EBV-miR-BART8-3p induces epithelial-mesenchymal transition and promotes metastasis of nasopharyngeal carcinoma cells through activating NF-κB and Erk1/2 pathways

This study was approved by the Ethical Review Committee of Fujian Cancer Hospital (approval no. FJZLYY2016–00143). Written informed consent was obtained from all participants following a detailed description of the purpose of the study. All experiments described in this study were conducted in accordance with international and national laws, regulations and guidelines.

Six EBV-positive NPC biopsy specimens and 4 normal nasopharyngeal mucosal specimens were sampled for miRNA sequencing, and another 19 NPC specimens and 10 normal nasopharyngeal specimens were used to quantify EBV-BART8-3p and RNF38expression. All NPC was diagnosed by pathological examinations, and all specimens were stored in liquid nitrogen for the subsequent experiments.

Five-week-old female BALB/c nude mice were purchased from the Medical Experimental Animal Center of Guangdong Province (Guangzhou, China). All mice were housed in a clean facility in the Laboratory Animal Center of Fujian Medical University (Fuzhou, China) and given free access to clean water and food.

Two EBV-negative NPC cell lines CNE-1 and SUNE-1 and human embryonic kidney (HEK) 293 T cells were purchased from China Center for Type Culture Collection (Wuhan, China), and one EBV-positive C666–1cell line was kindly presented by Professor Hong lin Chen from the University of Hong Kong. All NPC cells were cultured in RPMI-1640 medium (HyClone; Logan, UT, USA) supplemented with 10% fetal bovine serum (FBS; Gibco, Grand Island, NY, USA) at 37 °C containing 5% CO₂.HEK293T cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM; HyClone, Logan, UT, USA) supplemented with 10% FBS at 37 °C containing 5% CO₂.

Paired-end (2 × 150 bp) RNA-seq assays were performed from ribosomal RNA depleted total RNA on Illumina HiSeq 2500 system (Illumina; San Diego, CA, USA) according to the standard manufacturer’s protocol. The raw sequence reads were aligned to human genome GRCh38 with the star aligner (v2.5.0b) [26]. Then the gene level expression was quantified by feature Counts (v1.4.4) [27] based on GENCODE gene model version p2 release22. Genes with at least 1 count per million (CPM) in at least 1 sample were kept other removed from further analysis. The gene level read counts data was normalized using the trimmed mean of M-values normalization (TMM) method [28] to adjust for sequencing library size difference. Differential gene expression between cancer and control groups was predicted by a linear model analysis using Bioconductor package limma [29]. To adjust for multiple tests, the false discovery rate (FDR) of the differential expression test was estimated using the Benjamini-Hochberg method [30]. Genes with FDR adjusted p value ≤0.05 and fold change (FC) ≥ 1.2 were considered significant. Heatmap and cluster dendrogram of the significant genes were plotted using R programming language (https://​www.​r-project.​org/​). Gene ontology (GO) and pathways enriched in the differentially expressed genes were identified by Fisher’s exact test (FET) based on the gene set annotation collections from MSigDB [31].

Total RNA was extracted from NPC specimens and normal nasopharyngeal mucosal specimens using TRIzol reagent (Invitrogen; Carlsbad, CA, USA). The RNA concentration was measured with a NanoDrop2000 Spectrophotometer (NanoDrop Technologies; Wilmington, DE, USA), and the integrity of purified RNA was determined using Agilent 2100 Bioanalyzer (Agilent Technologies; Palo Alto, CA USA). miRNA quantification was evaluated using Hot Start PCR. A small RNA library was built using the Next Multiplex Small RNA Library Prep Set for Illumina (NEB; Ipswich, MA, USA) according to the manufacturer’s protocol. Polyacrylamide gel electrophoresis was performed to purify small RNA and to enrich for molecules ranging from 18 to 30 nt. Then, cDNA was synthesized, digested and amplified to set cDNA libraries, followed by purification on a polyacrylamide gel and quantification. Finally, the cDNA libraries were sequenced using standard protocols on an Illumina Hiseq 4000 System (Illumina; San Diego, CA, USA). miRNA annotation was performed in the miRBase database (http://​www.​mirbase.​org). Sequencing data were aligned to the reference human genome (UCSC hg19) and EBV genome (GCF_000872045.1).

Reads mapped to known miRNAs were identified by searching miRBase database (v21). Novel miRNAs were predicted by miRDeep [32]. Known and novel miRNA expression levels were measured by the number of mapped reads. Similar to the gene level differential expression analysis, differentially expressed miRNAs between cancer and control groups were predicted by limma following library depth normalization by the TMM method.

The weighted network analysis begins with a matrix of the Pearson correlations between all miRNA pairs, then converts the correlation matrix into an adjacency matrix using a power function f(x) = x^β. The parameter β of the power function is determined in such a way that the resulting adjacency matrix (i.e., the weighted co-expression network), is approximately scale-free. To measure how well a network satisfies a scale-free topology, we use the fitting index [33] (i.e., the model fitting index R² of the linear model that regresses log(p(k)) on log(k) where k is connectivity and p(k) is the frequency distribution of connectivity). The fitting index of a perfect scale-free network is 1. For this dataset, we select the smallest β which leads to an approximately scale free network. The distribution p(k) of the resulting network approximates a power law: p(k) ∼ k^−γ.

To explore the modular structures of the co-expression network, the adjacency matrix is further transformed into a topological overlap matrix [33]. As the topological overlap between two genes reflects not only their direct interaction, but also their indirect interactions through all the other genes in the network, previous studies have shown that topological overlap leads to more cohesive and biologically meaningful modules [33, 34]. To identify modules of highly co-regulated microRNAs, we use average linkage hierarchical clustering to group genes based on the topological overlap of their connectivity, followed by a dynamic cut-tree algorithm to dynamically cut clustering dendrogram branches into miRNA modules [35]. To distinguish between modules, each module was assigned a unique color identifier, with the remaining, poorly connected miRNAs colored grey.

The EBV-miR-BART8-3p and RNF38 expression were quantified using qPCR assay. Briefly, total RNA was extracted from NPC biopsy specimens and normal nasopharyngeal mucosal specimens using TRIzol reagent (Invitrogen; Carlsbad, CA, USA), and transcribed into cDNA using miScript II RT Kit (Qiagen; Valencia, CA, USA) following the manufacturer’s instructions. qPCR assay was performed using the miScript SYBR Green PCR Kit (Qiagen). U6 snRNA and GAPDH were served as internal controls for quantifying miRNA and mRNA expression, respectively. The relative quantity of miRNA and mRNA expression was calculated using the 2^−ΔΔCtmethod. All determinations were repeated in triplicate.

Lentiviral particles (GV369 and Ubi-MCS-SV40-EGFP-IRES-puromycin) containing EBV-miR-BART8-3p precursors and lentiviral particles (GV280 and hU6-MCS-Ubiquitin-EGFP-IRES-puromycin) containing reverse complement of EBV-miR-BART8-3p and their control vectors were constructed by Shanghai Genechem Co., Ltd. (Shanghai, China).CNE-1 and SUNE-1 cells were transfected with a recombinant lentiviral vector GV369 to upregulate EBV-miR-BART8-3p expression (CNE-1-BART8-3p and SUNE-1-BART8-3p cells), and C666–1 cells were transfected with a lentiviral vector GV280 to downregulate EBV-miR-BART8-3p expression (C666–1-BART8-3p cells). The transfection efficiency was checked using qPCR assay.

For the rescue assay, CNE-1-BART8-3p cells and SUNE-1-BART8-3p cells were transfected with the RNF38 lentiviral vector GV358 (Shanghai Genechem Co., Ltd.; Shanghai, China) or a normal control (Shanghai Genechem Co., Ltd.; Shanghai, China).

For cell proliferation assays, cells were seeded onto 96-well plates (Corning, Inc.; Corning, NY, USA) at a density of 1500cells per well and were incubated at 37 °C containing 5% CO₂ for 1, 2 and 3 days. Subsequently, 10 μl of Cell Counting Kit-8 solution (Dojindo; Dojindo Molecular Technologies, Inc., Tokyo, Japan) was added to each well and incubated for 2 h. Then, the absorbance value (OD) was measured at 450 nm.

For colony-forming assays, cells were seeded onto 6-well plates (Corning, Inc.; Corning, NY, USA) at a density of 1000 cells per well and each group had three replicate wells. Following 2-week incubation, cell colonies were washed twice in PBS, fixed with 100% methanol and stained with crystal violet staining solution (Beyotime Institute of Biotechnology; Shanghai, China) for 15 min for counting.

Cells were seed onto 6-well plates (Corning, Inc.; Corning, NY, USA), and artificial wounds were created using a sterile 200 μl plastic tip following serum starvation for 24 h. Then, cells were washed with serum-free medium to remove debris and floating cells. Images of the wounds were captured at 0, 24 and 48 h under an inverted microscope.

For cell migration and invasion assays, cells in serum-free medium were transferred to the upper chamber of 8.0 μm pore size (Corning, Inc.; Corning, NY, USA) without or with Matrigel (BD Biosciences, Franklin Lakes, NJ, USA), and 20% FBS was added to the bottom chamber. SUNE-1 and CNE-1 cells were incubated in the upper chamber and removed 24 (cell migration assays) or 48 h after incubation (cell invasion assays), respectively, while C666–1 cells were incubated for 48 h in migration and invasion assays. Then, cells on the lower surface of the membrane were fixed and stained. Finally, images were captured and cell numbers were counted using microscopy (100 ×).

Western blotting analyses were performed by using standard protocols. Primary antibodies, including rabbit anti-RNF38 polyclonal antibody (1:250; Abcam, Cambridge, MA, USA), rabbit anti-Vimentin monoclonal antibody (1:1000; Abcam), rabbit anti-E-cadherin monoclonal antibody (1:1000; Abcam), rabbit anti-Snail monoclonal antibody (1:1000; Cell Signaling Technology, Inc.), rabbit anti-N-cadherin monoclonal antibody (1:5000; Abcam), rabbit anti-MMP2 polyclonal antibody (1:200; Santa Cruz Biotechnology, Inc.; Santa Cruz, CA, USA), rabbit anti-MMP9 monoclonal antibody (1:1000; Abcam), mouse anti-IKKα monoclonal antibody (1:200; Santa Cruz Biotechnology, Inc.), rabbit anti-p-IKKα/β monoclonal antibody (1:1000; Cell Signaling Technology, Inc.; Danvers, MA, USA), rabbit anti-IKKβ monoclonal antibody (1:1000; Cell Signaling Technology, Inc.), rabbit anti-IκBɑ monoclonal antibody (1:1000; Cell Signaling Technology, Inc.), rabbit anti-p-IκBɑ monoclonal antibody (1:1000; Cell Signaling Technology, Inc.), mouse anti-IκBβ monoclonal antibody (1:200; Santa Cruz Biotechnology, Inc.), rabbit p-IκBβ monoclonal antibody (1:1000; Cell Signaling Technology, Inc.), mouse anti-NF-κB monoclonal antibody (1:200; Santa Cruz Biotechnology, Inc.), rabbit anti-p-NF-κB monoclonal antibody (1:1000; Cell Signaling Technology, Inc.), rabbit anti-Erk1/2 monoclonal antibody (1:1000; Cell Signaling Technology, Inc.), rabbit anti-p-Erk1/2 monoclonal antibody (1:1000; Cell Signaling Technology, Inc.), rabbit anti-p-Mek1/2 monoclonal antibody (1:1000; Cell Signaling Technology, Inc.), rabbit anti-p-c-Raf monoclonal antibody (Ser259) (1:1000; Cell Signaling Technology, Inc.), rabbit anti-TAK1 monoclonal antibody (1:1000; Cell Signaling Technology, Inc.), and rabbit anti-p-TAK1 monoclonal antibody (1:1000; Cell Signaling Technology, Inc.) were used. Membranes were incubated with the primary antibodies at 4 °C overnight, followed by incubation with the horseradish peroxidase (HRP)-conjugated secondary anti-rabbit or anti-mouse IgG antibody (1:3000; Santa Cruz Biotechnology, Inc.) for 1 h at room temperature, while rabbit anti-GAPDH polyclonal antibody (1:10000; Abcam) served as a loading control.

The potential target genes of EBV-miR-BART8-3p were predicted using two publicly available algorithms RepTar (http://​reptar.​ekmd.​huji.​ac.​il/​) and DIANA TOOLS (http://​diana.​imis.​athena-innovation.​gr/​DianaTools/​index.​php?​r=​tarbase/​index). Then, the predicted candidate targets were confirmed by the software RNAhybrid (https://​bibiserv.​cebitec.​uni-bielefeld.​de/) with low minimum free energy (MFE) (≤ − 20.0). To validate whether RNF38 was a direct target of EBV-miR-BART8-3p, wide-type or mutated luciferase reporter vectors of RNF38 were transfected into HEK293T cells with a miR-BART8-3p mimic or normal control. The Firefly and Renilla luciferase activity was measured using the Dual Luciferase Reporter Assay System (Promega; Madison, WI, USA)48 h post-transfection. All measurements were repeated in triplicate.

To evaluate the pulmonary metastatic ability of NPC cells in vivo, 100 μl of SUNE-1-BART8-3p cells at a density of 1.0 × 10⁶ cells/ml or an equal amount of control cells were intravenously injected into the tail vein of the nude mice. All mice were subjected to fluorescent imaging on anLT-9MACIMSYSPLUS whole-body imaging system (Lighttools Research; Encinitas, CA, USA) 6-weeks post-injection and then sacrificed. Pulmonary metastatic lesions were sampled and quantified.

All statistical analyses were performed using the software SPSS version 24.0 (SPSS, Inc.; Chicago, IL, USA) and Graph Pad Prism 7 (GraphPad; La Jolla, CA, USA) unless stated specifically. Differences of means were tested for significant significance with Student’s t-test, and the correlation between EBV-miR-BART8-3p expression and RNF38 mRNA expression was evaluated with Spearman’s correlation analysis. A P value less than 0.05 was considered statistically significant.

We first performed differential expression analysis of the miRNA sequencing data from 6 human NPC biopsy specimens and 4 normal nasopharyngeal mucosal specimens (termed control) to systematically uncover miRNAs associated with NPC pathogenesis. The clinical data of NPC cases are showed in Table 1 and TNM stage was listed according to The Union for International Cancer Control / The American Joint Committee on Cancer (UICC/AJCC) 7th of NPC. This dataset has 1315 miRNAs which includes 40 EBV BART miRNAs and 2 EBV BHRF1 miRNAs (i.e., EBV-miR-BHRF1–1 and EBV-miR-BHRF1–2-3p). We identified 30 downregulated miRNAs and 56 upregulated miRNAs in NPC versus control using the criteria of an FC of > 1.2 and an FDR of < 0.05 (Fig. 1a; Additional file 1: Table S1). The two EBV BHRF1 miRNAs and one EBV BART miRNA (EBV-miR-BART19-5p) were not differentially expressed but all the rest 39 differentially expressed EBV BART miRNAs were all upregulated and were highly significantly enriched in the overall 56 upregulated miRNAs (corrected Fisher’s Exact Test (FET) p = 9.63E-60, 22.90 fold enrichment (FE)). Interestingly, the top 35 most significantly upregulated miRNAs identified were all EBV BART miRNAs (Additional file 1: Table S2), indicating that EBV BART miRNAs may play an important role in the development of NPC.

To explore the relationship among the miRNAs, especially the EBV BART miRNAs, we applied the well-established weighted gene co-expression network analysis (WGCNA) [33, 36‐40] to the whole miRNA sequencing data. Fig. 1b shows a heat map of the weighted miRNA co-expression network. Twenty five modules with sizes between 3 and 206 microRNAs were identified (Additional file 1: Table S3). Only the turquoise and blue modules are enriched for differentially expressed miRNAs (DEMs) in NPC versus control. Note that the turquoise module comprised of 206 microRNAs includes 39 of all the 40 EBV BART microRNAs (corrected FET p = 1.32E-30, 6.2 FE). The only EBV BART miRNA that is not in the turquoise module is EBV-miR-BART19-5p, the only EBV BART miRNA was not differentially expressed in NPC versus control. Interestingly, the blue module is only enriched for the downregulated miRNAs (corrected FET p = 0.014, 3 FE) while the turquoise module is enriched for both upregulated (corrected FET p = 2.18E-44, 6.27 FE) and down-regulated ones (corrected FET p = 1.52E-5, 3.6 FE), as shown in Additional file 1: Table S4.

To understand the downstream pathways regulated by miRNAs that may potentially drive NPC pathogenesis, we sought out to analyze the matched transcriptomic data from the same set of samples to identify DEGs in NPC versus control. We identified 2294 downregulated and 1286 upregulated genes in NPC versus control (Additional file 1: Table S5) at FDR ≤ 0.05 and FC ≥ 1.2. The genes upregulated in NPC were most enriched for the cell cycle related functional ontology terms (2.9-FE, corrected FET P = 2.21E-31),while those downregulated in NPC versus normal control were most enriched for immune response related pathways including lymphocyte activation (3.3-FE, corrected FET P = 1.91E-20), immune system process (1.8-FE, corrected FET P 1.75E-18), leukocyte activation (3-FE, corrected FET P = 1.75E-18), cell activation (2.6-FE, corrected FET P = 6.59E-18), and immune response (2.0-FE, corrected FET P = 4.59E-18), as shown in Additional file 1: Table S6 and Additional file 2: Figure S1.

Meanwhile, we identified the genes significantly negatively correlated with each miRNA. These miRNA-correlation gene (MCG) signatures were then tested for enrichment of the DEG signature between NPC and normal control. Surprisingly, the MCG signatures for all the EBV BART microRNAs up-regulated in NPC versus control were highly significantly enriched for the DEGs (corrected FET P < 1.53E-272, FE > 6.1; Additional file 1: Table S7), suggesting that the activation of EBV BART miRNAs dramatically shut down the immune defense system in NPC.

EBV-miR-BART8-3p is the most up-regulated EBV BART miRNAs in NPC (FC = 1782.9, p = 4.24E-12, FDR = 5.48E-09; Fig. 1c, Additional file 1: Table S2). To confirm BART8-3p expression by RNA sequencing, we further examined BART8-3p expression in NPC and normal nasopharyngeal mucosal specimens (CTRL). Data indicated that BART8-3p was nearly undetectable in CTRL, while strongly over-expressed in NPC (Fig. 2a). Of note, among top 10 upregulated EBV BART miRNAs, BART 16, 7, 9, 2, 1, 22, 3 and 6, have been reported to be associated with the development of NPC [41]. BART8–5 was suggested to cause progression of nasal NK-cell lymphoma [42]. We wonder whether BART8-3p plays a role in the carcinogenesis of NPC. Based on miRNA sequencing results and previous studies, we focused on exploring potential function of BART8-3p in NPC.

To investigate the function of EBV-miR-BART8-3p in NPC, two EBV-negative NPC cell lines CNE-1 and SUNE-1 were transfected with lentiviral vectors containing EBV-miR-BART8-3p precursors, and qPCR assay showed higher EBV-miR-BART8-3p expression in NPC cells transfected with EBV-miR-BART8-3p precursors than in those transfected with control vectors (P < 0.001) (Fig. 2b). Wound-healing and migration and invasion assays revealed that upregulation of EBV-miR-BART8-3p expression promoted NPC cell migration (P < 0.01) and invasion (P < 0.001) compared to control vectors-transfected cells (Fig. 2c and d), while downregulation of EBV-miR-BART8-3p suppressed EBV-positive NPCC666–1 cell migration (P < 0.01) and invasion (P < 0.01) (Fig. 2e and f). However, no significant differences were detected between GV369- or GV280-transfected NPC cells and control vectors-transfected cells in terms of cell proliferation or cell colonies (Additional file 2: Figure S2).

To assess the pulmonary metastatic ability of EBV-miR-BART8-3p in vivo, we modeled pulmonary metastasis of NPC in nude mice by inoculation of SUNE-1-BART8-3p cells or control vector-transfected SUNE-1 cells into the tail veins of nude mice. We found more metastatic tumor nodules in the lung of mice injected with SUNE-1-BART8-3p cells than injection with control vector-transfected SUNE-1 cells (P < 0.05) (Fig. 3a and b), and a greater weight of lungs was measured in mice inoculated with SUNE-1-BART8-3p cells than injection with control vector-transfected SUNE-1 cells (P < 0.05) (Fig. 3c and d). Taken together, these results suggest that EBV-miR-BART8-3p upregulation promotes NPC cell migration, invasion and metastasis.

Since EMT is closely associated with NPC metastasis [14, 15], we then examined the role of EBV-miR-BART8-3p in EMT of NPC cells. The expression of EMT- and metastasis-related markers was determined in CNE-1 and SUNE-1 cells. Western blotting showed that upregulation of EBV-miR-BART8-3p caused a reduction in the E-cadherin expression and an increase in the Snail, N-cadherin, Vimentin, MMP2 and MMP9 expression (Fig. 4b). In addition, the NPC cells overexpressing EBV-miR-BART8-3p displayed a mesenchymal-like morphology, while control vectors-transfected NPC cells showed an epithelial-like morphology (Fig. 4b).

Conversely, downregulation of EBV-miR-BART8-3p had the opposite effects in EBV-positive NPC C666–1 cells. EBV-miR-BART8-3p knockdown resulted in a reduction in Snail, N-cadherin and Vimentin expression and an increase in E-cadherin expression, and impeded the EMT-like process in NPC cells (Fig. 4). Taken together, our findings suggest that EBV-miR-BART8-3p drives an EMT process and upregulates metastasis-related proteins in NPC cells.

To address the mechanisms underlying miR-BART8-3p-driving metastasis in NPC, we determined the expression of key proteins in the NF-κB and Erk1/2 signaling pathways. Western blotting showed that upregulation of EBV-miR-BART8-3p increased the total and phosphorylation expression of TAK1, p65, IKKα, IKKβ and IKKα/β, and reduced the IκBβ, p-IκBβ, IκBα and p-IκBα expression; in addition, the Erk1/2, p-c-Raf, p-MEK1/2 and p-Erk1/2 expression was also elevated following EBV-miR-BART8-3p upregulation (Fig. 5a). On the contrary, downregulation of EBV-miR-BART8-3p resulted in opposite effects in C666–1 cells (Fig. 5b). Collectively, our data indicate that EBV-miR-BART8-3p mediates NPC cell metastasis through activating the NF-κB and Erk1/2 signaling pathways.

Then, we predicted the target genes of EBV-miR-BART8-3p. Bioinformatics analysis showed that the 3′-UTR region of the gene RNF38 was complementary with the seed sequences of EBV-miR-BART8-3p and RNF38 had the lowest MFE (Fig. 6a). Firstly, we quantified endogenous RNF38 protein expression using Western blotting, and the RNF38 protein was found to be significantly downregulated in both CNE-1-BART8-3p and SUNE-1-BART8-3p cells (Fig. 6b). Subsequently, we cloned the wild-type or mutant-type 3′-UTR of RNF38 into luciferase reporter vectors, and the luciferase activity was found to be reduced significantly in HEK293 cells transfected with the wild-type 3′-UTR of RNF38 (P < 0.001), while no changes were seen in the cells transfected with the mutant-type (Fig. 6c), suggesting that RNF38 is regulated by EBV-miR-BART8-3p.

To further quantify the RNF38 expression, we determined RNF38 mRNA expression in 19 NPC specimens and 10 normal nasopharyngeal specimens, and examined the correlation between EBV-miR-BART8-3p and RNF38 expression in the19 NPC specimens. qPCR assay showed significantly lower RNF38 expression in NPC specimens than in normal nasopharyngeal specimens (P < 0.001) (Fig. 6d), and Spearman’s correlation analysis revealed a negative correlation between EBV-miR-BART8-3p and RNF38 expression (P = 0.001) (Fig. 6e). Taken together, these data indicate that EBV-miR-BART8-3p inhibits RNF38 expression in NPC through directly targeting RNF38.

To further confirm the functions of RNF38 in the tumor-promoting effects of EBV-miR-BART8-3p, we rescued RNF38 expression in CNE-1-BART8-3p and SUNE-1-BART8-3p cells. The introduction of exogenous RNF38 significantly reduced NPC cell migration and invasion (Fig. 7a). Consistent with this biological phenotype, reinstallation of RNF38 reversed EMT and expression of metastasis-related markers by EBV-miR-BART8-3p (Fig. 7b). These data strongly support our hypothesis that RNF38 is an important mediator of EBV-miR-BART8-3p-induced NPC cell migration and invasion (Fig. 7c).