RERG suppresses cell proliferation, migration and angiogenesis through ERK/NF-κB signaling pathway in nasopharyngeal carcinoma

Background Nasopharyngeal carcinoma (NPC) is a malignancy of the head and neck that is prevalent in Southeast Asia and southern China. Recent studies in epigenetics suggest that DNA methylation plays a pivotal role in the onset and progression of cancer. Combining the methyl-DNA binding domain capture technique and cDNA microarray analysis, we identified a unique hypermethylated gene, RERG (Ras-like estrogen-regulated growth inhibitor), that was down-regulated in NPC tissues. RERG is a tumor suppressor gene that was first reported in breast cancer. However, the functions of RERG are largely unknown in other tumor types. Methods RERG expression was assessed in human subjects (NPC primary tissues and non-cancer tissues) and cell lines (NPC cell lines and an immortalized epithelial cell line NP460). Further, we investigated the methylation rate of RERG in both human subject and cell lines. 5-Aza-2’-deoxycytidine (Aza) or combined with trichostatin A (TSA) were treated to three NPC cell lines (HK1, C666-1 and HK1_EBV). In addition, the role of RERG in NPC cells and its underlying mechanisms were explored by overexpression of RERG in NPC cell lines. Results RERG was significantly down-regulated in NPC cancer nests compared to normal nasopharyngeal epithelium cells. Furthermore, the RERG promoter was frequently methylated in NPC tissues and cell lines. The RERG methylation rate yielded an area under the curve (AUC) of receiver operating characteristic (ROC) curve was 0.897 (95%CI: 0.818–0.976). The down-regulation of RERG was restored in NPC cells treated with Aza and TSA. In addition, ectopic expression of RERG in NPC cell lines resulted in a significant suppression of cell proliferation, clonogenicity, migration and invasion. RERG-overexpressing cells showed significantly slower growth and less angiogenesis in tumor xenografts in nude mice. RERG suppressed the ERK/NF-κB signaling pathway and inhibited tumor growth and angiogenesis with down-regulation of MMPs and IL8 in tumors of nude mouse xenografts. Conclusions Our results suggest that RERG is frequently silenced by promoter CpG methylation in NPC, and acts as a functional tumor suppressor by suppressing the ERK/NF-κB signaling pathway. These findings support the potential use of RERG as a novel molecular target in NPC therapy. Electronic supplementary material The online version of this article (doi:10.1186/s13046-017-0554-9) contains supplementary material, which is available to authorized users.


Background
Nasopharyngeal carcinoma (NPC) is a common head and neck malignancy and important health issue in Southeast Asia and southern China although it is rare in Japan and Western countries [1]. It is closely associated with Epstein-Barr virus (EBV) [2] and shows highly invasive and metastatic features [3]. Evidence in recent years suggests that additional genetic and epigenetic abnormalities are necessary to drive the tumorigenic process. Epigenetic changes, which involve DNA methylation and histone modifications, play critical roles in the onset and progression of cancer and other diseases [4,5]. Hypermethylation of CpG islands is a well-recognized epigenetic event in cancer [6]. Hypermethylated genomic regions frequently mediate the silencing of tumor suppressor genes (TSGs) [6,7]. Recent studies indicate that epigenetic processes occur very early in cancer development and may contribute to cancer initiation [8]. Epigenetic inactivation of potential TSGs by aberrant promoter DNA methylation has been increasingly recognized to play a key role in the tumorigenesis of NPC by altering gene functions during cell proliferation, apoptosis, and differentiation [2,7,9]. The discovery of epigenetic changes may prove to be extremely useful for early detection and prevention of cancer, including NPC [10,11].
Ras-like estrogen-regulated growth inhibitor (RERG) was initially identified as a candidate tumor suppressor gene, and is regulated by estrogen in breast tumors [12,13]. RERG is located on chromosome 12p12 and encodes a small GTP-binding and hydrolyzing protein (GTPase) of the Ras superfamily. RERG is widely expressed in multiple normal tissues, while RERG expression is lost in breast, kidney, ovary, and colon tumor tissues [12,14]. It has been reported that RERG is a prognostic marker in breast cancer, and its expression has correlated inversely with proliferation, patient survival, and the development of distant metastases [15]. Significantly hypermethylated RERG was also observed in colorectal adenocarcinoma [16,17], and breast cancer [18]. These findings suggest that RERG acts as a putative TSG, but the functions of RERG in NPC have yet to be identified.
Our genome-wide analyses using methyl-capture sequencing and cDNA microarrays have indicated that RERG is a candidate TSG in NPC. In the present study, we investigated the epigenetic regulation and potential tumor suppressor function of RERG in NPC. RERG was found to be epigenetically inactivated in NPC, and ectopic expression of RERG reduced cell proliferation, clonogenicity, migration and invasion. In addition, RERG suppressed angiogenesis, functioning as a repressor of ERK/NF-κB activation and inhibiting the expression of matrix metalloproteinases (MMPs), interleukin 8 (IL8) and interleukin 6 (IL6) in NPC cells. Overall our data demonstrate that RERG acted as a functional TSG by deactivating ERK/NF-κB signaling effectors, and RERG is frequently silenced in NPC.

NPC primary tumor biopsies and non-cancerous nasopharyngeal epithelia (NNE)
This study was performed with ethical review committee approval (2009-07-07) from the First Affiliated Hospital of Guangxi Medical University, China, and ethical approval (No. 1116) from Mie University, Japan. As shown in Table 1, primary NPC tumor biopsies were obtained from 52 newly diagnosed and untreated cases (39 males and 13 females, 44.5 ± 13.9 y.o.). The diagnoses were made by experienced pathologists according to the World Health Organization (WHO) classification, and revealed that 98.1% of cases were non-keratinizing carcinoma. A total of 17 NNE samples obtained by tonsillectomy were used as controls (11 males and 6 females, 40.1 ± 14.9 y.o.). Biopsy samples were stored in a freezer at -80°C. Biopsies of 13 untreated primary NPC tumors (7 males and 6 females, 39.5 ± 7.9 y.o.) and 9 from NNE subjects (4 males and 5 females, 52.2 ± 13.4 y.o.) were obtained for immunohistochemistry (IHC) staining, and all NPC cases were nonkeratinizing carcinoma.

RNA extraction and quantitative RT-PCR
Total RNA from the biopsy samples and cell lines was isolated with TRIzol reagent (Invitrogen, Carlsbad, CA, USA). Reverse transcription was performed using the miScript II RT kit (Qiagen, Hilden, Germany). Quantitative RT-PCR (qRT-PCR) was performed using the miScript SYBR Green PCR kit (Qiagen) and QuantiTect Primer Assay (Qiagen) by using StepOnePlus Real-Time PCR Systems (Applied Biosystems, Waltham, MA, USA). mRNA expression levels were normalized against the corresponding levels of GAPDH. Relative expression levels of NPC relative to NNE were calculated using the ΔΔCt method [29].
qPCR to quantify DNA methylation rate Genomic DNA was extracted from cell lines and biopsy tissues using a QIAamp DNA mini kit (Qiagen). To quantify gene promoter methylation, EpiTect Methyl II PCR Assay (SA Bioscience, Qiagen, Hilden, Germany) was used according to the manufacturer's instructions. In briefly, genomic DNA was exposed to digestion performed by using EpiTect Methyl II DNA Restriction Kit (SA Bioscience, Qiagen). Four reaction digestions were labelled as: no-enzyme (Mo), methylation sensitive enzyme (Ms), methylation dependent enzyme (Md) and methylation sensitive and dependent enzymes (Msd). The reaction digestions were incubated at 37°C overnight. After incubation, the reactions were stopped by heating-inactivating the enzymes at 65°C for 20 min. Methyl qPCR was carried out for each of the four digestions (Mo, Ms, Md and Msd) using the RT 2 SYBR Green ROX qPCR Mastermix (SA Bioscience, Qiagen) and Epi-Tect Methyl II PCR Primer Assay (SA Bioscience, Qiagen).

DNA bisulfite treatment and promoter methylation analysis
One microgram of genomic DNA was modified by sodium bisulfite by the EpiTect Bisulfit kit (Qiagen). Bisulfite genomic sequencing (BGS) were performed as previously described [30]. BGS was carried out for 40 cycles using Platinum Taq DNA Polymerase High Fidelity (Invitrogen) with BGS primers (RERG-BGS-F: GGAGTTTGGAGGTTTGGAAAT and RERG-BGS-R: CAAAAACAAATACCAATAACCC). Amplified BGS products were subcloned and transformed into JM109 competent Escherichia coli cells (Promega, Madison, WI, USA). At least 5 clones for each sample were randomly chosen for sequencing.

5-Aza-2'-deoxycytidine (Aza) and trichostatin A (TSA) treatment
NPC cells (1 × 10 5 ) were seeded into 100-mm dishes and allowed to grow overnight. The culture medium was then replaced with fresh medium containing Aza at a final concentration of 10 μmol/L (Sigma-Aldrich, St. Louis, MO, USA). Cells were allowed to grow for 72 h with the Aza-containing medium changed every 24 h, and some were further treated with the histone deacetylase (HDAC) inhibitor TSA (Sigma-Aldrich) for an additional 24 h. Cells were then harvested for RNA and DNA extraction. Three NPC cell lines (HK1, C666-1 and HK1_EBV), which showed inactivation of RERG, was transfected with the pCMV6-RERG plasmid or the empty-vector pCMV6-Entry plasmid using Fugene HD (Promega). After 48 h, cells were harvested for further experiments and selected with G418 (400 μg/mL; Millipore, Darmstadt, Germany) for 2 weeks to obtain stably transfected cell lines. Transfected cell lines were verified by qRT-PCR and western blotting. The stable cells were maintained in medium containing 200 μg/mL G418.

Cell proliferation assay
The effect of RERG on NPC cell proliferation was measured using an MTT assay or a CCK8 assay. NPC cells transfected with empty vector or RERG vector were seeded in 96-well plates at a proportionate density and incubated at 37°C. Subsequently, cell proliferation was assessed every 24 h for the indicated duration. Briefly, 10 μL of MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide, 5.0 mg/mL; Sigma-Aldrich) or CCK8 (cell counting kit-8, Dojindo Laboratories, Kumamoto, Japan) was added to each well, and the plates were incubated for 4 h at 37°C. Optical density (OD) was determined at 570 nm (MTT assay) or 450 nm (CCK8 assay) on a Bio-Rad model 680 microplate reader (Bio-Rad Laboratories, Hercules, CA, USA). OD values reflect the relative number of viable cells.

Colony formation assay
HK1 cells stably expressing RERG or empty vector and HK1_EBV cells transfected with RERG or empty vector after 48 h were counted and plated at 500 cells/dish into 60-mm dishes. Then cells were incubated at 37°C and at an atmosphere of 5% CO 2 for 2 weeks. Once the cells formed visible colonies, the colonies were washed twice with phosphate-buffered saline and fixed with 70% ethanol for 20 min. The cells were appropriately stained with Giemsa's solution (Merck, Darmstadt, Germany) and allowed to air dry at room temperature. The experiments were triplicated and the numbers of colonies were microscopically counted.

In vitro cell migration and invasion assay
Cell migration assays were performed using the CytoSelect Cell Migration Assay kit (membrane filter pore size, 8 μm; Cell BioLabs, San Diego, CA, USA). Cell invasion assays were performed using the CytoSelect 24-Well Cell Invasion Assay (basement membrane, membrane filter pore size, 8 μm; Cell BioLabs). HK1 cells stably expressing RERG or empty vector were suspended (0.15 × 10 6 transfected cells/well) in serum-free medium and placed in the upper chamber, and media containing 10% FBS was placed in the lower well of the migration or invasion plate. After incubation for 24 h and removal of non-migratory cells or non-invasive cells, cells that migrated or passed through the filter were stained, photographed, and counted under a microscope (magnification, ×100 and × 200) for each filter in 3 areas.

In vivo tumorigenicity assay
Male BALB/c athymic nu/nu mice (5-week-old) (weight range: 17-19 g; Japan SLC, Hamamatsu, Japan) were used to examine tumorigenicity in vivo. The animal experiments were performed according to the Mie University guidelines for laboratory animals (approval No. . Nude mice were housed in ventilated caging conditions under a 12-h dark/light cycle at constant humidity and temperature. Animals were permitted free access to sterile water and standard laboratory chow. HK1 cells (2 × 10 6 cells/mouse) stably transfected with RERG were injected subcutaneously into the left flank near the forelimb or an equal number of empty vector stable transfected HK1 cells were injected into the right flank. Tumor growth was measured every 3 days for 3 weeks. Tumor volume (mm 3 ) was calculated with a caliper (model 530-312; range 0-150 mm; Mitutoyo, Kawasaki, Japan) and evaluated using the formula: V = L × W × H × π/6, where L, W, and H represent tumor diameter in 3 mutually perpendicular planes. Mice were sacrificed by cervical dislocation at the third week after subcutaneous injection. Tumor grafts were then excised and weighed. The tumors were kept in 4% formamide and RNAlater solution for further analysis of IHC and mRNA expression levels, respectively.

Western blot analysis
Western blot was done according to the standard protocol as described previously [31]. Cultured cells were harvested at 70-80% confluence and lysed using RIPA buffer (Cell Signaling Technology Inc., Dancers, MA, USA) supplemented PMSF (Phenylmethylsulfonyl fluoride, NACALAI TESQUE, INC., Kyoto, Japan). The list of antibodies used in this study is summarized in Additional file 1: Table S1. Signals were detected by LAS4000mini (Fujifilm, Tokyo, Japan), and band intensities of western blots were quantitatively measured integrated grayscale densities in windows of identical size incorporating for each band using ImageJ software (ver. 1.48).
For digital microscopy assessment of microvessel density, the intensity and area of endothelial or periendothelial cell staining were quantitatively measured using ImageJ software (ver. 1.48). The proportion of blood vessels exhibiting endothelial (CD34-positive) or periendothelial (α-SMA -positive) cells was defined as the area fraction (%) at a magnification of × 200.

Statistical analysis
All statistical analyses were performed with SPSS v16.0 (SPSS, Chicago, IL, USA). The unpaired Student's t-test was used to compare the average of 2 groups, and the paired Student's t-test was used to compare the size and weight of tumor xenografts in the left and right flanks of nude mice. Statistical differences between IHC scores were determined by the Mann-Whitney U test. Receiver operating characteristic (ROC) curves were generated to confirm the accuracy of diagnosis by RERG methylation rate, and sensitivity and specificity were computed. P <0.05 was considered as statistically significant.

RERG is down-regulated and hypermethylated in NPC primary tumors
We performed the methyl-DNA binding domain capture technique and cDNA microarray analysis in seven NPC patients compared to five NNE patients (unpublished data). In the cDNA microarray, 5 genes of RAS type GTPase family were found to be significantly altered. Among these, 3 genes decreased under half and 2 genes increased more than two-fold compared to the NNE group (Additional file 2: Figure S1A). In the Methylated-DNA capture sequencing data, 8 genes of RAS type GTPase family that have an increased more than twofold level of CpG island methylation relative to NNE group (Additional file 2: Figure S1B). RERG, RASL11A and RASL11B were marked with DNA promoter methylation and downregulated in NPC tissues (Additional file 2: Figure S1C). RERG gene expression was most significantly decreased in cDNA microarray and highly methylated in NPC, which we chose it for our further study.
We evaluated the expression levels of RERG in NPC primary tumors by qRT-PCR and IHC staining. RERG expression was significantly lower in NPC primary tumors than NNE tissues (Fig. 1a). IHC staining was then performed to evaluate the expression of RERG proteins in NPC patients and NNE subjects. Moderate expression of RERG (Fig. 1b) was observed in the membrane and cytoplasm of normal nasopharyngeal epithelium cells, while very weak or no expression of RERG was observed in NPC cancer nests. The IHC score of NPC tissues was significantly lower than that of NNE tissues (Fig. 1b). These results suggest that RERG is down-regulated in NPC primary tumors.
A genome database, accessed using the UCSC Genome Browser (https://genome.ucsc.edu/), revealed that the RERG promoter contains a CpG island (CGI). As CGI methylation is a well-recognized epigenetic mechanism of TSGs silencing in cancer [33], we sought to determine the quantitation of methylation at the promoter of RERG. We performed methyl qPCR in 46 NPC patients and 13 NNE subjects (Fig. 1c). The methylation rate of RERG promoter was significantly higher in the NPC group (Average 32.6%) than in the no-cancer group (1.0%) (Fig. 1d, P < 0.001).
In addition, ROC curves was conducted to evaluate whether the methylation of RERG could be used to be a potential marker for screening NPC. The area under the curve (AUC) of ROC plots for RERG methylation rate in the detection of NPC was 0.897 (95%CI: 0.818-0.976) (Fig. 1e). At a meth-index cutoff value of 5.2%, the methylation rate of RERG gave 78.3% sensitivity and 100% specificity. The current results indicated that DNA methylation rate of RERG holds some promise for NPC screening.

RERG is silenced in NPC cell lines with DNA hypermethylation and restored by pharmacologic demethylation
We also investigated the mRNA expression of RERG in cell lines. RERG was found to be silenced to a greater extent in NPC cell lines HK1, C666-1 and HK1_EBV compared to an immortalized epithelial cell line (NP460) (Fig. 2a).
We then determined the methylation rate of RERG promoter in cell lines. As expected, RERG was highly methylated in all 3 NPC cell lines compared to the immortalized normal epithelial cell line, NP460 (Fig. 2b).
To further investigate methylation status, we used BGS to determine 25 individual CpG sites from -145 to +133 bp from the transcription start site of the RERG promoter. All 25 CpG sites were obviously methylated in NPC cell lines (HK1 and C666-1), but were seldom methylated in the normal cell line, NP460 (Fig. 2c).
These results indicate that RERG methylation is a common event in NPC.
To investigate whether CpG methylation directly mediates RERG down-regulation, 3 NPC cell lines were treated with the DNA methyltransferase inhibitor Aza either alone or in combination with the HDAC inhibitor Fig. 1 RERG was down-regulated and hypermethylated in NPC primary tumors. a Expression levels of RERG were measured in NPC primary tumor biopsies (n = 16) and NNE tissues (n = 13) by qRT-PCR. GAPDH was used as an internal control. **: P < 0.01 analyzed by Student's t-test. b RERG was down-regulated in NPC clinical samples (n = 13) compared to NNE epithelium (n = 9) by IHC. Left: Representative photographs of IHC analyses of the expression of RERG. Original magnification is × 200, inner enlarged magnification is × 400. Bar represents 50 μm and 20 μm, respectively. Right: Graphs represent means ± SD of IHC scores of RERG in tissues of NPC and NNE. *: P < 0.05 by Mann-Whitney U test. c Each methylation rate (%) of RERG in NPC clinical samples and NNE samples by methyl qPCR. d Means ± SD of methylation rates of RERG in NPC group (n = 46) and NNE group (n = 13). ***: P < 0.001 by Student's t-test. e ROC analysis of DNA methylation index of RERG TSA (A + T). We found that RERG expression levels were significantly restored after the demethylation treatment (Fig. 2d), suggesting that promoter methylation might contribute to the silencing of RERG in NPC cells.
In addition, the mRNA expression levels of RERG were up-regulated in demethylated HK1 (2.0 and 1.85 fold with Aza alone and Aza + TSA, respectively) and C666-1 cells (0.57 and 0.62 fold with Aza alone and Aza + TSA, Fig. 2 RERG was down-regulated and hypermethylated in NPC cell lines, and restored by pharmacologic demethylation. a Gene expression was detected in tumor cell lines (HK1, C666-1, HK1_EBV) and an immortalized normal epithelia NP460 cells (n = 4) by qRT-PCR. GAPDH was used as an internal control. b RERG was highly methylated in NPC cell lines (HK1, C666-1, HK1_EBV) compared to NP460 (n = 3) by methyl qPCR. c Bisulfite genomic sequencing of the 25 CpG sites within the RERG promoter region in 2 NPC cell lines (HK1 and C666-1) and an immortalized epithelial cell line (NP460). Five clones were randomly selected and sequenced for each sample. Each row represents analysis of an individual promoter allele. Open circles indicate unmethylated cytosines, and filled circles indicate methylated cytosines. d RERG expression of NP460 and NPC cell lines (n = 3) treated with 5-aza-2'-deoxycytidine (Aza) alone or combined with trichostatin A (TSA) (A + T) were determined by qRT-PCR. GAPDH was used as an internal control. *: P < 0.05, **: P < 0.01, ***: P < 0.001 analyzed by Student's t-test respectively) closed to the immortalized epithelial cell line NP460. RERG was highly up-regulated in demethylated HK1_EBV (8.66 and 11.51 fold with Aza alone and Aza + TSA, respectively) than NP460.
Overexpression of RERG suppresses cell proliferation, clonogenicity, migration, and invasion To assess the biological roles of RERG in NPC cells, we performed transfection of RERG-expression vector in 3 NPC cell lines (HK1, C666-1 and HK1_EBV). We established one stable RERG-expressing HK1 cells. The overexpression of RERG was confirmed by western blot (Fig. 3a, inset). Overexpression of RERG significantly inhibited NPC cell proliferation (Fig. 3a). Moreover, the proliferative effect of RERG was confirmed via ICC staining of PCNA, a marker of cell proliferation in HK1 stable cells. The number of PCNA-positive cells was significantly reduced by 15% in HK1-RERG cells (Fig. 3b). RERG also significantly reduced the colony formation (Fig. 3c), migration efficiency (Fig. 3d) and invasion (Fig. 3e) of NPC cells, when compared with emptyvector cells.

Overexpression of RERG suppresses MMPs and proangiogenic factors via the ERK and NF-κB signaling pathway in NPC cells
To explore the mechanism of tumor suppression by RERG, we compared the expression levels of major ERK and NF-κB signaling pathway members, prominent cell migration and invasion molecules (MMP-2 and MMP-9), and several pro-angiogenesis-related factors, such as vascular endothelial growth factor (VEGF), IL8 and IL6 in NPC cells. RERG reduced p-ERK and p-p65, and increased p-IKBα protein levels in NPC cell lines, as typically seen in HK1 stable cells and C666-1 cells (Fig. 4a). Also, MMPs were suppressed by the overexpression of RERG (Fig. 4b). The significant downregulation of IL8 and IL6 were confirmed in HK1 stable cell by western blotting (Fig. 4c). However, RERG showed limited effects on VEGF when compared to empty-vector cells (Fig.  S2A). As expected, RERG suppressed ERK and NF-κB signaling pathway and its downstream effectors. It has been reported that both Ras/ERK and NF-κB signaling are significantly associated with cell proliferation, survival, differentiation, and commonly involved in malignant transformation in a variety of human cancers [34]. Thus, these results suggested that RERG overexpression exerted a tumor suppressor role by attenuating the activation of ERK/NF-κB signaling pathway.

Overexpression of RERG suppresses tumorigenesis and angiogenesis in vivo
We further tested whether RERG is able to suppress the growth of NPC cells in vivo. Stable transfected HK1-RERG and HK1-Empty cells formed tumors in nude mice. A 100% tumor formation rate suggested that RERG did not affect tumor incidence in vivo. However, the subcutaneous tumor growth curve was significantly less steep for RERG-transfected cells than for controls (Fig. 5a). After 19 days, xenografts were removed from mice (Figs. 5b). The mean weight of tumors formed by RERG-transfected cells was significantly lower than that of controls (59.2 ± 17.3 mg vs 254.6 ± 84.9 mg; Fig. 5c), indicating that RERG can inhibit NPC tumorigenicity in vivo.
RERG expression was confirmed by IHC (Fig. 5d, e) and qRT-PCR (Additional file 3: Figure S2B). In a histopathological analysis, H&E staining showed that angiogenesis was suppressed in xenografts from the RERG group (Fig. 5d). To study the potential effect of RERG on angiogenesis, the development of microvessels in tumor sections of nude mice was examined by immunofluorescence staining with vascular endothelial cell marker CD34 and periendothelial cell marker α-SMA (Fig. 5d). Area fraction of CD34 and α-SMA was significantly lower in the mouse xenografts within the RERGoverexpressed tumors (Fig. 5e).

RERG mediates downregulation of MMPs and proangiogenic cytokines with the inhibition of NF-κB signaling
The NF-κB signaling pathway is known to promote tumor progression and angiogenesis [24,25]. To understand how NF-κB signaling pathway is regulated by RERG in vivo, we focused on the subunit p65 translocation into the cell nuclei. IHC staining indicated that p-p65 localized exclusively in the nuclei of control mouse xenografts and slightly in the RERG-overexpressed mouse xenografts tumors (Fig. 6a). Interestingly, a decrease in total p65 expression was also observed in RERG-overexpressed tumors (Fig. 6a). It is indicated that expression of RERG inhibited p65 nuclear translocation in vivo. We then confirmed pro-angiogenic cytokines (IL8 and IL6), TIMP-2, and MMPs (MMP-2 and MMP-9) in the isolated tumor tissues by IHC (Fig. 6a). Pro-angiogenic cytokines IL8 and IL6 were significantly downregulated in RERGoverexpressed tumors (Fig. 6b). IHC scores (Fig. 6b) indicated that overexpression of RERG significantly decreased the expression of MMPs (MMP-2 and MMP-9), and drastically up-regulated TIMP-2, which is an inhibitor of MMPs, including MMP-2 and MMP-9. Real-time qRT-PCR validated that the mRNA expression of IL8 and IL1β were downregulated by RERG in mouse xenografts tumors (Additional file 3: Figure S2B). On the other hand, the expression of TNFα and VEGF showed no significant changes in vivo (Additional file 3: Figure S2B, C). These results suggested that RERG efficiently inhibited the cell proliferation, migration, invasion and angiogenesis of NPC cells by suppressing the cell migratory and proangiogenic molecules (MMP-2, MMP-9, IL8, IL6 and IL1β) via the ERK/NF-κB signaling pathway (Fig. 6c), but not VEGF and TNFα.

Discussion
In this study, we found that RERG was frequently silenced in NPC tissues and cell lines. As mRNA expression of RERG was pharmacologically reversible, promoter Fig. 3 Overexpression of RERG inhibited cell proliferation, colony formation, migration and invasion. a Growth curves of RERG-transfected and empty-vector-transfected NPC cells (HK1, C666-1, HK1_EBV). Data are means ± SD (n = 5). The insets show protein levels of RERG in each cell line transfected with empty and RERG vector. b ICC staining of HK1 stable cells against PCNA (magnification, ×200). PCNA-positive percentages in cultured HK1-RERG cells and HK1-empty cells (control) are presented as means ± SD (n = 4). c Colony formation assays. Cell colonies were stained with Giemsa. Quantitative analyses of colony formation efficiency show the means ± SD (n = 3) of at least 2 independent experiments. d Migration assay. Cells were seeded in a migration chamber and cultured for 24 h. Migratory cells were stained, photographed under a microscope and quantitatively analyzed. e Invasion assay. Cells were seeded in an upper chamber and cultured for 24 h. Invasive cells were stained, photographed under a microscope and quantitatively analyzed. Data are shown as means ± SD (n = 3) of at least 2 independent experiments. *: P < 0.05, **: P < 0.01, ***: P < 0.001 by Student's t-test methylation seems to be the major mechanism of inactivating RERG, and histone modification may partly coenhance its demethylation effect. Epigenetic gene silencing is associated with the onset and progression of various cancers [7]. The frequent, predominant epigenetic inactivation of RERG in NPC points to the importance of this gene in tumorigenesis. We also found that RERG exerts its tumor suppressor functions by inhibiting ERK/NF-κB signaling pathway, resulting in suppression of tumor cell proliferation, clonogenicity, migration, invasion and angiogenesis (Fig. 6c). These results suggest that RERG suppresses cellular growth through signaling pathways in which these molecules participate. Although this and other studies have demonstrated that RERG is downregulated in multiple cancers, few studies have reported on its function and related mechanisms in tumorigenesis.
To the best of our knowledge this study is the first to reveal the underlying antitumor mechanisms of RERG in NPC.
RERG is a member of RAS GTPase superfamily, which had previously been implicated regulating Ras/MEK/ ERK signaling pathway activation [12]. The Ras/MEK//  for immunofluorescence analyses, graphs represent average area fraction (%) ± SD of microvessels/field by CD34 and α-SMA area fraction (%) in tumors from nude mice analyzed by Image J. Data are shown as means ± SD. **: P < 0.01, ***: P < 0.001 by Mann-Whitney U test or Student's t-test ERK pathway, one of mitogen-activated protein kinases (MAPKs), function in a variety of cellular regulation leading to cell growth and development, and is hyperactivated in a variety of human cancers [34], such as gastric adenocarcinoma, hepatocarcinoma, colon cancer [35,36] and NPC [37]. Phosphorylation of ERK proteins via the Ras/MEK/ERK signaling pathway cascade induces the activation of transcription factors NF-κB, AP-1, and ETS [20]. NF-κB is also a well-known master regulator of host inflammatory responses and NF-κB activation is a common event in human cancers, including NPC [38]. Activation of NF-κB influences the proliferation, survival, invasive, and metastatic properties of cancer cells by regulating the transcription of various important target genes [39]. Our results indicate that overexpression of RERG inhibited ERK/NF-κB in several Fig. 6 RERG suppressed MMPs and pro-angiogenic factors through inhibition of NF-κB signaling. a IHC analyses or immunofluorescence analyses of the expression of NF-κB (p65 and p-p65), IL8, IL6, TIMP-2, MMP-2, and MMP-9 in tumors from nude mice. Nuclei were counterstained by DAPI (blue) in the merged pictures of immunofluorescence analyses. Original magnification is × 200 inner enlarged magnification is × 400. Bar represents 50 μm and 20 μm, respectively. b IHC scores of NF-κB (p65 and p-p65), IL8, IL6, TIMP-2, MMP-2, and MMP-9 in tumors from nude mice. **: P < 0.01, ***: P < 0.001 by Mann-Whitney U test. c Schematic diagram for the molecular basis of RERG as a tumor suppressor gene in NPC, based on decreasing cell proliferation and inhibiting migration, invasion and angiogenesis through the ERK/NF-κB signaling pathway NPC cell lines, and reduced total p65 and activated p65 both in vitro and in vivo. Several reports indicated that the inhibition of NF-κB activity in NPC cells resulted in the down-regulation of downstream target genes such as MMPs and IL8 [40,41]. Therefore, it is conceivable that RERG affects the ERK/NF-κB signaling pathway in NPC cells.
The tumor microenvironment, composed of noncancer cells and their stroma, influences on metastasis as a dynamic process [42] and angiogenesis response to a tumor [43]. Various of pro-tumorigenic factors, including MMPs and cytokines are produced by tumor cells as well as by the tumor stroma. MMPs, a family of zinc metallo-endopeptidases capable of digesting the extracellular matrix and basement membrane, promote tumor invasion and metastasis [44]. Decreased MMP expression and activity can inhibit cancer cell adhesion, migration, invasion and angiogenesis. In this study, we confirmed that ectopic expression of RERG reduced MMP-2 and MMP-9 expression in vitro and in vivo, although some differences were observed between the results of in vitro and in vivo. In mouse xenografts tumors model, the MMPs can be secreted by cancer cells as well as cancer-associated fibroblasts (CAF). The crosstalk between cancer cells and tumor stroma in the tumor microenvironment enhances the production of proangiogenic protein, including MMPs. A possible explanation is that overexpression of RERG might suppress the crosstalk between cancer cells and tumor stroma in vivo, but not in vitro. Additionally, IHC showed that TIMP-2, an inhibitor of MMPs, was up-regulated in mouse xenografts. It was reported that TIMPs exert antiangiogenic activity by suppressing MMPs or by directly inhibiting endothelial cell proliferation [45]. An imbalance between MMPs and TIMPs might result in the deposition or degradation of the extracellular matrix [46]. In the present study, therefore, we speculated that the overexpression of RERG mediated the suppression of ERK/NF-κB pathway, reducing the expression of MMP-2 and MMP-9 in NPC. These results might provide a molecular basis for its ability to inhibit cell proliferation, migration and invasion.
In addition, we observed that in a mouse xenograft model, RERG inhibited microvessel development as well as tumor growth. CD34 (an endothelial cell marker) and α-SMA (a pericyte marker) was dramatically decreased in the mouse xenografts within the RERG-expression. Our previous IHC study suggested that α-SMA-positive stromal cells may recruit circulating endothelial progenitor cells into NPC stroma for angiogenesis [32]. As α-SMA is widely accepted as CAF marker, there is a possibility of the activation and the recruitment of the stromal cells, and it may be inhibited by the over-expression of the novel tumor suppressor gene (RERG). As overexpression of RERG reduced the expression of α-SMA in mouse xenograft model, it is suggested that RERG might involve in the suppressing the crosstalk between tumor cells and tumor microenvironment.
Angiogenesis is a highly regulated process involving sprouting from pre-existing vessels and maturation into new blood vessels [47]. It can be triggered and modified by various factors, including cytokines, growth factors and their receptors, chemokines, and matrix metalloproteinases [47]. However, there are no studies on the relationship between angiogenesis and RERG. VEGF is one of the most well-known pro-angiogenic factors, stimulating the sprouting and proliferation of endothelia cells [43]. IL8 is a pro-angiogenesis cytokine, which may promote both tumor growth and the formation of new blood vessels [48,49]. IL6 promotes the migration and invasion in NPC cell lines and mediated through regulation of the expression of MMP-2 and MMP-9 [50]. IL-1 α and β were detectable in the majority of primary NPC biopsies and a fraction of metastatic lesions and its absence in control nasopharyngeal tissues [51]. In this study, RERG decreased IL8, IL6 and IL1β expression and angiogenesis in tumor xenografts in nude mice. In addition, the protease MMPs were wellknown pro-angiogenic factors in the angiogenesis pathway [43,52,53]. RERG increased the expression of TIMP-2 and inhibited the expression of MMP-2 and MMP-9, providing support for a mechanistic explanation of the inhibition of tumor microvessel growth and size by RERG. MMPs, as well as cytokines, IL8 and IL6 were the most down-regulated pro-angiogenic factors, followed reexpression of RERG in NPC. However, the effect of VEGF and TNFα were not significant, indicating that they may not play important roles in suppressing angiogenesis via RERG regulation. This study provides the first evidence that RERG expression can reduce angiogenesis. In this study, we observed that RERG reduced cellular migration and invasion in NPC in vitro and suppressed tumor growth and angiogenesis in vivo, but the potential ability and the detailed mechanisms of RERG-inducing distant metastasis in vivo requires further study.

Conclusions
We demonstrated that RERG functioned as a TSG through suppression of ERK/NF-κB signaling pathway, and was frequently silenced by promoter CpG methylation in NPC. This study provides insights into the possible mechanisms underlying the role of RERG in NPC carcinogenesis, and suggests that RERG might be employed as a target molecule in cancer therapy.

Additional files
Additional file 1: Table S1. List of primary antibodies used in the study. (DOCX 16 kb)