( Figure 5H ). Additionally, dual-luciferase reporter assays on 293T cells were conducted to reveal that luciferase activity distinctly decreased in 293T cells after transfection of miR-1224 mimics and the wild-type 3’-UTR of SATB1 reporters ( Figure 5I ). Moreover, overexpression of miR-1224 could suppress the SATB1 levels, and the inhibition of miR-1224 exhibited an opposite function that miR-1224 knockdown led to markedly increased SATB1 expression ( Figure 5J ). These findings verified that miR-1224 could target SATB1 in GC cells.

Since the above data validated that miR-1224 was aberrantly downregulated in GC, the mechanisms that contributed to its dysregulation were investigated. Considering that transcription factors (TFs) were the critical regulators for the modulation of miRNAs expression, FunRich software was applied to analyze the potential TFs that could regulate miR-1224 expression. The first twenty potential TFs, including SP1, SP4, TCF3, and KLF7, were presented in Figure 6A . Besides, the potential TFs-miR-1224 interaction network was generated using Cytoscape software, as illustrated in Figure 6B . Therefore, we uncovered the possible molecular mechanisms leading to miR-1224 dysregulation and the TFs-miR-1224 interaction network.

As suggested by the above findings, miR-1224 served as a tumor suppresser in GC. Next, the detailed mechanisms by which miR-1224 modulated the development and progression of GC were revealed. Since several essential signaling pathways (particularly Wnt/β-catenin signaling) acted as vital players in tumor growth, qPCR and western blot assays were conducted to evaluate the levels of Wnt/β-catenin signaling associated molecules. According to the results of qPCR assays, the mRNA levels of β-catenin and its two targeting genes (cyclin D1 and c-myc) dramatically decreased in GC cells after miR-1224 upregulation ( Figures 7A, B ). As certified by western blot assays, enforced expression of miR-1224 notably repressed the protein expression of c-myc, β-catenin, and cyclin D1 in BGC-823 and SGC-7901 cells. This was consistent with the results of qPCR assays ( Figure 7C ). These results demonstrated that enforced miR-1224 expression suppressed the activity of the Wnt/β-catenin pathway in GC.