Background
Recently, high-throughput genome and transcriptome sequencing and microarrays have indicated that apart from protein-coding genes, 75% of the human genomes is transcribed into noncoding RNAs [
1,
2]. LncRNAs are functionally catalogued as noncoding transcripts are more than 200 nucleotides in length, and have no potential protein-coding ability. The Encyclopedia of DNA Elements (ENCODE) Project Consortium revealed that more than 28,000 lncRNAs were transcribed in the whole genomes [
2]. The aberrant expression and deficiency or mutation of lncRNAs were reported to be involved in numerous complex diseases, including cancers [
3,
4]. Mounting evidence indicated that lncRNAs are implicated in a variety of biological processes, including chromatin interaction, transcription regulation, mRNA post-transcriptional regulation and epigenetic regulation [
5‐
7].
Additionally, increasing experimental evidence supports that lncRNA functions as competitive endogenous RNA (ceRNA), which compete for microRNA (miRNA) to up-regulate the expression of a target gene. The ceRNA hypothesis provide new insights into the function of a large amount of uncharacterized lncRNAs [
8]. It has been reported that muscle-specific long noncoding RNA (linc-MD1) regulate MAML1 and MEF2C expression by sponging miR-133 and miR-135 [
9]. Another study has shown that lncRNA BC032469 acted as a ceRNA for miR-1207-5p to up-regulate the expression of hTERT and promoted proliferation in gastric cancer (GC) [
10].
GC is the fourth most frequent malignancy and contributes to the second leading cause of cancer mortality. Although effective medical treatments such as surgery, chemotherapy and radiation have been improved, GC patients are usually diagnosed with advanced stage, resulting in a low five-year survival rate [
11‐
13]. Currently, GC is still a globe health problem, which highlights the need for further studies of molecular mechanism of GC and identify effective therapeutic targets. Emerging evidence have shown that aberrant expression of many lncRNAs were observed in gastric cancer and significantly associated with carcinogenesis, diagnosis and prognosis of gastric cancer [
14,
15]. However, the role of lncRNA and its molecular mechanism involved in GC remain largely obscure. To systematically identify lncRNAs involved in the carcinogenesis of GC, we analyzed and integrated the results of our lncRNA microarray and Gene Expression Omnibus (GEO) database. Among the deregulated lncRNAs, we selected and investigated
MT1JP, a lncRNA located at 16q12.2 region. Here, we found that lncRNA
MT1JP acted as a competing endogenous RNA in regulating FBXW7 through sponging miR-92a-3p and inhibit cell proliferation, migration, invasion and promote cell apoptosis.
Methods
GC tissues
A total of 80 pairs of matched normal and GC tissues were collected from The Second Affiliated Hospital of Nanjing Medical University between February 2009 and October 2013. Five paired adjacent normal tissues and GC tissues were randomly selected in lncRNA microarrays study, and the remaining 75 paired gastric tissues were applied to qRT-PCR analysis. Additionally, another 330 paraffin-embedded GC tissues and corresponding follow-up information were obtained from Nantong Tumor Hospital between February 2008 and March 2013. All subjects have written informed consent and this study was approved by the Institutional Review Boards of Nanjing Medical University.
LncRNA microarrays
The lncRNA expression characteristics of GC were investigated by Arraystar Human LncRNA microarray V2.0, which contains 30,215 coding genes and 33,045 lncRNAs collected from several databases such as UCSC, Ensembl, RefSeq and the lncRNAs reported from literatures were also included. The microarray and data collection were conducted by KangChen Bio-tech (Shanghai, PR China). The details are as mentioned previously [
16]. In addition, non-coding RNA profiling GSE53137 from the same platform was downloaded from GEO database, which investigate lncRNAs expression in six pairs of human gastric adenocarcinoma and adjacent normal tissues. Paired
t-test was conducted to assess the differentially expressed lncRNAs between tumor and adjacent normal tissues (fold change > 2.0 and
P value < 0.05).
qRT-PCR analysis
The total RNA from GC tissue or cell lines were extracted using Trizol Reagent (Invitrogen, CA, USA) and mirVana miRNA Isolation Kit (Applied Biosystems) according to the manufacturer’s instructions. M-MLV reverse transcriptase (Invitrogen) was used for lncRNA
MT1JP reverse transcription. The expression of lncRNA
MT1JP and FBXW7 was detected by ABI 7900HT Real-Time PCR System (Applied Biosystem, Foster City, CA, USA), using SYBR Green assays (TaKaRa Biotechnology, Dalian, China) and GAPDH was used as the internal control. The expression of miR-92a-3p was measured using TaqMan MicroRNA Assays (Applied Biosystems) and U6 was treated as an internal control. All the primer sequences were available in Additional file
1: Table S2.
LncRNA coding capacity prediction
Coding Potential Assessment Tool (CPAT,
http://lilab.research.bcm.edu/cpat/) was used to assess the coding capacity of lncRNA
MT1JP. The CPAT conducted a logistic regression model by using the sequence features of open reading frame coverage, open reading frame size, hexamer usage bias and Fickett TESTCODE statistic. The CPAT chose 0.364 as a cutoff as human coding probability (CP). CP < 0.364 suggests noncoding sequence, whereas CP ≥ 0.364 indicates coding sequence [
17].
Nuclear-cytoplasmic fractionation
Nuclear/cytoplasmic fractionation was conducted by the Protein and RNA Isolation System (Ambion) according to the manufacturer’s protocols. U6 was treated as a nuclear control while GAPDH was a cytoplasmic control.
Cell proliferation, migration, invasion, apoptosis and cell cycle analysis
The Cell Counting Kit 8 (Dojindo) was used to measure cell viability. The spectrophotometric absorbance at 450 nm for each sample was detected using spectrophotometer Infinite M200 (Tecan). All the experiments were repeated three times in six replicates. The transwell assay was used to evaluate cell migration. Cell invasion was assessed using BioCoat Matrigel Invasion Chamber (BD Biosciences Discovery Labware). Cell numbers for cell migration and invasion in three random fields were counted. Cells were stained by Annexin V and propidium iodide using the Annexin V–FITC Apoptosis Detection kit (Invitrogen), and the percentage of apoptosis was examined with flow cytometry (BD Bioscience, San Jose, CA, USA). For detection of cell cycle, cells were stained with PI after 48 h transfection and examination were performed by FACS Calibur system (Beckman Coulter).
Luciferase reporter assay
The full-length lncRNA MT1JP cDNA was cloned into the BamHI and XhoI enzyme restriction sites of psiCHECK-2 vector (Promega) (psicheck-2-MT1JP-wild vector). The potential miR-92a-3p binding sites were mutated by the QuikChang site-directed mutagenesis kit (Agilent Technologies) (psicheck-2-MT1JP-mut vector). The psicheck-2-MT1JP-wild vector or psicheck-2-MT1JP-mut vector and miR-92a-3p mimics were co-transfected into BGC-823 and SGC-7901 cell by Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA). The wild and mutant miR-92a-3p was cloned into the BamHI and XhoI enzyme restriction sites of psiCHECK-2 vector (Promega). miR-92a-3p-wild or miR-92a-3p-mutant and lncRNA MT1JP overexpression vector were co-transfected into both SGC-7901 and BGC-823 cell by Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA). The luciferase activity was assessed by Dual-Luciferase Reporter Assay System (Promega, Madison, WI, USA) and the firefly luciferase activity was normalized by renilla luciferase activity.
All the cloned sequences were validated by DNA sequencing.
Twenty (10 nude mice in each group) five-week-old female athymic BALB/c nude mice were kept under pathogen-free conditions. BGC-823 cells that were transfected with an empty vector and the MT1JP vector were collected and resuspended at a concentration of 1.5 × 107 cells/mL. Then, 0.1 mL of the cells that were transfected with the MT1JP vector and an empty vector were subcutaneously injected into each posterior flank of the nude mice. The tumor weights and volumes were examined every 3 days. The mice were killed after 16 days post-injection, and the tumors from the mice were measured.
Immunohistochemistry (IHC)
Hematoxylin and eosin (H&E) staining was applied to select representative areas. The antibodies against Ki-67 (Abcam, Cambridge, MA, USA) were applied for IHC. The process of IHC was according to our previous study [
18].
Western blotting analysis
Protein was isolated from BGC-823 and SGC-7901 cell lines as previously stated [
19].
Forty micrograms of total protein s were run on a 14.7% polyacrylamide gel, and then transferred to polyvinylidene difluoride membranes (Hybondenhanced chemiluminescence; Amersham Pharmacia Biotech). The antibodies against FBWX7(1:1000) and β-actin (1:1000) were purchased from Abcam (Abcam, Cambridge, MA, USA). The antibodies against Caspase-9 antibody (1:1000) and Caspase-3 (1:1000) were purchased from Cell Signaling Technology (Cell Signaling Technology, USA). The protein was measured with a Phototope–horseradish peroxidase Western blot detection kit (Cell Signaling Technology, Inc.), and β-actin was treated as an internal control.
Statistical analysis
For continuous variables, the results were shown as mean ± SD. Student’s t-test was applied to compare the difference of means between two groups. Differentially expressed MT1JP between gastric cancer and normal tissues from TCGA database was also evaluated by Student’s t-test. We used Kaplan-Meier curve and log-rank test to evaluate the effect of lncRNA MT1JP on survival of GC patients. The relationship between lncRNA MT1JP and FBXW7 expression was assessed by Spearman Pearson correlation analysis. A two-sided P value < 0.05 was considered as statistically significant. All analyses were performed using SAS software (version 9.2; SAS Institute, Inc., Cary, NC, USA).
Discussion
There has been wide consensus that lncRNA play a key role in a variety of biological process, and aberrant expression or function of lncRNAs are commonly observed in a variety of cancers, including GC [
2,
20]. Recently, lncRNA microarray was widely performed to investigate the lncRNA expression signatures, and identified significantly altered lncRNAs in cancers. In this study, we applied lncRNA microarray and integrated a GEO dataset to depict the lncRNA profiles, and found that
MT1JP was the most significantly deregulated lncRNA with the largest Fold change. In the further validation study, the results showed that lncRNA
MT1JP was significantly down-regulated in GC tissues compared with adjacent normal tissues. In line with the our results, a previous study applied an mRNA/lncRNA microarray in 76 pairs of tumor and normal tissue sample from live, colon, lung and gastric cancer, and revealed that lncRNA
MT1JP had remarkably lower expression in all tumor samples [
21]. Furthermore, our study found that expression of lncRNA
MT1JP was prominently associated with lymph node metastasis and advance stage, suggesting a clinic pathological role of lncRNA
MT1JP in GC. We also detected the involvement of miR-92 and FBXW7 in disease stage and progression. The results showed miR-92 and FBXW7 were associated with the TNM stage (Additional file
1: Figure S7 and S8). In addition, log-rank results demonstrated that GC patients with higher expression of lncRNA
MT1JP had a well prognosis. These findings suggested that lncRNA
MT1JP may act as a promising prognosis biomarker for GC. Additionally, we employed the ROC curve to calculate the prognostic significance of combination of MT1JP and miR-92a. The result indicated combination of MT1JP and miR-92a could improve the predictive value for prognosis of GC patient (Additional file
1: Figure S9).
To further explore the effect of
MT1JP on cell phenotypes,
MT1JP overexpression vector was constructed and transfected into SGC-7901and BGC-823 cells. Compared with negative control, lncRNA
MT1JP overexpression significantly inhibited cell proliferation, migration and invasion and promoted cell apoptosis in both SGC-7901 cell and BGC-823 cell. Consistent with these findings in vivo, animal experimental study also confirmed that lncRNA
MT1JP overexpreesion inhibited tumor growth and metastasis. Intriguing, Liu et al. reported that knockdown of
MT1JP increased cell proliferation, migration and invasion while suppressed cell apoptosis in live cell line [
21]. These results suggest a tumor-suppressor role of lncRNA
MT1JP in GC and highlight the need for further study of the molecular mechanism of
MT1JP involving in GC.
Nuclear-cytoplasmic fractionation were conducted to assess the subcellular localization and showed that majority of lncRNA MT1JP were located in cytoplasmic.
Emerging evidence support the ceRNA hypothesis, and indicated that ceRNA regulation implicated in the carcinogenesis of GC [
22,
23]. It has been reported that lncRNA HOTAIR may act as a ceRNA to spong miR-331-3p and regulate the expression of HER2 in GC [
24]. Therefore, we speculated that lncRNA may function as a ceRNA, participate in carcinogenesis of GC. Bioinformatics analysis and luciferase reporter assay confirmed that lncRNA
MT1JP is a target of miR-92a-3p. Furthermore, our functional analysis also revealed that FBXW7 is a direct target of miR-92a-3p.
Given that both lncRNA MT1JP and FBXW7 interact with miR-92a-3p, suggesting that lncRNA MT1JP may regulate FBXW7 expression by competing bind to miR-92a-3p. In this study, the correlation analysis results showed that a significantly positive correlation between lncRNA MT1JP and FBXW7 expression was observed in GC tissues. Moreover, western blotting analysis also showed that miR-92a-3p mimic suppressed the expression of FBXW7, whereas lncRNA MT1JP overexpression substantially abolished this effect. Besides, although MT1JP was overexpressed, FBXW7 expression was not remarkably altered when knocked down miR-92a-3p. These results supported that lncRNA MT1JP involved in post-transcriptional regulation of FBXW7 by sponging miR-92a-3p.
Mounting evidence indicated that miR-92a-3p was often dysregulated in GC, and expression of miR-92a was remarkably related to the development of GC and had a prognosis value for GC [
25,
26]. In this study, compared with negative control, miR-92a-3p significantly increased cell proliferation, migration and invasion and inhibited cell apoptosis in BGC-823 cell. Intriguing, in line with our study, miR-92a-3p was reported to promote tumor growth in multiple tumors by targeting FBXW7 [
27‐
29]. FBXW7 is a member of SCF (complex of SKP1, CUL1 and F-box protein)-type ubiquitin ligase complex, which regulate target protein ubiquitination and degradation. The substrates of FBXW7 include several broadly investigated oncoproteins, such as MYC, cyclin E, mTOR and Notch. It is well known that FBXW7 is a tumor suppressor and its down-regulation was displayed in numerous human malignancies, including GC [
30‐
32]. FBXW7 was reported to regulate tumor apoptosis, growth arrest and epithelial-to-mesenchymal transition in GC, and it also played an important role in drug resistance [
33,
34]. The E-cadherin and MT1JP acted as the key proteins in the epithelial-to-mesenchymal transition. We use The Cancer Genome Atlas(TCGA) to explore the association between the FSP1and E-cadherin and MT1JP. The results indicated the expression of FSP1and E-cadherin were significantly associated with the expression of
MT1JP (Additional file
1: Figure S10), which suggested
MT1JP may play important role in epithelial-to-mesenchymal transition. In the present study, we have also shown that lncRNA
MT1JP regulate FBXW7 expression by competition for miR-92a-3p binding. Up-regulated lncRNA
MT1JP may lead to increase the expression of FBXW7, and consequently inhibited cell proliferation and promoted cell apoptosis in GC. These changes in GC cells could be reversed by miR-92a mimics and si-FBXW7 vector, according to rescue experiments in vitro.