|Home | About | Journals | Submit | Contact Us | Français|
Promyelocytic leukemia (PML) protein plays an essential role in the induction of apoptosis; its expression is reduced in various cancers. As the functional roles of PML in glioblastoma multiforme (GBM) have not been clarified, we assessed the expression of PML protein in GBM tissues and explored the mechanisms of PML- regulated cell death in GBM cells. We examined the PML mRNA level and the expression of PML protein in surgical GBM specimens. PML-regulated apoptotic mechanisms in GBM cells transfected with plasmids expressing the PML gene were examined. The protein expression of PML was significantly lower in GBM than in non- neoplastic tissues; approximately 10% of GBM tissues were PML-null. The PML mRNA levels were similar in both tissue types. The overexpression of PML activated caspase-8 and induced apoptosis in GBM cells. In these cells, PML decreased the expression of transactivated forms of NFκ B/p65, and c-FLIP gene expression was suppressed. Therefore, PML-induced apoptosis resulted from the suppression of the transcriptional activity of NFκB/p65. PML overexpression decreased phosphorylated IκBα and nuclear NFκB/p65 and increased the expression of the suppressor of cytokine signaling (SOCS-1). A proteasome inhibitor blocked the reduction of activated p65 by PML. The reduction of PML is associated with the pathogenesis of GBM. PML induces caspase-8– dependent apoptosis via the repression of NFκB activation by which PML facilitates the proteasomal degradation of activated p65 and the sequestration of p65 with IκBα in the cytoplasm. This novel mechanism of PML-regulated apoptosis may represent a therapeutic target for GBM.
Astrocytomas, the most common human primary brain tumors, are histologically divided into four grades according to guidelines established by the World Health Organization. Grade IV astrocytoma, glioblastoma (GBM), is the most common and most malignant.1,2 Despite multimodal therapies including surgical resection, radiotherapy, and chemotherapy, the prognosis of patients with GBM remains poor; their median survival is 1 year.3,4 The presence of genetic alterations in GBM renders the tumor resistant to various apoptotic stimuli such as ionizing irradiation and chemotherapeutic agents.5,6
In patients with acute promyelocytic leukemia (APL), the promyelocytic leukemia (PML) gene is fused to the retinoic acid receptor α (RARα) gene as a result of reciprocal t(15:17) chromosomal translocation.7,8 PML encodes a founding member of a growing family of proteins that all contain a distinctive C3H4 zinc-binding domain termed “RING finger.” The PML protein resides within subdomains of the nucleus, which are known as PML nuclear bodies (PML NBs). There it co-localizes with more than 30 different proteins including p53, pRb, Daxx, Sp100, CBP, and SUMO19 and regulates several important cellular functions including transcriptional and translational regulation,9,10 alternative telomere lengthening,11 Ras-induced premature senescence,12,13 the DNA damage response, and genomic stability.14,15
The expression of PML protein is reduced in various cancers.16–21 Gurrieri et al. showed that it was completely lost in 49% of human brain tumors although none of their GBMs manifested complete PML loss.19 In human osteogenic sarcoma cells PML has been reported to augment tumor necrosis factor–induced apoptosis by interacting with nuclear factor-kappa B (NFκB/p65) and sequestering it to PML NBs.22 It represses the target A20 gene downstream of NFκB by interfering with the binding of p65 to its promoter element.23 The functional roles of PML, however, have not been clarified in malignant glioma cells.
In this study we evaluated the expression profile of PML in GBM tissues. We propose a new mechanism of PML-regulated apoptosis in GBM cell lines.
The human GBM cell line U87MG was purchased from American Type Culture Collection (Manassas, VA, USA). The primary GBM cell line, Tokushima glioblastoma (TGB) cells, was obtained from a patient who granted prior informed consent for its use in this study. Cell lines were cultured in RPMI-1640 medium (Invitrogen, Carlsbad, CA, USA) with 5% fetal bovine serum (FBS) (Gibco-BRL, Grand Island, NY, USA) at 37°C in 5% CO2 and air.
Plasmids expressing Flag-tagged PML (Flag-PML) were generated as previously reported.24 To construct plasmids expressing Flag-tagged PML lacking amino acids 1–166 (Flag-PMLΔRB1), cDNA for PMLΔRB1 was PCR amplified from plasmids encoding Flag-PML with a cloned Pfu DNA polymerase (Stratagene) using specific primers: forward (5′-AAGCTGGAATTCGCAGAGCT-GCGCAACCAGTCG-3′), containing the EcoRI site and reverse (5′-CTTTTTCTCGAGAAGCTTCTAAATTA-GAAAGGGGTGGGGGTA-3′), containing the XhoI site. The PCR fragments were digested with EcoRI and XhoI and cloned into pcDNA3-Flag.
GBM cells were plated in six-well tissue culture plates (3 × 105 cells/well). After 24 h, cultured cells were transfected with Flag-PML or Flag-PMLΔRB1 using FuGENE 6 (Roche, Basel, Switzerland) according to the manufacturer’s protocol. Control cells were transfected with an empty pcDNA3 vector. The viability of TGB cells was evaluated by counting adherent cells 24, 48, and 72 h after transfection.
To examine the morphologic changes exhibited by apoptotic cells we applied Hoechst staining. Adherent and floating cells harvested 72 h after the induction of PML expression were incubated for 15 min with 10 μM bis-benzimide (Sigma-Aldrich, St. Louis, MO, USA) solution on ice. Apoptotic cells exhibiting bright-green condensed chromatin under a fluorescence Olympus IX71 microscope (Olympus, Oosaka, Japan) were counted.
The distribution of various cell-cycle phases was examined by flow cytometry. Adherent and floating cells were harvested 72 h after the induction of PML expression, fixed for at least 3 h in 70% ethanol at 4°C, washed twice with phosphate-buffered saline (PBS), resuspended in PBS containing 50 μg/ml propidium iodide (Dojindo, Kumamoto, Japan) and 20 μg/ml RNase A (QIAGEN, Valencia, CA, USA), and analyzed in an EPICS XL-MCL FACScan (Coulter Corp., Hialeah, FL, USA) using the Coulter cytological program. All experiments were performed in triplicate.
With permission from the ethics committee, tissue samples were obtained from the Department of Neurosurgery of the University of Tokushima Graduate School. Specimens were collected between March 2002 and May 2007. All patients gave prior written consent to use their brain tissue material and additional clinical data for research purposes.
Our samples were from normal and GBM tissues of patients undergoing brain surgery. We used normal brain tissues from patients without brain tumors and from non-neoplastic regions (NNRs) containing no obvious tumor cells from patients with malignant astrocytoma. GBM samples were prepared from surgical specimens of apparent GBM tumor tissues. All tissue samples were immediately frozen and stored at −80°C. We soaked samples for RNA extraction in RNAlater (QIAGEN) and samples for immunohistochemistry in PBS; frozen sections (7 μm) were prepared within 24 h and handled identically. The patients were 15 women and 12 men, whose mean age at diagnosis was 46.7 years. All samples (normal, n = 6; NNR, n = 16; GBM, n = 17) were diagnosed by neuropathologists according to the WHO classification.
Total RNA isolated from tumor samples and GBM cell lines was purified with the RNeasy Lipid Tissue Mini Kit (QIAGEN) following the manufacturer’s protocol and reverse-transcribed with the QuantiTect Reverse Transcription kit (QIAGEN). Real-time PCR was performed under the conditions recommended by the manufacturer on a Light Cycler Rapid Thermal Cycler (Roche Diagnostics, Lewes, UK). The forward and reverse primer sequence for PML was 5′-GATG-GCTTCGACGAGTTCAA-3′ and 5′-GGGCAGGT-CAACGTCAATAG-3′, respectively; for c-FLIP it was 5′-GTATATCCCAGATTCTTGGC-3′ and 5′-GGCT-TCCCTGCTAGATAAGG-3′, respectively; and for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) it was 5′-GGGTGTGAACCATGAGAAGTATGA-3′ and 5′-TGCTAAGCAGTTGGTGGTGC-3′, respectively. The primers were optimized at Nihon Gene Research Laboratories, Inc. (Sendai, Japan). PML and c-FLIP mRNA expression was normalized with the GAPDH mRNA content.
Immunohistochemical staining for PML was performed within 3 days of sample collection. After blocking non-specific protein binding with serum-free protein block (Dako Cytomation, Glostrup, Denmark) for 60 min, the sections were incubated (4°C, overnight) with the primary antibody (PML, Santa Cruz Biotechnology, Santa Cruz, CA, USA; dilution 1:100; solution 1% bovine serum albumin [BSA]/PBS). After washing in PBS, sections were incubated (room temperature, 1 h) with the secondary antibody (A11005, Alexa Fluoro 594 goat antimouse immunoglobulin G; dilution 1:200, solution 1% BSA/PBS) and analyzed under a fluorescence microscope (Olympus). Staining with 4′,6′-diamino-2- phenylindole (DAPI; Dojindo) for 30 min was performed to identify the cell nuclei. The sections were incubated with auto fluorescence eliminator reagent (Chemicon, Temecula, CA, USA) to eliminate autofluorescence for lipofuscin. PML-positive cells in three portions of each specimen were counted, and the ratio of positive cells per 100 cells was determined. Based on the classification protocol of Gurrieri et al.,19 we divided all immunostained samples into three groups: complete loss of PML (positive cells <10% of all cells), focal positivity (positive cells >10% but <50%), and diffuse positivity (positive cells >50%).
TGB cells were immunostained 48 h after transfection. They were washed in PBS, fixed for 10 min with 4% paraformaldehyde in PBS, and permeabilized for 5 min with 0.2% Triton-X 100 in PBS. After three PBS washes the cells were incubated (room temperature, 60 min) with primary antibodies against PML (Santa Cruz Biotechnology; dilution 1:100), Flag (Cell Signaling Technology, Beverly, MA, USA), or the active form of NFκB/p65 (active-p65) (Chemicon; dilution 1:100). The samples were further incubated for 60 min with Cy3-conjugated secondary antibody.
Tissue samples and GBM cells were homogenized in lysis buffer (Cell Signaling) with protease inhibitor cocktail (Roche, Tokyo, Japan). After 5-min centrifugation at 5,000 rpm, protein levels in the supernatants were assayed with BCA reagent (Pierce, Tokyo, Japan). After reduction in 60 mM Tris-HCl buffer (pH 6.8) including 10% glycerol, 2% sodium dodecyl sulfate (SDS), 100 mM dithiothreitol (0.002%), and bromophenol blue (0.1%), each protein sample (50 μg) was separated on a 10%–20% gradient or on 7.5% polyacrylamide/ SDS gels and electroblotted onto polyvinylidene fluoride membranes (Bio-Rad, Hercules, CA, USA). Blots were immersed in blocking buffer (5% nonfat dry milk in Tris-buffered saline) for 1 h and incubated overnight with primary antibodies recognizing PML (Santa Cruz Biotechnology), Survivin (Bio Vision, Mountain View, CA, USA), cleaved caspase-3 (Cell Signaling), cleaved caspase-8 (Santa Cruz Biotechnology), cleaved caspase-9 (Cell Signaling), IκBα (BD Biosciences, Franklin Lakes, NJ, USA), phospho-IκBα (Ser32/36) (Cell Signaling), active- p65 (Chemicon), NFκB/p65 (total-p65) (BD Biosciences), phospho-NF-κB/p65 (Ser536)(phospho-p65) (Cell Signaling), β-actin (Sigma-Aldrich), c-FLIP (Santa Cruz Biotechnology), and suppressor of cytokine signaling (SOCS)-1 (Santa Cruz Biotechnology). The membranes were then incubated for 1 h with the secondary antibody (GE Healthcare, Little Chalfont, Buckinghamshire, UK). Detection was with ECL reagent (Amersham, Piscataway, NJ, USA). Western blotting for each antibody was performed more than three times. To calculate the protein levels obtained by Western blot analysis, we used Image-J software.
At 47 h after transfection with Flag-PML or Flag-mock, TGB cells were treated for 1 h with 3 μM of the proteasome inhibitor MG132 (Calbiochem, San Diego, CA, USA). We collected whole cell lysates 48 h post- transfection and performed Western blot analysis.
To compare the difference between two samples we used Student’s t-test; p values <0.05 were considered statistically significant.
To investigate the expression profile of PML in GBM tissues we analyzed PML gene expression by RT-PCR and PML protein expression by immunohistochemical and Western blot analysis. PML mRNA was expressed at similar levels in each tissue (Fig. 1A). Immunohistochemically, 11.1% of the GBM samples manifested complete loss of PML protein, and 77.7% were focally positive. In contrast, all normal and 42.9% of NNR tissues exhibited diffuse positive staining (Fig. 1B, C). PML was diffusely expressed in all examined regions of normal tissue. In Western blot analysis, the expression of PML protein was significantly lower in GBM than in normal tissues (p < 0.001, Fig. 1D). PML mRNA was not correlated with PML protein expression in individual brain tissues (data not shown). These data suggest that the degradation of PML protein is facilitated in GBM and that the reduction of PML is associated with its pathogenesis.
The reduction in the expression of PML protein suggests that, as in other tumors, PML acts as a tumor suppressor in GBM. To clarify the role of PML in GBM, we introduced plasmid encoding PML into TGB, a primary cell line extracted from a GBM operated on at our department. Endogenous PML expression was very low in TGB cells (Fig. 2Aa). PML NBs were formed in PML-transfected cells (Fig. 2Ad) but not in mock-transfected or untransfected cells (Fig. 2Aa, b). PMLΔRB1, a mutant form of PML lacking the RING finger and B1 box, was diffusely localized in nuclei without the formation of PML NBs (Fig. 2Ac). The transfection efficacy was 80% or higher in both cell types expressing PML and PMLΔRB1.
To investigate the effect of PML on cellular proliferation, we performed a cell growth assay. Starting at 48 h post-transfection, PML markedly and PMLΔRB1 scarcely influenced cell growth (Fig. 2B). To examine whether the inhibition of cell growth resulted from apoptosis or cell-cycle arrest, we analyzed the cell-cycle profiles with flow cytometry at 72 h post-transfection. Flow cytometric analysis showed that the subG1 population was significantly increased in PML-transfected cells compared to mock- and PMLΔRB1-transfected cells (p < 0.01, Fig. 2C, D). There was no difference in the G1 or G2M population. To confirm the induction of apoptosis by PML, PML was overexpressed in TGB cells, and the cells were subjected to Hoechst staining 72 h after transfection. The ratio of Hoechst-positive cells was significantly increased in PML- compared to mock-transfected cells (p < 0.05, Fig. 2E, F). These results indicate that PML inhibits cell growth in GBM cells, primarily by inducing apoptosis.
To clarify the molecular mechanisms underlying the proapoptotic effect of PML in GBM cells we assessed several apoptosis-related proteins in TGB and U87MG cells. At 48 h post-transfection of both cell lines with the PML gene, cleaved caspase-8 and -3 were increased (Fig. 3A), suggesting activation of the extrinsic apoptosis pathway mediated by caspase-8. However, the expression of cleaved caspase-9, SAPK/JNK, or the phosphor-ylated form of SAPK/JNK was not affected by PML overexpression (Fig. 3A, B). NFκB negatively regulates the extrinsic apoptosis pathway by c-FLIP gene activation.25–27 We examined whether PML could modulate NFκB transactivation in GBM cells. PML expression significantly reduced the expression of activated p65 but not total p65 (p < 0.01, Fig. 4A–C). Furthermore, p65 localized in nuclei was also decreased (Fig. 4D). The phosphorylation of p65 at Ser-536 is also required for efficient transcriptional activation of NFκB.28 To further confirm the suppression of NFκB activity by PML we examined whether PML expression affects the phosphorylation of p65. As expected, the expression of phosphorylated p65 was reduced in cells over-expressing PML (p < 0.05, Fig. 5B, D). c-FLIP mRNA and protein were significantly reduced by PML (p < 0.01, Fig. 5A–C). In addition, c-FLIP mRNA in PML- overexpressing U87MG cells was decreased (data not shown). These findings suggest that PML overexpression induces caspase-8–mediated apoptosis via the downregulation of c-FLIP due to NFκB inactivation.
Next we investigated the mechanisms underlying the PML-induced reduction of transcriptionally active p65. The expression of phosphorylated IκBα but not of IκBα was reduced (p < 0.01, Fig. 6A–C), and the expression of SOCS-1 was increased (p < 0.05, Fig. 6A, D). Because SOCS-1 targets p65 for proteasomal proteolysis,29 we examined whether PML augments the proteasomal degradation of activated p65. The proteasome inhibitor MG132 blocked the reduction of activated p65 by PML overexpression (Fig. 6E). These results suggest that PML represses the transcriptional activity of p65 by inhibiting the nuclear translocation of cytosolic p65 controlled by IκBα and by promoting the ubiquitin-proteasomal degradation of active p65 via an increase in SOCS1 expression.
In this study we found that the expression of PML protein was reduced in GBM tissues; whereas some were PML-null, there was no strong correlation between the expression of PML and PML mRNA levels (Fig. 1). The exposure of U251MG cells to MG132 increased their PML expression (data not shown). Our findings suggest that proteasomal degradation of PML is facilitated in GBM cells. Gurrieri et al.19 also showed that the primary mechanism underlying the decrease in PML protein was a proteasome-dependent degradation of the PML protein rather than a gene mutation. Thus, the reduction of PML protein appears to be associated with the pathogenesis of GBM.
PML−/− mice were resistant to γ-irradiation, and PML−/− cell lines were resistant to apoptotic stimulation.30,31 The overexpression of PML reportedly induced apoptosis in various cancer cells.32–34 We found that PML overexpression induced apoptosis in GBM cells (Fig. 2), indicating that the inhibition of growth by PML overexpression may result primarily from the induction of apoptosis. In addition, the overexpression of PMLΔRB1 failed to result in the formation of PML NBs or to induce apoptosis (Fig. 2), suggesting that the formation of PML NB is necessary for the induction of apoptosis by PML. These results suggest that the reduced PML expression in GBM tissues may contribute to the resistance of these tumors to various proapoptotic stimuli.
The regulatory mechanisms by which PML induces apoptosis are not fully understood. In TGB and U87MG cells, the ectopic expression of PML led to the activation of caspase-8 but not of caspase-9, resulting in the subsequent activation of caspase-3 (Fig. 3A). However, the expression of survivin and the phosphorylated form of SAPK/JNK, a regulator of caspase-9–mediated apoptosis, was not altered (Fig. 3B). Caspase-8 is involved in an extrinsic apoptotic pathway activated by a death receptor. Taken together, these observations indicate that PML-induced apoptosis is mediated by an extrinsic apoptotic pathway.
NFκB mainly constitutes a heterodimer of p65/RelA and p50; it is normally sequestered in the cytoplasm via its noncovalent interaction with a family of inhibitory proteins termed “IκBs.” Various extracellular stimuli activate IκB kinase (IKK) and IKK phosphorylates IκB. The phosphorylation of IκB induces degradation of IκB interacting with NFκB dimers via the ubiquitin-mediated pathway, and freed NFκB dimers are translocated to the nucleus,35 resulting in the transcriptional activation of NFκB. In response to TNF engagement, p65 can be phosphorylated at Ser-536 in its transactivation domain by IKK. The phosphorylation of p65 is also required for the efficient transcriptional activation of NFκB28; it stimulates the expression of target genes such as c-FLIP, Bcl-XL, survivin, and A20.25,26 c-FLIP contains a two-death effector domain (DED) and a catalytically inactive caspase-like domain; its ability to interact with Fas-associated death domain and procaspase-8 through homotypic DED-mediated interactions results in apoptotic inhibition via interference with procaspase-8 activation.27
We found that PML reduced the expression of active p65 translocated to nuclei and phospho-p65 (Figs. 4 and and5).5). Consequently, c-FLIP, one of the target genes of NFκB, was downregulated (Fig. 5) in PML-over-expressing GBM cells. NFκB is activated aberrantly in human malignant astrocytoma cell lines and tissues.36 p65/RelA-deficient mice were shown to die of hepatocyte apoptosis during embryonic development.37 Cur-cumin, an inhibitor of NFκB activity, induces apoptosis in human malignant astrocytoma cell lines.38 In addition, Sulfasalazine, an inhibitor of NFκB activity, inhibited the growth of experimental human GBMs in nude mice brains.39 These reports suggest that the inhibition of NFκB activity can induce apoptosis in GBM cells. Based on our previous and present findings, we posit that PML inhibits the transcriptional activity of NFκB in GBM cells, and the subsequent reduction of c-FLIP expression leads to caspase-8–dependent apoptosis.
The transcriptional activity of NFκB is also regulated by proteasomal proteolysis of p65 targeted by SOCS-1.29 The SOCS family encompasses a large superfamily of proteins, the SOCS box, which regulates the ubiquitination and turnover of SOCS and SOCS-associated proteins.40 We demonstrated that PML increased the expression of SOCS-1 and decreased the expression of phospho-IκBα (Fig. 6A). Exposure of control- and PML-expressing cells to MG132 inhibited the reduction of active p65 by PML (Fig. 6E). Furthermore, we observed remarkable apoptosis in U87MG cells treated with 3 μM MG132 for 6 h (data not shown). Taken together, our results indicate that activated p65 is targeted for proteasomal proteolysis and that PML enhances this degradation, probably through an increase of SOCS-1 expression. The inhibition of IκBα phosphorylation may also contribute to the inactivation of NFκB by PML, although the mechanisms by which PML regulates IκBα phosphorylation have not been elucidated.
Our study demonstrated that a reduction in PML protein, but not in PML mRNA, is associated with the pathogenesis of GBM. PML reduction may be attributable to proteasomal degradation. Although the mechanisms regulating PML degradation remain unclear, we present a new pathway; that is, in GBM cells, PML induces caspase-8-dependent apoptosis via the repression of NFκB activity. The deregulation of NFκB and IκBα phosphorylation is a hallmark of cancer and inflammatory diseases. New strategies targeting these constitutively activated signaling pathways may represent promising therapeutic means.
We are indebted to Yoko Sugimoto and Rika Okabe at the Department of Neurosurgery, Institute of Health Biosciences, University of Tokushima Graduate School, for technical assistance. This study was supported by a grant from the Ministry of Health and Welfare of Japan (C: 16591443).