Background
Nasopharyngeal carcinoma (NPC) is the most common primary malignancy in the nasopharynx [
1,
2]. Although NPC is rare worldwide [
3], this malignancy remains highly prevalent in endemic regions, notably in southern China [
4]. Currently, radiotherapy remains the treatment of choice for NPC, and concomitant chemoradiotherapy has been proved to increase the survival [
5,
6]. Although 80 to 90% 5-year local control rates have been achieved for NPC, there are still 15 to 30% of patients developing distant metastasis [
7]. Elucidation of the mechanisms underlying NPC metastasis in NPC is therefore of great significance to improve the prognosis and treatment outcomes.
Epstein-Barr virus (EBV) has been accepted as an etiological factor for NPC. Almost 100% of non-keratinizing NPC, are associated with EBV infections [
1], which accounts for 95% of all NPCs in endemic regions [
2]. A set of EBV latent genes have been identified to play an important role in the development of NPC, including three latent membrane proteins (LMP1, LMP2A and LMP2B), EBV nuclear antigen 1 (EBNA1) and EBV-encoded RNAs (EBERs) [
3]. However, only three latent proteins (EBNA1, LMP1 and LMP2A) are expressed in type II latency of EBV infection, a predominant form of latency observed in NPC [
4]. Whereas EBV-encoded microRNAs (EBV-miRNAs) are detected in most of clinical NPC specimens and cells during all the forms of latency [
5‐
8].
EBV is the first identified human tumor-causing virus that encodes miRNAs. EBV-miRNAs consist of more than 10% of total miRNAs in NPC specimens and EBV-positive NPC C666–1 cells, and increasing evidence shows that EBV-miRNAs contribute to cancer survival [
9‐
13], metastasis [
14‐
18], immune evasion [
19,
20] and latency [
21‐
24]. Currently, EBV-miRNAs are divided into two clusters based on their locations, including
BamHI fragment H rightward open reading frame 1 (BHRF1) and
BamHI-A rightward transcripts (BART) miRNAs. EBV BART miRNAs contain 22 miRNA precursors (termed EBV-miR-BART1 to 22) that produce totally 44 mature miRNAs [
6]. Previous studies showed that EBV BART miRNAs were highly overexpressed in clinical specimens and plasma samples of NPC patients and were involved in the development and progression of NPC [
5,
25]. However, so far the role of EBV-miR-BART8-3p in NPC progression remains unknown. This study aimed to investigate the role of EBV-miR-BART8-3p in NPC, and explore the underlying mechanisms.
Materials and methods
Ethical approval
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.
Clinical specimens
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.
Animals
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.
Cell lines and culture
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.
RNA sequencing and data analysis
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].
miRNA sequencing and data analysis
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.
microRNA co-expression network analysis
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.
qPCR assay
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 transfection
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).
Cell proliferation and colony-forming assays
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.
Wound healing assay
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.
Cell migration and invasion assays
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 analysis
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.
Animal experiments
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.
Statistical analysis
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.
Discussion
EBV infection has been proved to be strongly associated with a variety of lymphoid and epithelial malignancies [
3], notably NPC. Since the EBV-miRNAs were firstly discovered in 2004, there are 44 mature EBV BART miRNAs identified to date [
43]. Our report showed that a vast majority (97.5% or 39) of the 40 EBV BART miRNAs profiled in this study were significantly upregulated in NPC and further study suggested that these upregulated EBV BART miRNAs were all co-expressed in the same gene module according to WGCNA. In addition, the analysis of DEGs and MCG signatures revealed that EBV BART miRNAs may be involved in immune response, which is little known until now [
19,
20].
So far, a few EBV BART miRNAs have been investigated and partially defined in NPC. EBV-miR-BART20-5p, EBV-miR-BART5 and EBV-miR-BART cluster I (miR-BART1-5p, miR-BART16 and miR-BART17-5p) were reported to inhibit apoptosis of NPC cells by targeting BAD, PUMA and LMP1, respectively [
9,
10,
13] while miR-BART22 and miR-BART2-5p have been found to contribute to immune evasion through targeting LMP2A and MICB, respectively [
19,
20]. Additionally, miR-BART18–5p and miR-BART2 have shown regulatory effects on viral latency by inhibiting MAP3K2 and BALF5 [
22,
23]. However, there is little knowledge on the roles of EBV BART miRNAs in metastasis of NPC [
26], and the oncogenic and tumor suppressive activities of EBV BART miRNAs remain unknown. Previous studies have demonstrated that miR-BART1, miR-BART7-3p, miR-BART9 and miR-BART10-3p induce tumor metastasis by targeting PTEN, E-cadherin and BTRC, respectively [
14‐
17]. However, the role of EBV-miR-BART8-3p is still unknown in NPC. Our findings showed that EBV-miR-BART8-3p was the most upregulated EBV BART miRNA and it contributed to NPC metastasis by targeting
RNF38. A recent study reported that miR-BART6-3p inhibits NPC cell metastasis and invasion by targeting long non-coding RNA [
18]. Therefore, like human miRNAs, EBV BART miRNAs also constitute a complex miRNA regulatory network and have shown both positive and negative effects on metastatic progression in NPC [
44]. It is therefore of great significance to investigate how EBV BART miRNAs are generated and whether they act as oncogenes or tumor suppressors in metastatic progression.
EMT is considered to be a key part of tumor migration and invasion [
45]. Until now, few studies have shown the crucial role of EBV BART miRNAs during the process of EMT in NPC, including EBV-miR-BART1, EBV-miR-BART7-3p, EBV-miR-BART9 and EBV-miR-BART10-3p [
14‐
17]. To our knowledge, the present study is the first report demonstrating that EBV-miR-BART8-3p upregulation facilitated EMT, leading to the metastasis of NPC cells. In addition, upregulation of EBV-miR-BART8-3p increased the expression of EMT-related markers and metastasis-related proteins, while downregulation of EBV-miR-BART8-3p attenuated EMT and reduced the expression of EMT-associated proteins. Further studies, however, to investigate the proportions of EBV BART miRNAs and the mechanisms underlying the role of EBV BART miRNAs during the process of EMT in NPC seem justified.
Increasing evidence suggests that NF-κB signaling plays a crucial role in maintaining EBV latency in NPC cells through regulating EBV BART miRNAs and lncRNAs [
46], and NF-κB activation is essential in the pathogenesis and progression of NPC [
47,
48]. Moreover, NF-κB and Erk1/2 signaling pathways have been found to be vital for EMT and metastasis, along with the upregulation of MMPs [
45,
49‐
51]. It is therefore hypothesized that NF-κB and Erk1/2 signaling is involved in the process of EBV-miR-BART8-3p-induced metastasis in NPC. Our data supported our hypothesis that upregulation of EBV-miR-BART8-3p induced NPC metastasis by activating the NF-κB and Erk1/2 signaling pathways, while downregulation of EBV-miR-BART8-3p had the opposite effects. Of note, previous studies have shown that EBV-miR-BART1 activates PI3K-AKT, FAK-p130
Cas and MAPK-ERK1/2 pathways, miR-BART10-3p activates β-catenin/Snail signaling and miR-BART7 activates PI3K/AKT and p-GSK-3β-ser9 signaling to promote NPC metastasis [
14‐
16]. In agreement with EBV-miR-BART1, we found that Erk1/2 signaling was closely associated with the metastasis in NPC. Interestingly, we also found, for the first time, that NF-κB signaling contributed to EBV-miR-BART8-3p-induced NPC metastasis. Further studies to examine the mechanisms underlying the contribution of NF-κB signaling to the metastasis of NPC are warranted.
RING finger protein family is involved in a variety of diverse biological processes, including oncogenesis, signal transduction, apoptosis, development and viral infection [
52].
RNF38, located at chromosome 9 (9p 13), is a member of the RING finger protein family [
52]. It is reported that chromosome 9p loss was a key event during the pathogenesis of NPC and frequently deleted in multiple cancers, such as lung cancer [
53,
54], hepatocellular carcinoma [
55], squamous cell head and neck cancer [
56] and NPC [
57]. Previous studies have shown that RNF38 might modify p53 [
58,
59] and was associated with the neuronal activity [
60]. However, the role of
RNF38 in cancer remains unknown until now. Our data showed that RNF38 might act as a tumor suppressor in NPC.
Our previous study reported that circulating EBV BART miRNAs might be used as biomarkers for early diagnosis and prognosis of NPC [
25]. Therefore, we hypothesize that EBV-miR-BART8-3p may be predictive of prognosis in NPC. Further understanding of the role and the potentially clinical value of
RNF38 in the carcinogenesis of NPC and the potential clinical value of EBV-miR-BART8-3p in NPC requires further investigations. Since a growing number of miRNAs and their targeting genes have been identified [
61] and they form complex regulatory miRNA network as we demonstrated here, further study of regulatory relationships between EBV-miR-BART8-3p and its target genes in NPC are warranted. Moreover, future studies should also examine the relationships between upregulated EBV BART miRNAs and deactivated immune response genes as well as the relationships between up-regulated genes and cell cycle related processes in NPC.