Background
Gastric cancer (GC) is a hazard to public health, causing the second largest number of cancer-related deaths globally [
1]. Eastern Asian populations are particularly susceptible compared to other world populations [
2]. Due to early diagnosis and comprehensive treatment strategies, which include chemotherapy, radiotherapy, targeted therapy and surgical intervention, the overall survival rate has improved slightly [
3,
4]. However, patients in advanced stages, who often appear with local infiltration, refractory proliferation, and even distant metastasis, might encounter poor outcomes [
5]. Therefore, it is of significance to investigate the underlying mechanism(s) of GC initiation and progression, as well as to identify effective biomarkers that are indicative of GC diagnosis and prognosis.
Long noncoding RNAs (lncRNAs) are a group of transcripts that are over 200 nucleotides in length. Formerly considered to be “transcriptional junk”, lncRNAs are currently considered as “novel regulators” in various cancers. They can coordinate gene expression by acting as decoys for transcriptional factors, scaffolds for chromatin modifying complexes, or compete with other genomic elements in binding to miRNAs [
6‐
8], and play crucial roles in the development, progression, invasion and metastasis of multiple cancers [
9‐
11]. Additionally, lncRNAs serve as valuable biomarkers for cancer diagnosis and prognosis [
12‐
14]. LncRNA ZEB1 antisense 1 (ZEB1-AS1) is transcribed from a shared bi-directional promoter of the zinc finger E-box binding homeobox 1 gene (ZEB1). An increasing number of studies have provided evidence that ZEB1-AS1 has oncogenic properties and serves as a promising biomarker in multiple cancers; GC included. For example, ZEB1-AS1 activates prostate cancer by regulating ZEB1 and the expression of the downstream molecule [
15]. It has also been associated with predictions of unfavorable prognoses and the promotion of tumor metastasis in hepatocellular cancer [
16] and gliomas [
17], and is further associated with the progression of esophageal squamous cell carcinoma and patient survival [
18]. In GC, ZEB1-AS1 plays a cancer-promoting role and is related to poor prognosis [
19], while in osteosarcoma, ZEB1-AS1 epigenetically activates ZEB1 to facilitate tumor progression [
20]. Nevertheless, the regulatory mechanism involved in GC and the prospect of the clinical application of ZEB1-AS1 in identifying GC is still limited and warrants further investigation.
Another type of noncoding RNA, microRNAs (miRNAs), are 19–25 nucleotides in length, are well characterized, and primarily participate in gene regulation via mRNA transcript degradation or translation inhibition. They dominate biological behaviors, such as proliferation, migration, invasion, and apoptosis, of tumor cells [
21,
22]. A considerable number of miRNAs are critical regulatory elements that facilitate or suppress GC occurrence and progression and might serve as therapeutic targets and novel biomarkers for GC [
23‐
26].
Recently, some studies suggested that lncRNAs engage in crosstalk with mRNAs and act as competing endogenous RNAs (ceRNAs) by sequestering shared miRNAs [
8,
27‐
29]. It has been suggested that lncRNAs, mRNAs and miRNAs communicate with each other and form competitive endogenous networks, which could be of profound significance to tumor mechanism investigations. For instance, lncRNA-KRTAP5-AS1 and lncRNA-TUBB2A regulate CLDN4 and influence tumor formation and metastasis via a ceRNA-mediated regulatory network in GC [
30], while HOTTIP promotes small cell lung cancer via a HOTTIP/miR-574-5p/EZH1-associated ceRNA network [
31]. Several studies have documented that ZEB1-AS1 can aggravate malignant behaviors of cancer cells through miRNA-mediated mechanisms [
32‐
34]. However, it is still uncertain whether ZEB1-AS1 can interact with miRNAs to form a ceRNA network in GC.
In the present study, significantly down-regulated miRNAs predicted to bind to ZEB1-AS1 were selected by bioinformatics analysis and included miR-204, miR-610, miR-149. Subsequently, it was found that ZEB1-AS1 negatively modulates miR-204-5p, miR-610, and miR-149-3p in vitro. Furthermore, we evaluated the functional effects of miR-149-3p on GC cell proliferation, migration, and invasion and suggest the existence of a ZEB1-AS1/miR-149-3p axis. Collectively, ZEB1-AS1 can interact with specific miRNAs, form a ceRNA regulatory network, and, in part, promote GC progression through a ZEB1-AS1/miR-149-3p axis.
Materials and methods
Tissue samples and cell lines
Eighty-four fresh GC tissue samples from patients who received surgical re-sectioning between January 2010 and October 2012, together with 47 precancerous gastric lesions (dysplasia) and 59 healthy gastric mucosal tissue samples from volunteers who underwent pathological biopsy under gastroscopy, were provided by the China Medical University Cancer Institute (Shenyang, China). Each specimen was frozen at − 80 °C in liquid nitrogen until studied. Every participant provided informed consent, with surgical patients subjected to post-operative follow-up for 65 months. The study was carried out under the approval of the China Medical University ethics committee in accordance with the Helsinki Declaration (1975). Clinicopathological data involved in the study is listed in Table
1. The Chinese Academy of Sciences Cell Bank (Shanghai, China) provided the cell lines used in the study, which included a normal human gastric cell line (GES-1) and five gastric cancer cell lines (SGC-7901, MGC-803, MKN-45, HGC-27, and AGS). All cells were cultured in RPMI 1640 medium (HyClone; GE Healthcare Life Sciences, Logan, UT, USA) supplemented with 10% fetal bovine serum (HyClone; HyClone Laboratories Inc. Victoria, Australia), 100 U/mL penicillin and 100 µg/mL streptomycin. The cell growth conditions were maintained at a constant temperature of 37 °C with 5% CO
2 and invariable humidity.
Table 1
Correlation between clinicopathological characteristics and ZEB1-AS1 or miR-149-3p
Age (years) | | | | 0.507 | | | 0.268 |
< 60 | 35 | 19 | 16 | | 20 | 15 | |
≥ 60 | 49 | 23 | 26 | | 22 | 27 | |
Gender | | | | 0.124 | | | 0.826 |
Male | 47 | 27 | 20 | | 24 | 23 | |
Female | 37 | 15 | 22 | | 18 | 19 | |
Tumor size (cm) | | 0.02* | | | 0.188 |
≤ 5 | 38 | 12 | 26 | | 22 | 16 | |
> 5 | 46 | 30 | 16 | | 20 | 26 | |
Differentiation | | | 0.127 | | | 0.275 |
Well/moderate | 43 | 18 | 25 | | 24 | 19 | |
Poor | 41 | 24 | 17 | | 18 | 23 | |
Invasion depth | | | | 0.387 | | | 0.027* |
T1–T2 | 36 | 16 | 20 | | 23 | 13 | |
T3–T4 | 48 | 26 | 22 | | 19 | 29 | |
Lymph node metastasis | | | 0.01* | | | 0.126 |
No | 39 | 12 | 27 | | 23 | 16 | |
Yes | 45 | 30 | 15 | | 19 | 26 | |
TNM stage | | | | 0.048* | | | 0.04* |
I–II | 37 | 14 | 23 | | 25 | 12 | |
III–IV | 47 | 28 | 19 | | 17 | 30 | |
CA-199 (tissue) (U/mL) | | | | 0.825 | | | 0.268 |
< 37 | 35 | 17 | 18 | | 20 | 15 | |
> 37 | 49 | 25 | 24 | | 22 | 27 | |
Cell transfection
For the establishment of stably transfected cells, lentiviruses carrying ZEB1-AS1 and ZEB1-AS1-sh-RNA vectors were constructed (Wanleibio, Shenyang, China) and respectively transfected into SGC-7901 and MGC-803 cells according to the manufacturer’s protocol. Following 48 h of transfection, cells were selected with 10 μg/mL puromycin, measured through a fluorescent inverted microscope, with qPCR performed to determine the transfection efficiency (Additional file
1: Fig. S1). The miR-149-3p mimic and the miR-149-3p negative control (miR-NC) elements (GenePharma Corporation, Shanghai, China) were transiently and separately transfected into the same cell lines using Lipofectamine TM 3000 transfection reagent (Invitrogen, Carlsbad, CA, USA). Cells and transfection complexes were co-incubated for 5 h. Transfection efficiency in these cells was also validated by qPCR. The sequences involved in the study are detailed in Additional file
2: Table S1 and Additional file
3.
RNA isolation, cDNA synthesis and real-time PCR
Total RNA was isolated from frozen samples and cell lines with RNAiso Plus reagent (TaKaRa, Dalian, China) and complementary DNA (cDNA) generated from reverse transcription of 1 µg of total RNA using PrimeScript RT reagent Kit with gDNA Eraser (TaKaRa). Real-time PCR was performed with SYBR
® Premix Ex TaqTM kit (TaKaRa) on an ABI 7500 system (Applied Biosystems, Foster City, CA, USA), according to the following PCR conditions: 30 s at 95 °C for initial denaturation, followed by 45 cycles of 5 s at 95 °C for denaturation, 10 s 60 °C for annealing and 30 s 72 °C for extension. For miRNA detection, reverse transcription was accomplished with miRcute plus miRNA First-Strand cDNA Synthesis Kit (Tiangen, Beijing, China). Real-time PCR was conducted with a miRcute plus miRNA qPCR Detection Kit (Tiangen). The primer sequences are shown in Additional file
2: Table S1. RNA expression was normalized to U6 or GAPDH; relative RNA expression was calculated through the 2
−ΔΔCt method.
Cell proliferation assay
The CCK-8 (Cell Counting Kit-8) assay was performed to detect cell viability following the manufacturer’s instructions (Dojindo Laboratories, Kumamoto, Japan). In brief, 4 × 103 transfected cells were seeded into 96-well plates in quintuplicate and incubated for 24, 48, 72, and 96 h. Then, 10 μL CCK8 working solution was added to the medium for 4 h at the fixed time points. Absorbances were measured on a microplate reader at 450 nm. The experiment was performed in triplicate.
Wound healing assay
The transfected cells were seeded separately into six-well plates (5 × 105 cells/well) and cultured overnight. Once a monolayer of cells had formed, a 200-μL pipette tip was used to scratch the cell layer. After cells present in suspension and cell debris were washed out with PBS, cells were cultured in serum-free medium to permit wound healing. Phase-contrast images were taken at the same position under an inverted microscope at 0, 24, and 48 h after scratching. Three independent experiments were conducted.
Transwell migration and invasion assays
For the transwell assay, transwell chambers with 8-μm porous membranes (Corning, NY, USA) were used. A total of 5 × 104 transfected cells were added with 200 μL serum-free medium to the upper chambers, the membranes of which were pre-coated with Matrigel (BD Bioscience, San Jose, CA, USA) for invasion assays but uncoated when performing migration assays. A volume of 750 μL medium containing 10% FBS was added to the lower chamber. Following at least 24 h of incubation at 37 °C with 5% CO2, upper-chamber cells were removed using a cotton swab. Cells traversing the membranes were fixed in 4% paraformaldehyde and stained with 0.1% crystal violet for 20 min. Cell counts were completed with at least five random visual fields conducted per membrane under a light microscope. The assays were performed in triplicate.
Luciferase reporter assay
The miR-149-3p-binding sites in ZEB1-AS1 (ZEB1-AS1 wild type or wtZEB1-AS1) and the corresponding mutant sites (ZEB1-AS1 mutant type or mutZEB1-AS1) were respectively amplified by PCR and cloned into the pmirGLO plasmid (Promega, Madison, WI, USA) which contains a luciferase gene. Consequently, miR-149-3p mimics or miR-NC were co-transfected with luciferase reporter plasmids into 293T cells. Luciferase activity was analyzed 48 h post-transfection using the Dual-Luciferase Reporter Assay System (Promega) and normalized against Renilla luciferase activity.
Statistical analysis
SPSS 13.0 (SPSS Inc, Chicago, IL, USA) was used for statistical analyses. A X2 test was utilized to assess categorical variables with Student’s t-tests and analysis of variance (ANOVA) carried out for appropriate comparisons. The Kaplan–Meier method with a log-rank test was adopted to analyze overall survival (OS). Prognostic factors were evaluated using Cox regression analysis. The diagnostic value of each biomarker was tested via receiver operating characteristic (ROC) curve analysis. Spearman’s correlation analysis was applied to examine the correlation between ZEB1-AS1 and miR-149-3p. P-values less than 0.05 were considered statistically significant.
Discussion
It is well recognized that genes don’t work in isolation, but rather communicate with one other, thereby forming regulatory networks. LncRNAs, which are newly described and emerging regulatory elements, appeal to researchers as they participate in gene regulation in multiple manners. This includes engaging in competitive binding of specific miRNAs in order to form ceRNA networks. LncRNA-mediated sequestering of miRNAs could result in the increased expression of mRNAs which share miRNA Response Elements (MREs) with lncRNAs [
43,
44]. Thus, the specific, common miRNAs-mediated crosstalk between lncRNAs and mRNAs weaves a competitive endogenous network [
44], which extends the current dimensions and understanding of gene regulation. ZEB1-AS1 attracted our attention due to the fact that it can function as a tumor promoter in several cancers in miRNA-mediated manners [
32‐
34]. Therefore, it is rational to hypothesize that ZEB1-AS1 could be involved in a miRNA-regulated ceRNA network in GC.
In the present study and by performing bioinformatics analysis, we speculate that ZEB1-AS1 can interplay with eight down-regulated miRNAs in GC. Among these miRNAs, the anti-tumor effects of miR-610, -204, and -149 on GC have been previously described, thereby supporting the reliability of our bioinformatics prediction. In addition, enrichment analysis for common target genes of the three miRNAs was performed. The enriched pathways included the PI3 K-Akt, MAPK, and Ras signaling pathways, which are explicitly related to GC progression and metastasis [
45‐
47]. Hence, we theorized that ZEB1-AS1 can indirectly modulate these crucial pathways by sequestering the three miRNAs. In agreement with the bioinformatics conclusion, in vitro assays demonstrating the effect of gaining or losing ZEB1-AS1 indicated that it could regulate the expression of miR-149-3p, miR-204-5p and miR-610. Currently, our findings suggest that ZEB1-AS1 may sequester these three miRNAs to form a ceRNA network in GC. This finding is in keeping with previous research results. For example, the SNHG1 lncRNA sequestered miR-302/372/373/520 and consequently activated TGFBR2 and RAB11A in invasive pituitary tumors [
48], while DANCR acted as a decoy for miR-335-5p and miR-1972, thus promoting ROCK1-associated proliferation and metastasis in osteosarcomas [
49]. Our study extends the mechanism by which ZEB1-AS1 exerts its effect in GC. We furthermore infer that ZEB1-AS1 can act as a key therapeutic target in GC. In addition, in vitro assays showed that the expression patterns of miR-149-5p and miR-204-3p were variable and did not always match that of ZEB1-AS1. This implies that bioinformatics predictions need to be confirmed by additional tests.
The role of miR-149-3p in tumors remains controversial. For example, Bellazzo et al. report that it facilitates the aggressiveness of tumor cells in prostate cancer, bladder carcinoma, and breast carcinoma, to name a few [
50]. However, Yang et al. observed that miR-149-3p blocks cell proliferation, migration, and invasion in bladder cancer [
42]. Furthermore, its effects on GC cell proliferation, migration, and invasion are not fully elucidated. The exact function(s), underlying mechanism(s), and clinical significance relating to ZEB1-AS1 and miR-149-3p in GC therefore deserves further clarification.
In the present study, functional assays confirmed that miR-149-3p could suppress GC cell proliferation, migration, and invasion, thus indicating that miR-149-3p acts as a tumor repressor in GC. Its role in oncogene or tumor suppression was, however, found to be dependent on the type of tumor present. Quantitative real-time PCR analysis of ZEB1-AS1 and miR-149-3p first showed that the up-regulation of ZEB1-AS1 was accompanied by down-regulation of miR-149-3p in GC tissue. This observation was supported by the negative correlation found between the two factors through Pearson analysis. We speculated that the two factors could exert opposing functions in GC. Moreover, the in vitro malignant behaviors (i.e., proliferation, invasion, and migration) induced by ZEB1-AS1 could be abrogated by miR-149-3p. The luciferase reporter assay consistently indicated ZEB1-AS1 was directly bound by miR-149-3p. These results further confirm that, interacting with miR-149-3p, ZEB1-AS1 can form a ceRNA network. What is more, we theorized that ZEB1-AS1 can modulate GC occurrence, progress, and metastasis through a ZEB1-AS1/miR-149-3p axis; a thought reinforced by the outcomes from related studies. For example, LINC01088 targets miR-24-1-5p to inhibit ovarian epithelial cell tumor [
51], while the lncRNA HNF1A-AS1 represses the miR-34a/SIRT1/p53 axis to promote colon cancer metastasis [
52].
In addition, our study has revealed that high expression of ZEB1-AS1 or low expression of miR-149-3p could serve as independent indicators of an unfavorable prognosis in GC patients. Similarly, an increasing number of studies support the belief that ZEB1-AS1 might be attributed to a poor prognosis in GC [
19], colorectal cancer [
53], gliomas [
17], esophageal squamous cell carcinoma [
18], and hepatocellular carcinoma [
16]. To the best of our knowledge, no report has previously described miR-149-3p as a possible prognostic factor in GC; the implication thereof is indicated here for the first time. Furthermore, with reliable ROC curve analyses and AUC outputs with good sensitivity and specificity, we conclude that ZEB1-AS1 could potentially be used as a marker for the early diagnosis of GC. Diagnostic capacity might be improved should ZEB1-AS1 and miR-149-3p be used in combination. To our knowledge, our study is also the first to evaluate the early diagnostic ability of ZEB1-AS1 and miR-149-3p in GC.
Authors’ contributions
DQD, MHM and JL devised the experiment scheme; MHM, CZ, JXA and KZW performed the experiment; MHM, YL, ZZ and CDZ collected the human samples; YL, KZW and ZZ provided the data; MHM, CZ and JL analyzed the data; MHM wrote the manuscript. All authors read and approved the final manuscript.