RNF135, RING finger protein, promotes the proliferation of human glioblastoma cells in vivo and in vitro via the ERK pathway

Ring finger protein 135 (RNF135), located on chromosome 17q11.2, is a RING finger domain-containing E3 ubiquitin ligase that was identified as a bio-marker and therapy target of glioblastoma. In our study, we confirmed that RNF135 was up-regulated in glioblastoma tissues compared with normal brain (NB) tissues, and that RNF135 knockdown inhibited proliferation and migration and led to cell cycle arrest in the G0/G1 phase in vivo. By lowering RNF135 expression, phosphorylated Erk and cell cycle protein CDK4 were down-regulated, while p27Kip1 and p21Waf1/Cip1 were up-regulated in U87 and U251 cells in vitro. In addition, using the immunofluorescence double labelling method, we found that RNF135 and P-Erk were co-localized in the cytoplasm and were highly expressed in glioblastoma samples compared with NB tissues. Moreover, the growth of U87 cell-transplanted tumours in nude mice was inhibited while transduced with Lv-shRNF135. Taken together, our findings demonstrate the biological effects of RNF135 in glioblastoma cell proliferation, migration and cell cycle, and its role in the progression of glioblastoma may be associated with the ERK signal transduction pathway.

malignant peripheral nerve sheath tumours, and it was also suggested that RNF135 was involved in the increased malignant risk in NF1 micro deletion patients 19 . However, the function and expression levels of the RNF135 gene in human glioblastoma remain unknown.
In the present study, we confirmed that RNF135 expression was increased in human patient samples. A lentivirus-mediated RNA interference (RNAi) system was used to explore the effect of the RNF135 gene on the migration, proliferation and cell cycles of U87 and U251 cells in vitro. Furthermore, the possible mechanisms involved in this process were also investigated. Our findings provide new insights into the role of RNF135 in the development of glioblastoma, and implicate the potential application of RNF135 in glioblastoma therapy.

Results
RNF135 is highly expressed in glioblastoma tissues compared with NB tissues. To evaluate the role of RNF135 in glioblastoma, quantitative real-time PCR and Western blotting were employed to detect the expression of RNF135 in 28 fresh glioblastoma samples and 12 NB tissues. The data of quantitative real-time PCR showed that glioblastoma tissues presented much higher transcription levels of RNF135 than NB tissues (P < 0.05) (Fig. 1A). Similarly, the RNF135 protein levels in glioblastoma tissues were also higher than those in NB tissues (Fig. 1B) (P < 0.05). Furthermore, to measure the expression levels of RNF135 in different pathological grade gliomas and normal brain tissues, we examined the RNF135 expression levels in 154 archived paraffin-embedded samples, including 14 astrocytic tumours, 17 oligodendroglioma, 66 anaplastic astrocytoma and 45 glioblastoma and 12 NB tissues, using the immunohistochemical (IHC) staining assay (Fig. 1C). The IHC assay demonstrated that the RNF135 protein was highly expressed in approximately 66.9% of glioma tissues but only highly expressed in 25% of NB tissues (P = 0.009) ( Table 1).

Figure 1. RNF135 is highly expressed in human malignant glioma tissues (tumour) compared with normal brain tissues (NB). (A)
The mRNA expression of RNF135 is increased in tumour tissues (28 clinical cases) compared with NB tissues (12 clinical cases) by Q-RT-PCR assay. GAPDH was used as a loading control. (B) Western blot assay was used to test the RNF135 protein expression levels between NB tissues (12 clinical cases) and glioblastoma tissues (28 clinical cases) with GAPDH as the loading control. Data are presented as the mean ± SD for three independent experiments. (C) Representative IHC images showed RNF135 expression was decreased in 15 normal brain tissues (a,b) compared with 142 primary glioma samples (c,d). (Original magnification: 400×,100×).
Scientific RepoRts | 6:20642 | DOI: 10.1038/srep20642 Relationship between RNF135 expression and clinicopathologic characteristics in glioma patients. The clinicopathologic characteristics of the patients are shown in Table 2. There was no remarkable association between RNF135 expression and the patients' age and gender in the randomly selected glioma cases. Nevertheless, the levels of RNF135 expression were correlated with the grade of pathology classification (WHO I-II vs WHO III-IV) and histologic type in glioma patients (P = 0.000). In addition, to confirm whether the RNF135 expression levels and some other clinical factors were the independent factors of prognostic prediction in glioma patients, we performed univariate and multivariate analyses using the Cox proportional hazard regression model. The results showed that the RNF135 expression level (P = 0.000; HR: 2.958; multivariate Cox regression analysis) and the radiochemotherapy (p = 0.02; HR: 0.573; multivariate Cox regression analysis) both were indicative of significant prognostic factors of glioma in patient survival time and the RNF135 expression levels was a more important factor (Table 3).
High expression of RNF135 predicts a negative survival in glioma patients. To explore the prognostic value of RNF135 expression in glioma, the relationship between RNF135 expression levels and 142 glioma  patients' survival was assessed by Kaplan-Meier analysis and log-rank test. Low levels of RNF135 expression in glioma tissues correlated significantly with a better survival than those with high-expression (P = 0.000, log-rank test) (Fig. 2).
Successful establishment of steadily down-regulated RNF135 expression in U87 and U251 cell lines. To determine the effect of RNF135, the lentiviral shRNA vector was used to stably and specifically lower RNF135 expression in U87 and U251 cells established from high-grade tumours. The RNF135 mRNA and protein expression levels were measured by quantitative real-time PCR and Western blotting. Respectively, the RNF135 mRNA ( Fig. 3a,b) and protein expression levels (Fig. 3c,d) were significantly attenuated compared with the negative vector control [pLVTHMGFP-Control (PLV-Ctr)] (P < 0.05).  Stably down-regulated RNF135 expression suppresses cell proliferation and migration in vitro. To investigate the biological implication of RNF135 cells in glioblastoma, we conducted MTT, wound healing and Boyden chamber analysis experiments to determine the effect of decreased RNF135 expression on glioblastoma cell growth and migration in vitro. We performed the MTT assay to measure the effect of RNF135 on U87 and U251 cell line growth and proliferation. The results from this assay showed that cells transfected with shRNF135 grew significantly slower than shPLV-Ctrl cells (P < 0.05) (Fig. 4a,b). Next, to examine the effect of RNF135 on glioblastoma cell migration, we performed wound healing and Boyden chamber assays. The results of the wound healing assay showed that shRNF135 U87 and U251 cells presented a lower speed into the gap than PLV-Ctrl cells (P < 0.05) (Fig. 3c-e).The Boyden chamber assay revealed fewer U87 and U251 cells on the lower surface than PLV-Ctrl cells. The strained cells were counted using Image J software (P < 0.05) ( Fig. 4f-h). The results inferred that knockdown of RNF135 expression could reduce the migration effect on U87 and U251 cells.
Knockdown of RNF135 could arrest the cell cycle in the G0/G1 phase in U87 and U251 cells. We explored the effect of low expression of RNF135 on cell cycle distribution in U87 and U251 cells by flow cytometry plus propidium iodide (PI) staining. The results revealed 71.62% and 64.74% of G0/G1 phase cells in the shRNF135 U87 and U251 cells, respectively, and 62.45% and 50.71% of G0/G1 phase cells in the PLV-Ctr cells, respectively (P < 0.05) (Fig. 5A). These data indicate that the cell cycle was arrested in the G0/G1 phase after knockdown of RNF135 expression in U87 and U251 cells. Moreover, the expression levels of cell cycle regulation proteins p27 Kip1 , p21 Waf1/Cip1 and CDK4 were determined using Western blot analysis (Fig. 5B). Our results suggested that low RNF135 expression could reduce CDK4 expression, while enhancing the expression of p27 Kip1 and p21 Waf1/Cip1 . To measure the expression of p27 Kip1 , p21 Waf1/Cip1 and CDK4 in glioma tissues, we performed the IHC staining assay (Fig. 6). In this assay, we found that CDK4 was highly expressed in high-grade glioma tissues compared with low-grade glioma tissues ( Fig. 6a,b); however, p27 Kip1 and p21 Waf1/Cip1 showed opposite results ( Fig. 6c-f).

Knockdown of RNF135 could attenuate the activation of p-Erk.
According to previous studies that the ERK and p38 pathways play an important role in modulating the cell cycle, we further investigated the effect of RNF135 on the ERK and p38 pathways. The results revealed that decreased RNF135 significantly inhibited the activation of P-Erk. However, RNF135 had no effects on the expression levels of P-p38 or the total protein levels of Erk and p38 (Fig. 7A). This work indicated that RNF135 could modulate the Erk pathway in glioblastoma.
We examined the degree of co-expression of RNF135 and P-Erk in glioblastoma samples and NB tissues with fluorescent double-labelling ( Fig. 7B). We observed that the red fluorescence-labelled P-Erk was localized in the cytoplasm; the green fluorescence-labelled RNF135 also appeared in the cytoplasm; the blue fluorescence counterstained by DAPI was in the nucleus. Next, we used the picture analysis software to construct the 3 fluorescence patterns; yellow fluorescence appeared in the position at which the red and green fluorescence occurred simultaneously. This finding illustrated that RNF135 and P-Erk were co-localized in the cytoplasm. Furthermore, this assay confirmed that RNF135 and P-Erk were up-expressed compared with NB tissues.

Knockdown of RNF135 suppresses cell tumourigenicity in vivo.
To confirm the growth-increasing effects of RNF135, a nude mouse xenograft model was established by implanting shRNF135-U87 cells. The average weights of the tumours removed from mice injected with shRNF135 U87 cells and shPLV-Ctr U87 cells were 1.564 g and 2.970 g, respectively (P < 0.05) (Fig. 8A,B). Moreover, the results of immunohistochemistry staining implied that the expression of RNF135 in the shRNF135-xenografted tumours was reduced compared with the normal expression of RNF135 in shPLV-Ctr-xenografted tumours (Fig. 8C).

Discussion
GBM is the most malignant primary brain tumour in adults. Classified as grade IV by the World Health Organization (WHO), GBM is known as the most frequent, devastating and malignant astrocytic glioma 20,21 .
Recently, the main treatment of GBM has included surgical resection, chemotherapy and radiotherapy. Despite these very aggressive methods, the prognosis is still unfavourable. The median survival is less than 15 months, and the 5-year overall survival is below 2% for inevitable tumour reappearance 22 . Recently, researchers have started their investigation of molecular biological technologies to develop new approaches. E3 ubiquitin-protein isopeptide ligases have been identified as bio-markers and therapy targets of glioblastoma. RNF135 is a RING domain-containing E3 ubiquitin ligase. The RNF135 protein comprises an N-terminal RING finger domain, and C-terminal SPRY and PRY motifs. The gene RNF135, which has been identified by the genome project previously, was found to be a cause of human genetic diseases such as neurofibromatosis 15 . Researchers have hypothesized that a change in the dosage of RNF135 may contribute to RNF135-involved human disease 23 .
It was reported that RNF135 haploinsufficiency leads to an overgrowth syndrome as well as learning disabilities in human beings 17,24 . In the present study, we first investigated the RNF135 effect in glioblastoma. First, we confirmed that the RNF135 mRNA was highly expressed in 28 glioblastoma samples compared with 12 normal brain tissues using quantitative real-time PCR. The same applied to the protein expression level. Additionally, the IHC staining assay revealed that RNF135 was highly expressed in 66.9% of glioma patients and up-expressed in 25% of normal tissues. Furthermore, the RNF135 expression levels were correlated with the grade of glioma (Table 2), and the multivariate analysis showed that the RNF135 expression level was the most important factor predicting the overall survival in patients with glioma (Table 3), with lower RNF135 expression correlated with a better survival than higher RNF135 expression (Fig. 2).
Because it contains a ring finger and PY motif, RNF135 can bind numerous proteins, indicating its breadth of functions, such as involvement with the EGFR and TGF-β 25 . Many ring finger proteins belong to the family of ubiquitin ligases, which promote the ubiquitination of proteins. This process plays a very important role in a wide spectrum of biological processes involved in development and disease pathogenesis 26 . Previous research has shown that RNF149 attenuates the increase in cell growth induced by wild-type BRAF 14 . It has been confirmed that Ring finger protein 43 (RNF43) is up-expressed in colorectal cancer and mediates cancer cell proliferation 27 , and RNF43, as a novel tumour-associated antigen, is a new target for cancer immunotherapy 28 . It was also demonstrated that RING finger proteins may play a significant role in the cancer biological process. To explore the biological function of RNF135, we knocked down RNF135 expression using shRNA in U87 and U251 cell lines. Next, we conducted a series of experiments and found that knocking down RNF135 with shRNA could significantly inhibit U87 and U251 glioma cell line proliferation and migration in vitro compared with that in the PLV-Ctrl groups.
Previous studies have shown that RING finger proteins play important roles in the cell cycle 29,30 . CDKs, p27 Kip1, p18 INK4C and p21 Waf1/Cip1 are cell cycle regulatory proteins 31 . When the cell cycle is arrested at the G0/G1 phase, the cell cycle proteins CDK2 and CDK4 are decreased, while the p21Waf1/Cip1 is highly expressed 32 . In the FACS flow cytometry results, the shRNF135 U87 and U251 cells showed 71.62% and 64.74% of G0/G1 phase cells, whereas only 62.45% and 50.71% of G0/G1 phase cells were in the PLV-Ctrl cells. The data confirmed that knockdown of RNF135 could arrest the cell cycle at the G0/G1 phase in U87 and U251 cells. A previous study showed that the RNF135 deficient mice displayed no apparent abnormalities at seven month 33 .However, in our study, knockdown RNF135 with siRNA arrested cell cycle at G0/G1 stage in U87 and U251 glioma cell lines. The discrepancy may be resulted from several reasons. First, it's well known that when a gene was knockout, a related gene maybe upregulated to compensate for the lost function. Thus, it's possible that the function of RNF135 in proliferation was compensated by another gene in the deficient mice. Second, the glioma cell and embryonic stem cells (ESCs) grew in different microenvironment and were regulated by different signal pathways. So, RNF135 may play different roles at cell proliferation in different cells. Actually, the discrepancy between cell and animal models was not uncommon. For example, some studies showed that the TRPC6 deficient mice is normal 34 , while knockdown TRPC6 arrested cell cycle at G2/M stage in U87 and U251 glioma cell lines 35 .Moreover, the Western blot assay showed that knockdown of RNF135 could attenuate the expression of the cell cycle protein CDK4 but enhance the expression of p27 Kip1 and p21 Waf1/Cip1. Previously, researchers have reported that the Erk and p38 pathways are involved in the regulation of cell cycle progression 36,37 , cell growth, proliferation and migration [38][39][40][41] . Moreover, it was inferred that Cbl-c, which belongs to the family of RING finger ubiquitin ligases (E3s), decreased downstream ERK activation by RETMEN2A 42 . Furthermore, MEK kinase 1 (MEKK1) demonstrated two functions, as an upstream activator of JNK and ERK with its kinase domain and as an E3 ligase with the plant homeodomain (PHD) domain resembling the RING finger domain, providing a negative effect to inhibit ERK1/2 activity 43 . If the concentration was high enough, MEKK1 could increase ERK2 and p38 activities 44 . In this study, we observed that decreased RNF135 expression significantly attenuated the activation of P-Erk, whereas P-p38 and p38 were not affected. Furthermore, we found that RNF135 and P-Erk were highly co-expressed in glioblastoma tissues compared with normal brain tissues. Thus, RNF135's effect on tumour progression and malignancy may be via the Erk pathway.

Conclusion
In summary, RNF135 may have significant value as a progression indicator for patients who have glioblastoma. Evidence has confirmed that attenuated RNF135 expression could lead to suppressed cell growth and migration via inactivation of the Erk pathway in U87 and U251 glioma cells.

Materials and Methods
Cell culture and reagents. The human glioma U87 and U251 cell lines were supplied by the Cell Bank    (Fig. 8C a) and shRNF135 (Fig. 8C b). Cell proliferation assay. The MTT assay was performed to assess the proliferation of cells. U87 and U251 cells were seed at a certain density (1 × 104 cells per well) in 96-well culture plates, and cultured in media for 1-6 days. Next, 10 μ l of MTT reagent (5 mg/ml) was added to each well, and the plates were then incubated at 7 °C for 4 h. The formative formazan crystals were dissolved by adding 100 μ l of SDS-isobutanol-HCl solution (10% SDS, 5% isobutanol, and 12 μ M HCl) overnight at 37 °C. The absorbance was measured at 490 nm with a microplate reader (Bio-Rad, San Diego, CA, USA).
Flow cytometry analysis of the cell cycle. Cells were harvested and seeded in 6-well plates. After 24 h of incubation, cells were digested and collected, following by centrifugation at 1000 rpm for 5 min at 4 °C. The deposited cells were then suspended in ice-cold PBS and washed, and then re-suspended with 70% ethanol for 30 min at 4 °C. Next, the cells were washed and re-suspended in 100 mL of PBS containing the final concentration of 50 mg/mL RNase A (Sigma, US) and 0.25% Triton X-100 for 30 min at 4 °C. Finally, the cells were stained with 10 mg/mL propidium iodide (PI, lot: 1685935, life technologies, USA) for 30 min. Next, the cells were immediately analysed in a FACScan flow cytometer (Becton Dickinson, CA, USA). All of the works were assayed three times, and the fraction of each cell cycle phase was measured.
Immunohistochemistry. Glioma and NB tissues were cut into parts, removed into tubes and fixed in 10% formaldehyde. Tissue sections were embedded in paraffin. Pure xylene was used to dewax paraffin sections by soaking sections into pure xylene. Tissue sections were then re-hydrated in a series of water-diluted ethanol solutions. The re-hydrated tissues were blocked in serum and incubated with rabbit anti-human RNF135 antibody at 4 °C overnight. Tissues were washed twice and then incubated in biotin-labelled goat anti-rabbit antibody for 10 min at room temperature. After incubation with horseradish peroxidase (HRP), 3,3-diaminobenzidine (DAB, Cat: 15111A03, Zhong Shan Jin Qiao, China) was used to perform the peroxidase reaction. Haematoxylin was used to stain the samples, which were then counted under a microscope (Original magnification: 400×, 100×).

Evaluation of staining.
Tissue sections stained by immunohistochemistry were analysed by two pathologists blinded to the clinical parameters. Observation of RNF135 expression in the cytoplasm was performed separately. In the cytoplasm, the staining score was determined based on the cytoplasm staining intensity together with the number of positive cytoplasm-staining cells. The obtained percentage of positive staining in the cytoplasm was defined as follows: 0: <10%, 1: 10-25%, 2: 26-75%, and 3: >76%. A final staining score of 0-1 in the cytoplasm was considered low expression of RNF135. By contrast, a final staining score of 2-3 in the cytoplasm was considered high expression of RNF135.
Immunofluorescence double labelling method. To investigate the glioma samples and NB tissues expressing the phosphorylated ERK (Thr202/Tyr204) and RNF135, a double-labelling procedure using a rabbit polyclonal antibody was carried out as described previously 45,46 . The antibodies included anti-RNF135, anti-P-Erk, Goat anti rabbi lgG antibody and goat anti mouse lgG antibody (Original magnification: 400×, 100×).
Establishment of U87 and U251 cell lines with stable low expression of RNF135 short hairpin RNA. The pLVTHM-GFP lentiviral RNAi expression system was used to perform the lentiviruses expressing human RNF135 short hairpin RNA (TOP: G AT CC GC AG GC CC TG TC TT CT GG AA AG CA TT TT CA AG AG AA-AT GC TT TC CAGAAGACAGGGCCTGTTTTTTC; BOTTOM: A AT TGAAAAAACAGGCCCTGTCTTCTGG  AAAGCATTTCTCTTGAAAATGCTTTCCAGAAGACAGGGCCTGCG). Cells were infected with lentiviral particles containing specific or negative vectors. The selected polyclonal cells with GFP signals were analysed using FACS flow cytometry. The transfection efficiency of RNF135 was confirmed by quantitative real-time PCR analysis and Western blot assay.
Wound Healing Assay. Cell migration and proliferation were analysed by wound-healing activity assays in vitro. The stably shRNA-transfected U87 and U251 cells were harvested and cultured in 6-well plates. When the cells became confluent, a cell-free scratched gap approximately 500 μ m in width was created with a pipette tip in the wells of the plates, and then the cells were incubated for another 6 h and 12 h. We compared the length of the healed wound with the length of the initial wound. The lengths from two different wounded regions were measured with Image J (public domain software, http://rsb.info.nih.gov/ij/) (Original magnification: 100×).
Boyden chamber migration assay. To perform the transwell migration assay, 1 × 10 5 cells/ml of U87 and U251 cells were collected and diluted in FBS-free DMEM medium. Next, 100 μ l of cells were seeded into the Boyden chamber (0.8 μ m; Corning, USA) in 24-well plates, and 600 μ l of DMEM containing 10% FBS medium was added to the wells outside the chamber. After 6 h of incubation at 37 °C, cells were washed twice with PBS. The top surface cells of the transwell were removed, and the lower surface cells were fixed with methanol. The fixed cells were stained with Giemsa solution and counted stained cells under a microscope using Image J software (public domain software, http://rsb.info.nih.gov/ij/) (Original magnification: 100×).
Tumourigenesis in nude mice in vivo. U87 cells were harvested and diluted to the concentration of 1 × 10 8 cells/ml. Next, 200 μ l of cells at the density of 1 × 10 8 cells/ml were implanted s.c. under the right shoulder of 4-to 6-week-old female BALB/c nu/nu mice. The mice were randomly divided into two groups (ten mice per group) and were administered an intratumoural injection of sh-RPS15A-U87 cells (1 × 10 7 pfu) or sh-PLV-Ctl-U87 cells (1 × 10 7 pfu). After 25 days of breeding, all of the mice were sacrificed, respectively, and the tumour tissues were excised and weighed. All of the animal experiments were conducted according to the principles and procedures of the National Institutes of Health Guide. Quantitative Real-Time PCR. According to the protocol reported previously 10 , RNA was extracted from glioma and NB tissues, as well as U87 cells, using the RNA isoPlus ® Reagent Kit (Cat: RR820A, Takara Biotechnology) according to the manufacturers' protocols. The PrimeScript ® RT Reagent Kit (Cat: RR037A, Takara, Shiga, Japan) was used to reverse transcribe RNA into cDNA. The cDNA was amplified using the SYBR ® Premix Ex Taq ™ Kit under 7500 Real-Time PCR System (Applied Biosystems, 7500 Real Time PCR System, Thermo, US). The cycling conditions were as follows: 40 cycles of 95 °C for 30 s, 95 °C for 5 s, 60 °C for 34 s. The comparative Δ Δ Ct method was used to analyse the data, and GAPDH was used as the loading control for the target genes. The primers were as follows: RNF135 (Forward: 5′-TACTGGGAAGTGGACACTAGGAATT-3′, reverse: 5′-CTTGACCATGTGCCATGCA-3′); GAPDH (Forward: 5′-GCACCGTCAAGGCTGAGAAC-3′, reverse: 5′-TGGTGAAGACGCCAGTGGA-3′).
Western blot analysis. Tissues and cell total proteins were collected using RIPA lysis buffer containing phosphatase and protease inhibitors (Roche, Swiss) in accordance with the manufacturer's instructions. The preparation of tissues for Western blotting was reported as described previously 11 . Glioma and NB tissues were homogenized first and then lysed in lysis buffer with shaking at 4 °C for 30 min. Proteins were quantified by the BCA protein assay agent (Cat: PC0020, Solarbio, China). Proteins were separated by SDS-PAGE, and then transferred to PVDF membranes (Cat: IPVH00010, EMD Millipore Corporation Billerica Ma01821, USA), blocked with 5% fat-free dry milk or 5% BSA in TBST and immunoblotted with primary antibodies at 4 °C overnight. The following day, the membranes were incubated with second antibodies at room temperature for 2 h. The detection of proteins on the membranes was determined using the enhanced ECL reagent by the Bio-Spectrum Gel Imaging System with image lab software (UVP, USA). Protein expression was quantified by Gel-Pro Analyzer 3.0 software. Statistical analysis. The qualifying data obtained were represented as the mean ± SD of at least three independent experiments. SPSS 17.0 software (SPSS, Chicago, USA) and Graph Pad Prism 5.0 (www.graphpad.com, USA) were used for statistical analysis. One-way ANOVA test or two-tailed Student's t-test were used for comparisons between groups. Chi-squared test or Fisher's exact test was used to analyse the difference between classified variables. Kaplan-Meier analysis and the log-rank test were used for Survival analysis. The relationship between variables and patient survival was performed using univariate and multivariate Cox proportional hazards models. P values less than 0.05 were considered to be statistically significant.