miR-140-5p suppresses the proliferation, migration and invasion of gastric cancer by regulating YES1

Background The aberrant expression of microRNA-140-5p (miR-140-5p) has been described in gastric cancer (GC). However, the role of miR-140-5p in GC remains unclear. In this study, the prognostic relevance of miR-140-5p in GC was investigated and YES1 was identified as a novel target of miR-140-5p in regulating tumor progression. Methods miR-140-5p level was determined in 20 paired frozen specimens through quantitative real-time PCR, and analyzed in tissue microarrays through in situ hybridization. The target of miR-140-5p was verified through a dual luciferase reporter assay, and the effects of miR-140-5p on phenotypic changes in GC cells were investigated in vitro and in vivo. Results Compared with that in adjacent normal tissues, miR-140-5p expression decreased in cancerous tissues. The downregulated miR-140-5p in 144 patients with GC was significantly correlated with the reduced overall survival of these patients. miR-140-5p could inhibit GC cell proliferation, migration and invasion by directly targeting 3′–untranlated region of YES1. miR-140-5p could also remarkably reduce the tumor size in GC xenograft mice. Conclusions miR-140-5p serves as a potential prognostic factor in patients with GC, and miR-140-5p mediated YES1 inhibition is a novel mechanism behind the suppressive effects of miR-140-5p in GC. Electronic supplementary material The online version of this article (doi:10.1186/s12943-017-0708-6) contains supplementary material, which is available to authorized users.


Background
Gastric cancer (GC) is a frequently diagnosed cancer and considered a major contributor to health problems worldwide [1,2]. Although diagnostic approaches and treatment strategies have improved gradually, the prognosis of patients with GC in advanced stages remains poor. Thus, the causative mechanisms of GC tumorigenesis and metastasis should be elucidated.
Aberrant signaling events frequently occur in human malignancies and contribute to oncogenesis. For example, upregulated Src signaling, which increases proportionally with progressive TNM stages, has been reported in various human cancers [3], and members of Src family tyrosinekinases (SFKs) are prototypical non-receptor protein tyrosine kinases. YES proto-oncogene 1(YES1), a notable member of SFKs, serves as a seminal regulator of cell growth, adhesion, and differentiation. The major role of YES1 in tumor progression is supported by the decreased YES1 expression correlated with the impaired growth abilities of several malignancies, including malignant mesothelioma, rhabdomyosarcoma, and pancreatic cancer [4][5][6]. YES1 also provides chemotherapeutic resistance in several cancers. For instance, YES1 can promote the nuclear translocation of the epidermal growth factor receptor that induces resistance to cetuximab in the non-small cell lung cancer line NCI-H226 [7]. However, the expression level of YES1 in GC tissues remains controversial. Seki et al. observed the upregulation of YES1 expression in GC tissues compared with that in adjacent normal tissues, but Wang et al. obtained an opposite conclusion [8,9]. In addition to these inconsistent findings, the underlying causes of aberrant YES1 expression in GC are unclear.
MicroRNAs (miRNAs) are an endogenous group of small non-coding RNAs that post-transcriptionally repress the expression of protein-coding genes. miRNAs function as specific repressors of translation or cleavage inducers of RNA transcripts by complementarily binding to the corresponding 3′ -untranslated region (3′-UTR) of an mRNA transcribed from a protein-coding gene, miRNAs regulate up to 60% of human protein-coding genes [10], but specific miRNA subsets or clusters are dysregulated in various tumors [11][12][13]. An example of these miRNAs is miR-140-5p, whose various roles have been widely explored. miR-140-5p has been characterized as a tumor suppressor in colorectal cancer, biliary tract cancer and hypopharyngeal cancer [14][15][16]. In GC, high miR-140-5p levels have been detected in the plasma of patients, whereas low miR-140-5p levels have been found in 20 tissue samples and a HGC-27 cell line [17,18]. Nevertheless, the prognostic relevance and functions of miR-140-5p in GC are largely unknown. Considering that various protein coding genes are targeted by one miRNA, researchers should determine the mechanisms of miR-140-5p mediated phenotypic changes in GC cells.
This study aimed to evaluate the prognostic relevance of miR-140-5p in GC and identify whether YES1 is a direct target of miR-140-5p in the regulation of tumor progression.

Patients and tissue samples
Tissue specimens (tumor, adjacent gastric mucosa, and normal tissues from a benign disease) were collected and immediately frozen in liquid nitrogen by a pathologist after informed consent was obtained from patients in the First Affiliated Hospital of Anhui Medical University (Hefei, China). The tissue specimens consisted of 20 paired cancer and adjacent normal tissues from 144 patients with GC and 43 normal gastric mucosa tissues from patients with a benign disease. Histological cancer types were evaluated by at least two independent pathologists in accordance with the TNM staging guide (2010) released by The American Joint Committee on cancer (AJCC). The patients' characteristics are presented in Additional file 1: Tables S1 and Additional file 2: Table S2. None of the patients received preoperative chemoradiotherapy. Clinical follow-up consisted of telephone calls and reviews of clinical data, including age, gender, tumor location and size. The dates of patients' death were documented and the overall survival (OS) was defined as the time from the operation date to the date of death. All of the experiments in this study were approved by the Ethics Committee of Anhui Medical University.

RNA extraction and quantitative real-time PCR (qRT-PCR) analysis
Total RNA was extracted from fresh tissue specimens or cells and assayed for mature miRNAs by using TRIzol reagent (Invitrogen, USA). cDNAs were generated using MMLV reverse transcriptase (Takara, Japan). (qRT-PCR) was run in an IQ5 Real-time PCR detection system (Applied Biosystems, USA), with a SYBR Premix Ex Taq kit (Takara, Japan). The sequences of the primers were listed as follows: miR-140-5p forward primer 5-T GCGGCAGTGGTTTTACCCTATG-3′ and reverse primer 5′-CCAGTGCAGGGTCCGAGGT -3′. U6 forward primer 5-´TGCGGGTGCTCGCTTCGGCAGC-3′ and reverse primer 5′-CCAGTGCAGGGTCCGAGGT -3′. The reactions were incubated at 94°C for 4 min followed by 40 cycles of 94°C for 30 s, 58°C for 30 s and 72°C for 30 s. Relative expression was achieved by comparative CT calculation (2 -ΔΔCt method) and normalized to U6-snRNA. All of the experiments were performed at least in triplicate.

miRNA In situ hybridization (ISH)
The tissue sections were dewaxed, rehydrated, digested with 15 μg/mL proteinase K, and incubated with a double-DIG-labeled LNA microRNA probe (Exiqon, Denmark) for miR-140-5p with positive U6 (Exiqon, Denmark) and negative scrambled LNA probes as controls. The staining intensities of the tumor cells or normal epithelial cells were scored by two independent pathologists in accordance with the following descriptions: 0, negative, 1, weakly positive, 2 moderately positive and 3 strongly positive [19][20][21]. Scores ≥2 were designated as "high expression", whereas scores <2 were defined as "low expression".

Dual luciferase reporter assay
HumanYES1 3′ UTR (Accession number NM_014143 for YES1), and mutant YES1 3′ UTR were cloned with the following primers: underlined letters corresponded to the mutated sequences and bold letters indicated the sites cut by PmeI or Xbal).
YES1-3′ UTR wild-type: YES1-3′ UTR mutant: The fragment of the wild type or mutant YES1 3′ UTR was ligated into a pmirGLO -dual-luciferase miRNA target expression vector (Promega, USA). The constructed vectors were validated by sequencing, cloned, and purified. AGS was then co-transfected with the constructed vectors and miR-140-5p mimics or miR-NC mimics (GenePharma, Shanghai, China) and then cultivated for 48 h. Luciferase activity was subsequently measured using the Dual Luciferase Reporter Gene Assay Kit (Beyotime, Jiangsu, China) in accordance with the manufacturer's protocol. The relative light unit of the firefly luciferase was recorded with a GloMax 96 microplate luminometer (Promega, USA). The ratio of firefly to Renilla was used to normalize the firefly luciferase values.

Cell proliferation assay
After transfection was conducted, MTT assay was performed for 24, 48, or 72 h in accordance with the manufacturer's manual. In brief, 100 μL of a culture medium containing 1 × 10 4 GCcells were added to each well of a 96well plate. After 12 h of cultivation, 10 μL of MTT solutions were added into each well and incubated for 4 h. Afterward, 100 μL of a detergent reagent was added, and the plate was incubated in the dark at room temperature for 2 h. Absorbance at 490 nm in each well was recorded using a microplate reader (Bio-tek, USA).

Colony formation assays
GC cells were suspended in DMEM containing 10% FBS and 0.3% low-melting-point agarose. A total of 500 cells were then seeded into each well of a 12-well plate coated with 0.5% agarose. After 14 days of cultivation, colonies were stained with 0.1% crystal violet and counted.
Wound healing assays GC cells were suspended and seeded into each well of a12-well plate at a density of 1 × 10 5 /cm 2 . After the cells attached to the bottom and formed a monolayer, linear wounds were scratched with a pipette tip at the center of the monolayer The detached cells were carefully washed off with PBS thrice. Afterward, 1 mL of DMEM (Gibco BRL, USA) containing2%FBS (Gibco BRL, USA) was added into each well, and the wounds were photographed at 0,2,48, and 72 h under a microscope (Olympus, Japan) and the widths of each scratch wound were recorded.

Transwell assays
Invasion assay was performed in 24-well Transwell chambers (Millipore, USA) in accordance with the manufacturer's manual. In brief, the invasion chamber was pre-coated with 10 μL of fibronectin(Sciencell, USA) and loaded with 50 μL of Matrigel (BD, USA). After the coated plate was incubated at 37°C for 2 h, 0.5 mL of cell suspension without serum (1 × 10 5 cells) was added to the invasion chamber and 500 μL DMEM with 10% FBS, (Gibco BRL, USA) was added to the lower chamber. The plate was then incubated at 37°C for 24 h, and the cells on the outer surface of the insert bottom were fixed, stained with crystal violet, and counted under a microscope at 100× magnification.

In vivo xenograft study
A mouse xenograft model was established in 6 week-old BALB/c-nu mice (Model Animal Research Institute of the Chinese Academy of Medical Sciences, Beijing, China) in accordance with the institutional guidelines. In brief, 5 × 10 6 BGC823 cells suspended in 100 μL of PBS were injected subcutaneously into the left dorsal flanks of the mice. When the tumor size reached approximately 50 mm 3 , 18 mice were equally divided into the following three groups: group A as blank control, group B as miR-140-5p treatment, and group C as control mimic treatment. Each tumor in group B or C was injected with50 μL solution containing 7 nmol miRNA mimics, 3 μL of Lipofectamine2000 (Invitrogen), and 40 μL of serum-free medium Opti-MEM (Invitrogen). After the treatments were administered, the length (L) and width (W) of the tumors were measured every 3 days, and the mice were weighed every 4 days. Tumor volume (V) was calculated using the following eq. V = 0.5 × L × W 2 . On day 19 the mice were sacrificed and the tumors were dissected, weighed and photographed. The tumor tissues were fixed in 4% paraformaldehyde for further studies.

Immunohistochemistry
The tumor tissues of the mice were fixed, embedded in paraffin. And sliced into consecutive tissue sections. Next, 5 μm thick tissue sections were deparaffinized, dehydrated, and heated in citrate buffer (pH 6.0) for 15 min at 95°C. To block the nonspecific bindings of the first antibody, we added 1% bovine serum onto the slides for 20 min at room temperature. The tissue sections were then incubated with YES1 antibodies (1:50; Saierbo, Tianjin, China) overnight at 4°C, and subsequently incubated with the secondary HRP-labeled goat anti-rabbit IgG antibodies (1:20,000, Saierbo, Tianjin, China). Finally, the sections were visualized with a DAB kit (ZSGB-bio, Beijing, China) and counterstained with hematoxylin (Beyotime, Jiangsu, China).

Statistical analysis
Data were statistically analyzed using GraphpadPrism5. The relative expression of miR-140-5p in the GC tissues and adjacent normal tissues was presented as median (interquartile range) through a nonparametric test (Wilcoxon's test). Other measurement data were described as means ± standard error of means.
The correlations between miR-140-5p level and the pathologic parameters were analyzed through Pearson's χ2 test, and the survival and OS curves were plotted using the Kaplan-Meier method. Student's t test was used to evaluate the differences between two groups, and one-way (ANOVA) was preformed to compare more than two groups. All P-values were two-sided and differences were considered significant when P < 0.05. *, **, or *** corresponded to P < 0.05, P < 0.01, or P < 0.001, respectively.

miR-140-5p is downregulated in GC tissues and correlated with the patients'OS
We investigated miR-140-5p expression in 20 fresh GC tissues and matched adjacent normal tissues through qRT-PCR and found that the miR-140-5p expression was significantly downregulated in the tumor tissues (P < 0.001) (Fig. 1a-b). Subsequent ISH analysis of the miR-140-5p levels in the 144 GC tissues and 43 normal gastric tissues revealed similar results (Fig. 1c). The staining intensities of miR-140-5p were reduced in 32.6% of the GC tissues (47 of 144).By comparison, 11.6% (5 of 43) of the normal gastric tissues showed downregulations (P = 0.007). A Cox proportional hazards regression model was established to examine the prognostic significance of miR-140-5p and other clinical parameters. In addition to tumor differentiation, invasion depth and lymph node metastasis, miR-140-5p level was an independent favorable predictor of OS (Additional file 2: Table S2). Further correlation analysis demonstrated that the downregulated miR-140-5p levels were correlated with the reduced OS of the patients with GC (P = 0.0084; Fig. 1d). The median survival times (20 months) of the patients with low miR-140-5p expression were shorter than those (59 months) with high miR-140-5p expression.
miR-140-5p directly targets one evolutionarily conserved sequence within the 3′UTR of YES1 To investigate the potential target of miR-140-5p, we combined the results from prediction databases (Tar-getScan, miRDB, and ComiR) and found that YES1 was a possible target (Fig. 2a). We then utilized a panel of five GC cell lines and one normal gastric cell line to verify the predicted results. We initially screened the miR-140-5p expression in these cell lines (Fig. 2b) and then confirmed the inverse correlation between miR-140-5p level and YES1 protein expression in AGS and BGC823 cells (P < 0.05; Fig. 2c-d). Next, we performed luciferase reporter assay to investigate whether the 3′-UTR of the YES1 gene was a direct target of miR-140-5p in GC. The transient transfection of AGS cells with miR-140-5p and YES1 3′-UTR plasmids revealed that the relative luciferase intensity was significantly decreased compared with that of the miRNA control. By contrast, the vector containing the site mutated 3′-UTR sequence was not affected by miR-140-5p (Fig. 2e). These results indicated that 3′-UTR of YES1 was a direct target of miR-140-5p.

miR-140-5p suppresses the cell proliferation, migration, and invasion of GC in vitro
To determine the effects of miR-140-5p on GC cell proliferation, we performed MTT assays and colony formation assays. Compared with those of the control groups in vitro, the transfection-induced overexpression of miR-140-5p in AGS and BGC823 cells significantly suppressed the proliferation ability of GC cells over time (Fig. 3a). The colony formation assays demonstrated that the AGS and BGC823 cells transfected with miR-140-5p showed remarkably decreased colony formation abilities in comparison to those in the control groups (Fig. 3b). Next, we investigated whether miR-140-5p affects GC cell migration and invasion. Wound healing assays showed that the miR-140-5p overexpressing cells were less effective than the non-transfected cells in closing an artificial wound over a confluent monolayer (Fig. 3c). We also examined the influence of miR-140-5p on GC cell invasion by Transwell assay (Fig. 3d). and observed that the numbers of GC cells invading through the Matrigel were remarkably decreased after these cells were treated with miR-140-5p compared with those in the negative control. These results suggested that miR-140-5p attenuated the migration and invasion of GC cells.

miR-140-5p regulates genes associated with tumor proliferation, migration, and invasion in vitro
Western blot assays were performed to detect the expression of the markers associated with the tumor proliferation, migration, and invasion of AGS and BGC823 cells and to clarify the mechanisms of the inhibitory effects of miR-140-5p on GC cells. After miR-140-5p treatment was administered, the protein levels of Ki-67, CyclinD1, vimentin, Snail, MMP2, and MMP3 were remarkably decreased, whereas the protein level of E-cadherin was significantly increased in both cells. However, the increased level of survivin was observed only in BGC823 cells, whereas no remarkable changes were found in AGS cells. We further verified that the level of SOX4, which has been reported as a direct target of miR-140-5p in HGC-27, was reduced in both cells (Fig. 3e).

Reconstitution of YES1 rescues the miR-140-5p-mediated inhibition in GC cells
To examine whether miR-140-5p elicits inhibitory effects on GC cells through YES1, we transfected the constructed YES1 expression vectors with miR-140-5p ( Fig. 4a-b). Subsequent MTT and colony forming assay revealed that reintroducing YES1 restored the miR-140-5p-mediated growth repression of GC cells (Fig. 4c-d). Likewise, the repressive effect of miR-140-5p on GC cell migration and invasion could be rescued by reconstituting YES1 (Fig. 5a-c). Collectively, these data provided evidence that miR-140-5p partially repressed the proliferation, migration and invasion of GC cells by inhibiting YES1.

miR-140-5p suppresses the growth of BGC-823 xenograft tumors
On the basis of our in vitro results, we further explored the inhibitory potential of miR-140-5p on GC growth in vivo. BGC-823 cells were subcutaneously implanted into the flank of nude mice and the size of the xenograft tumors was measured after miR-140-5p mimics or control mimics with a blank control were intratumorally injected. The growth curve tended toward lower tumor volumes in the miR-140-5p group than in the control groups (Fig. 6a). Subsequently, the xenograft tumors were harvested and weighed immediately after the mice were sacrificed. Consistent with the tumor volume, the mean tumor mass was significantly lower in the miR-140-5p group than in the control groups (Fig. 6b-c). The YES1 levels in the tumor tissues were than examined through immunohistochemistry staining. In line with the in vitro results, YES1 protein staining was evidently weaker in the miR-140-5p-treated tumors than in the control tumors (Fig. 6d).

Discussion
Compared with that in adjacent normal tissues, miR-140-5p was downregulated in GC tissues. Similar results were also observed in the GC cell lines compared with those in the nonmalignant cell line. This study is the first to show the correlation between the downregulated expression of miR-140-5p in GC and the reduced mean survival of patients with this disease. We also demonstrated that miR-140-5p directly targeted 3′ UTR-YES1 and the downregulated miR-140-5p was partly responsible for the upregulation of YES1. Moreover, the forced miR-140-5p A previous study similarly concluded that miR-140-5p levels decrease in GC tissues compared with those in normal tissues. In addition to YES1, other targets, such as SOX4, are downregulated by increased miR-140-5p in HGC-27 [18]. However, these data were obtained from 20 GC samples and 1 GC cancer cell line. Studies with large sample sizes are required to further corroborate these findings. In our study, the ISH staining results of TMA from the144 patients are consistent with the conclusions of other groups. We demonstrated that low miR-140-5p levels in GC tissues predicted the poor clinical outcomes of patients, and we suggested that miR-140-5p could be a potential prognostic factor. Unfortunately, we did not conserve the blood sample of the patients enrolled in the TMA study and thus failed to verify the previous description of serum miR-140-5p levels in patients with GC. To achieve translational application, we will further investigate the correlation of serum miR-140-5p levels with patient prognosis. Previous studies suggested the diversity of miR-140-5p mediated effects on tumor cell phenotypes. For instance, miR-140-5p suppresses the self-renewal and growth of breast cancer stem cells [22] and inhibits the invasion of esophageal cancer cells rather than their proliferation [23]. However, studies have yet to clarify whether miR-140-5p can induce phenotypic changes, in addition to its inhibitory effects on tumor cell proliferation, in GC. Cancer cell lines might also respond differently to the same treatment because of their genetic variances. As such, additional functional experiments are necessary to support previous conclusions. In the two cell lines observed in our study, miR-140-5p inhibited not only the proliferation of GC cells but also their migration and invasion. These results were further validated by the reduced levels of phenotype-related proteins, including Ki-67 and cyclinD1 (proliferation) [24], vimentin and Snail (migration) [25], and MMP2 and MMP3 (cell invasion) [26,27]. E-cadherin levels were also increased, and this finding implied that the cell-to-cell adhesion abilities of GC cells are enhanced [28]. Considering the same phenotypic changes of both cell lines, the inconsistent changes of survivin (cell division) levels were possibly due to the different cell line backgrounds; this topic requires further research. Moreover, the upregulation of survivin in BGC823 cells, which indicates the increased abilities of cell division, seemed to be contrary to the growth inhibition phenomenon after miR-140-5p treatment [29]. Thus, we speculate that cell division process might not be the key procedures interfered by miR-140-5p. Additionally, in vitro assays may be unable to show the actual biological processes of tumor cells in complex microenvironments. Thus, we detected the suppressive effect of miR-140-5p by using a nude mouse model. Consistent with our in vitro findings, our observations in the mouse model revealed that miR-140-5p impeded the growth of GC tumor grafted in nude mice. YES1 is suppressed by miRNAs in the tumor progression of oral cancer, ovarian cancer, and hepatocellular carcinoma, and this observation indicates that YES1 may be regulated by miRNAs in GC [30][31][32]. Although in silico analysis suggests that YES1 is a target gene of miR-140-5p, neither the putative interaction nor the functional role of YES1 in GC has been defined. Our results show that miR-140-5p directly targeted the 3′-UTR of the YES1 gene, because its overexpression was significantly associated with luciferase activity suppression. A significant decrease in YES1 protein expression was validated after miR-140-5p was overexpressed in GC cells. Likewise, miR-140-5p injections reduced the immunohistochemical staining density of theYES1 protein in the transplanted tumors of nude mice. Considering the importance of YES1 in solid cancer, we further characterized its role by overexpressing the YES1 protein. YES1, which is a SFK family member, plays a pleiotropic role in oncogenic cell functions. Rosenbluh et al. .confirmed that YES1 promotes the phosphorylation of YAP1 and the proliferation of cancer cells by activating the Wnt/β-catenin pathway [33]. YES1 upregulation similarly reverses the miR-140-5p-mediated inhibition of GC cell growth and colony formation. Liu et al. reported that activating YES1 enhances the metastatic behaviors of melanoma cells [34]. Overexpressing YES1 reverses the inhibitory effects of miR-140-5p on GC cell migration and invasion. We also demonstrated the role of YES1 in the multiple oncogenic activities of GC cells. Interestingly, the co-transfection of YES1-overexpressing vectors with miR-140-5p did not fully revert the suppressive behavior of GC induced by miR-140-5p. This phenomenon could be possibly attributed to the ability of miR-140-5p to directly target other important oncogenic genes, such as SOX4. Previous studies confirmed that miR-140-5p suppresses GC cell proliferation by targeting SOX4, and this finding was supported by our results, which demonstrated that miR-140-5p overexpression decreased the protein levels of SOX4 in AGS and BGC823 cells. Unfortunately, SOX4 has been reportedly associated with the proliferation phenotype of HGC-27 only, and this observation may