Nasopharyngeal carcinoma is an Epstein–Barr virus (EBV)-related malignancy. Recently, we found that the EBV-encoded miRNA BART2-5p was increased in the serum of patients with preclinical nasopharyngeal carcinoma and that the copy number positively correlated with disease progression. In this study, we established its role in nasopharyngeal carcinoma progression and explored underlying mechanisms and clinical significance. BART2-5p was an independent unfavorable prognostic factor for progression-free survival and its circulating abundance positively associated with distant metastasis. Ectopic expression of BART2-5p promoted migration and invasion of EBV-negative nasopharyngeal carcinoma cells, whereas genetic downregulation of BART2-5p in EBV-positive nasopharyngeal carcinoma cells decreased aggressiveness. Mechanistically, BART2-5p targeted RND3, a negative regulator of Rho signaling. Downregulation of RND3 phenocopied the effect of BART2-5p and reconstitution of RND3 rescued the phenotype. By suppressing RND3, BART2-5p activated Rho signaling to enhance cell motility. These findings suggest a novel role for EBV miRNA BART2-5p in promoting nasopharyngeal carcinoma metastasis and its potential value as a prognostic indicator or therapeutic target.

Significance:

This study shows that EBV-encoded BART2-5p miRNA suppresses expression of the RND3 Rho family GTPase, consequently promoting ROCK signaling, cell motility, and metastatic behavior of NPC cells.

Nasopharyngeal carcinoma, a malignancy with highly metastatic potential originating from the nasopharynx, is remarkable for its distinct racial and geographic distribution (1). The incidence of nasopharyngeal carcinoma is less than 1 per 100,000 persons-years in most parts of the world, however, it climbs to about 15–30 per 100,000 persons-years in endemic areas like Southern China and Southeast Asia (2, 3). Nasopharyngeal carcinoma has several etiology factors including Epstein–Barr virus (EBV) infection, genetic susceptibility, and environmental factors. Although nasopharyngeal carcinoma is sensitive to radiotherapy, the prognosis is still not satisfactory due to distant metastasis (3, 4). Unfortunately, the mechanism underlying metastasis is not fully elucidated. EBV, which was identified in 1964, is the first human virus clearly implicated in oncogenesis (5). Up to now, it is found to be associated with many human malignancies including Burkitt lymphoma, post-transplant lymphoma, Hodgkin disease, gastric cancer, and nasopharyngeal carcinoma (6). Many viral proteins, especially LMPs and EBNAs, make a great contribution to the pathogenesis of EBV-associated malignancies (7, 8).

MiRNA is a class of small RNAs that is about 22 nucleotides long and modulates gene expression by degrading mRNA or repress translation (9). MiRNAs are frequently dysregulated in various human cancers and reported to contribute to almost all the cancer hallmarks (10). However, most studies mainly focus on the role of human miRNAs before virus-encoded miRNA was first identified in EBV in 2004 (11). The BART (BamHI A rightward transcripts) region of EBV encodes more than 40 miRNAs and some of them are actively transcribed in nasopharyngeal carcinoma (12, 13). These highly abundant miRNAs have been found to play critical roles in pathogenesis of nasopharyngeal carcinoma, including antiapoptosis (14), prometastasis (15, 16), immune escape (17), and metabolism (18). Previously, we identified 12 highly expressed EBV BART miRNAs from five paired nasopharyngeal carcinoma/normal tissue and found their copy number in the tumor tissue was positively correlated with that in serum (19). Later, we systematically reviewed the diagnostic value of these upregulated circulating miRNAs in larger cohorts and found BART2-5p was positive in most patients with nasopharyngeal carcinoma but not healthy controls (20). Interestingly, the microarray data from another group was consistent with ours, indicating the distinct presence of BART2-5p in nasopharyngeal carcinoma tissue (16). Although BART2-5p was reported to facilitate immune escape from natural killer cells by targeting MICB (21) or inhibiting EBV transition from latent to lytic viral replication (22), its biological role in the progression of nasopharyngeal carcinoma remained elusive.

In this study, we observed that patients with higher copy number of BART2-5p in circulation were more likely to develop metastasis and had shorter progression-free survival. Nasopharyngeal carcinoma cells with ectopic expression of BART2-5p showed enhanced capability of migration and invasion in vitro. More importantly, nasopharyngeal carcinoma cells stably expressing BART2-5p developed more metastatic lesions in tail vein injection and spleen injection metastasis models. Further study disclosed that BART 2-5p could activate Rho signaling via direct binding with the pathway negative regulator RND3. These findings provide novel insights into the role of BART2-5p in the progression of nasopharyngeal carcinoma.

Cell lines and cell culture

Six cell lines were used in this study. SUNE1 was derived from poorly differentiated nasopharyngeal carcinoma and kept in our laboratory (23). HK1, which was derived from a differentiated squamous carcinoma of the nasopharynx (24), was kindly provided by Professor Mu-Sheng Zeng (Sun Yat-sen University Cancer Center, Guangzhou, P.R. China). EBV (+) nasopharyngeal carcinoma cell line C666-1 and three carcinoma cell lines, CNE2, HONE-1 EBV (−), and HONE-1 EBV (+) were kindly offered by Professor Sai Wah Tsao (The University of Hong Kong, Hong Kong, China; ref. 25). All cell lines were authenticated by short tandem repeat typing and cultured in RPMI1640 (Gibco) with 10% FBS (Gibco) in a humidified 5% CO2 incubator at 37°C for less than 2 months. Mycoplasma contamination was not detected in all these cell lines by PCR. Primers for Mycoplasma detection were listed in Supplementary Table S1.

Clinical sample collection

Serum samples were offered by nasopharyngeal carcinoma tissue bank at The University of Hong Kong, which collected clinical specimens retrospectively from Queen Mary Hospital (Hong Kong SAR, China) between January 1, 2010 and December 31, 2014. This project was approved by Institutional Review Board of The University of Hong Kong according to Declaration of Helsinki (2013). Cases enrolled in this study were randomly selected from the tissue bank and all diagnosed with pathologic evidence. Their clinical stage was determined according to the 7th Unio Internationale Contra Cancrum (UICC)/American Joint Committee on Cancer (AJCC) staging system. Their clinicopathologic data are recorded in Supplementary Table S2. Primary nasopharyngeal carcinoma tissues were retrospectively collected from 70 patients at Sun Yat-sen University Cancer Center (Guangzhou, China). All patients were from Y.Q. Xiang's group. This project was approved by the Institutional Clinical Ethics Review Board at Sun Yat-sen University Cancer Center according to Declaration of Helsinki (2013) and informed written consent was obtained from all the individuals. All diagnosis was confirmed by pathologists and the clinical stage was also determined according to the 7th UICC/AJCC staging system. Other clinical information was also recorded in Supplementary Table S2. The fresh sample was lysed in Qiazol (Qiagen) for total RNA extraction. The extracted RNA was stored in −80°C.

RNA oligos and cell transfection

The control mimic (cel-39), BART2-3p mimic, and BART2-5p mimic were synthesized by Integrated DNA Technology. Their inhibitors were synthesized by GeneCopoeia. Sequence: Anti-BART2-3p (5′-UUUAUUUUCUCCAAAUCGCUCCUU-3′) and Anti-BART2-5p (5′-GCAAGGGCGAAUGCAGAAAAUA-3′). The culture medium was changed to fresh RPMI1640 with 10% FBS 24 hours before transfection. The mimics and inhibitors were transfected to cells with Lipofectamine 2000 (Invitrogen) at a final concentration of 50 nmol/L. The medium was changed to fresh medium again 6 hours after transfection.

In vitro migration and invasion assays

Cell migration was assessed by transwell assay. Briefly, cells were serum starved for 4–6 hours before the assay for cell-cycle synchronization and better response to chemoattractant treatment. Then, 1 × 105 transfected or virus-infected nasopharyngeal carcinoma cells were resuspended in 100 μL (for migration) or 500 μL (for invasion) serum-free RPMI1640 medium and loaded to the inserts with 8 μm pores (Corning) for migration or Corning BioCoat Matrigel Invasion Chamber (Corning) for invasion. RPMI1640 medium (600 or 700 μL) with 10% FBS was added to the bottom compartment, respectively. 24 hours later, the inserts were put into wells containing 75% ethanol for fixation and Crystal Violet for staining. The cells on top surface were removed, followed by counting cells on bottom surface in five random optical fields with microscope.

Lentiviral construction and transduction

The lentiviral plasmids for control and stable BART2 expression were purchased from GeneCopoeia. The lentiviral vector for RND3 reconstitution and knockdown were Plenti 6 (Invitrogen) and Pll3.7 (Addgene), respectively. Primers for molecular cloning are listed in Supplementary Table S1. The lentiviral plasmids were cotransfected with ViraPower Lentiviral Packaging Mix (Invitrogen) into 293FT cells for virus production. Seventy-two hours after transfection, the media containing virus were collected and filtered using a 0.22-μm filter. For transduction, virus medium was mixed with fresh medium at 1:1 ratio and added to 6-well plates with 1 μg/mL polybrene (Sigma-Aldrich). Twelve hours later, the medium was changed with fresh medium. Twenty-four hours later, puromycin or blasticidin was added to the medium at a final concentration of 3 or 20 μg/mL, respectively, for 5 days selection.

In vivo metastasis assay

All experiments were approved by the Committee on the Use of Live Animals in Teaching and Research of The University of Hong Kong. Male BALB/c nude mice that were 4–5 weeks old with 18–20 g body weight were purchased from Lab of Animal Units of The University of Hong Kong. The in vivo metastasis model was established by two ways: tail vein injection and spleen injection. For the first model, 2 × 105 nasopharyngeal carcinoma cells stably expressing vector or BART2 were suspended in 200 μL serum-free RPMI1640 medium and injected into the tail vein of nude mice (6 mice in each group). Eight weeks later, the mice were sacrificed and their lungs were harvested for further examination. For spleen injection model, four groups of mice, with 9 mice in each group, received spleen injection of SUNE1/HK1-Vec/BART2 cells. Mice were anesthetized with pentobarbital sodium by intraperitoneal injection and 2 × 105 nasopharyngeal carcinoma cells with or without BART2 suspended in 40 μL medium with Matrigel were injected in the spleen. The leakage was largely prevented as the Matrigel soon becomes solid at 37°C. After confirming that no bleeding or leakage was occurring, the spleen was then returned to the peritoneal cavity. The muscle layers and skin incision were closed by absorbable suture and nylon suture, respectively. The status of mice was checked every 3 days and all deaths were recorded. Five aspects of an animal's conditions, including body weight, physical appearance, measurable clinical signs, unprovoked behavior, and response to external stimuli were checked for each time. If rapid weight loss, persistent anorexia or dehydration, ulcerated tumors or large tumors that interfere with normal movement, and any conditions interfering with eating or drinking, the animals were euthanized immediately with over dose pentobarbital sodium. Eight weeks later, the rest of mice were sacrificed for in vivo imaging and their major organs were fixed for IHC staining.

IHC staining

IHC staining was performed based on paraffin sections. For each paraffin block, slides from five different layers, which were about 30 μm in depth, were examined for metastasis under microscopy. As nasopharyngeal carcinoma cells infected by lentivirus expressed GFP, anti-GFP antibody (Abcam) was used to locate the metastatic lesions formed by nasopharyngeal carcinoma cells. Colonies consisting of more than 50 GFP-positive cells were considered as a metastatic site. Anti-RND3 antibody (Abcam) was used to assess the RND3 level in vivo.

Bioinformatic prediction of miRNA targets

The candidate targets of BART2-5p were initially predicted by Targetscan Custom (http://www.targetscan.org/vert_50/seedmatch.html) and miRDB (http://mirdb.org/). The 3′untranslated region (UTR) of the potential target genes predicted by both were obtained from NCBI and subjected to PITA analysis (https://genie.weizmann.ac.il/pubs/mir07/mir07_prediction.html). The cut-off point for selection was ΔΔG score < −10.

Luciferase reporter assay

The 3′UTR of eight potential targets and MICB (served as a positive control) were cloned into pmirGLO Dual-Luciferase miRNA Target Expression Vector (Promega). The four mutant forms of RND3 3′UTR were derived from plasmid PCR and homologous recombination (Vazyme). Primers for construction of mutated plasmids are listed in Supplementary Table S1. For luciferase reporter assay, wild-type or mutated 3′UTR pmirGLO was cotransfected with 50 nmol/L control mimic or BART2-5p mimic into SUNE1 cells. Luciferase activity was measured with Dual-Glo Luciferase Assay System (Promega) at indicated time points.

Reverse transcription and qPCR

The method for quantification of serum miRNAs was described in published guideline (26) and our recent work (20). Briefly, serum miRNAs were extracted with miRNeasy RNA Isolation Kit (Qiagen) according to the manual. Reverse transcription of miRNA was performed with TaqMan miRNA Transcription Kit and microRNA-Specific Stem-Loop Primers (Applied Biosystems). qPCR was done triplicate on ABI 7900 Real-time PCR machine with the following conditions: 95°C for 5 minutes, 40 cycles of 15 seconds at 95°C and 1 minute at 60°C. The exact copy number of circulating BART2-5p can be calculated from established standard curve. For the study of association of BART2-5p and RND3 in clinical samples, total RNA was extracted from nasopharyngeal carcinoma tissue with QIAzol. RNA integrity was evaluated by Agilent 2100 Bioanalyzer, RIN ≥ 7 was considered qualified for downstream assays. After DNase I treatment (Life Technologies), 500 ng of total RNA was reverse transcribed with PrimeScript RT Reagent Kit (Takara) for RND3 quantification. The primers used are listed in Supplementary Table S1. qPCR was based on SYBR Green method and the premix reagent was brought from Roche (Switzerland). BART2-5p was reverse transcribed and quantified with TaqMan miRNA Transcription Kit (Applied Biosystems) and microRNA-Specific Stem-Loop Primers (Applied Biosystems). 18S, RNU6B, and spiked-in miRNAs served as internal control for data normalization of tissue samples and serum samples, respectively. Controls without template were included as negative control. Melting curve for each well was checked after the experiment to ensure specific amplification. To evaluate amplification efficiency, standard curve was established with serial diluted samples: y = −3.531x + 41.272, R² = 0.976 (BART2-5p); y = −3.206x + 35.676, R² = 0.98 (RND3). Amplification efficiency was 91.9% and 105.1%, respectively, which can be calculated from the curve, meeting the premise of 2−ΔΔCt method. Therefore, the fold changes were calculated by 2−ΔΔCt method.

Ribonucleoprotein immunoprecipitation assay

Ribonucleoprotein immunoprecipitation (RIP) assay was performed with RIP-Assay Kit for microRNA (MBL) according to the user manual. In brief, nasopharyngeal carcinoma cells (SUNE1 and CNE2) stably expressing vector or BART2 were lysed in RIP lysis buffer and precleared by protein G agarose beads (Pierce). Precleared cell lysate (500 μL) was incubated with protein G agarose beads immobilized with human anti-Argonaute 2 (Ago2) antibody (MBL) or isotype control mouse IgG (MBL) at 4°C overnight with rotation. Next, the post-immunoprecipitation beads were washed three times with 1 mL wash buffer and 100 μL mixture was dispensed to a new tube for Western blotting to check immunoprecipitation efficiency. The remaining mixture was used for following RNA isolation. The retracted RNA was reverse transcribed with PrimeScript RT Reagent Kit (Takara). RT-PCR was performed to analyze the enrichment of RND3 mRNA on the RNA-induced silencing complex.

Western blot analysis

Cells were harvested and lysed with RIPA Buffer (Thermo Fisher Scientific) supplemented with Protease Inhibitor (Roche) and Phosphatase Inhibitor (Cell Signaling Technology). Proteins were separated by SDS gel in electrophoresis and transferred to polyvinylidene difluoride membrane (Millipore). The membrane was incubated with the primary antibody listed in Supplementary Table S3 at 4°C overnight, followed by washing and probing with horseradish peroxidase (HRP)-conjugated secondary antibody at room temperature for 1 hour. The membrane was developed with Luminata Forte Western HRP Substrate (Millipore).

F-Actin staining

To visualize F-Actin, Alexa Fluor 488 Phalloidin (Invitrogen) was used for staining according to the manufacture's protocol. In brief, the attached cells were washed twice with prewarmed PBS and fixed with 3.7% formaldehyde solution in PBS for 10 minutes at room temperature, then washed with PBS twice again. 0.1% Triton X-100 in PBS was used to permeabilize cells. Subsequently, the cells were stained with Alexa Fluor 488 Phalloidin in PBS supplemented with 1% BSA at room temperature for 30 minutes, washed with PBS twice, and mounted with DAPI-containing medium (Abcam). High-resolution pictures were taken with the LSM700 confocal microscope in the Faculty Core Facility of Li Ka Shing Faculty of Medicine.

Statistical analysis

All experiments were performed in triplicate unless specified. Data were shown as mean ± SEM. Comparisons between two groups were performed with χ2 test (categorical variables) or Student t test (continuous variables). The correlation between BART2-5p and RND3 in clinical samples was analyzed with Pearson correlation coefficient. Two-way ANOVA was used to analyze luciferase activity of RND3. The progression-free time was analyzed by Kaplan–Meier method and P value was calculated with log-rank analysis. Multivariate survival analysis was performed with a Cox regression model. All the data were analyzed by SPSS 19.0 software. All the tests were two-sided and considered statistically significant if P < 0.05.

High level of circulating BART2-5p is associated with poor prognosis of nasopharyngeal carcinoma

In previous study (19), we discovered high abundance of BART2-5p in both nasopharyngeal carcinoma tissue and circulation in a small cohort of patients with nasopharyngeal carcinoma, which was later confirmed by Cai and colleagues in 2015 (16). In our recent attempt to develop a new biomarker to facilitate nasopharyngeal carcinoma early diagnosis, we found this miRNA was elevated in circulation of two large cohorts of patients with nasopharyngeal carcinoma and even 90% of preclinical patients, suggesting its important biological role in nasopharyngeal carcinoma (20). Hence, we further looked into clinical association of this miRNA by collecting serum samples from 150 patients with nasopharyngeal carcinoma at Queen Mary Hospital (Hong Kong SAR, China) and quantifying the level of BART2-5p in these samples. The result suggested that patients with advanced stages (stage III/IV) had significantly higher copy number of BART 2-5p in circulation than early-stage patients (stage I/II; Fig. 1A). Meanwhile, with the median Ct value as the cut-off point, the patients were divided into two groups. Those patients with lower Ct value, which indicated higher copy number in circulation, tended to have shorter progression-free survival time (PFS; Fig. 1B). In addition, we analyzed the association between circulating abundance of BART2-5p and PFS in early-stage patients or advanced-stage patients with Cox regression model. PFS and Ct value of BART2-5p (higher Ct value indicates lower copy number in circulation) were the two variables in the model. For early-stage patients, HR [95% confidence interval (CI)], 0.678 (0.460–0.982), P < 0.05; for advanced-stage patients, HR (95% CI), 0.434 (0.187–0.883), P < 0.05. This result suggested that circulating abundance of BART2-5p were significantly associated with PFS either in early- or advanced-stage patients. Furthermore, multivariate analysis revealed that high-copy number of BART2-5p was an independent and unfavorable prognostic indicator in nasopharyngeal carcinoma (HR, 2.184; 95% CI of HR, 1.119–4.263; P = 0.022; Supplementary Table S4). Next, we associated the circulating abundance of BART2-5p with clinical features including age, gender, clinical stage, recurrence, and metastasis. The result indicated that high circulating level of BART2-5p was positively correlated with recurrence (P = 0.017) and metastasis (P = 0.013; Supplementary Table S5).

Figure 1.

BART2-5p is enriched in patients with advanced nasopharyngeal carcinoma and correlated with shorter survival. A, Circulating level of BART2-5p in 150 patients with nasopharyngeal carcinoma from Queen Mary Hospital was quantified by qPCR. The copy number distribution in early- and advanced-stage patients is shown. B, The survival data of these patients was also analyzed with median Ct value 33 as the cut-off point for classification. Blue line and green line indicate patient group with lower or higher level of circulating BART2-5p, respectively. ***, P < 0.001.

Figure 1.

BART2-5p is enriched in patients with advanced nasopharyngeal carcinoma and correlated with shorter survival. A, Circulating level of BART2-5p in 150 patients with nasopharyngeal carcinoma from Queen Mary Hospital was quantified by qPCR. The copy number distribution in early- and advanced-stage patients is shown. B, The survival data of these patients was also analyzed with median Ct value 33 as the cut-off point for classification. Blue line and green line indicate patient group with lower or higher level of circulating BART2-5p, respectively. ***, P < 0.001.

Close modal

BART2-5p promotes nasopharyngeal carcinoma aggressiveness in vitro

The BART2 miRNA was transduced into two EBV (−) nasopharyngeal carcinoma cell lines (SUNE1 and HK1) and carcinoma cell line CNE2 by lentivirus to establish stable cell lines. The relative abundance of BART2-5p in vector, BART2-5p cells, and nasopharyngeal carcinoma tissue was determined by qPCR and shown in Supplementary Fig. S1A. The EBV (−) nasopharyngeal carcinoma cell lines did not express BART2-5p originally but after lentivirus transduction they had comparable level of BART2-5p as nasopharyngeal carcinoma tissue. In transwell migration assay, BART2 overexpression significantly increased cell migration in the EBV (−) nasopharyngeal carcinoma cell lines (Fig. 2A, left; Supplementary Fig. S1B). Meanwhile, BART2 was found to promote nasopharyngeal carcinoma cell invasion dramatically in Matrigel invasion chamber (Fig. 2A, right; Supplementary Fig. S1B). As BART2 could be further processed into two mature miRNAs in cells, we used siRNA to knockdown them, respectively, to determine which one was responsible for the enhanced capability of migration and invasion. Upon transfection of the siRNAs, cell migration and invasion were evaluated by transwell assay and the knockdown efficiency was confirmed by qPCR at the same time. The assay results suggested that knockdown of BART2-5p instead of BART2-3p significantly decreased cell motility of SUNE1-BART2, HK1-BART2, and CNE2-BART2 cells compared with that of the control and no synthetic effect of BART2-3p and BART2-5p was observed (Fig. 2B and C; Supplementary Fig. S1C).

Figure 2.

BART2-5p significantly enhanced cell migration and invasion in vitro and promoted lung metastasis in tail vein injection model. A, The eight representative images show the result of SUNE1 and HK1 cells with or without stable expression of BART2 in cell migration and invasion.Five random fields of vision were selected to count the migrated/invaded cells for each well and the statistical result of which is shown in the bar chart. B, Knockdown efficiency of BART2-3p and BART2-5p upon transfection of corresponding inhibitors was quantified by qPCR. C, SUNE1 or HK1 cells transiently transfected with negative control, inhibitor of BART2-3p, inhibitor of BART2-5p, or inhibitors of both were evaluated for invasion capability in Matrigel-coated invasion chambers. The bar chart shows the result. D, SUNE1 and HK1 cells with or without stable expression of BART2 were injected into the tail vein of nude mice. All mice were sacrificed for examination of lung metastasis after 8 weeks. Nasopharyngeal carcinoma cells with BART2 expression formed more and larger metastatic lesions in lung (left). The number of metastatic nodules on the lung surface were counted and compared (right). E, IHC staining was performed to examine micrometastasis in the lung. As all nasopharyngeal carcinoma cells were GFP labeled, they were found under microscope. Compared with the minor metastasis formed by nasopharyngeal carcinoma cells without BART2 expression, BART2-expressing cells developed into larger metastatic lesions. NC, negative control; N.S, not significant; Vec, vector. **, P < 0.01; ***, P < 0.001.

Figure 2.

BART2-5p significantly enhanced cell migration and invasion in vitro and promoted lung metastasis in tail vein injection model. A, The eight representative images show the result of SUNE1 and HK1 cells with or without stable expression of BART2 in cell migration and invasion.Five random fields of vision were selected to count the migrated/invaded cells for each well and the statistical result of which is shown in the bar chart. B, Knockdown efficiency of BART2-3p and BART2-5p upon transfection of corresponding inhibitors was quantified by qPCR. C, SUNE1 or HK1 cells transiently transfected with negative control, inhibitor of BART2-3p, inhibitor of BART2-5p, or inhibitors of both were evaluated for invasion capability in Matrigel-coated invasion chambers. The bar chart shows the result. D, SUNE1 and HK1 cells with or without stable expression of BART2 were injected into the tail vein of nude mice. All mice were sacrificed for examination of lung metastasis after 8 weeks. Nasopharyngeal carcinoma cells with BART2 expression formed more and larger metastatic lesions in lung (left). The number of metastatic nodules on the lung surface were counted and compared (right). E, IHC staining was performed to examine micrometastasis in the lung. As all nasopharyngeal carcinoma cells were GFP labeled, they were found under microscope. Compared with the minor metastasis formed by nasopharyngeal carcinoma cells without BART2 expression, BART2-expressing cells developed into larger metastatic lesions. NC, negative control; N.S, not significant; Vec, vector. **, P < 0.01; ***, P < 0.001.

Close modal

EBV miRNA BART2-5p promotes nasopharyngeal carcinoma metastasis in vivo

We further evaluated metastatic capability of nasopharyngeal carcinoma cells in vivo by establishing two metastasis models: tail vein injection model and spontaneous spleen–liver metastasis model. Initially, 2 × 105 SUNE1, HK1, or CNE2 cells transduced with vector/BART2 were injected into the tail vein of nude mice. All mice were sacrificed 2 months later and their lungs were excised for further examination. The result revealed that BART2-expressing nasopharyngeal carcinoma cells formed more and larger metastatic lesions on the surface of the lungs (Fig. 2D; Supplementary Fig. S1D). As all the nasopharyngeal carcinoma cells transduced with lentivirus were labeled with GFP, IHC staining was performed to visualize inner metastatic sites under microscope. As shown in Fig. 2E; Supplementary Fig. S1E, BART2-expressing nasopharyngeal carcinoma cells formed larger metastatic lesions. The classical process of metastasis includes following steps: local invasion, intravasation, survival in the circulation, extravasation, and colonization; however, tail vein metastasis model lacks the first two steps of the process as cancer cells were directly injected into circulation. Therefore, we proceeded to evaluate nasopharyngeal carcinoma cell metastatic capability in spleen injection model in which cancer cells would grow into a tumor in spleen and then experience a complete process of metastasis. Vector/BART2-expressing cells were suspended in mixture of serum-free RPMI1640 and Matrigel and then inoculated to the spleen of nude mice. As these cells were also labeled with GFP, metastasis could be observed by in vivo imaging. At the endpoint, BART2-expressing cells, instead of vector-carrying cells, were found to develop extensive metastasis in the abdominal cavity in imaging. Major organs of mice were excised and checked carefully for overt metastasis on organ surface. The result indicated that BART2 could promote nasopharyngeal carcinoma metastasis in multiple organs including lung, liver, intestine, and kidney (Fig. 3A; Supplementary Fig. S1F). Subsequently, those organs were fixed for IHC staining to examine minor metastasis under microscope (Fig. 3B; Supplementary Fig. S1G). Particularly, some mice with extensive metastasis were sacrificed because of animal welfare during the observation period, their survival data were also recorded for following survival analysis, which suggested that mice inoculated with BART2-expressing cells tended to have significantly shorter survival (Fig. 3C; Supplementary Fig. S1H). This spleen injection assay was repeated with SUNE1 and CNE2 cells for once and the number of cases with metastatic sites in different organs is summarized, respectively, in Supplementary Table S6. In brief, ectopic expression of BART2 enhanced incidence rate of metastasis in many organs, especially in lung and intestine. Although single disseminated cancer cells can be found in both organs, mice injected with control cells did not develop metastatic colonies with more than 50 cells. In contrast, around half the mice injected with BART2-expressing nasopharyngeal carcinoma cells developed such metastatic colonies, as examined under microscope. For liver metastasis, the frequency was higher and both control and BART2 cells were observed to develop overt metastasis on the surface. This is probably due to the fact that cells injected to the spleen could easily migrate to liver via splenic vein. However, mice with inoculation of BART2-expressing cells still had a higher frequency of metastasis, either being observed directly or checked under microscope, indicating a prometastatic role of BART2 in liver. Interestingly, a few mice injected with BART2-expressing cells but not control cells were found to have metastatic nodules on renal capsule, probably through implantation metastasis. Overall, these results suggested that BART2 could promote cell metastasis in vivo, either by hematogenous metastasis or implantation metastasis.

Figure 3.

BART2 promotes nasopharyngeal carcinoma metastasis in spontaneous spleen–liver metastasis model. A,In vivo imaging was used to visualize metastatic status in abdominal cavity of mice at the endpoint. Metastatic tumor with GFP florescence was visualized (left). Major organs were excised from sacrificed mice for examination of overt metastasis. Blue arrows, metastatic nodules on organ surface (right). B, Slides from different organs were subjected to IHC staining. Brown, GFP expression. Compared with single disseminated cancer cells found in organs of control mice, cells with BART2 expression developed into more and larger metastatic colonies in these organs. C, Survival analysis of mice in each group during the observation period. Each group had 9 mice. Vec, vector.

Figure 3.

BART2 promotes nasopharyngeal carcinoma metastasis in spontaneous spleen–liver metastasis model. A,In vivo imaging was used to visualize metastatic status in abdominal cavity of mice at the endpoint. Metastatic tumor with GFP florescence was visualized (left). Major organs were excised from sacrificed mice for examination of overt metastasis. Blue arrows, metastatic nodules on organ surface (right). B, Slides from different organs were subjected to IHC staining. Brown, GFP expression. Compared with single disseminated cancer cells found in organs of control mice, cells with BART2 expression developed into more and larger metastatic colonies in these organs. C, Survival analysis of mice in each group during the observation period. Each group had 9 mice. Vec, vector.

Close modal

BART2-5p targets RND3 directly in vitro and in vivo

To search potential targets of BART2-5p, we resorted to online bioinformatic predictor and literature retrieval. Targetscan Custom and miRDB were used to predict downstream targets of BART2-5p and 156 genes were hit by both. The 3′UTR of the 156 genes were then obtained on NCBI and sent for PITA analysis. Previously reported metastasis suppressor with ΔΔG score < −10 in PITA analysis were considered for further analysis. In this way, we identified eight potential targets and cloned their 3′UTR into pmir-GLO plasmid for luciferase assay (Fig. 4A). As MICB was once reported as the target of BART2-5p, it served as a positive control here (21). Compared with negative control, the luciferase signal did not decrease when cells were cotransfected with luciferase reporter vector harboring 3′UTR of 7/8 potential target genes and BART2-5p mimic except for RND3 (Fig. 4B). As cells cotransfected with luciferase reporter containing wild-type of 3′UTR of RND3 and BART2-5p mimic showed significantly decreased luciferase intensity, especially at 36 hours after transfection (Fig. 4C) we further tested the responsible binding sites by introducing mutations into each binding site predicted by bioinformatic analysis. Primers for mutated cloning are listed in Supplementary Table S1. In total, four potential binding sites were mutated in the 3′UTR of RND3 at position 931, 1497, 1522, and 1635 nucleotides (Supplementary Fig. S2A). Mutations at position 1497 and 1635 rather than 931 and 1522 rescued downregulated luciferase activity, suggesting BART2-5p suppress RND3 by binding at position 1497 and 1635 of its 3′UTR (Fig. 4D). Western blot analysis performed with SUNE1 and HK1 cells infected with different titers of lentivirus and collected at different time points upon infection also indicated that RND3 was downregulated in BART2-expressing cells (Fig. 4E). The downregulation of RND3 was also confirmed in CNE2 cells (Supplementary Fig. S2B). To demonstrate direct binding of BART2-5p and endogenous RND3, we further performed RIP assay by pulling down Ago2--microRNA–mRNA complex. RT-PCR result showed that RND3 mRNA was only enriched in complex pulled down by Ago2 antibody–conjugated beads from BART2-expressing cell lysates, indicating their direct interaction in cells (Fig. 4F; Supplementary Fig. S2C). Western blot analysis confirmed that Ago2–RNA complex was immunoprecipitated successfully and specifically in both cell lines (Fig. 4G; Supplementary Fig. S2D). In addition, expression of RND3 in lung metastasis samples from nude mice was determined by IHC staining. The IHC results indicated that expression of RND3 was decreased in metastatic tumors formed by BART2-expressing cells, compared with metastatic tumors formed by control cells (Fig. 4H; Supplementary Fig. S2E). To verify these findings in the clinic, the correlation of BART2-5p with RND3 was studied in primary nasopharyngeal carcinoma tissues by qPCR. The results demonstrated that expression of RND3 was negatively correlated with BART2-5p level in primary nasopharyngeal carcinoma samples (r = −0.643; P < 0.0001; Fig. 4I). Clinical information of the 70 nasopharyngeal carcinoma cases is attached in Supplementary Table S2.

Figure 4.

BART2-5p targets RND3 in nasopharyngeal carcinoma. A, The Venn diagram shows that 156 genes were predicted as the target genes of BART2-5p by both bioinformatic websites (miRDB and Target Scan). The 156 genes were further narrowed down by PITA analysis with cutoff ΔG score < −10. Among them, 8 genes reported to be associated with cell movement were selected for luciferase assay. B, Seven of eight potential targets failed to be validated in following luciferase assay as the luciferase signal did not decrease significantly when plasmids harboring their 3′UTR were cotransfected with BART2-5p. MICB, a gene previously reported to be the target of BART2-5p, served as a positive control here. C, Plasmid harboring RND3 3′UTR was cotransfected with 50 nmol/L negative control or BART2-5p mimics and luciferase activity was measured at indicated time points. D, Compared with negative control, BART2-5p transfection resulted in a decrease of luciferase signal when wild-type RND3 3′UTR was cotransfected. This effect could be abolished by introducing mutation at positions 1497 and 1635 rather than positions 931 and 1522. E, SUNE1 and HK1 cells were infected by lentivirus at multiplicity of infection (MOI) 5 or 10 and collected at 1, 2, and 3 weeks after infection for Western blot analysis to determine RND3 expression. F, The RT-PCR revealed that RND3 3′UTR was enriched on the RISC complex in BART2-expressing cells, but not in cells carrying vector. Input, cell lysate before immunoprecipitation. NC, negative control, precipitates pulled down by isotype control mouse IgG. Ago2, precipitates pulled down by anti-Ago2 antibody. Ago2 is a core component of RISC. G, Quality control of RIP assay. Western blot analysis confirmed that Ago2 was successfully and specifically pulled down by anti-Ago2 antibody. Light and heavy chains of primary antibody were also detected. H, The staining of RND3 in metastatic colonies formed by BART2-expressing cells in lung was weaker than that formed by control cells, as shown in IHC. I, The abundance of BART2-5p and RND3 was determined in 70 nasopharyngeal carcinoma cases by qPCR. The results suggested that RND3 expression was negatively correlated with BART2-5p expression (r = −0.643; P < 0.0001). N.S, not significant; Vec, vector. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 4.

BART2-5p targets RND3 in nasopharyngeal carcinoma. A, The Venn diagram shows that 156 genes were predicted as the target genes of BART2-5p by both bioinformatic websites (miRDB and Target Scan). The 156 genes were further narrowed down by PITA analysis with cutoff ΔG score < −10. Among them, 8 genes reported to be associated with cell movement were selected for luciferase assay. B, Seven of eight potential targets failed to be validated in following luciferase assay as the luciferase signal did not decrease significantly when plasmids harboring their 3′UTR were cotransfected with BART2-5p. MICB, a gene previously reported to be the target of BART2-5p, served as a positive control here. C, Plasmid harboring RND3 3′UTR was cotransfected with 50 nmol/L negative control or BART2-5p mimics and luciferase activity was measured at indicated time points. D, Compared with negative control, BART2-5p transfection resulted in a decrease of luciferase signal when wild-type RND3 3′UTR was cotransfected. This effect could be abolished by introducing mutation at positions 1497 and 1635 rather than positions 931 and 1522. E, SUNE1 and HK1 cells were infected by lentivirus at multiplicity of infection (MOI) 5 or 10 and collected at 1, 2, and 3 weeks after infection for Western blot analysis to determine RND3 expression. F, The RT-PCR revealed that RND3 3′UTR was enriched on the RISC complex in BART2-expressing cells, but not in cells carrying vector. Input, cell lysate before immunoprecipitation. NC, negative control, precipitates pulled down by isotype control mouse IgG. Ago2, precipitates pulled down by anti-Ago2 antibody. Ago2 is a core component of RISC. G, Quality control of RIP assay. Western blot analysis confirmed that Ago2 was successfully and specifically pulled down by anti-Ago2 antibody. Light and heavy chains of primary antibody were also detected. H, The staining of RND3 in metastatic colonies formed by BART2-expressing cells in lung was weaker than that formed by control cells, as shown in IHC. I, The abundance of BART2-5p and RND3 was determined in 70 nasopharyngeal carcinoma cases by qPCR. The results suggested that RND3 expression was negatively correlated with BART2-5p expression (r = −0.643; P < 0.0001). N.S, not significant; Vec, vector. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Close modal

BART2-5p activates cytoskeleton reconstruction to promote cell motility via ROCK signaling pathway

To determine whether reduced RND3 was responsible for enhanced cell motility, first we used short hairpin RNA (shRNA) to knockdown endogenous RND3 for phenocopy study. The knockdown efficiency was examined by Western blot analysis and nasopharyngeal carcinoma cells knocked down by RND3-sh1&2 were selected for following study due to higher knockdown efficiency (Fig. 5A; Supplementary Fig. S2F, left). In transwell assay, knockdown of RND3 by sh1 and sh2 dramatically promoted cell migration and invasion in SUNE1, HK1, and CNE2 (Fig. 5B; Supplementary Fig. S2F, middle and right). We further tested whether restoration of RND3 could rescue the phenotype generated by BART2-5p. The successful reconstitution of RND3 in vector/BART2-expressing nasopharyngeal carcinoma cells was confirmed by Western blot analysis (Fig. 5C; Supplementary Fig. S2G, left). Reconstitution of RND3 significantly undermined nasopharyngeal carcinoma cell migration and invasion, as shown in Fig. 5D and Supplementary Fig. S2G. To examine whether the findings above could be reproduced in EBV (+) cells, we quantified the level of BART2-5p in C666-1 EBV (+) and HONE-1 EBV (+). To assess cell motility, we performed migration and invasion assays after knocking down of BART2-5p by siRNA or overexpression of RND3. Compared with the control, knockdown of BART2-5p significantly decreased migrated cells in C666-1 and HONE-1 EBV (+) cells, but not in HONE-1 EBV (−) cells, while overexpression of RND3 undermined cell migration in all three cell lines (Fig. 5E; Supplementary Fig. S2H). For invasion, knockdown of BART2-5p or ectopic expression of RND3 led to significantly less invaded cells in C666-1 and HONE-1 EBV (+) cells but not in HONE-1 EBV (−) cells (Fig. 5F; Supplementary Fig. S2I). As RND3 was reported as a negative regulator of ROCK signaling, serving as a metastasis suppressor (27), F-Actin staining was performed to visualize cytoskeleton reconstruction. Compared with the control, nasopharyngeal carcinoma cells transfected with BART2-5p mimic formed more stress fibers at 6 hours after transfection, however, the stress fibers diminished if RND3 expression vector was transfected into the cells 24 hours earlier or 1 μmol/L ROCK inhibitor Y-27632 was coadministered to culture medium for 15 minutes incubation at the end point (Fig. 5G; Supplementary Fig. S3A). This result suggested that BART2-5p promoted cytoskeleton reconstruction via activating ROCK signaling pathway.

Figure 5.

BART2-5p activates ROCK to promote cell motility. A, Both SUNE1 and HK1 cells were infected by lentivirus particles carrying scramble or shRNAs targeting RND3. Two different shRNAs were used to knockdown RND3. Knockdown efficiency was examined by Western blot analysis here. B, Compared with cells infected with scramble lentivirus, cells with knockdown of RND3 had significantly enhanced cell motility, as shown in transwell assay. Migrated or invaded cells were counted on the basis of five random fields for each well and compared statistically. The result of comparison is shown in the bar charts. C,RND3 was overexpressed in EBV (−) nasopharyngeal carcinoma cell lines with or without stable expression of BART2 by lentivirus infection. Western blot analysis confirmed successful restoration of RND3. D, Reinstatement of RND3 by lentivirus attenuated nasopharyngeal carcinoma cell migration and invasion significantly. A statistical analysis was conducted after counting number of migrated or invaded cells in each well, the result of which is shown in the bar charts (right). E, Cell migration of C666-1 was significantly attenuated after knockdown of BART2-5p with siRNA or RND3 overexpression. Migrated cells were counted on the basis of five random fields for each well and compared statistically. F, Invaded C666-1 cells were significantly decreased after knockdown of BART2-5p with siRNA or overexpression of RND3. Invaded cells were counted on the basis of five random fields for each well, the result of which is shown in the bar chart. G, Two nasopharyngeal carcinoma cell lines (SUNE1 and HK1) were transiently transfected with control mimic (Cel- 39), BART2-5p mimic, BART2-5p mimic + pcDNA3.1Vec, BART2-5p mimic + RND3 or BART2-5p, and Y-27632 treatment. For control cells and BART2-5p–transfected cells, F-actin staining was performed to visualize stress fiber formation at 6 hours after transfection. For cells transfected by BART2-5p and pcDNA3.1Vec or RND3-expressing vector, vector or RND3 was transfected 24 hours earlier than transfection of BART2-5p to allow gene expression in cells. Pictures were taken at 6 hours after transfection of BART2-5p. ROCK inhibitor Y-27632 (1 μmol/L) was added to the culture medium for a 15-minute incubation at 6 hours after transfection of BART2-5p before fixation. Con, control; Vec, vector. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 5.

BART2-5p activates ROCK to promote cell motility. A, Both SUNE1 and HK1 cells were infected by lentivirus particles carrying scramble or shRNAs targeting RND3. Two different shRNAs were used to knockdown RND3. Knockdown efficiency was examined by Western blot analysis here. B, Compared with cells infected with scramble lentivirus, cells with knockdown of RND3 had significantly enhanced cell motility, as shown in transwell assay. Migrated or invaded cells were counted on the basis of five random fields for each well and compared statistically. The result of comparison is shown in the bar charts. C,RND3 was overexpressed in EBV (−) nasopharyngeal carcinoma cell lines with or without stable expression of BART2 by lentivirus infection. Western blot analysis confirmed successful restoration of RND3. D, Reinstatement of RND3 by lentivirus attenuated nasopharyngeal carcinoma cell migration and invasion significantly. A statistical analysis was conducted after counting number of migrated or invaded cells in each well, the result of which is shown in the bar charts (right). E, Cell migration of C666-1 was significantly attenuated after knockdown of BART2-5p with siRNA or RND3 overexpression. Migrated cells were counted on the basis of five random fields for each well and compared statistically. F, Invaded C666-1 cells were significantly decreased after knockdown of BART2-5p with siRNA or overexpression of RND3. Invaded cells were counted on the basis of five random fields for each well, the result of which is shown in the bar chart. G, Two nasopharyngeal carcinoma cell lines (SUNE1 and HK1) were transiently transfected with control mimic (Cel- 39), BART2-5p mimic, BART2-5p mimic + pcDNA3.1Vec, BART2-5p mimic + RND3 or BART2-5p, and Y-27632 treatment. For control cells and BART2-5p–transfected cells, F-actin staining was performed to visualize stress fiber formation at 6 hours after transfection. For cells transfected by BART2-5p and pcDNA3.1Vec or RND3-expressing vector, vector or RND3 was transfected 24 hours earlier than transfection of BART2-5p to allow gene expression in cells. Pictures were taken at 6 hours after transfection of BART2-5p. ROCK inhibitor Y-27632 (1 μmol/L) was added to the culture medium for a 15-minute incubation at 6 hours after transfection of BART2-5p before fixation. Con, control; Vec, vector. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Close modal

BART2-5p activates ROCK signaling pathway via suppressing RND3

To investigate molecular mechanism of RND3 in regulating nasopharyngeal carcinoma cell motility and invasiveness, we examined the status of ROCK signaling pathway under different conditions. Here, two key components of ROCK pathway have been examined by Western blot analysis: MYPT1 (myosin phosphatase target subunit) and MLC2 (myosin light chain 2). MYPT1 is a negative regulator of cytoskeleton remodeling as it could remove the phosphate group from phosphorylated MLC2. MYPT1 itself, loses activity if it is phosphorylated at Thr 853 by ROCK1 (28). As shown by Western blot analysis, EBV (−) nasopharyngeal carcinoma cells infected by lentivirus particles carrying either BART2 or RND3 shRNA had increased level of phosphorylated MLC2 (Ser 19), indicating cytoskeleton reconstruction was activated. Also, BART2 expression or knockdown of RND3 led to the phosphorylation of MYPT1 at Thr 853. As a result, it promoted the phosphorylation of MLC2 as MYPT1 lost its phosphatase activity (Fig. 6A; Supplementary Fig. S3B). In addition, we examined the status of this signaling pathway after the phenotype was rescued by reconstitution of RND3 or ROCK inhibitor Y-27632. Being consistent with the phenotype observed, either reinstallation of RND3 or Y27632 treatment would dephosphorylate MLC2 at Ser 19 and MYPT1 at Thr853, suggesting disassembling stress fibers (Fig. 6B; Supplementary Fig. S3C). For EBV (+) nasopharyngeal carcinoma cell lines, knockdown of BART2-5p by siRNA led to derepression of RND3 and dephosphorylation of MLC2, MYPT1 in C666-1, and HONE-1 EBV (+) cells instead of HONE-1 EBV (−) cells (Fig. 6C; Supplementary Fig. S3D). In addition, a decreased level of MLC2 and MYPT1 phosphorylation was observed in all three cell lines with RND3 overexpression (Fig. 6C; Supplementary Fig. S3D). Taken together, these findings suggested EBV miRNA BART2-5p could suppress RND3 to activate ROCK signaling, thus promoting cell motility.

Figure 6.

BART2-5p activates ROCK via suppressing RND3. A, SUNE1 and HK1 stable cell lines of BART2 expression or RND3 knockdown were lysed for Western blot analysis. Phosphorylated MLC2 and MYPT1 were increased by either BART2 overexpression or RND3 knockdown. B, Western blot analysis shows induced activation of ROCK signaling by BART2 could be abolished by reinstatement of RND3 or Y-27632 treatment. C, Western blot analysis indicated that siBART2-5p restored RND3 expression to inhibit ROCK signaling. Overexpression of RND3 had similar effects. D, A schematic diagram illustrating the proposed role of BART2-5p in nasopharyngeal carcinoma metastasis. Vec, vector.

Figure 6.

BART2-5p activates ROCK via suppressing RND3. A, SUNE1 and HK1 stable cell lines of BART2 expression or RND3 knockdown were lysed for Western blot analysis. Phosphorylated MLC2 and MYPT1 were increased by either BART2 overexpression or RND3 knockdown. B, Western blot analysis shows induced activation of ROCK signaling by BART2 could be abolished by reinstatement of RND3 or Y-27632 treatment. C, Western blot analysis indicated that siBART2-5p restored RND3 expression to inhibit ROCK signaling. Overexpression of RND3 had similar effects. D, A schematic diagram illustrating the proposed role of BART2-5p in nasopharyngeal carcinoma metastasis. Vec, vector.

Close modal

Nasopharyngeal carcinoma is a unique malignancy due to its distinct geographic distribution and close association with EBV infection. The effects of EBV viral proteins (e.g., LMPs and EBNAs) on nasopharyngeal carcinoma development and progression have been widely investigated (29, 30). In 2004, the first virus-encoded miRNA was identified in the BamH1 A region of EBV (11), and currently more than 40 mature miRNAs have been reported (31). Comprehensive profiling of these BART miRNAs has been conducted by us and many other groups (13, 32), and the results indicated most BART miRNAs were present in nasopharyngeal carcinoma tissues, despite their variance in abundance. Some BART miRNAs with rich abundance in nasopharyngeal carcinoma tissue has aroused the interest of scientists and their functions were investigated in recent years. They have been reported to suppress apoptosis (33), contribute to immune evasion (34, 35), or metastasis of nasopharyngeal carcinoma (15, 16). In addition, some of BART miRNAs enriched in circulation could be developed as biomarkers for nasopharyngeal carcinoma diagnosis (36).

In our previous work, we found BART2-5p could be developed as a good biomarker as it is abundant in most patients with early-stage nasopharyngeal carcinoma and even goes up 1 year prior to clinical diagnosis, suggesting its important biological role in nasopharyngeal carcinoma development and progression. Notably, BART2-3p, another mature form of BART2, were found to be expressed at a low level and almost cannot be detected in nasopharyngeal carcinoma, suggesting it might not play important roles in nasopharyngeal carcinoma development and progression (13, 37). Another clue came from the analysis of survival data of patients with nasopharyngeal carcinoma with/without high abundance of circulating BART2-5p. High circulating level of BART2-5p was closely correlated with recurrence and metastasis, serving as an independent unfavorable prognostic factor in nasopharyngeal carcinoma. We later demonstrated that BART 2-5p contributed to cell migration and invasion in vitro and metastasis in vivo. To demonstrate RND3 was the target of BART2-5p, luciferase activity assay and RIP assay were performed. By means of phenotype rescue and phenocopy, RND3 was found to be a negative regulator of ROCK signaling pathway which could be inhibited by BART2-5p.

The important role of ROCK signaling pathway in cell migration was first elucidated by Kimura and colleagues at the end of 1990s (38). In brief, Rho-associated protein kinases (ROCK1 and ROCK2) could be autoinhibited by its carboxy terminal region in quiescent state. The inhibitory effect could be removed once they bind with activated Rho. ROCK is a serine/threonine kinase, which could phosphorylate myosin light chain directly or myosin light chain phosphatase (MYPT1), both of which promote stress fiber assembly and cell migration. RND3, which was validated as the target gene of BART2-5p, belongs to the superfamily of RasGTPases. Interestingly, unlike other members in the superfamily, RND3 always bind to GTP as it lacks essential amino acids for GTP hydrolysis (39). Consequently, the regulation of its activity is controlled at transcriptional, posttranscriptional, and translational levels instead of the classic GTP-GDP cycle (40). Many studies have revealed the effects of RND3 on actin reconstruction (41), cell polarity (42), cell apoptosis (43, 44), cell-cycle arrest (45), and cell differentiation (46, 47). In particular, it was described as a ROCK inhibitor as it binds with its kinase region (amino-terminal part) and might therefore interfere with the kinase activity. In cancer, the effects of RND3 have been reported as a double-edged sword. For example, RND3 was reported as a metastasis suppressor in hepatocellular carcinoma (HCC; ref. 48) and esophageal squamous cell carcinoma (ESCC; ref. 49). However, RND3 could promote metastasis in gastric cancer (50) and melanoma (51). In this study, RND3 acted as a metastasis suppressor, being consistent with its role in HCC and ESCC. The contradictory results were probably due to the cell lineage, variant baseline expression of RND3 in cells, and different stages of malignant progression.

Our study also has certain limitations. Metastasis is a complex process including multiple stages. We have demonstrated that BART2-5p could promote distant metastasis (lung metastasis) in the spleen injection metastasis model and enhanced cell motility due to activated ROCK signaling. However, successful formation of metastatic sites cannot be attributed to stronger capability of cell movement only. Other important factors, such as antiapoptosis effect, immune evasion, and angiogenesis switch after colonization were not evaluated in this study. Also, the high level of circulating BART2-5p suggested it might be a messenger in host–pathogen interaction. The EBV or nasopharyngeal carcinoma cells may regulate the functions of distant recipient cells via circulating miRNA. Unfortunately, we cannot point out the recipient cells at present. Another limitation was caused by cell identity. As some researchers pointed out that some nasopharyngeal carcinoma cell lines, including CNE2 and HONE-1, were contaminated by Hela, data based on these two cell lines are placed in Supplementary Figures, although evidence supported a prometastatic role of BART2-5p in these two carcinoma cell lines (52). Despite the fact that BART2-5p was encoded by EBV genome, probably it is still functional in non-EBV–infected cells.

In summary, we have demonstrated the important role of BART2-5p in nasopharyngeal carcinoma metastasis and its underlying mechanism. Here, BART2-5p activated ROCK signaling pathway via suppressing RND3, a negative regulator of ROCK kinase. Downstream genes MYPT1 would be inactivated and MLC2 would be activated by phosphorylation, as a result, stress fiber formation was induced to promote cell motility and invasion (Fig. 6D). This may provide a novel insight into the metastasis mechanism of virus-associated malignancies. More importantly, as miRNA silencing technology has been developed (53), targeting the oncogenic viral miRNAs could be a promising therapeutic option for the treatment of virus-associated malignancies in the near future.

No potential conflicts of interest were disclosed.

Conception and design: C. Jiang, J.-P. Yun, D.L.W. Kwong, X.-Y. Guan

Development of methodology: C. Jiang, D.L.W. Kwong

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Jiang, L. Li, Y.-Q. Xiang, D.L.W. Kwong

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C. Jiang, X.-Y. Guan

Writing, review, and/or revision of the manuscript: C. Jiang, L. Li, J.-P. Yun, D.L.W. Kwong, X.-Y. Guan

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.L. Lung, T. Zeng, J. Lu, S.W. Tsao, M.-S. Zeng

Study supervision: D.L.W. Kwong, X.-Y. Guan

This work was supported by grants from the Hong Kong Research Grant Council: Area of Excellence scheme (AoE/M-06/08), General Research Fund (GRF17161116), and Collaborative Research Fund (C7027-16G); the National Key R&D Program of China (2017YFC1309000); National Natural Science Foundation of China/Research Grants Council Joint Research Scheme (N_HKU735118); and the Shenzhen Peacock Team Project (KQTD201533117210153 and KQTD2018041118502879). X.-Y. Guan is Sophie YM Chan Professor in Cancer Research.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Yu
MC
. 
Nasopharyngeal carcinoma: epidemiology and dietary factors
.
IARC Sci Publ
1991
;
105
:
39
47
.
2.
Chang
ET
,
Adami
HO
. 
The enigmatic epidemiology of nasopharyngeal carcinoma
.
Cancer Epidemiol Biomarkers Prev
2006
;
15
:
1765
77
.
3.
Wei
WI
,
Sham
JS
. 
Nasopharyngeal carcinoma
.
Lancet
2005
;
365
:
2041
54
.
4.
Lee
AW
,
Ma
BB
,
Ng
WT
,
Chan
AT
. 
Management of nasopharyngeal carcinoma: current practice and future perspective
.
J Clin Oncol
2015
;
33
:
3356
64
.
5.
Epstein
MA
,
Achong
BG
,
Barr
YM
. 
Virus particles in cultured lymphoblasts from Burkitt's lymphoma
.
Lancet
1964
;
1
:
702
3
.
6.
Kutok
JL
,
Wang
F
. 
Spectrum of Epstein-Barr virus-associated diseases
.
Annu Rev Pathol
2006
;
1
:
375
404
.
7.
Tsao
SW
,
Tsang
CM
,
To
KF
,
Lo
KW
. 
The role of Epstein-Barr virus in epithelial malignancies
.
J Pathol
2015
;
235
:
323
33
.
8.
Vockerodt
M
,
Yap
LF
,
Shannon-Lowe
C
,
Curley
H
,
Wei
W
,
Vrzalikova
K
, et al
The Epstein-Barr virus and the pathogenesis of lymphoma
.
J Pathol
2015
;
235
:
312
22
.
9.
Bartel
DP
. 
MicroRNAs: genomics, biogenesis, mechanism, and function
.
Cell
2004
;
116
:
281
97
.
10.
Esquela-Kerscher
A
,
Slack
FJ
. 
Oncomirs - microRNAs with a role in cancer
.
Nat Rev Cancer
2006
;
6
:
259
69
.
11.
Pfeffer
S
,
Zavolan
M
,
Grasser
FA
,
Chien
M
,
Russo
JJ
,
Ju
J
, et al
Identification of virus-encoded microRNAs
.
Science
2004
;
304
:
734
6
.
12.
Cai
X
,
Schafer
A
,
Lu
S
,
Bilello
JP
,
Desrosiers
RC
,
Edwards
R
, et al
Epstein-Barr virus microRNAs are evolutionarily conserved and differentially expressed
.
PLoS Pathog
2006
;
2
:
e23
.
13.
Cosmopoulos
K
,
Pegtel
M
,
Hawkins
J
,
Moffett
H
,
Novina
C
,
Middeldorp
J
, et al
Comprehensive profiling of Epstein-Barr virus microRNAs in nasopharyngeal carcinoma
.
J Virol
2009
;
83
:
2357
67
.
14.
Choy
EY
,
Siu
KL
,
Kok
KH
,
Lung
RW
,
Tsang
CM
,
To
KF
, et al
An Epstein-Barr virus-encoded microRNA targets PUMA to promote host cell survival
.
J Exp Med
2008
;
205
:
2551
60
.
15.
Cai
L
,
Ye
Y
,
Jiang
Q
,
Chen
Y
,
Lyu
X
,
Li
J
, et al
Epstein-Barr virus-encoded microRNA BART1 induces tumour metastasis by regulating PTEN-dependent pathways in nasopharyngeal carcinoma
.
Nat Commun
2015
;
6
:
7353
.
16.
Cai
LM
,
Lyu
XM
,
Luo
WR
,
Cui
XF
,
Ye
YF
,
Yuan
CC
, et al
EBV-miR-BART7-3p promotes the EMT and metastasis of nasopharyngeal carcinoma cells by suppressing the tumor suppressor PTEN
.
Oncogene
2015
;
34
:
2156
66
.
17.
Lung
RW
,
Tong
JH
,
Sung
YM
,
Leung
PS
,
Ng
DC
,
Chau
SL
, et al
Modulation of LMP2A expression by a newly identified Epstein-Barr virus-encoded microRNA miR-BART22
.
Neoplasia
2009
;
11
:
1174
84
.
18.
Ye
Y
,
Zhou
Y
,
Zhang
L
,
Chen
Y
,
Lyu
X
,
Cai
L
, et al
EBV-miR-BART1 is involved in regulating metabolism-associated genes in nasopharyngeal carcinoma
.
Biochem Biophys Res Commun
2013
;
436
:
19
24
.
19.
Wong
AM
,
Kong
KL
,
Tsang
JW
,
Kwong
DL
,
Guan
XY
. 
Profiling of Epstein-Barr virus-encoded microRNAs in nasopharyngeal carcinoma reveals potential biomarkers and oncomirs
.
Cancer
2012
;
118
:
698
710
.
20.
Jiang
C
,
Chen
J
,
Xie
S
,
Zhang
L
,
Xiang
Y
,
Lung
M
, et al
Evaluation of circulating EBV microRNA BART2-5p in facilitating early detection and screening of nasopharyngeal carcinoma
.
Int J Cancer
2018
;
143
:
3209
17
.
21.
Nachmani
D
,
Stern-Ginossar
N
,
Sarid
R
,
Mandelboim
O
. 
Diverse herpesvirus microRNAs target the stress-induced immune ligand MICB to escape recognition by natural killer cells
.
Cell Host Microbe
2009
;
5
:
376
85
.
22.
Barth
S
,
Pfuhl
T
,
Mamiani
A
,
Ehses
C
,
Roemer
K
,
Kremmer
E
, et al
Epstein-Barr virus-encoded microRNA miR-BART2 down-regulates the viral DNA polymerase BALF5
.
Nucleic Acids Res
2008
;
36
:
666
75
.
23.
Sizhong
Z
,
Xiukung
G
,
Yi
Z
. 
Cytogenetic studies on an epithelial cell line derived from poorly differentiated nasopharyngeal carcinoma
.
Int J Cancer
1983
;
31
:
587
90
.
24.
Huang
DP
,
Ho
JH
,
Poon
YF
,
Chew
EC
,
Saw
D
,
Lui
M
, et al
Establishment of a cell line (NPC/HK1) from a differentiated squamous carcinoma of the nasopharynx
.
Int J Cancer
1980
;
26
:
127
32
.
25.
Glaser
R
,
Zhang
HY
,
Yao
KT
,
Zhu
HC
,
Wang
FX
,
Li
GY
, et al
Two epithelial tumor cell lines (HNE-1 and HONE-1) latently infected with Epstein-Barr virus that were derived from nasopharyngeal carcinomas
.
Proc Natl Acad Sci U S A
1989
;
86
:
9524
8
.
26.
Kroh
EM
,
Parkin
RK
,
Mitchell
PS
,
Tewari
M
. 
Analysis of circulating microRNA biomarkers in plasma and serum using quantitative reverse transcription-PCR (qRT-PCR)
.
Methods
2010
;
50
:
298
301
.
27.
Riento
K
,
Villalonga
P
,
Garg
R
,
Ridley
A
. 
Function and regulation of RhoE
.
Biochem Soc Trans
2005
;
33
:
649
51
.
28.
Velasco
G
,
Armstrong
C
,
Morrice
N
,
Frame
S
,
Cohen
P
. 
Phosphorylation of the regulatory subunit of smooth muscle protein phosphatase 1M at Thr850 induces its dissociation from myosin
.
FEBS Lett
2002
;
527
:
101
4
.
29.
Sivachandran
N
,
Sarkari
F
,
Frappier
L
. 
Epstein-Barr nuclear antigen 1 contributes to nasopharyngeal carcinoma through disruption of PML nuclear bodies
.
PLoS Pathog
2008
;
4
:
e1000170
.
30.
Tsao
SW
,
Tramoutanis
G
,
Dawson
CW
,
Lo
AK
,
Huang
DP
. 
The significance of LMP1 expression in nasopharyngeal carcinoma
.
Semin Cancer Biol
2002
;
12
:
473
87
.
31.
Lo
AK
,
Dawson
CW
,
Jin
DY
,
Lo
KW
. 
The pathological roles of BART miRNAs in nasopharyngeal carcinoma
.
J Pathol
2012
;
227
:
392
403
.
32.
Gourzones
C
,
Jimenez
AS
,
Busson
P
. 
Profiling of Epstein-Barr virus-encoded microRNAs in nasopharyngeal carcinoma reveals potential biomarkers and oncomirs
.
Cancer
2012
;
118
:
4634
.
33.
Marquitz
AR
,
Mathur
A
,
Nam
CS
,
Raab-Traub
N
. 
The Epstein-Barr Virus BART microRNAs target the pro-apoptotic protein Bim
.
Virology
2011
;
412
:
392
400
.
34.
Lisnic
VJ
,
Krmpotic
A
,
Jonjic
S
. 
Modulation of natural killer cell activity by viruses
.
Curr Opin Microbiol
2010
;
13
:
530
9
.</bib
>
35.
Iizasa
H
,
Wulff
BE
,
Alla
NR
,
Maragkakis
M
,
Megraw
M
,
Hatzigeorgiou
A
, et al
Editing of Epstein-Barr virus-encoded BART6 microRNAs controls their dicer targeting and consequently affects viral latency
.
J Biol Chem
2010
;
285
:
33358
70
.
36.
Zhang
G
,
Zong
J
,
Lin
S
,
Verhoeven
RJ
,
Tong
S
,
Chen
Y
, et al
Circulating Epstein-Barr virus microRNAs miR-BART7 and miR-BART13 as biomarkers for nasopharyngeal carcinoma diagnosis and treatment
.
Int J Cancer
2015
;
136
:
E301
12
.
37.
Chen
SJ
,
Chen
GH
,
Chen
YH
,
Liu
CY
,
Chang
KP
,
Chang
YS
, et al
Characterization of Epstein-Barr virus miRNAome in nasopharyngeal carcinoma by deep sequencing
.
PLoS One
2010
;
5
.
DOI: 10.1371/journal.pone.0012745
.
38.
Amano
M
,
Chihara
K
,
Kimura
K
,
Fukata
Y
,
Nakamura
N
,
Matsuura
Y
, et al
Formation of actin stress fibers and focal adhesions enhanced by Rho-kinase
.
Science
1997
;
275
:
1308
11
.
39.
Aspenstrom
P
,
Ruusala
A
,
Pacholsky
D
. 
Taking Rho GTPases to the next level: the cellular functions of atypical Rho GTPases
.
Exp Cell Res
2007
;
313
:
3673
9
.
40.
Riou
P
,
Villalonga
P
,
Ridley
AJ
. 
Rnd proteins: multifunctional regulators of the cytoskeleton and cell cycle progression
.
BioEssays
2010
;
32
:
986
92
.
41.
Guasch
RM
,
Scambler
P
,
Jones
GE
,
Ridley
AJ
. 
RhoE regulates actin cytoskeleton organization and cell migration
.
Mol Cell Biol
1998
;
18
:
4761
71
.
42.
Lesiak
A
,
Pelz
C
,
Ando
H
,
Zhu
M
,
Davare
M
,
Lambert
TJ
, et al
A genome-wide screen of CREB occupancy identifies the RhoA inhibitors Par6C and RND3 as regulators of BDNF-induced synaptogenesis
.
PLoS One
2013
;
8
:
e64658
.
43.
Ongusaha
PP
,
Kim
HG
,
Boswell
SA
,
Ridley
AJ
,
Der
CJ
,
Dotto
GP
, et al
RhoE is a pro-survival p53 target gene that inhibits ROCK I-mediated apoptosis in response to genotoxic stress
.
Curr Biol
2006
;
16
:
2466
72
.
44.
Zhu
Y
,
Zhou
J
,
Xia
H
,
Chen
X
,
Qiu
M
,
Huang
J
, et al
The Rho GTPase RhoE is a p53-regulated candidate tumor suppressor in cancer cells
.
Int J Oncol
2014
;
44
:
896
904
.
45.
Villalonga
P
,
Guasch
RM
,
Riento
K
,
Ridley
AJ
. 
RhoE inhibits cell cycle progression and Ras-induced transformation
.
Mol Cell Biol
2004
;
24
:
7829
40
.
46.
Liebig
T
,
Erasmus
J
,
Kalaji
R
,
Davies
D
,
Loirand
G
,
Ridley
A
, et al
RhoE is required for keratinocyte differentiation and stratification
.
Mol Biol Cell
2009
;
20
:
452
63
.
47.
Maria Vieites
J
,
Torre Rde
L
,
Ramirez Mdel
C
,
Torres
MI
,
Sanchez-Pozo
A
,
Gil
A
, et al
Exogenous nucleosides accelerate differentiation of rat intestinal epithelial cells
.
Br J Nutr
2008
;
99
:
732
8
.
48.
Grise
F
,
Sena
S
,
Bidaud-Meynard
A
,
Baud
J
,
Hiriart
JB
,
Makki
K
, et al
RND3/RhoE Is down-regulated in hepatocellular carcinoma and controls cellular invasion
.
Hepatology
2012
;
55
:
1766
75
.
49.
Zhao
H
,
Yang
J
,
Fan
T
,
Li
S
,
Ren
X
. 
RhoE functions as a tumor suppressor in esophageal squamous cell carcinoma and modulates the PTEN/PI3K/Akt signaling pathway
.
Tumour Biol
2012
;
33
:
1363
74
.
50.
Feng
B
,
Li
K
,
Zhong
H
,
Ren
G
,
Wang
H
,
Shang
Y
, et al
RhoE promotes metastasis in gastric cancer through a mechanism dependent on enhanced expression of CXCR4
.
PLoS One
2013
;
8
:
e81709
.
DOI: 10.1371/journal.pone.0081709.
51.
Klein
RM
,
Aplin
AE
. 
RND3 regulation of the actin cytoskeleton promotes melanoma migration and invasive outgrowth in three dimensions
.
Cancer Res
2009
;
69
:
2224
33
.
52.
Strong
MJ
,
Baddoo
M
,
Nanbo
A
,
Xu
M
,
Puetter
A
,
Lin
Z
. 
Comprehensive high-throughput RNA sequencing analysis reveals contamination of multiple nasopharyngeal carcinoma cell lines with HeLa cell genomes
.
J Virol
2014
;
88
:
10696
704
.
53.
Murphy
BL
,
Obad
S
,
Bihannic
L
,
Ayrault
O
,
Zindy
F
,
Kauppinen
S
, et al
Silencing of the miR-17∼92 cluster family inhibits medulloblastoma progression
.
Cancer Res
2013
;
73
:
7068
78
.

Supplementary data