Bortezomib Sensitizes Primary Human Astrocytoma Cells of WHO Grades I to IV for Tumor Necrosis Factor–Related Apoptosis-Inducing Ligand–Induced Apoptosis

Malignant gliomas are the most common and aggressive human brain tumors. Astrocytoma constitutes the most frequent glioma subgroup and can be subdivided into benign pilocytic (WHO grade I) and malignant diffuse astrocytoma (1). Diffuse astrocytomas are composed of low-grade diffuse astrocytoma (WHO grade II), anaplastic astrocytoma (WHO grade III), and glioblastoma multiforme (WHO grade IV), the most frequent and most aggressive intracranial tumor in adults. Astrocytomas containing a conspicuous proportion of oligodendroglial tumor cells are termed mixed oligoastrocytoma and are clinically related to pure astrocytomas. They emerge as WHO grade II (nonanaplastic oligoastrocytoma) or grade III (anaplastic oligoastrocytoma) tumors (1).

Despite highly sophisticated surgical approaches and improved chemotherapeutic and radiotherapeutic options, the survival rates of patients with diffuse astrocytoma of, e.g., ∼12 months for glioblastoma multiforme is still disappointing (2). Therefore, there is an urgent need for new therapeutic strategies.

Tumor necrosis factor–related apoptosis-inducing ligand (TRAIL) efficiently induces apoptosis in glioma cell lines (3) but not in the majority of normal cells (4, 5) thereby representing a novel and promising anticancer cytokine. However, the therapeutic potential of TRAIL for gliomas has only been thoroughly investigated in cancer cell lines. Only very little is known about TRAIL sensitivity of primary glioma cells. Furthermore, many tumor cell lines (6, 7) and especially most primary tumor cells of hematologic origin (8) are TRAIL resistant and require sensitization for TRAIL-induced apoptosis. Bortezomib, the first proteasome inhibitor which is currently used as an anticancer drug in clinical trials (reviewed in ref. 9), sensitized some but not all TRAIL-resistant tumor cell lines, including glioma cell lines, for TRAIL-induced apoptosis (10, 11). However, no data exist thus far on the sensitization of primary glioma cells for TRAIL-induced apoptosis by bortezomib.

Here we assessed the apoptosis-inducing potential of bortezomib/TRAIL cotreatment in primary human glioma cells. We found that primary tumor cells from all 13 investigated glioma patients were TRAIL resistant but could be efficiently sensitized by bortezomib cotreatment. The concomitant activation of different levels of the TRAIL signaling pathway by bortezomib/TRAIL cotreatment seems to be responsible for its highly synergistic antitumor effect. Here, we provide the first native TRAIL death-inducing signaling complex (DISC) analysis of freshly isolated primary cells from solid tumors.

The human glioma cell line LN215 was maintained in DMEM containing 10% FCS (Invitrogen).

Surgical glioma specimens were obtained under the guidance of a neuronavigation system and intraoperative magnetic resonance imaging, which allowed for a selection of vital tumor tissue for consecutive analyses. For tumor cell isolation, fresh nonnecrotic surgical specimens were washed in PBS and mechanically disaggregated into small pieces, which were evenly distributed in a cell culture flask (Sarstedt) coated with AmnioMax medium (Invitrogen) and incubated at 37°C with 5% CO₂. After tumor cell outgrowth, tumor pieces were removed and cells were covered with supplemented AmnioMax medium. The study was approved by the Ethics Committee, Medical Faculty, University of Leipzig. Detailed information on glioma patients is given in Table 1.

For Western blot, the following primary antibodies were used: anti-polyubiquitin antibody (Biomol, FK1), β-actin (Sigma, clone AC15), Bak (PharMingen), Bax (Santa Cruz Biotechnology), Bcl-2 (Merck Biosciences, clone AB-1), Bid (full-length, Biosource), tBid (Biosource), caspase-3 (CPP32, R&D), caspase-8 (C15, Axxora), caspase-9 (MBL, clone 5B4), caspase-10 (MBL International, clone 4C1), Fas-associated death domain (BD Biosciences), cellular FLICE inhibitory protein (cFLIP; Axxora, NF6), poly(ADP-ribose) polymerase (Biomol, clone C-2-10), TRAIL receptor 1 (TRAIL-R1; ProSci), and TRAIL-R2 (ProSci). Horseradish peroxidase–conjugated secondary antibodies were obtained from Southern Biotechnology Associates. For fluorescence-activated cell sorting analysis, we used anti–APO-1 (immunoglobulin G1), HS101 (TRAIL-R1), HS201 (TRAIL-R2), HS301 (TRAIL-R3), and HS401 (TRAIL-R4; Axxora). Biotinylated secondary goat Fab anti-mouse antibodies (Southern Biotechnology Associates) and streptavidin-phycoerythrin (PharMingen) were used. The pan-caspase inhibitor z-VAD(oMe)-fmk was obtained from Axxora. Bortezomib (Velcade) was purchased from Millenium Pharmaceuticals. All other chemicals were of analytic grade and purchased from Sigma or Merck. An isoleucine zipper–tagged form of human TRAIL (iz-TRAIL) was used (5).

Preparation of cell lysates, immunoblotting, and fluorescence-activated cell sorting analysis of tumor cell lines and primary tumor cells were done as described (5, 12).

Cell viability was quantified by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay. Late apoptotic cell death was flow cytometrically quantified according to propidium iodide uptake in combination with characterization of cell shrinkage (decrease in forward scatter) and cellular fragmentation (increase in side scatter). Each measurement was done in triplicate as described before (5, 6). To verify apoptotic cell death, we checked for characteristic cell morphology (light microscopy), nuclear condensation/fragmentation (Hoechst 33342 staining), and caspase inhibition by z-VAD-fmk as described (5).

Clonogenicity assay. Primary tumor cells were cotreated with bortezomib and TRAIL. Dead cells were washed off with PBS after 16 h. Surviving cells were cultured for additional 14 days with medium being replaced once per week without any further death stimulus. At the indicated time points, cell viability was quantified by 3-[4,5-dimethylthiazol-2-yl-2,5-diphenyltetrazolium] bromide assay.

Primary tumor cell lines were seeded and pretreated at 37°C with medium or 1 μg/mL TRAIL. Unbound TRAIL was removed after 90 min by five washing steps with PBS. Cells were subsequently sensitized overnight with 25 nmol/L bortezomib either alone or in combination with additional TRAIL (1 μg/mL). Control cells were left untreated, pretreated with TRAIL, or incubated with bortezomib overnight as indicated. Cell death was quantified by propidium iodide exclusion assay.

Biotinylated iz-TRAIL (Bio-iz-TRAIL) was generated for ligand affinity purification experiments essentially as described before for a leucine zipper–tagged form of TRAIL (6).

The Bcl-2 open reading frame was excised from pMIG-Bcl-2 (ref. 13; obtained from Addgene, Inc.) and inserted into pIRES2 (Clontech) to yield pIRES2-Bcl-2. Subsequently, a CMV-Bcl-2-IRES-GFP–containing cassette was excised from pIRES2-Bcl-2 and inserted into pWPTS-eGFP (kindly provided by D. Trono, University of Geneva, Geneva, Switzerland) to replace the EF1-α-IRES-GFP cassette, yielding pLVIE-Bcl-2.

Lentiviral particles were prepared according to published protocols (14, 15) using pLVIE-Bcl-2, the packaging vector psPAX2, and the envelope plasmid pMD2G. Target cells were transduced by diluting the virus-containing supernatant 1:2 onto the target cells in the presence of 2 μg/mL polybrene (Sigma).

For generation of the retroviral cFLIP small interfering RNA (siRNA) expression vector pSRNG-FLIP911, sense and antisense oligomers containing the cFLIP targeting sequence 5′-gatccccGGAGCAGGGACAAGTTACAttcaagagaTGTAACTTGTCCCTGCTCCtttttggaaa-3′ (sense strand) starting at position +911 were annealed and cloned into the pSuper.retro.neo.GFP retroviral vector using HindIII and BglII restriction sites. Viral particles were produced by transient transfection of the producer cell line Phoenix-Ampho together with the vesicular stomatitis virus G protein–expressing plasmid pMD2G. Virus-containing supernatant was harvested after 48 h and used to infect target cells in the presence of 2 μg/mL polybrene.

Transduced cells were sorted by green fluorescent protein (GFP) expression using a fluorescence-activated cell sorter. Bcl-2 overexpression and cFLIP down-regulation were verified by Western blot analysis.

Primary human glioma cells are TRAIL resistant but can be sensitized for TRAIL-induced apoptosis by bortezomib. Primary tumor cells were isolated from surgical specimens of 13 astrocytoma and oligoastrocytoma patients (Table 1) comprising all four WHO grades of malignancy. We established cell culture conditions for primary tumor cells, which provided constant proliferation rates and preserved the heterogeneous tumor cell phenotype for up to eight passages of in vitro culture. In contrast to most glioma cell lines, primary glioma cells from all investigated astrocytoma and oligoastrocytoma patients were highly TRAIL resistant but could be very efficiently sensitized for TRAIL-induced apoptosis by cotreatment with low concentrations of bortezomib (Fig. 1A). Apoptotic cell death was confirmed by Hoechst 33342 staining and inhibition of cell death by the pan-caspase inhibitor z-VAD-fmk (data not shown). Longer exposure with bortezomib or TRAIL alone or higher concentrations of bortezomib (50 nmol/L) neither induced significantly more cell death in primary glioma samples nor increased the sensitization to TRAIL (see below and data not shown).

To avoid artifacts of long-term cell culture of primary tumor cells, all experiments were done between passages 1 and 8 (maximum of 8 weeks in vitro culture time). During this period, primary human glioma cells showed a constant behavior toward treatment with TRAIL and bortezomib (Fig. 1B), showing the stability of the primary tumor cell culture model.

To investigate the occurrence of biochemical apoptotic events induced by TRAIL treatment, bortezomib-sensitized versus nonsensitized primary glioma cells were treated with TRAIL for different periods and cell lysates were analyzed for the cleavage of caspases and caspase substrates (Fig. 1C). Only in bortezomib-treated and not in nonsensitized primary glioma cells did TRAIL induce the rapid cleavage of caspase-8 and, consequently, the processing of Bid, caspase-9, caspase-7, and caspase-3 and the effector caspase substrate poly(ADP-ribose) polymerase. Accordingly, active caspase-3 was immunohistochemically detected only in sensitized but not in nonsensitized primary glioma cells on TRAIL treatment (data not shown).

Bortezomib enhances TRAIL DISC formation upon TRAIL triggering in primary glioma cells. Bortezomib treatment enhanced the surface expression of TRAIL-R1 (patients #14 and #17) or TRAIL-R2 (most prominently in patients #16, #17, and #20) in some primary glioma cells (Fig. 2A). TRAIL-R3 and TRAIL-R4 could not be detected on the surface of nonsensitized or sensitized glioma cells (data not shown).

To investigate the contribution of TRAIL death receptor up-regulation to TRAIL sensitization in tumor samples with enhanced TRAIL receptor expression, we did a “wash kill” experiment in these cells as described before (6). Primary glioma cells were pretreated with TRAIL to occupy the TRAIL receptors already present on the cell surface. Unbound TRAIL was thoroughly washed off after 90 min and cells were treated overnight with bortezomib either alone or with additional TRAIL. On treatment with bortezomib alone, TRAIL receptors which appear on the cell surface after removal of unbound TRAIL cannot be activated. In this case, only the preexisting but not the up-regulated TRAIL receptors are occupied by the ligand. Pretreatment with TRAIL followed by bortezomib without re-adding TRAIL still resulted in the death of ∼40% of the cells (patient #13; Fig. 2B) indicating that bortezomib-induced TRAIL sensitization is partly independent of TRAIL-R1/TRAIL-R2 up-regulation even in tumor samples which show TRAIL death receptor up-regulation on bortezomib treatment. However, because the remaining cells only died when additional TRAIL was administered, TRAIL death receptor up-regulation also contributes to bortezomib-induced TRAIL sensitization in these cells. Similar results were obtained with another sample (patient #16; data not shown).

To investigate the influence of bortezomib on TRAIL DISC formation, the native TRAIL DISC was immunoprecipitated from sensitized versus nonsensitized primary glioma cells of WHO grades I, II, and IV (Fig. 2C, lanes “+”). As a control, TRAIL receptors were precipitated from total cell lysates (when no TRAIL DISC can be formed; Fig. 2C, lanes “−”). Although TRAIL treatment induced TRAIL DISC formation in all nonsensitized primary gliomas, only little procaspase-8 (p55/p53) and cleaved caspase-8 (p43/41) were detectable in the TRAIL DISC. On sensitization with bortezomib, significantly more TRAIL-R1, TRAIL-R2, FADD, procaspase-8, caspase-8 p43/p41, procaspase-10 (only in patient #18), and caspase-10 p47/43 (only in patient #18) could be precipitated from primary tumor cells.

In contrast, full-length cFLIP long form (cFLIP_L) was hardly detectable, whereas the cleaved form of cFLIP_L (cFLIP_L p43) was apparent in all three investigated TRAIL DISCs. A background-normalized quantification of caspase-8 and cFLIP_L/cFLIP_L-cleaved in the TRAIL DISC of glioblastoma multiforme patient #20 was done. It revealed a 10-fold stronger caspase-8 recruitment to the sensitized versus the nonsensitized TRAIL DISC. In contrast, recruitment of cFLIP_L and the cleaved p43 fragment to the TRAIL DISC was only 1.4-fold enhanced by bortezomib. This results in an increase of the caspase-8/cFLIP ratio in the TRAIL DISC of sensitized cells.

Bortezomib influences the expression of intracellular regulators of TRAIL sensitivity. Total cell lysates of bortezomib-treated primary glioma cells of all four WHO grades of malignancy were analyzed by Western blot for the expression of regulators of TRAIL sensitivity (Fig. 3). In some tumor samples and to a different extent, bortezomib treatment resulted in accumulation of Bid (patients #27 and #3), Bak (patients #27 and #18), or Bax (patient #27; Fig. 3). Quantification of the cFLIP_L/β-actin ratio shows a cFLIP_L down-regulation in all four glioma samples, most prominently in patient #3 (cFLIP_L/β-actin ratio on 0, 12.5, and 25 nmol/L bortezomib, respectively, for patient #27: 1.1, 1.1, and 0.8; patient #17: 0.9, 0.5, and 0.5; patient #18: 0.9, 0.8, and 0.6; and patient #3: 1.1, 0.8, and 0.4). FADD, caspase-8, and caspase-10 were abundantly expressed in all primary glioma cells and were not significantly influenced by bortezomib (data not shown).

Bcl-2 and cFLIP expressions influence apoptosis induction by bortezomib/TRAIL cotreatment. Compared with primary glioma cells, human glioma cell lines displayed higher sensitivity to TRAIL (60% cell death with 1,000 ng/mL in two of three investigated cell lines) when administered alone but comparable sensitivity to bortezomib (0-40% cell death with 50 nmol/L; data not shown). We found that LN215 cells, like primary glioma cells, were TRAIL resistant but could be sensitized by bortezomib for TRAIL-induced apoptosis (Fig. 4A). LN215 cells showed enhanced TRAIL-R2 expression on the cell surface, down-regulation of cFLIP_L, and slightly enhanced Bak expression on bortezomib treatment (Fig. 4A). LN215 cells stably expressing cFLIP siRNA exhibited enhanced TRAIL sensitivity (Fig. 4B) whereas stable Bcl-2 overexpression partially protected LN215 cells from bortezomib-induced toxicity. Interestingly, Bcl-2–overexpressing LN215 cells could still be sensitized by bortezomib for TRAIL-induced apoptosis albeit to a lesser extent than wild-type cells (Fig. 4C). Similar results were obtained with these cells in viability assays (data not shown).

Bortezomib/TRAIL cotreatment massively reduces the clonogenic survival of primary glioma cells. To investigate the clonogenic capacity after cotreatment with bortezomib and TRAIL, primary glioma cells were treated with bortezomib and TRAIL either alone or in combination. After 24 h, dead cells were removed and cell viability of regrown tumor cell clones was quantified at days 1, 2, 7, and 14 after treatment by 3-[(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium] bromide assay and visualized by crystal violet stains at day 14 (patient #18, Fig. 5). Neither incubation with bortezomib nor TRAIL alone resulted in a significant reduction of viable cells. In contrast, combinatorial treatment with bortezomib and 100 ng/mL TRAIL reduced tumor cell viability to <5% at day 14 after treatment (Fig. 5). This massive reduction in the clonogenic potential of primary glioma cells by bortezomib/TRAIL cotreatment was also observed in primary glioblastoma multiforme cells (patient #20; data not shown).

Many tumor cell lines including glioma cell lines have been reported to be sensitive to TRAIL. In contrast, investigating freshly isolated tumor cells from 13 astrocytoma and oligoastrocytoma patients of all four WHO grades of malignancy, we found that all primary glioma cells (Fig. 1) as well as primary medulloblastoma, meningioma, and esthesioneuroblastoma cells (data not shown) were TRAIL resistant. This may dampen the expectations of clinical trials using TRAIL as a monotherapeutic agent (16, 17). Even more worrying is the recent finding that TRAIL treatment of TRAIL-resistant pancreatic cancer and cholangiocarcinoma cells even increased tumor cell migration and metastatic spread in vitro and in vivo (18). Therefore, data derived from glioma cell lines are a deceptive predictor for the clinical benefit of TRAIL-based therapies, and a careful characterization of primary glioma cells on their TRAIL sensitivity is urgently needed.

Many TRAIL-resistant cancer cell lines could be sensitized for TRAIL-induced apoptosis by chemotherapeutic drugs including proteasome inhibitors (19). Here, we show that primary glioma cells from all investigated patients could be very efficiently sensitized for TRAIL-induced apoptosis by 16-h cotreatment with as little as 12.5 to 25 nmol/L bortezomib (Fig. 1). Furthermore, in combination with 25 nmol/L bortezomib, 100 ng/mL TRAIL massively reduced the clonogenic survival of primary glioma cells in long-term experiments (Fig. 5) but was not toxic to normal cells like primary human hepatocytes (20). Both TRAIL and bortezomib monotherapy are already in clinical trials, which will accelerate future clinical studies on their combination. Our study shows that concentrations of bortezomib far below the measured peak plasma level of bortezomib-treated patients (21) are sufficient to sensitize primary glioma cells for TRAIL. In contrast, applying bortezomib alone for 48 h, 100 nmol/L to 1 μmol/L bortezomib were needed to reduce the viability of ex vivo glioblastoma multiforme cells from two patients by only 60% (11). In contrast to bortezomib, other chemotherapeutic drugs were far less efficient (doxorubicin) or failed (etoposide) to sensitize primary glioma cells for TRAIL-induced apoptosis (data not shown) although they have been described to efficiently sensitize glioma cell lines for TRAIL-induced apoptosis (3, 22).

In accordance to other tumor cells (6, 23), some of the investigated primary glioma cells showed an enhanced surface expression of TRAIL-R1 and/or TRAIL-R2 on bortezomib treatment, which contributed to the sensitization for TRAIL-induced apoptosis (Fig. 2). Performing the first native TRAIL DISC analysis of primary glioma cells of WHO grades II, III, and IV, we found that bortezomib treatment resulted in enhanced TRAIL DISC formation on TRAIL triggering (Fig. 2C). Due to a reduction of the cFLIP_L protein level by bortezomib (Fig. 3), less cFLIP_L was recruited to the sensitized TRAIL DISC in comparison with FADD or caspase-8 (Fig. 2C), resulting in a change of the ratio between antiapoptotic and proapoptotic molecules in the TRAIL DISC of sensitized cells. This is in accordance with reports on cancer cell lines in which bortezomib sensitized for TRAIL-induced apoptosis by cFLIP_L down-regulation (24). Accordingly, as reported for other cell lines (6, 25), siRNA-mediated cFLIP down-regulation sensitized LN215 cells for TRAIL-induced apoptosis (Fig. 4B). In a recent report, three primary glioblastoma multiforme samples were TRAIL sensitive (26). TRAIL resistance in another three primary samples correlated with high expression of cFLIP_S (26). In our study, however, all primary tumor cells were TRAIL resistant regardless of their cFLIP_S expression level. PED/PEA-15 plays an important role in the regulation of TRAIL sensitivity in glioma cell lines (27). Representing an important regulatory difference to tumor cell lines, PED/PEA-15 was not recruited to the TRAIL DISC of primary glioma cells (data not shown) and does not seem be involved in the regulation of the TRAIL pathway in primary cells. Bortezomib-induced growth inhibition in glioma cell lines correlated with reduced transcriptional activity of nuclear factor κB (11). However, recent data show that TRAIL sensitization of glioma cells by proteasome inhibition is nuclear factor κB independent (10).

Bortezomib treatment has been shown to sensitize tumor cells for TRAIL-induced apoptosis by involvement of the mitochondrial pathway (28). Interestingly, some primary gliomas, although to a different extent, showed enhanced Bid, Bax, and/or Bak expression on bortezomib treatment (Fig. 3). The basal levels of these proteins differed considerably in the tumor samples (Fig. 3), underlining the heterogeneity of the investigated gliomas. Investigating the involvement of the intrinsic apoptosis pathway by bortezomib/TRAIL cotreatment, we found that Bcl-2 overexpression in LN215 glioma cells inhibited bortezomib-mediated toxicity but only marginally the sensitization of bortezomib for TRAIL-induced apoptosis (Fig. 4C).

Because all investigated tumor cells were efficiently sensitized for TRAIL-induced apoptosis, bortezomib cotreatment represents a powerful sensitizing tool for TRAIL-based therapies even in a heterogeneous population of cancer patients. We recently showed that bortezomib/TRAIL cotreatment at concentrations similar to those used in the present study are not toxic to primary human hepatocytes (6). It will be interesting to test potential adverse effects of bortezomib/TRAIL cotreatment on nonneoplastic normal brain tissue.

Taken together, having established a stable and reproducible in vitro model for the analysis of TRAIL-induced apoptosis in primary glioma cells, we showed that (a) primary glioma cells, in contrast to most glioma cell lines, are highly TRAIL resistant; (b) cotreatment with low concentrations of bortezomib and TRAIL efficiently sensitized all investigated primary human astrocytoma and oligoastrocytoma cells of all four WHO grades of malignancy for TRAIL-induced apoptosis; (c) this synergy involves regulatory mechanisms at the TRAIL DISC and presumably also at the intrinsic apoptosis pathway in a very heterogeneous cellular background; and, of clinical importance, (d) bortezomib/TRAIL cotreatment was superior to any single treatment in the reduction of clonogenic survival of primary tumor cells in vitro.

Thus, bortezomib/TRAIL cotreatment could represent a novel and effective therapeutic approach for glioma patients. In vivo experiments have to confirm the therapeutic potential of bortezomib/TRAIL cotreatment in primary glioma cells and are subject of our current studies.

We thank Bärbel Moos, Jutta Mohr, and Rainer Baran-Schmidt for excellent technical assistance, Till Wenger (DKFZ Heidelberg) for help with lentiviral transduction, and Klaus Hexel for cell sorting.