TCF-3-mediated transcription of lncRNA HNF1A-AS1 targeting oncostatin M expression inhibits epithelial-mesenchymal transition via TGFβ signaling in gastroenteropancreatic neuroendocrine neoplasms

Long noncoding RNAs play key roles in several cancers, but their potential functions in gastroenteropancreatic neuroendocrine neoplasms remain to be investigated. We performed GeneChip assay to explore differentiated lncRNAs in gastric NENs and peri-cancerous tissues. The regulation of HNF1A-AS1 on biological behavior of GEP-NENs cells and in vivo xenograft model was confirmed by CCK8, colony formation assay, transwell, western blot and qRT-PCR. We next detected the potential transcription factors and the binding sites between them with bioinformatic analysis. qRT-PCR was performed to analyze the exact relationship between them. HNF1A-AS1 expression was decreased in gastric NENs tissues (p < 0.01). Over-expression of HNF1A-AS1 suppressed cellular proliferation, migration and invasion. Knockdown of transcription factor 3 inhibited the expression of HNF1A-AS1 and promoted cellular migration and invasion. Oncostatin M was identified as the downstream target of HNF1A-AS1. Inhibition of transforming growth factor-β activity inhibited HNF1A-AS1/Oncostatin M-mediated epithelial-mesenchymal transition. Our data suggest that transcription factor 3/HNF1A-AS1/Oncostatin M axis inhibits the tumorigenesis and metastasis of gastroenteropancreatic neuroendocrine neoplasms via transforming growth factor-β signaling.


INTRODUCTION
Neuroendocrine neoplasms (NENs) are a group of human malignancies originating from the diffuse neuroendocrine system [1]. Gastroenteropancreatic neuroendocrine neoplasms (GEP-NENs) account for 60-70% of all NENs, and involve the largest hormoneproducing organ of our body [2]. Epidemiological studies have shown a steady increase in incidence up to 6.98/100,000 people during the past few decades [3]. The inferior quality of living standards for individuals and more global burden to our society have been emerging due to the long course of NENs [4][5][6][7].

AGING
Mutation of mammalian target of rapamycin (mTOR) signaling was a common pathogenic factor of GEP-NENs [14,17,18]. The inhibitors of mTOR signaling such as everolimus are currently standard drugs for well differentiated GEP-NENs [19]. However, the effect of everolimus is still unsatisfactory. Thus, there may be other signaling pathways which participate in the development and progression of GEP-NENs.
Epithelial-mesenchymal transition (EMT) is essential for cellular reprogramming during development [20,21]. EMT increases the capacity of cancer cells to initiate and promotes the tumorigenesis via angiogenesis. HNF1A-AS1 inhibited malignant progression of laryngeal squamous cell carcinoma via EMT [22,23]. LncRNAs mediated EMT was closely linked to breast cancer invasion and metastasis via transforming growth factor-β (TGF-β) signaling [24,25]. Some EMT markers such as Snail1 were associated with the invasion and metastasis of GEP-NENs [26].

HNF1A-AS1 inhibited GEP-NENs tumorigenesis in vivo
The QGP-1 cells transfected with HNF1A-AS1 overexpression plasmid were subcutaneously inoculated into nude mice (n = 6 mice per group). At 49 days after injection, the tumors formed in HNF1A-AS1 over-expression group were smaller than those in  AGING control group ( Figure 4A). The average tumor volume and weight were decreased either (p < 0.05, Figure 4B, 4C). HNF1A-AS1 expression in over-expression group was much higher than control group (p < 0.01, Figure 4D).

TCF3 was the potential transcription factor of HNF1A-AS1
We took intersection between 92 potential transcription factors from TFbind and 40 upstream genes from RNAreg. Totally of 8 potential transcription factors probably binding to HNF1A-AS1 were selected. We analyzed the promoter region of HNF1A-AS1 to predict score of these potential transcription factors and detect the binding sites with JASPAR program ( Figure 5A). TCF3 could potentially bind to HNF1A-AS1 and the putative binding sites were confirmed (˗158 to ˗167bp, AGCACGTGCA, Figure 5B). TCF3 was decreased in GEP-NENs cells compared with normal cells by qRT-PCR (p < 0.05, Figure 5C). TCF3 knockdown with lentivirus down-regulated HNF1A-AS1 level (p < 0.05, Figure 5D).

HNF1A-AS1 suppressed cellular invasion via targeting oncostatin M (OSM)
Through GeneChip array analysis, we identified 280 discrepant coding genes (p < 0.05, Fold change >2, 140 up-regulated and 140 down-regulated), as presented in the heat map and volcano plot ( Figure 7A, 7B). The GO analysis was performed based on these 280 differential genes (p = 0.05, FDR = 0.05). We found that receptor AGING binging was one of the top ten GO terms in molecular function analysis ( Figure 7C). In cellular component analysis, extracellular space was the most represented component ( Figure 7D). Then, we chose OSM, phosphodiesterase 3B (PDE3B), Tenascin C (TNC) from these 280 discrepant genes, which had been reported as ligands to gp130, leukemia inhibiting factor receptor (LIFR), Receptor beta (OSMR-β). Interestingly, OSM expression presented the most discrepant following dysregulation of HNF1A-AS1. HNF1A-AS1 failed to alter OSMR-β expression (p < 0.05, Figure 7E, 7F). We evaluated OSMR-β and OSM expression in human AGING samples. Both OSMR-β and OSM mRNA levels were increased in gastric NENs tissues (p < 0.01, Figure  7G). At last, we added 50 ng/ml of OSM to STC-1 cells and found that it could reverse HNF1A-AS1induced decrease of migration and invasion (p < 0.01, Figure 7H).

DISCUSSION
LncRNAs exert important functions in various biological processes such as cell differentiation, proliferation and apoptosis [27,28]. By far, the role of lncRNA HNF1A-AS1 in GEP-NENs development has not been clarified.
In this study, we found that HNF1A-AS1 level in gastric NENs tissues was decreased with GeneChip assay. HNF1A-AS1 over-expression could suppress proliferation, migration and invasion in vitro and in vivo. These effects induced by HNF1A-AS1 might be dependent on induction of EMT.
Transcription factors could bind to lncRNAs to regulate the progression of tumorigenesis and metastasis [29,30]. We found that TCF-3 was upstream transcription factor to regulate HNF1A-AS1 transcription. Knockdown of TCF3 decreased the expression of HNF1A-AS1 and promoted cellular migration and invasion. TCF-3 had been proven to take part in human cancer development and progression [31][32][33]. HNF1A-AS1 enhanced cell proliferation and metastasis in osteosarcoma through activation of the Wnt/β-catenin signaling pathway [34]. In the present study, we found that knockdown of TCF3 could bind to the specific site of HNF1A-AS1 promoter region which induced downregulation of HNF1A-AS1 and promoted tumorigenesis and metastasis in GEP-NENs. AGING OSM had been considered as a pleiotropic cytokine. It exerted its biological function by binding to two different OSM receptor complexes, OSMR-β and LIFR [35]. Aberrant expression of OSM promoted cellular invasion and induced mesenchymal phenotype in many solid tumors including osteosarcoma, gliomas and breast cancer [36][37][38][39]. We found that HNF1A-AS1 could inhibit OSMR-β and OSM expression. Previous studies had reported that OSM promoted cancer cell plasticity through cooperation with STAT3-SMAD3 signaling [40]. It has also been reported as a novel inhibitor of TGFβ-induced matricellular protein expression [41][42][43]. In our study, we discovered that TGFβ signaling was essential for cellular invasion induced by HNF1A-AS1/OSMR.
In summary, HNF1A-AS1 was down-regulated in GEP-NENs tissues. TCF3-mediated HNF1A-AS1 inhibited cellular proliferation and invasion via OSM/TGFβ signaling. HNF1A-AS1 may be served as a potential target for further diagnosis and treatment in human GEP-NENs.

Tissue specimens and cell lines
The gastric NENs tissues and peri-cancerous tissues (>5 cm distant from cancer tissues) were obtained from three patients who underwent surgical resection at the First Affiliated Hospital of Nanjing Medical University. All patients did not receive any local or systemic treatment before surgery. All experiments were approved by the Research Ethics Committee of Nanjing Medical University. Written informed consent was obtained from all participants.

Gene expression profiles and data analysis
Total RNA was extracted and cDNAs were prepared according to the standard Following labeling, 5.5 μg of cDNA were hybridized for 16 h at 45°C on GeneChip Human Transcriptome Array 2.0. The arrays were scanned using the GeneChip ® Scanner 3000 7G. Raw data were analyzed using Robust Multiarray Analysis (RMA) with a log base 2 (log2) transformation. The threshold was fold change > 2 and p-value < 0.05.

Construction of TCF3 knockdown lentivirus
The short hairpin RNA (shRNA) sequence targeting TCF3 (CCGGCTCCTAATGTCAACCGAGAACTC GAGTTCTCGGTTGACATTAGGAGCTTTTT) was obtained from GeneChem, Inc., Recombinant lentiviral vectors were constructed according to previous studies. The transfection efficiency of shTCF3 was determined using reverse transcription-quantitative (qRT-)PCR and western blot analysis after 72 h.

qRT-PCR
RNA was extracted from each fraction using TriPure Isolation Reagent (Roche, USA). The separation of nuclear and cytosolic fractions was conducted using the NE-PER Nuclear and Cytoplasmic Extraction Reagents kit (Thermo Fisher Scientific) according to the manufacturer's protocol. cDNA was synthesized using a Reverse Transcription Kit (Takara, Dalian, China). qRT-PCR analysis was conducted with Essential DNA Green Master (Roche, USA). The results were normalized to GAPDH. Primers for amplification of HNF1A-AS1 were designed as follows: Forward:5′ -TCAAGAAATGGTGGCTAT-3′, Reverse:5′ -GCTCTG AGACTGGCTGAA-3′.

Cell viability and proliferation assay
Cell viability was assessed by CCK-8 assay. Cells were seeded at density of 3000 per 100 μl medium per well in 96 well culture plates. At different time points after transfection, 10 μl of CCK-8 (Dojindo) reagent was added to wells and further incubated for 4 h at 37°C. The absorbance was measured at 450 nm after 1-4 h.

AGING
Suspended pre-treated cells were seeded in culture dishes at a density of 300 cells per well and incubated at 37°C until visible colonies emerging (2-3 weeks). Cells were fixed and stained before counting the colonies.

Cellular migration and invasion assays
Pre-treated cells were seeded into the upper Transwell  chambers 36 h after transfection for migration assays (8 μm pore size, Millipore) and invasion assays with the Matrigel-coated (BD, Franklin Lakes, NJ, USA) filters in 24-well plates. Cells in the upper chambers were cultured with 300ul DMEM/F12 or RPMI-1640 medium without FBS. The lower chambers were DMEM/F12 or RPMI 1640 medium containing 10% FBS. Then, cells that passed through the filters were stained and photographed from random fields.
At 24 h after transfection, wound healing assay was performed using Ibidi cell migration technology (Ibidi, Martinsried, Germany). Cells were seeded at density of 3-7 × 10 5 /ml per well and scraped two perpendicular straight lines. Cells were photographed at different time points. The migration rates were shown as the percentage of area reduction of wound closure by ImageJ.

Protein extraction and western blotting
Cell proteins were extracted from cells as usual. Samples from cell lysates were resolved by SDS-PAGE and then transferred to polyvinylidene fluoride (PVDF) membrane and incubated with specific antibodies against E-Cadherin, N-cadherin, β-catenin, GAPDH and other markers (Cell Signaling Technology, MA, USA). The ECL chromogenic substrate was used to detect specific bands.

Tumor formation assay
All animals were approved by the Committee on the Ethics of Animal Experiments of the Nanjing Medical University. The athymic male BALB/c nude mice (4 weeks old) were obtained from the Shanghai Laboratory Animals Center of the Chinese Academy of Sciences (Shanghai, China). A total of 5 × 10 6 QGP-1 cells transfected with pcDNA3.1-HNF1A-AS1 or pcDNA3.1 empty vector (EV) were subcutaneously injected into male BALB/c nude mice. Seven weeks after cell injection, the mice were sacrificed. The volume and weight of each excised subcutaneous tumor was measured. The tumor volumes were calculated by the following formula: 0.5 × length × width 2 .

Statistical analysis
All the experiments were carried out at least three times independently. Bars shown represented mean ± SEM. The differences between independent groups were analyzed by Student′s t-test using SPSS software, the correlation between TCF-3 levels and HNF1A-AS1 was assessed using Spearman's correlation coefficient, and a p value < 0.05 was considered to be significant.

AUTHOR CONTRIBUTIONS
J.B. and J.X. designed the experiments and wrote the manuscript. J.X., Q.L., Y.W. and J.P. performed the experiments. J.X. and J.B. performed the bioinformatics analysis. J.B., J.X., X.L. and Q.T. analyzed data.

ACKNOWLEDGMENTS
This study was supported by Medical Key Talents Project of Jiangsu Province (Grant No. ZDRCA2016008) and the "333" Project of Jiangsu Province (Grant No. BRA2017535).