Role of protein arginine methyltransferase 5 in group 3 (MYC-driven) Medulloblastoma

The human medulloblastoma cell lines Daoy (HTB-186), D-283 (HTB-185) and D-341 (HTB-187) were purchased from American Type Culture Collection (ATCC, Manassas, VA, USA). HD-MB03 (ACC-740) human medulloblastoma cell line was purchased from Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ, Braunschweig, Germany). ONS-76 (IFO50355) human medulloblastoma cell line was obtained from Sekisui-XenoTech (Kansas, USA). These Cell lines were authenticated by their respective companies using short tandem repeat profiling. All cell lines were tested for mycoplasma contamination using MycoSensor PCR Assay Kit (Santa Clara, CA, USA). In this study, Daoy and ONS-76 were used as SHH medulloblastoma subgroup cell lines without MYC-amplification, whereas, D-341, HD-MB03 and D-283 were used as Group 3 medulloblastoma cell lines with MYC-amplified status. All these cell lines were cultured and maintained using Eagle’s minimal essential medium (EMEM) or RPMI-1640 media supplemented with 10% heat-inactivated FBS and 1% penicillin/streptomycin (Invitrogen, Carlsbad, CA, USA) in a humidified incubator at 5% CO2 and 95% air atmosphere at 37 °C. The experiments were performed using no more than 10 passages for each cell line. The cell lysate of human normal brain cerebellum was purchased from BioChain Institute Inc. (Newark, CA, USA).

The R2 Genomics Analysis and Visualization Platform (www.​r2.​amc.​nl) was used to investigate PRMT5 mRNA expression and its correlation with patient survival across medulloblastoma subgroups using publicly available datasets. The expression of PRMT5 mRNA in medulloblastoma was analyzed using a total 491 medulloblastoma tumors (5 independent cohorts) and 9 normal cerebellum samples. The survival analyses with respect to PRMT5 expression in medulloblastoma patients were performed using a separate cohort of 612 medulloblastoma samples from Cavalli (763 samples) dataset.

Both control (Scrambled, sc-37,007) and PRMT5 siRNAs ((sc-41,073) were purchased from Santacruz Biotechnology (Dallas, TX, USA). Each siRNA was dissolved in RNase-free water at 10 μM stock concentration and stored at -20 °C. The PRMT5 inhibitor EPZ015666 was purchased from Selleckchem Company (Houston, TX, USA). This inhibitor was dissolved in DMSO at 10 mM stock concentration and stored at -20 °C.

Control (scrambled) and PRMT5 siRNA (a pool of 3 target-specific 19–25 nt siRNAs with 50 nM) were transiently transfected into medulloblastoma cells using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. Following 72 h of transfections, cells were subjected to downstream analyses using western blotting and MTT assay.

To examine the effects of PRMT5 inhibition on medulloblastoma cell growth, twenty thousand cells of each medulloblastoma cell line were plated in 96-well plates 24 h before the experiment. Then, these cells were transfected with PRMT5 siRNAs or treated with PRMT5 inhibitor for 72 h according to the experimental plan and the growth of these cells was determined using an MTT assay as described previously [24].

The effect of PRMT5 inhibitor to induce apoptosis in medulloblastoma cells at 72 h, was determined using an Annexin-V:FITC flow cytometry assay kit (BD Biosciences, San Jose, CA, USA) following the manufacturer’s instructions. For cell cycle analysis, the control and PRMT5 inhibitor-treated medulloblastoma cells for 24 and 48 h, were fixed with 75% ethanol and stained with propidium iodide using a propidium iodide flow cytometry kit (Abcam, Cambridge, UK).

To determine protein stability, medulloblastoma cells were treated with 50 μg/ml cycloheximide (Sigma Aldrich, St. Louis, MO, USA) following siRNA transfection for 72 h. Following transfection, cell lysates from the indicated time points of cycloheximide treatments were subjected to western blotting.

For co-immunoprecipitation, 500 μg protein lysate was precleared with 50 μl of protein A-Sepharose beads (Cell Signaling Technology, Danvers, MA, USA) for 1 h at 4 °C. Immunoprecipitation was performed in the presence of 8 μg of the indicated primary antibodies at 4 °C overnight. Immune complexes were captured by adding 50 μl of protein A-Sepharose beads and rotated at 4 °C for 2 h. After the supernatant was discarded, protein A-Sepharose beads were washed with PBS and lysed in 1x Laemmli buffer and then subjected to western blotting.

The expression levels of indicated proteins in medulloblastoma cells were determined using western blot analyses as described previously [24]. The primary human antibodies for cMYC (sc-40), PRMT5 (sc-376,937), histone H3 (sc-8654) and β-Actin (sc-130,301) were purchased from Santacruz Biotechnology (Dallas, TX, USA). H4R3me2s (61188) and H3R8me2s (ab130740) antibodies were from Active Motif (Carlsbad, CA, USA) and Abcam (Cambridge, UK), respectively. Immunoreactivity was detected using appropriate peroxidase-conjugated secondary antibodies (Jackson Lab, ME) and visualized using an ECL detection system (Pierce, IL).

Methanol-fixed HD-MB03 cells on glass cover slips, and an antigen-retrieved medulloblastoma tumor section were washed with PBS and blocked in 1% BSA in PBS for 30 min. The tumor cells were then co-incubated with PRMT5 (rabbit, 1:100) and MYC (mouse, 1:100) antibodies overnight at 4 °C. Following three washes with PBS, the cells were further co-incubated with fluorochrome-conjugated anti-rabbit (Alexa-488) and anti-mouse (Alexa-647) secondary antibodies (Invitrogen, Carlsbad, CA) for 1 h at room temperature. The cells were then washed three times with PBS and the cover slips were mounted on glass slides and visualized under confocal microscope. DAPI was co-incubated with the secondary antibodies to facilitate the visualization of the nuclei. Confocal images were taken using a Zeiss LSM 5 Pascal confocal microscope (Carl Zeiss, Oberkochen, Germany) using a 40x objective in the UNMC Confocal Microscopy facility.

Frozen samples of normal cerebella and medulloblastoma tumor specimens were collected from the Children’s Hospital and Medical Center, Omaha and the University of Nebraska Medical Center after Institutional Review Board (IRB) approval. Normal cerebellum specimens were obtained from patients at autopsy. All normal and tumor samples were from the pediatric age group.

Normal cerebellum and medulloblastoma tumor sections were deparaffinized with xylene and rehydrated with water. Antigen retrieval was performed using citrate buffer at 95 °C for 20 min. Sections were treated with 3% hydrogen-peroxide for 30 min to block peroxidase activity. Sections were blocked using 5% goat serum with 0.3% Triton-X-100 in PBS and incubated with PRMT5 (1:100) and MYC (1:100) rabbit-antibodies (Abcam, Cambridge, UK) overnight at 4 °C. Next day, primary antibodies were washed with PBS three times and incubated with appropriate HRP-conjugated secondary antibodies for 1 h at room temperature. Following three washes with PBS, detection was performed using a DAB Peroxidase Substrate Kit (Vector Labs, Burlingame, CA, USA) followed by counterstaining with hematoxylin. Sections were mounted in Paramount solution and visualized under an EVOS FL Auto Imaging System (Life Technologies, Carlsbad, CA, USA). Staining intensity was scored from 0 to 3, where signal detected at 10X was 3+, at 20X was 2+, at 40X was 1+, and no detection was 0. The percentage positive cells was scored from 1 to 4 scale, where < 25% scored 1, 25–50% scored 2, 50–75% scored 3; and > 75 scored 4. Composite score (0–12) was derived from the staining intensity and % positive cells.

All experiments were repeated at least two times and the mean and standard error values calculated. Differences (p-value) were calculated using independent Student t-tests or analysis of variance (ANOVA) and p-values < 0.05 were considered significant. The IC₅₀ values of inhibitor EPZ015666 for each medulloblastoma cell line were determined using GraphPad Prism V6 software (provided in Table 1).

The aberrant expression of PRMT5 has been associated with a variety of cancers including glioblastoma and neuroblastoma. In addition, PRMT5 expression has correlated with MYC or MYCN protein in these cancers [20‐23]. However, its expression and function in medulloblastoma have not been reported. These studies prompted us to examine the correlation between MYC levels and PRMT5 expression in medulloblastoma. We first examined the clinical relevancy of PRMT5 in medulloblastoma by analyzing its mRNA expression in 491 medulloblastoma (from independent 5 cohorts) and 9 normal cerebellum samples using the R2 platform (www.​r2.​amc.​nl). Our analyses using these data showed a significant overexpression of PRMT5 in medulloblastoma compared to normal cerebellum tissues (Fig. 1a). We further analyzed the PRMT5 expression across medulloblastoma subgroups using a cohort that has maximum number (223) of samples with all 4 molecular subgroups. We observed significantly higher expression of PRMT5 in Group 3 (MYC-driven) medulloblastoma compared to other 3 subgroups (Fig. 1b). We next compared PRMT5 expression against patient survival. To this end, we performed survival analyses with respect to PRMT5 expression, using 612 medulloblastoma samples from the Cavalli (763 samples) dataset. Our survival analyses showed that high levels of PRMT5 expression correlated with poor survival of medulloblastoma patients, a pattern recapitulated in Group 3 medulloblastoma patients (Fig. 1c and d). These data suggest that PRMT5 expression is not simply deregulated in medulloblastoma, but is also a poor prognostic marker, particularly in Group 3 (MYC-driven) tumors.

We next analyzed the correlation between PRMT5 and MYC mRNA expression across medulloblastoma subgroups using Pfister (n = 223) cohort at the R2 genomic analysis platform. Results from this analysis showed that high expression of PRMT5 was strongly correlated (R-value = 0.531; p-value = 2.51e-05) with high MYC in Group 3 medulloblastoma. Although results showed some degree of correlation (R-value = 0.069–0.284; p-value = 0.111–0.606) of these genes in the other three medulloblastoma subgroups, none of these medulloblastoma subgroups showed a significant correlation between PRMT5 and MYC expression (Additional file 1: Figure S1). We further examined the correlation between PRMT5 and MYC expression at the protein levels by western blotting in non-MYC (Daoy, ONS-76) and three MYC-driven (D-283, D-341, HD-MB-03) medulloblastoma cell lines compared to normal cerebellum. We observed that PRMT5 protein levels were significantly (p < 0.01) higher in MYC-driven medulloblastoma cell lines compared to non–MYC medulloblastoma cell lines and normal human cerebellum cells (Fig. 2a and b). Stronger PRMT5 band intensity seemed associated with higher MYC expression in medulloblastoma cell lines. To further authenticate this correlation at the protein level, we examined the expression of PRMT5 and MYC by immunohistochemistry in Group 3 medulloblastoma tumor samples (n = 6) compared to normal pediatric cerebellum (n = 4). We found that the protein levels of PRMT5 and MYC were significantly (p = 0.004) higher with more than 75% staining in Group 3 medulloblastoma samples than normal pediatric cerebellum tissues (Fig. 2c). We observed very poor immunostaining of these proteins in normal cerebellum with less than ‘1’ intensity score. We not only observed intensely high expression of PRMT5 and MYC protein in Group 3 medulloblastoma but also, a positive correlation with their predominantly nuclear co-expression. These results consistently suggest a positive correlation and co-operative role of the PRMT5-MYC oncogenic axis in poor prognosis medulloblastoma.

The strong correlation between PRMT5 and MYC expression prompted us to explore possible influence of PRMT5 on MYC function. To investigate the role of PRMT5 on MYC function, we determined the effect of short-interfering RNA (siRNA)-mediated knockdown of PRMT5 on MYC expression and cell survival in MYC-amplified medulloblastoma cell lines. We used two cell lines D-341 and HD-MB03 in this study as both lines have previously been reported as well-established Group 3 medulloblastoma cell lines with high MYC expression. Consistently, our western blot results further confirmed strongly higher expression of MYC protein in these two cell lines compared to other medulloblastoma lines (Fig. 2a). Using these two lines, we first verified the knockdown of PRMT5 protein expression after siRNA transfections and then assessed the consequences of knockdown. We found that knockdown of PRMT5 efficiently suppressed the expression of MYC protein in both cell lines by approximately 35–45%, compared to control scrambled siRNA (Fig. 3), suggesting an on-target effect of PRMT5. Concurrently, PRMT5 knockdown significantly reduced cell growth (Fig. 3) in both cell lines by approximately 40–45%, compared to control scrambled siRNA. We also observed that D-341 cell line showed relatively lesser knockdown of PRMT5 and MYC protein by 5 and 13%, respectively, compared to HD-MB03 cell line. However, the impact of knockdown on inhibition of D-341 cell growth was relatively (~ 5%) higher compared to HD-MB03 cell growth, indicating differential knockdown activity between two MYC-amplified cell lines. One possible explanation for this differential response in cell lines could be that they have different growth pattern in culture in vitro, as we observed that D-341 cells grow more slower with mixed spheroids and monolayer cells compared to mostly monolayer HD-MB03 cells.

Previous studies [20, 23] in neuroblastoma and glioblastoma have demonstrated that PRMT5 can physically interact with MYC and regulate its stability at the post-translational level. To confirm whether endogenous PRMT5 physically interacts with MYC protein in medulloblastoma, we performed a co-immunoprecipitation experiment in MYC-driven HD-MB-03 medulloblastoma cells using MYC and PRMT5 antibodies. Our results presented in Fig. 4a, showed the presence of PRMT5 in MYC-immunoprecipitated complexes from cell extracts. We further confirmed this interaction by detecting MYC in the reverse PRMT5-immunoprecipitated complexes (Fig. 4b). To further support of association between PRMT5 and MYC, we examined co-localization of these two proteins in HD-MB03 cells and a tumor specimen of a Group 3 medulloblastoma patient, using immunofluorescence-coupled with confocal microscopy. As shown in Fig. 4c, the merged immunofluorescent-staining of MYC and PRMT5 demonstrated a significant co-localization pattern in both HD-MB03 and primary tumor cells. The results further showed that both MYC and PRMT5 were predominantly co-localized in the nucleus, and this localization pattern was consistent with their immunohistochemical co-expression in Group 3 medulloblastoma primary tumors shown in Fig. 2c. Together, the results of co-immunoprecipitation and co-localization of MYC and PRMT5 suggest that endogenous PRMT5 forms a complex with MYC in medulloblastoma cells harboring MYC amplification. The observation of physical interaction between PRMT5 and MYC indicates a potential functional role of this novel protein complex in medulloblastoma.

To determine the consequence of this physical interaction on post-translational MYC stability and influence on MYC expression, we performed cycloheximide (CHX) chase experiments in PRMT5 knocked-down HD-MB03 cells and measured the half-life of MYC protein. As shown in Fig. 4d and e, MYC has a half-life of approximately 60 min after CHX treatment in cells transfected with control scrambled siRNA, whereas PRMT5 knockdown dramatically decreased its half-life to approximately 35 min. There was approximately 25 min earlier degradation of MYC protein in PRMT5 knocked-down cells compared to control siRNA treated cells. Together, these results suggest that PRMT5 physically interacts with and stabilizes MYC in medulloblastoma cells at the post-translation level.

Given the potential for anti-neoplastic effects with PRMT5 knockdown, via reduction in cell viability and MYC expression, we investigated the therapeutic potential of PRMT5 inhibition using a recently developed potent PRMT5 inhibitor (EPZ015666) [25]. We first determined the growth inhibitory efficacy of EPZ015666 against three MYC-amplified (D-283, D-341, HD-MB03) and two non-MYC amplified (Daoy, ONS-76) medulloblastoma cell lines. Cells were treated with inhibitor (0.1–10 μM) in a dose-dependent manner for 72 h and growth of cells was assessed using an MTT assay. Our MTT results clearly demonstrated that EPZ015666 significantly induced the dose-dependent growth inhibition of all MYC-driven medulloblastoma cell lines at low micromolar potency with IC50 of ~ 1.5–2.5 μM (Fig. 5a, Table 1). However, there was minimal effect of EPZ015666 on growth inhibition of non-MYC amplified medulloblastoma cells even at higher doses, suggesting anti-neoplastic specificity of EPZ015666 to MYC-dependent tumors.

We next determined the ability of EPZ015666 to induce apoptosis in a representative MYC-amplified medulloblastoma cell line HD-MB03. The results of the apoptosis analyses (Fig. 5b) using Annexin-V assay demonstrated a dose-dependent induction of apoptosis by EPZ015666 and showed consistency with MTT growth results. Since PRMT5 is known to act during the G1 cell cycle phase, we sought to investigate whether inhibiting PRMT5 by EPZ015666 reduced medulloblastoma cell growth by disrupting the cell cycle. The cell cycle results (Fig. 5c) using PI staining of DNA, demonstrated that treatment of EPZ015666 at 24 h and 48 h arrested medulloblastoma cells in the G1 cell cycle phase in a dose-dependent manner. Since we observed a doubling time of the HD-MB03 cell line of between 28 to 36 h, we examined the impact of EPZ015666 on HD-MB03 cell cycle at 24 and 48 h, close to the doubling time points. Lastly, using western blot analyses, we confirmed that treatment of HD-MB03 cells with optimum doses of EPZ015666 efficiently downregulated the expression levels of PRMT5 and its key target symmetric dimethylated histone H3 (H3R8me2s) and H4 (H4R3me2s), including suppressed MYC expression (Fig. 5d). Taken together, these results suggest that inhibiting the specific interaction of PRMT5 with MYC arrests medulloblastoma cell growth and favors apoptosis in MYC-dependent tumors.