A positive feedback loop between TAZ and miR-942-3p modulates proliferation, migration, epithelial-mesenchymal transition process and glycometabolism in human bladder cancer

Background: Transcriptional co-activator with PDZ-binding motif (TAZ) has been reported to involve in tumor progression, epithelial-mesenchymal transition (EMT) process and glycometabolism modulation. Herein, the underlying molecular mechanisms of TAZ-induced biological effects in bladder cancer were discovered; Methods: qRT-PCR, western blot and immunohistochemistry were performed to determine the level of TAZ in bladder cancer cells and tissues; CCK-8 assay, Colony formation assay, wound healing assay and Transwell assay were performed to evaluate the functions of TAZ, miR-942-3p and GAS1. qRT-PCR and western blot were used to determine the expression levels of related genes. Chromatin immunoprecipitation and dual-luciferase reporter assay conrmed the interaction between TAZ and miR-942. In vivo tumorigenesis assay and colorimetric assay of glycolysis were also conducted; Results: We determined the upregulation and vital roles of TAZ in bladder cancer. TAZ-induced upregulation of miR-942-3p amplied upstream signaling by inhibiting the expression of large tumor suppressor 2 (LATS2, a TAZ inhibitor). MiR-942-3p attenuated the suppression of cell proliferation, EMT process and glycolysis induced by TAZ knockdown. Further, miR-942-3p resulted in restrained expression of growth arrest-specic 1 (GAS1) to modulate biological functions; Conclusion: Our study identied a novel positive feedback loop between TAZ and miR-942-3p that regulates biological functions in bladder cancer cells via GAS1 expression, and illustrated that TAZ and miR-942-3p might be potential therapeutic targets for bladder cancer treatment.

regulates biological functions in bladder cancer cells via GAS1 expression, and illustrated that TAZ and miR-942-3p might be potential therapeutic targets for bladder cancer treatment.

Background
Bladder cancer is the ninth most common malignant tumor worldwide and ranks 13th in mortality from cancer every year. There were approximately 430,000 new-set cases in 2012 [1][2][3][4] . Bladder cancer can be classi ed into to non-muscle invasive and muscle-invasive bladder cancer according to the depth of tumor in ltration 5,6 . Although timely surgical intervention and chemotherapy can restrict the progression and development of tumor, the 5-year overall survival rate of muscle-invasive bladder cancer patients is 60% due to distant metastasis 7 . Therefore, it is meaningful to clarify the molecular mechanisms underlying bladder cancer progression for discovering the novel and precise therapeutic targets and improving prognosis in bladder cancer.
In the past decades, numerous studies have con rmed the critical roles of Hippo pathway in tissue homeostasis, cell proliferation, apoptosis and multiple biological processes. Surprisingly, impairment of the Hippo pathway leads to a remarkable impact on cancer development, progression and metabolic phenotype [8][9][10][11][12][13] . The core of the Hippo pathway is composed of a kinase cascade, transcription coactivators, and DNA-binding partners. The pathway is regulated by intrinsic cell machineries and various cellular signals 14 . The upstream serine/threonine kinases Mst1/2 (mammalian sterile twenty-like) can phosphorylate and activate Lats1/2 (large tumor suppressor) via a complex with the adaptor protein Sav1. Then activated Lats1/2, together with Mob1, suppresses transcriptional coactivators TAZ (transcriptional coactivator with PDZ-binding motif) and its paralogue YAP (Yes-associated protein) through phosphorylation 15 . TAZ interact with the TEA domain DNA-binding family of transcription factors (TEAD) to recruit to their target promoters and regulate gene expression 16 . In mammals, the transcriptional activation of TEADs requires transcriptional coactivators, such as TAZ, YAP and p160 family of nuclear receptor coactivators 17 .
Epithelial-mesenchymal transition (EMT) plays essential roles during normal mammalian development in which epithelial cells acquire mesenchymal features. However, EMT is associated with tumorigenesis and metastasis and is unintentional in cancer progression [18][19][20] . Therefore, the EMT-related signaling pathways have been studied as a novel focus for cancer therapy in past decades [21][22][23][24] . Interestingly, emerging studies have demonstrated the regulatory role of the components in Hippo signaling pathway in EMT process [25][26][27][28][29] .
MicroRNAs regulate the level of protein-coding genes by binding to the speci c sequence of mRNAs 30 . A growing number of researches have reported that microRNAs involve in multiple aspects of biologic cellular processes, including cancer development and progression, making them novel therapeutic targets for treatment 24,31,32 . Moreover, the dysregulation of miRNAs and their in uences on tumorigenesis, development, and progression have been discovered in bladder cancer 33,34 . GAS1 (Growth arrest-speci c 1) is a well-known cell growth suppressor 35 and involves in tumorigenesis and progression [36][37][38] . Aberrant expression of GAS1 reduce tumorigenicity in human brain tumor-initiating cells 39 , while downregulation of GAS1 is a potential biomarker of clear cell renal cell carcinoma 40 .
Interestingly, GAS1 has been reported to serve as a novel biomarker and inhibit proliferation, EMT process and glycolysis in colorectal cancer 41 .
In the current study, we investigated the abundant expression of TAZ in both bladder cancer cell lines and tissues. Besides, TAZ knockdown suppressed the proliferation, migration, EMT process and glycolysis of bladder cancer cells. Mechanically, we identi ed a positive feedback loop between TAZ and miR-942-3p, which enhanced upstream signals and modulated biological and metabolic phenotypes via regulating GAS1 expression. Collectively, our results indicated that TAZ, miR-942-3p and GAS1 are novel therapeutic targets in clinical intervention of bladder cancer patients.

Luciferase reporter assay
For verifying the transcriptional activity of TAZ-TEAD on miR-942 promoter, 293T cells were plated in 96well plates (5000 cells) and were co-transfected with Fire y luciferase plasmid miR-942 promoter, TEAD2 and TAZ plasmid (Transheep, China). pRL-CMV Renilla luciferase were also co-transfected to normalize the luciferase activities. Plasmids were transfected by Lipofectamine ™ 3000 Transfection Reagent (Invitrogen, USA). Dual-luciferase Reporter Assay Detection Kit (Promega, USA) was performed for cell luciferase activities detection.
For con rmation the relationship between miR-942-3p and GAS1 or LATS2, a luciferase reporter vector (pGL3-Fire y_Luciferase-Renilla_Luciferase) with full length of the 3'-UTR of GAS1 or LATS2 was constructed and the mutant vectors were also generated (GeneChem, China). 293T cells were seeded and co-transfected with a luciferase vector and miR-942-3p mimic or negative control. The Dual-luciferase Reporter Assay Detection Kit (Promega, USA) was used for the re y and Renilla luciferase activities measure.
Migration and invasion assays 10 4 cells in serum-free medium were seeded into the upper chambers. Speci cally, for invasion assay, Matrigel (BD, MA, USA) pre-coated upper chambers were used. The bottom chambers were added with 10% fetal bovine serum containing medium. Migrated and invasive cells were xed with fomalin and stained with crystal violet. The stained cells were observed and counted by microscopy (200×) in ve randomly selected elds.
Wound healing assay Bladder cancer cells were plated into Culture-Insert (ibidi, Germany) according to manufacturer's instructions. After 24 h, Culture-Inserts were removed and the cells were incubated in serum free medium. Photos were captured at 0 and 24 h after insert removed and the migration rate of the cells was measured by Image J software.

Apoptosis analysis
Cells with different condition were counted and stained with Annexin V-FITC and PI. The apoptotic rate of cells was detected by ow cytometry (Becton Dickinson, USA) after incubation for 30 min at room temperature.

Western blot analysis and antibodies
Total proteins were extracted by using RIPA buffer (C1053, APPLYGEN, Beijing, China), and protein concentration measurement was performed. Proteins were then separated by SDS-PAGE and transferred to PVDF membranes (Millipore, Billerica, MA). After blocked for 2 h in 5% milk, the membrane was rinsed with TBST for 3 times and was incubated with primary antibodies at 4°C overnight. The membrane was then incubated with secondary antibody (anti-mouse or anti-rabbit IgG Cell Signaling Technology, USA) for 1 h at room temperature. A Bio-Rad detection system was used to detect the bands. The antibodies used in this study were as follows: anti-GAPDH (5174), anti-TAZ (70148), anti-E-cadherin (3195), anti-Snail (3879), anti-N-cadherin (13116), anti-Vimentin (5741) and anti-Fibronectin (26836) were all obtained from Cell Signaling Technology. Anti-LATS2 (ab110780), anti-GAS1 (ab236618), anti-PFKFB3 (ab181861), anti-HK2 (ab209847) and anti-GLUT1 (ab115730) were all obtained from Abcam.

Glycolysis process evaluation
The uptake of Glucose in different cells was determined by Glucose Uptake Colorimetric Assay Kit (Biovision, CA, USA). The production of lactate was detected by using Lactate Colorimetric Assay Kit (Solarbio, Beijing, China). All the procedures were carried out according to the manufactures' instructions.
Other in vitro experiments CCK-8 assay, colony formation assay and HE staining were performed according to previously described methods 24 .
In vivo studies For establishment of xenograft tumorigenesis model, 5×10 6 of T24 cells with different treatments were injected subcutaneously into nude mice. The mice were sacri ced and the tumors were excised, weighed, and xed with formalin for immunohistochemical examination 25 days later. Tumor sizes were measured daily and calculated by the following formula: Total tumor volume (mm 3 ) = L×W 2 /2 ("L" = length and "W" = width). All animal-related procedures were approved by the Animal Care and Use Committee of The First A liated Hospital of the School of Medicine of Zhejiang University.

Statistical analysis
All data were expressed as the mean ± standard deviation (SD) and were analyzed with SPSS. A paired ttest was used to analyze the difference expression in both cells and tissues. A two-tailed p-value < 0.05 was regarded as statistically signi cant in this study.

TAZ expression is markedly elevated in bladder cancer cells lines and tissue
First, we investigated TAZ expression difference between human bladder cancer tissues and their matched adjacent normal tissues (cohort 1, n=20). As shown in Fig. 1a and 1b, TAZ was overexpressed in the bladder cancer tissues at both mRNA and protein levels. Meanwhile, the data in TCGA database represented the similar trend (Fig. 1c). We then detected TAZ level in SV-HUC-1 and bladder cancer cell lines. The results revealed that TAZ was markedly upregulated in bladder cancer cell lines compared with normal human uroepithelium cells (Fig. 1d and 1e). Furthermore, we performed immunohistochemical assay and found that TAZ was higher in bladder cancer tissue than that in adjacent normal tissue in cohort 2 (n=30, Fig. 1f).
TAZ knockdown suppresses proliferation, colony formation, migration, invasion and mesenchymal transformation of bladder cancer cells To evaluated the potential roles of TAZ in bladder cancer cells, we used a short interfering RNA (siRNA) to deplete TAZ expression. The e ciency of siRNAs was evaluated by western blotting in T24 cells (Fig. 2a). CCK-8 assay and colony-formation assay demonstrated that TAZ knockdown suppressed cell proliferation and colony-forming ability ( Fig. 2b and 2c). Wound healing assay indicated that TAZ was involved in bladder cancer cell migration (Fig. 2d). In addition, depletion of TAZ signi cantly decreased migrated cells in Transwell migration and Matrigel invasion assays in T24 and EJ cells (Fig. 2e). Previous researches have reported that cell migration and EMT process are closely associated with tumor migration, invasion, progression and metastasis [18][19][20]22 . Interestingly, we found that TAZ expression was positively correlated with that of mesenchymal markers (Vimentin, N-cadherin, Fibronectin and Snail) and negatively correlated with that of epithelial marker (E-cadherin) (Fig. 2f). The above results indicated the key role of TAZ in EMT process. As we predicted, TAZ inhibition led to downregulation of N-cadherin, Vimentin, Fibronectin and Snail and upregulation of E-cadherin at protein level and mRNA level (Fig. 2g  and 2h). Our current results demonstrated that TAZ served as a vital regulator in various biological functions of bladder cancer cell.

Inhibition of TAZ suppresses glycolysis via regulating PFKFB3, HK2 and GLUT1 expression in bladder cancer cells
Numerous researches have reported that Hippo pathway is involved in glucose metabolism 12,13 .
Meanwhile, the speci c metabolic phenotype is well accepted as a characteristic of tumor cells. Therefore, we evaluated the role of TAZ in glycometabolism of bladder cancer cells. Fig. 2i illustrated the process of aerobic glycolysis. As shown in Fig. 2j and 2k, inhibition of TAZ signi cantly decreased uptake of glucose and production of lactate, which indicated the suppression of glycolysis. In addition, we also detected the expression levels of several glycolysis-related genes to clarify the underlying regulatory mechanism of TAZ on glycolysis. The results demonstrated that TAZ knockdown could decrease the levels of PFKFB3, HK2 and GLUT1 at mRNA level and protein level ( Fig. 2l and 2m). These results illustrated that TAZ might function as a key factor in aerobic glycolysis of bladder cancer.

MiR-942-3p is regulated by TAZ-TEAD
To clarify the greater details about regulatory mechanism of TAZ in bladder cancer, we veri ed miRNAs regulated by TAZ in T24 bladder cancer cells by RNA sequencing (Fig. 3a and Supplemental File 2). Among the downregulated miRNAs induced by TAZ knockdown, we found that miR-942-3p has been reported to be a key factor in cell proliferation and invasion 42,43 and selected miR-942-3p as our research subject. qRT-PCR was used to con rm the remarkable downregulated miR-942-3p in TAZ-knockdown bladder cancer cells (Fig. 3b). Furthermore, TAZ overexpression could enhance the miR-942-3p expression (Fig. 3c). Besides, the expression levels of TAZ and miR-942-3p were positively correlated both in cohort 1 and TCGA database ( Fig. 3d and 3e). We also found the alternations of miR-942-3p in LATS1 and LATS2 or TEAD2 knockdown cells, which indicated that Hippo signaling pathway play a key role in TAZ/miR-942-3p regulatory mechanism ( Fig. 3f and 3g). TAZ interacts with the TEAD to recruit to their target promoters and regulate gene expression 16,17,44 . Therefore, we predicted the binding site of TEADs in the promoter region of miR-942 in JASPAR (Fig. 3h). We then performed chromatin immunoprecipitation (ChIP) to investigate the interaction between TAZ and miR-942 promoter. qRT-PCR detection for miR-942 promoter was performed and analyzed by employing different primers (indicated in Fig. 3h). As shown in Fig. 3i, TAZ could enrich the DNA segments containing the predicted TEADs binding site, which could be ampli ed by using primer 3 and 4. Luciferase reporter assay further con rmed that the ability of TAZ-TEAD2 to activate the promoter of miR-942 depending on the presence of the TEAD-binding sequence (Fig. 3j). In summary, these results veri ed that miR-942 was a direct target of TAZ-TEAD.
MiR-942-3p acts as a tumor promotor in bladder cancer We rst detected the expression level of miR-942-3p in bladder cancer cells and tissues. The results con rmed the upregulation of miR-942-3p in bladder cancer tissues (cohort 1) and cell lines (Fig. 4a-c), which indicated that miR-942-3p was associated with tumorigenesis and progression in bladder cancer. To verify this hypothesis, we transfected T24 and EJ cells with Pre-miR-942 or miR-942-3p sponge. CCK-8 assay and colony formation assay revealed that miR-942-3p overexpression enhanced cell viability and proliferation, while miR-942-3p inhibition led to the opposite results ( Fig. 4d and 4e). Besides, wound healing, Transwell migration and invasion assays were performed to evaluate the migratory and invasive abilities. The results showed that miR-942-3p acted as a promoter in bladder cancer cell migration and invasion ( Fig. 4f and 4g). We also detected EMT markers by western blotting and qRT-PCR in miR-942-3p overexpression or knockdown cells to investigate whether EMT process was regulated by miR-942-3p. As shown in Fig. 4h and 4i, miR-942-3p enhanced the expression of E-cadherin and suppressed the expression of N-cadherin, Vimentin, Fibronectin and Snail at protein level and mRNA level. Conversely, miR-942-3p inhibition exhibited opposite effects in bladder cancer cells. Our data demonstrated that miR-942-3p was a TAZ-induced and abundantly expressed miRNA in bladder cancer, which played a signi cant role in cell migration, invasion and EMT process.

GAS1 and LATS2 act as direct targets of miR-942-3p in bladder cancer
We then identi ed the molecular mechanism of miR-942-3p in bladder cancer by predicting the target genes of miR-942-3p in several databases (TargetScan, miRWalk, miRDB and miRTarBase). Firstly, four lists of genes overlapped, and four genes were targeted by miR-942-3p in all the databases (Fig. 5a). GAS1 is a well-known cell growth suppressor and involves in tumorigenesis and progression [39][40][41]45 . Interestingly, apart from the above 4 genes, we found that LATS2, as a signi cant component of Hippo signaling pathway, was targeted by miR-942-3p in TargetScan. Therefore, GAS1 and LATS2 were selected for further veri cation. The potential binding sites of miR-942-3p within the 3'-untranslated regions (3'-UTRs) of GAS1 and LATS2 were showed in Fig. 5b. To validate whether miR-942-3p binds to these 3'-UTRs, we constructed luciferase reporter plasmids containing the full length 3'-UTR of GAS1 or LATS2 and their mutant types and co-transfected them with miR-942-3p mimics. MiR-942-3p mimics induced a remarkable reduction in luciferase activity of the wild type GAS1 and LATS2 3'-UTRs. Besides, the mutated forms of the above 3'-UTRs eliminated the reduction in luciferase activity induced by miR-942-3p mimics in wild type 3'-UTRs (Fig. 5c). In addition, GAS1 and LATS2 were negatively correlated with miR-942-3p in TCGA database and bladder cancer cell lines at mRNA level and protein level, respectively ( Fig.  5d and 5e). These ndings demonstrated that miR-942-3p speci cally targets GAS1 and LATS2.
A TAZ/miR-942-3p positive feedback loop regulates the oncogenic effects of TAZ Since LATS2 inhibits the activity of TAZ in Hippo signaling pathway, we considered that TAZ and miR-942-3p formed a positive feedback loop in bladder cancer. To verify this regulatory mechanism, we rst investigated the biological functions in TAZ-knockdown cells with or without overexpressing miR-942-3p. Notably, upregulation of miR-942-3p partly attenuated the suppression of cell viability and proliferation induced by TAZ depletion (Fig. 6a and 6b). Transwell migration, Matrigel invasion assays along and wound healing assay indicated the rescue of migration and invasion abilities induced by miR-942-3p in TAZ-depleted cells (Fig. 6c and 6d). Additionally, the alternations in E-cadherin, Vimentin, N-cadherin, Fibronectin and Snail at protein level and mRNA level demonstrated that EMT process were reversed by miR-942-3p in TAZ knockdown cells (Fig. 6e and 6f). Moreover, the miR-942-3p-mediated expression levels of LATS2, GAS1 and TAZ were also indicated in Fig. 6e. We then veri ed whether miR-942-3p could reverse the metabolic phenotype induced by TAZ knockdown. The results of glucose uptake and lactate production assays showed that glycolysis was enhanced by miR-942-3p in TAZ-de cient cells (Fig. 6g  and 6h). PFKFB3, HK2 and GLUT1 were also upregulated by miR-942-3p at mRNA level and protein level ( Fig. 6i and 6j). In summary, our results proved that a positive feedback loop between TAZ and miR-942-3p was involved in biological functions, EMT process and glycometabolism of bladder cancer cells.

GAS1 suppresses the aggressive characteristics and glycolysis of bladder cancer cells
We assessed the effects of GAS1 in bladder cancer cells by constructing stable GAS1-overexpressing cells via transducing lentivirus. CCK-8 assay and colony formation assay showed that GAS1 inhibited cell viability and proliferation of bladder cancer cells (Fig. 7a and 7b). Moreover, overexpressing GAS1 signi cantly reduced the relative percentage of wound closure (Fig. 7c). Transwell migration and invasion assays further re ected the attenuation of migration and invasion induced by GAS1 overexpression in T24 and EJ (Fig. 7d). To demonstrate whether GAS1 suppressed migration and invasion by regulating EMT process, we assessed the level of EMT markers expression at protein level and mRNA level. The results indicated that overexpression of GAS1 enhanced E-cadherin expression and decreased the levels of Vimentin, N-cadherin, Fibronectin and Snail ( Fig. 7e and 7f). Furthermore, the detections of glucose uptake and lactate production showed that GAS1 inhibited cellular glycolysis ( Fig. 7g and 7h). qRT-PCR and western blotting indicated the downregulations of PFKFB3, HK2 and GLUT1 at mRNA level and protein level (Fig. 7i and 7j), which were consistent with previous results induced by TAZ inhibition. In addition, both TCGA database (Fig. 7k) and immunohistochemical detection for GAS1 in cohort 2 demonstrated the dysregulation of GAS1 in bladder cancer (Fig. 7l).

MiR-942-3p and TAZ promotes the growth of xenograft tumors in vivo
To verify the relationship between TAZ and miR-942-3p in vivo, we injected the cells with different condition (negative control, TAZ knockdown or sh-TAZ with miR-942 overexpression) into nude mice subcutaneously. Tumor volumes were measured every ve days after injection. As shown in Fig. 8a-c, depletion of TAZ remarkably suppressed tumor growth rate and weight compared with the control group. Besides, overexpression of miR-942-3p abrogated the inhibited effects induced by TAZ knockdown. MiR-942-3p levels of tumor tissue in three groups were con rmed by performing qRT-PCR (Fig. 8d). We next assessed the expression level of TAZ, LATS2 and GAS1 in vivo. Immunochemical assay and western blotting indicated that TAZ knockdown upregulated the levels of LATS2 and GAS1 in tumor tissue, and miR-942-3p could reverse the trend (Fig. 8e and 8f). These results con rmed that miR-942-3p is vital for tumor growth mediated by TAZ in vivo. To illustrate our results more clearly, we created a schematic of TAZ/miR-942-3p regulatory model in Fig. 8g.

Discussion
TAZ is a key executor of the Hippo signaling pathway to regulate cellular proliferation, differentiation and tissue homeostasis 46,47 . In the past decades, a great deal of evidence indicated that dysregulation of TAZ contributed to cancer initiation and progression 48,49 . The present study con rmed that TAZ was overexpressed in bladder cancer cell lines and tissue and was related to cell proliferation, migration and invasion.
EMT are trans-differentiation programs in which epithelial cells acquire mesenchymal features. Interestingly, recent studies indicated that different state of EMT regulated tumor migration, invasion, metastasis and resistance to clinical therapy. Moreover, EMT might lead to the emergence of cancer stem cells and triggered tumor initiation 18,20,25,27,50 . Hippo signaling pathway has been reported to regulate EMT process and mesenchymal characteristics [51][52][53][54] . In the light of this, we examined EMT markers such as E-cadherin, Vimentin, N-cadherin, Fibronectin and Snail and found that TAZ knockdown reversed EMT process. It is well known that numerous pathways and factors are involved in modulation of EMT and therefore further study is needed to validate the exact regulatory mechanism in TAZ-induced EMT process.
TAZ and Hippo pathway are also participated in metabolism modulation, such as to regulate glycolysis, lipogenesis, and glutaminolysis [10][11][12][13] . Interestingly, a growing body of evidence has veri ed that glycometabolism might modulate cell growth, migration and progression in bladder cancer [55][56][57] . Actually, while normal cells acquire energy for physiological processes from the oxidation of pyruvate, cancer cells rely on aerobic glycolysis to generate energy and compound for their aberrant growth and metastasis.
This metabolic characteristic of cancer cells is termed as the Warburg effect. The Warburg effect not only ensures the supply of energy and nutrients, but also provides an acidic microenvironment that enhances migration and invasion 58 . Therefore, it is meaningful to clarify the underlying mechanism of glycolysis in bladder cancer for clinical intervention and treatment. Our current results suggested that TAZ was vital for glycolysis of bladder cancer cells by regulating the expression of PFKFB3, HK2 and GLUT1, which serves as key components in glycolysis. However, the deeper relationship between glycolysis and other biological effects mediated by TAZ is worth exploring and verifying in the future study.
The TEAD protein family consists of four paralogous factors that function as nuclear DNA-binding proteins to modulate transcriptional activity of downstream genes in response to Hippo signaling pathway. Of note, the modulated role of TEAD depends on binding with TAZ or YAP in nucleus whose nuclear importation is mediated by LATS1/2 16,17,44 . MiRNAs participate in numerous biological processes, including tumor initiation, progression and metastasis 24,[32][33][34]59 . Nevertheless, the underlying mechanism of dysregulated miRNAs in bladder cancer remains unclear. Therefore, we speculated that TAZ might exert its biological functions via enhancing the expression of miRNAs, and following miRNA sequencing, qRT-PCR, ChIP and luciferase reporter assays illustrated that miR-942-3p was regulated by TAZ-TEADs in bladder cancer cells. MiR-942 has been con rmed to play a key role in cancer 42,43 . Our results found that miR-942-3p was abundantly expressed in bladder cancer cell lines and tissue and served as a tumor promoter. Moreover, GAS1 and LATS2 were veri ed as downstream targets of miR-942-3p directly. LATS2 could suppress the transcriptional activation of Hippo pathway-related gene by phosphorylating TAZ 9,14,44 . Interestingly, GAS1 acted as a key regulator in cell migration, EMT and glycometabolism in colorectal cancer cells 41 . Based on previous literatures and these results, our further experiments identi ed a positive feedback loop between TAZ and miR-942-3p and the biological effects of GAS1.

Conclusion
In summary, the present study veri ed a novel positive feedback loop between TAZ and miR-942-3p that regulates GAS1 expression and modulates biological behaviors, EMT process and glycometabolism in bladder cancer. TAZ, miR-942-3p and GAS1 might be potential therapeutic targets in clinical interventions of bladder cancer.

Consent for publication
Zhejiang Province also supported this work, grant number: LY18H160013.
Authors' contributions XDJ, GQD and FFW conceived of the study and carried out its design. FFW, MJF, YSC, XJZ and NH performed the experiments. FFW, MJF, HSW, YZ, JXL and SHH conducted the statistical analyses. FFW and YLY wrote the paper and XDJ and ZHH revised the paper. All authors read and approved the manuscript.