Long noncoding RNA NORAD, a novel competing endogenous RNA, enhances the hypoxia-induced epithelial-mesenchymal transition to promote metastasis in pancreatic cancer

Expression profiles were obtained from the GEO database (www.​ncbi.​nim.​nih.​gov/​geo/​) as mentioned previously [24]. Expression data from 55 pancreatic tumor tissues and their matched non-tumor tissues from two human microarray datasets GSE15471 [25] and GSE16515 [26] (39 pairs from GSE15471 and 16 pairs from GSE16515) were obtained and compared using the robust multiarray average (RMA) algorithm. Next, we used Venn analysis to analyze aberrant lncRNAs. Afterward, GSE32688 [27] was analyzed for the expression of miR-125a-3p and RhoA. In addition, box plots (Additional file 1: Figure S1) were made to graphically view the value distributions among datasets. By analyzing the boxplots, these databases were confirmed to be median-centered and cross-comparable.

The potential microRNA binding sites within NORAD were predicted by computer-aided algorithms obtained from Starbase v2.0 (http://​starbase.​sysu.​edu.​cn) [28, 29]. MicroRNA-mRNA binding sites were obtained from miRTarBase (http://​mirtarbase.​mbc.​nctu.​edu.​tw) [30].

PDAC specimens and the corresponding adjacent non-tumor tissues were obtained from patients who had undergone curative surgery at Shanghai Jiao Tong University School of Medicine Affiliated Ruijin Hospital between 2013 and 2015. None of them had received radiotherapy or chemotherapy before their surgery. Staging had been performed clinically, radiographically, and pathologically. The fresh primary PDAC tissues and adjacent non-tumor pancreas tissue samples were collected and snap-frozen in liquid nitrogen immediately after resection, then stored at −80 °C for further use.

HEK-293Ts and the pancreatic cancer cell lines SW1990, Capan-1, PANC-1, AsPC-1, CFPAC-1, MIAPaCa-2, and BxPC-3 were purchased from ATCC (www.​atcc.​org). Premalignant HPDE cells (HPV16-immortalized normal pancreatic ductal epithelium) were generously provided by Shanghai Cell Bank of the Chinese Academy of Sciences. All cell lines were tested and authenticated by DNA typing at the Shanghai Jiao Tong University Analysis Core. All cells were cultured in DMEM, IMDM or RPMI 1640 (Gibco) with 10% Fetal Bovine Serum (FBS, Gibco) at 37 °C under 5% CO2 in a humidified chamber.

During hypoxic treatment conditions [31], cells were cultured in a modular incubator chamber (Billups-Rothenberger) that was flushed with 1% O₂, 5% CO₂, and 94% N₂. Cells were harvested after 48 h. During normoxic control treatments, cells were cultured in an incubator with a humidified atmosphere containing 5% CO₂ and 20% O₂.

Total RNA was isolated from cells with TRIzol reagent (Invitrogen, Carlsbad, CA, USA). Next, cDNA was synthesized with a reverse transcription kit (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. qRT-PCR was performed on the 7500 Fast Real-time PCR system (Applied Biosystems) using a standard protocol from the SYBR Green PCR Kit (Toyobo). The results were normalized to the expression of GAPDH. Primers used for qRT-PCR assays are listed in Additional file 2: Table S1. For the detection of miRNA expression, reverse transcription was performed and microRNAs were detected with stem-loop primers purchased from RiboBio (Guangzhou, China). U6 snoRNA was used as the endogenous control. Relative fold changes were calculated using the 2^-△△Ct method. All PCR assays were repeated three times.

NORAD cDNA was cloned into the mammalian expression vector pcDNA3.1 (Invitrogen). NORAD cDNA fragments containing either the predicted potential microRNA binding sites (wildtype, wt) or scrambled microRNA binding site sequences (mutation, mut) were amplified by PCR. Moreover, plasmids containing the firefly luciferase reporter were constructed with RhoA-3′-UTR-wildtype ( RhoA-3′-UTR-wt-luc) and RhoA-3′-UTR-mutation (mutated miR-125a-3p binding region; RhoA-3′-UTR-mut-luc). Mimics and inhibitors of hsa-miR-125a-3p and rno-miR-239b-5p were purchased from RiboBio (Guangzhou, China). rno-miR-239b-5p was used as a negative control.

shRNA sequences targeting NORAD were designed, inserted in lentiviral plasmids and then transfected into HEK-293Ts with virus packaging plasmids to produce the lentivirus. The viral supernatants were collected at 48 and 72 h after transfection and added into PANC-1 and SW1990 cells to construct the stably transfected cell lines. Puromycin was added to the media 48 h after infection and maintained for at least 1 week to select stably transfected cell lines (PANC-1/SW1990-sh-NORAD or PANC-1/SW1990-sh-NC).

PDAC cells (1 × 10⁶ cells/well) were treated with different reagents, seeded in six-well plates and cultured until they reached confluence. Wounds were made in the cell monolayer by making a scratch with a 20 μL pipette tip. Plates were washed once with fresh medium after 24 h in culture to remove non-adherent cells. Following this wash, plates were photographed.

Approximately 1 × 10⁵ PDAC cells treated with different reagents were suspended in serum-free medium and seeded in either Matrigel-coated chambers with 8-μm pores (BD Biosciences, Franklin Lakes, NY, USA) for migration assays or in chambers (Corning Costar, NY, USA) that were not coated with Matrigel for invasion assays. For both assays, medium containing 10% FBS was added to the lower chamber as a chemoattractant. After 12 h, the cells that had invaded through the membrane to the lower surface were fixed with methanol, stained with 0.5% crystal violet solution and counted in 5 random fields of view (200×). RhoA pathway specific inhibitor CCG-1423 [32] was purchased from selleck and used as recommended.

Athymic BALB/C mice (male, 4–6 weeks old) were purchased from the Chinese Academy of Sciences (Shanghai, China) and were maintained in a specific pathogen-free facility. Orthotopic implantation of pancreatic cancer cell lines was performed as described previously [33]. For experimental metastasis assays, the SW1990-sh-NORAD and SW1990-sh-NC cell lines were labeled with firefly luciferase. Each mouse was then injected with 100 μL of cell suspension (1 × 10⁶ cells) into the pancreas. Animal health conditions and metastatic progression were monitored. Metastasis was quantified using a noninvasive bioluminescence In Vivo Imaging System (IVIS, Xenogen) as described previously [34]. All mice were sacrificed and both the primary tumor and metastatic foci were observed by general observation, H & E staining and immunohistochemical staining.

Recombinant plasmids or an empty plasmid encoding the firefly luciferase reporter were transiently transfected into SW1990-sh-NORAD and SW1990-sh-NC cells using Lipofectamine 2000 (Invitrogen, USA) The renilla luciferase reporter pRL-CMV plasmid (Promega) was also transfected into these cells as an internal control. Each group above was transfected with miR-125a-3p mimic or a negative control. Luciferase activities in the indicated cells were measured using the Dual-Luciferase Reporter Assay System (Promega, USA) 48 h after transfection according to the manufacturer’s instructions.

Cells were lysed with RIPA cell lysis buffer in the presence of a protease inhibitor cocktail (Sigma, USA) in order to harvest total cell protein. Standard Western blotting was carried out as described previously [35]. In brief, the BCA Protein Assay Kit (Pierce, USA) was used to quantify the protein concentration of each cell lysate. Equal amounts of protein samples were loaded onto a 10% SDS-PAGE gel and then transferred onto PVDF membranes. After blocking with skim milk, the membranes were incubated with primary antibodies diluted in TBST buffer overnight at 4 °C and then with the HRP-conjugated secondary antibody for 2–3 h at room temperature. Finally, the images of protein bands were captured by a Tanon detection system using ECL reagent (Thermo).

Data were presented as the mean ± one standard deviation (SD) from at least three separate experiments. The Student t test, X² test, log-rank test, nonparametric Mann–Whitney U test, or Fisher exact test was used for comparisons between groups. The Kaplan–Meier method was used to estimate overall survival (os) and recurrence rate. Differences were deemed statistically significant at P < 0.05. The results are shown as the mean ± SEM.

This study was approved by the Clinical Research Ethics Committee of Shanghai Jiao-Tong University. All patients were fully informed of the experimental procedures before tissue acquisition. All animal experimental procedures were performed in compliance with the institutional ethical requirements and were approved by the Shanghai Jiao-Tong University School of Medicine Committee for the Use and Care of Animals.

To identify eligible lncRNAs in PDAC, we analyzed microarray expression profiles. Two human PDAC microarray datasets (GSE15471 and GSE16515) were selected and analyzed for consistently aberrant lncRNAs between PDAC and normal pancreas tissues using GEO2R. Fold changes in robust multiarray average (RMA) signal intensities (which represent expression differences between tumor and normal tissue) were calculated and cutoff values were set at |RMA fold change| > 1.5 and a p value < 0.05. Subsequently, we identified 272 consistent non-coding RNAs that were significantly upregulated or downregulated by Venn analysis (Fig. 1a).

As is already known, intratumoral hypoxia plays a crucial role in pancreatic cancer metastasis. After the above analyses, aberrant lncRNAs (Fig. 1b) with potential roles in the hypoxia pathway were chosen for further study.

To evaluate hypoxia-induced alterations in lncRNA expression, the pancreatic cancer cell line SW1990 was cultured under hypoxic or normoxic conditions for 48 h. We found that NORAD was significantly increased after hypoxic stimulation (Fig. 1c). Additionally, we found that NEAT1, PVT1, NBR2, HOTAIR, MALAT1 and NORAD increased during hypoxia, while MEG3 decreased during hypoxia.

As NORAD was upregulated under hypoxia-stimulating conditions, we hypothesized that superfluous NORAD may participate in the development of hypoxia-induced malignant phenotypes.

To investigate further, we selected SW1990 and PANC-1, two pancreatic cell lines with highest NORAD expression among 8 cell lines (shown in Additional file 3: Figure S2), and knocked down NORAD using shRNAs and then assessed tumor cell migration and invasion abilities using wound-healing assays as well as transwell migration and invasion assays. As shown in Fig. 2a-d, knockdown of NORAD in PANC-1 or SW1990 cells markedly suppressed cell migration and invasion compared with their respective control cells.

To further quantify metastatic and dissemination potential in vivo, we established an orthotopic metastatic model, as mentioned above. SW1990-sh-NORAD and SW1990-sh-NC cells were labeled with firefly luciferase and injected into the pancreas of nude mice. After 3 weeks, the scope of metastasis was quantified by non-invasively detecting the bioluminescent signal on a bioluminescence In Vivo Imaging System (Fig. 2e-f). Signals from the group injected with SW1990-sh-NC cells were significantly stronger than those from the group injected with SW1990-sh-NORAD cells (p = 0.043). Moreover, more distant micrometastatic foci were observed in the negative control group than the sh-NORAD group by H & E staining. It is very obvious that metastasis was repressed when knocking down NORAD (Fig. 2g).

Accumulating evidence has suggested that lncRNAs might function as ceRNAs by binding to miRNAs and functionally liberating other RNA transcripts [17]. To examine whether NORAD operates by a similar mechanism in PDAC cells, we used Starbase v2.0 to predict the potential miRNA binding sites in NORAD and found a sequence complementary to miRNA-125a-3p (Fig. 3a). Furthermore, we found that the expression of NORAD was negatively associated with the expression of miR-125a-3p in 33 PDAC tissues (Fig. 3b). We then tested the effects of shRNA-mediated downregulation of NORAD on the expression of miR-125a-3p. Notably, miR-125a-3p was upregulated in the sh-NORAD-treated SW1990 and PANC-1 cells compared with negative control-treated cells (Fig. 3c). Moreover, we selected BxPC-3 and Canpan-1, two pancreatic cell lines with lowest NORAD expression among 8 cell lines (shown in Additional file 3: Figue S2), and overexpress NORAD with plasmids containing wildtype NORAD (NORAD-wt) or NORAD with mutation in its MRE (NORAD-mut) respectively. After we overexpressed NORAD in BxPC-3 and Canpan-1, we found a significant decrease in miR-125a-3p. This phenomenon disappeared when the MRE of NORAD was mutated (Fig. 3d), suggesting that the lncRNA NORAD can repress the expression of miR-125a-3p via direct binding at the MRE.

Using miRTarBase, we found that RhoA was a potential target of miR-125a-3p. Additionally, we found that RhoA expression is negatively correlated with miR-125a-3p in GSE32688 [27] (data provided in Additional file 4: Table S2) and that miR-125a-3p can complement the seed region of RhoA. Interestingly, the MRE of NORAD was found to be similar to the seed region of RhoA and the expression levels of NORAD and RhoA were found to be positively correlated (Fig. 4a-c). RhoA is downregulated after NORAD knockdown in vitro and vivo (Fig. 4d). And we also found that RhoA is increased by NORAD-wt, while NORAD-mut failed to increase RhoA (Fig. 4e). This suggests that NORAD may enhance RhoA by repressing miR-125a-3p. To confirm the binding in predicted sites of RhoA-3′-UTR, we mutated binding sites in RhoA and then performed reporter assays. We found that miR-125a-3p could inhibit the activity of RhoA-3′-UTR-wt, but had no influence on the activity of RhoA-3′-UTR-mut (Fig. 4f). In addition, we found that downregulation of NORAD could significantly inhibit the activity of RhoA-3′-UTR. This effect could be partially retrieved using a miR-125a-3p inhibitor (rather than miR-239b-5p), which suggests that NORAD competes with the RhoA-3′-UTR for miR-125a-3p (Fig. 4g).

As we know, hypoxic stimuli can trigger the expression of RhoA and its downstream effector ROCK1, leading to cell metastasis, and RhoA plays a crucial role in mediating the epithelial-mesenchymal transition (EMT) [36]. To further investigate the molecular mechanisms underlying these events, we explored whether NORAD regulates EMT in cancer cells. As shown in Fig. 5a-b, knockdown of NORAD significantly decreased the expression levels of the mesenchymal cell markers N-cadherin, Vimentin, and ZEB1 but increased the expression levels of the epithelial cell marker E-cadherin. And overexpression of NORAD resulted in RhoA increasing and E-cadherin decreasing. This suggests that the more NORAD expressed the more RhoA protein will be produced, and then triggering more aggressive EMT and cell migration and invasion. Furthermore, as Fig. 5b-d shown, when we treated cell lines with RhoA pathway specific inhibitor CCG-1423 [32], RhoA downstream protein ROCK1 decreased and aggressive EMT and invasive behaviors caused by NORAD overexpression were partly redressed. In another words, blocking RhoA activity with the RhoA inhibitor can impede the flow of EMT induced by NORAD, indicating that NORAD promotes EMT via regulating RhoA.

The expression of NORAD was significantly higher in 75 pancreatic cancer samples than in 55 normal pancreas tissues (GSE15471 and GSE16515, p < 0.001, Fig. 6a). We then assayed NORAD expression in a panel of 33 PDAC with complete clinicopathological information (Additional file 5: Table S3), along with 33 paired normal pancreas tissues. NORAD was significantly upregulated in PDAC specimens compared with normal pancreas tissues (Fig. 6b, p<0.01). The expression of NORAD was also upregulated in most tumor tissues compared with the non-tumor tissues from the same donor (Fig. 6c), thus linking the upregulation of NORAD to pathological changes in pancreatic tissue.

Next, we evaluated the clinical significance of NORAD in PDAC. Thirty-three PDAC patients from our hospital were divided into two groups (high: n = 16, low: n = 17) according to the level of NORAD expression.

The Kaplan–Meier estimator was used to estimate recurrence-free survival (RFS) and overall survival (OS) rates and p values were calculated using the log-rank test. As shown in Fig. 6d-e, PDAC patients with higher NORAD expression have shorter overall survival (p = 0.0335) and recurrence-free survival (p = 0.0419) rates. The median overall and recurrence-free survival rates were sharply decreased (RFS: 9.35 versus 19.8 months; OS: 14.35 versus 22.8 months) in the high-NORAD group, which indicates that NORAD upregulation is associated with poor prognosis.