LncRNA MAGI2-AS3 inhibits bladder cancer progression by targeting the miR-31-5p/TNS1 axis

In this study, we performed bioinformatics analysis to identify the competing endogenous RNAs (ceRNAs) that regulate bladder cancer (BCa) progression. RNA-sequencing data analysis identified 2451 differentially expressed mRNAs, 174 differentially expressed lncRNAs, and 186 microRNAs (miRNAs) in BCa tissues (n=414) compared to the normal urothelial tissues (n=19) from the TGCA database. CeRNA network analysis of the differentially expressed lncRNAs and mRNAs showed strong positive correlation between lncRNA MAGI2-AS3 and Tensin 1 (TNS1) mRNA in BCa tissues. Bioinformatics analysis also showed that both MAGI2-AS3 and TNS1 mRNA sequences contain miR-31-5p binding sites. Furthermore, we observed significantly lower MAGI2-AS3 and TNS1 mRNA expression and higher miR-31-5p expression in the BCa tissues and cell lines (T24 and J82) compared with their corresponding controls. Functional and biochemical experiments in BCa cell lines including luciferase reporter assays showed that MAGI2-AS3 upregulated TNS1 by sponging miR-31-5p. Transwell assays showed that the MAGI2-AS3/miR-31-5p/TNS1 axis regulated migration and invasion ability of BCa cell lines. Moreover, immunohistochemical staining of paired BCa and normal urothelial tissues showed that low expression of TNS1 correlated with advanced tumor (T) stages and lymph node metastasis in BCa. In conclusion, our study demonstrates that the MAGI2-AS3/miR-31-5p/TNS1 axis regulates BCa progression.


INTRODUCTION
Bladder cancer (BCa) is the 10 th most common cancer with 549,000 new cases and 200,000 deaths reported worldwide in 2018 [1]. In the United States, 81400 new cases and 17980 BCa-related deaths are projected for 2020 [2]. The standard treatments for BCa patients include surgery, radiotherapy and chemotherapy, but, the prognosis of patients with advanced-stage BCa remains poor [3]. Therefore, new diagnostic and prognostic biomarkers are required to guide early diagnosis and effective treatments of BCa patients to improve their survival outcomes.
Several reports have shown that competing endogenous RNAs (ceRNAs) such as small non-coding RNAs, pseudogenes, long noncoding RNAs (lncRNAs) and circular RNAs (circRNAs) modulate expression of their target genes by binding to microRNAs (miRNAs) [4][5][6][7]. LncRNAs are transcripts without any protein-coding potential that are approximately 200 nucleotides in length and act by sponging miRNAs and enhancing the expression of their target mRNAs [8][9][10]. The regulatory role of lncRNAs has been widely studied in BCa. For example, lncRNA DANCR enhances the malignancy of BCa cells by sponging miR-149 and increasing the expression of MSI2 [11]. LncRNA SPRY4-IT1 promotes proliferation and metastasis of BCa cells by sponging miR-101-3p and increasing the expression of EZH2 [12].
Tensin 1 (TNS1) is a focal adhesion molecule that has been implicated in the migration of normal and tumor cells [13]. The role of TNS1 is controversial in different AGING cancers. Low expression of TNS1 promotes metastasis and invasion of prostate cancer and breast cancer cells [14,15]. However, high expression of TNS1 promotes metastasis and invasion of colorectal cancer cells [16,17]. In this study, we performed bioinformatics analyses and functional in vitro experiments to determine the regulatory mechanisms underlying the expression of TNS1 in BCa tissues and cell lines.

LncRNA MAGI2-AS3 and TNS1 mRNA expression is decreased and miR-31-5p expression is upregulated in BCa tissues from the TGCA database
We analyzed the RNA-sequencing data from 412 patients with BCa and normal bladder tissues using log 2 FC >2.0 and adjusted P value (FDR) < 0.01 as threshold parameters to identify differentially expressed mRNAs, lncRNAs, and miRNAs. The volcano maps of differentially expressed lncRNAs, miRNAs, and mRNAs are shown in Figure 1A-1C. We identified 2451 differentially expressed mRNAs (818 up-regulated and 1633 down-regulated; Supplementary Next, we constructed a ceRNA regulatory network to identify functionally significant interactions between differentially expressed lncRNAs, miRNA and mRNAs in BCa tissues. We analyzed the correlations between differentially expressed lncRNAs and mRNAs using GDCRNATools and shinyCorPlot functions in R language and ranked lncRNA-mRNA pairs as shown in Table 1. The correlation between MAGI2-AS3 and TNS1 was ranked second. Since the functional relationship between MAGI2-AS3 and TNS1 has not been investigated in BCa, we chose them for further investigation. The distribution map of MAGI2-AS3 and TNS1 in BCa patients is shown in Figure 1D. We identified miR-31-5p as the top ranked putative target of MAGI2-AS3 with a complementary strand sequence, 5' AUCUUGCC-UAGAACGG-3' (Supplementary Table 4). Starbase analysis showed that TNS1 was a potential downstream target gene of miR-31-5p (Supplementary Table 5). The analysis of BCa patient tissues from the TGCA database showed that the expression of MAGI2-AS3 and TNS1 was significantly downregulated and miR-31-5p was significantly upregulated in BCa tissues compared to the normal urothelial tissues ( Figure 1E-1G).

MAGI2-AS3, miR-31-5p and TNS1 expression levels in BCa tissues and cell lines
We performed qRT-PCR analysis to verify the expression of MAGI2-AS3, miR-31-5p and TNS1 in 45 pairs of BCa and adjacent normal urothelial tissues. QRT-PCR results showed that MAGI2-AS3 and TNS1 levels were down-regulated and miR-31-5p levels were up-regulated in BCa tissues compared to the adjacent normal urothelial tissues (Figure 2A-2C). Moreover, MAGI2-AS3 and TNS1 levels were significantly reduced and miR-31-5p levels were increased in the BCa cell lines (T24 and J82) compared to the normal urothelial cell line, SV-HUC-1 ( Figure 2D-2F). Furthermore, western blot analysis showed that TNS1 protein expression was significantly downregulated in the BCa cell lines compared to the normal urothelial cell line, SV-HUC-1 ( Figure 2G). These results demonstrate the regulatory relationship between MAGI2-AS3, miR-31-5p and TNS1 in BCa tissues.
tensin-like (CTEN) [18][19][20]. TNS1 is a focal adhesion molecule that binds the actin cytoskeleton to the integrins and also forms a signaling complex through its multiple binding domains [19,20]. We postulated that TNS1 may be related to tumor progression in BCa. Therefore, we analyzed the expression of TNS1 in 45 pairs of BCa and adjacent normal urothelial tissues at different T stages and the results are summarized in  Table 2. Overall, the expression of TNS1 was significantly lower in the BCa tissues compared to normal bladder tissues; moreover, the expression of TNS1 was significantly lower in the advanced T stages (T3-4) compared to the patients in the lower T stages (T1-2) ( Figure 5C). Based on the chi-square test, we found low expression of TNS1 correlated with lymph node metastasis in BCa patients (Table 2). Therefore, our results demonstrate that low TNS1 expression correlates with tumor progression and poor prognosis in BCa patients.

DISCUSSION
LncRNA are non-protein coding RNA molecules that are ≥200 nucleotides in length and play significant roles in critical cellular functions under normal physiological and pathophysiological conditions [21,22]. Several lncRNAs have been identified as diagnostic and prognostic biomarkers in many cancer types. For example, lncRNA PCA3 is highly expressed in prostate cancer [23]. PCA3 levels in urine show higher sensitivity in prostate cancer diagnosis than the standard PSA tests [24,25]. LncRNA MALAT1 expression inversely correlates with the progression and metastasis of breast cancer [26,27].
The roles of several lncRNAs have been investigated in BCa [28][29][30][31][32][33]. UCA1 is very sensitive and specific for rhe diagnosis of BCa [29], and it can also affect the progress of BCa [30,31]. TUG1 can mediate cell proliferation affect the metastasis and grading of BCa [32,33]. In this study, we demonstrate that lncRNA MAGI2-AS3 expression was significantly reduced in  BCa tissues compared to the normal bladder tissues. Moreover, we demonstrate that the MAGI2-AS3/miR-31-5p/TNS1 axis regulates proliferation, metastasis and invasion of BCa cells. The role of MAGI2-AS3 has also been investigated in other cancers. MAGI2-AS3 inhibits invasion and metastasis of breast cancer cells by sponging mir-374a [34]. MAGI2-AS3 also regulates proliferation and metastasis of hepatocellular carcinoma cells through the mir-374b-5p/SMG1 axis [35].
Tensin protein family members regulate metastasis in many cancers [18]. TNS1 is involved in cell migration and invasion [36]. TNS1/miR-548j axis regulates invasion and metastasis of breast cancer cells [14]. TNS1 expression also correlates with bone metastasis of prostate cancer [15]. In the present study, we demonstrate that low expression of TNS1 is associated with advanced T stages and lymph node metastasis. Therefore, TNS1 is a potential prognostic biomarker for BCa.
In conclusion, our study demonstrates that the MAGI2-AS3/miR-31-5p/TNS1 axis regulates BCa progression. Hence, MAGI2-AS3, miR-31-5p, and TNS1 are potential prognostic biomarkers that can predict survival outcomes of BCa patients. Further studies are needed to verify the findings of our study.

Data retrieval from the TCGA database
We searched the TCGA database (https://portal.gdc.cancer.gov) using TCGA-BLCA as project ID and downloaded clinical data for 412 BCa patients. We also downloaded RNA-seq data for 414 primary BCa and 19 normal urothelial tissue samples, and miRNA-seq data for 417 primary BCa and 19 normal urothelial tissue samples.

Identification of differentially expressed mRNAs and lncRNAs
We removed the duplicate samples, non-primary tumor and non-solid normal tissue samples from the RNA-seq and miRNA-seq datasets. Then, we integrated and normalized the gene expression data of the BCa patients in comparison to normal urothelial tissues by using the calcNormFactors function and screened for differentially expressed mRNAs and lncRNAs using the R programming language.

Target gene prediction and functional enrichment analysis
We used the Starbase database to determine the direct relationship between lncRNAs and miRNAs as well as miRNAs and mRNAs, and the target binding sites in each pair. We used the STRING (https://string-db.org) database to identify related proteins and performed AGING functional enrichment analysis using DAVID (https://david.ncifcrf.gov).

Clinical samples from BCa patients
We obtained 45 pairs of cancer and adjacent normal tissues from 45 BCa patients who underwent radical cystectomy at the Xiangya Hospital. These patients had not received any radiotherapy or chemotherapy before surgery. The tissues were stored in liquid nitrogen before analysis. This study was approved by the Ethics committee of Xiangya Hospital. We obtained signed written informed consent from all patients.

Cell culture
The human normal urothelial cell line (SV-HUC-1) and BCa cell lines (T24 and J82) were purchased from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China) and cultured with Dulbecco modified Eagle medium (DMEM) containing 10% fetal bovine serum (FBS) in a humidified chamber at 37° C and 5% CO2.

Quantitative real time PCR
We extracted total cellular or tissue RNA using TRlzol reagent (Thermo Fisher Scientific, Shanghai, China). Equal amounts of total RNAs were reverse transcribed into cDNA using the PCR kit (Takara, Dalian, China). Then, q-PCR was performed in the ABI7500 Real-Time PCR system and the expression of MAGI2-AS3, TNS1 and miR-31-5p was evaluated using the 2 -ΔΔCt method relative to corresponding controls.

Generation of MAGI2-AS3 silenced BCa cells
We knocked down the expression of MAGI2-AS3 by transfecting BCa cell lines with siRNAs against MAGI2-AS3 (Genepharma, Suzhou, China). Briefly, 2 × 10 5 BCa cells were seeded in each well in 6-well plates and transfected with si-NC (control) or si-MAGI2-AS3 using lipofectamine for 48 h according to manufacturer's instructions. The efficiency of MAGI2-AS3 knockdown was analyzed by qRT-PCR.

Generation of MAGI2-AS3 overexpressing BCa cells
We first cloned the MAGI2-AS3 DNA sequence into the GV358 lentiviral vector (Genechem, Shanghai, China). We then co-transfected the 293T cells with oeMAGI2-AS3 vector plasmid and helper plasmids for 48 h. We collected the supernatants and purified the lentiviruses carrying the oeMAGI2-AS3 plasmid by centrifugation and further concentrated the virus through ultrafiltration. The ultrapure oeMAGI2-AS3 plasmid and polybrene was incubated with T24 and J82 cells (2 × 10 5 cell/ml) for 48 h. The efficiency of MAGI2-AS3 overexpression in the BCa cells was confirmed by observing the GFP fluorescence under a fluorescence microscope.

Western blotting
We prepared total protein lysates using the RIPA lysis buffer and quantified the proteins using the BCA protein assay. Equal amounts of total cell protein lysates were separated by 10% sodium dodecyl sulfate (SDS) polyacrylamide gel electrophoresis and transferred onto the polyvinylidene fluoride (PVDF) membranes. The membranes were blocked with 5% skimmed milk for 10 minutes. Then, the membranes were incubated overnight with the primary antibody against TNS1 (Proteintech, Wuhan, China) at 4 °C. Then, the membranes were incubated with secondary antibodies (Proteintech, Wuhan, China). The blots were developed using ECL and the protein bands were captured on a film through scanning. The integrated optical density (IOD) value of the protein bands were evaluated using Total Lab Quant V11.5 (Newcastle upon Tyne, UK,)

Luciferase reporter assay
We PCR amplified the miR-31-5p sequence containing binding sites for wild-type MAGI2-AS3 and TNS1 according to the Starbase and cloned into the pGL3 reporter plasmid (Promega, Beijing, China). Then we constructed target plasmids containing wild-type or mutant MAGI2-AS3 and TNS1 constructs (Thermo Fisher Scientific, Shanghai, China). We cultured T24 cells in a 6-well plate for 24 h, and then transfected them with different combinations of reporter and target plasmids for 48 h. We used the GloMax 20/20 fluorescence detector (Promega, Beijing, China) to detect the fluorescence intensity of the reporter gene plasmid.

MTT cell proliferation assay
We incubated 5 × 10 3 cells / well in 96-well plates for 24, 48, 72, and 96 h. When the incubation time was reached, we added 10 μl of MTT solution into each well and incubated further for 4 h. We then added 150 μl DMSO to dissolve the crystals in each well, and measured the absorbance at 490 nm in a Fluoroskan microplate reader (Thermo Fisher Scientific, Shanghai, China).

Transwell migration and invasion assay
To determine the BCa cell migration ability, we added 5 × 10 4 T24 and J82 cells in 500 μl DMEM without FBS into the upper chambers and 500 μl DMEM medium with 20% FBS into the lower chamber and incubated the Transwell chambers in a humidified incubator at 5% CO 2 and 37° C for 48 h. Then, the cells in the upper chamber were removed and the cells in the lower chamber were fixed with methanol for 15-20 minutes, and stained with 0.1% crystal violet for 15 minutes. The average numbers of cells in five randomly selected areas were counted for each sample under a light microscope to determine their cell migration ability. To determine the invasive ability of the cells, we added Matrigel (ECM) between the Transwell chambers and repeated the experiment as described for the migration assay.

Immunohistochemistry (IHC)
The sections of BCa and normal adjacent tissues were deparaffinized and permeabilized. Then, the specimens were incubated with the anti-TNS1 antibody (Proteintech, Wuhan, China) overnight at 4 o C. Then, we incubated the specimens with the secondary antibody (Proteintech, Wuhan, China). The color was developed with DBA and the stained samples were imaged with a Nikon E200 microscope.

Statistical analysis
All experiments were performed at least in triplicates. The data are expressed as means ± standard deviation (S.D). Differences between samples were evaluated using two-tailed t-tests. Values of P < 0.05 were considered statistically significant. GraphPad software is used to determine the P value.

AUTHOR CONTRIBUTIONS
Congyu Tang performed the experiments and wrote the manuscript; Yi Cai, Huichuan Jiang, Zhengtong Lv, Changzhao Yang, Haozhe Xu and Zhi Li collected clinical samples and performed data analysis; Yuan Li designed and revised the manuscript. All the authors have read and approved the final manuscript.

Supplementary Tables
Supplementary Table 1