Overexpression of isocitrate dehydrogenase mutant proteins renders glioma cells more sensitive to radiation

Abstract

Mutations in isocitrate dehydrogenase 1 (IDH1) or 2 (IDH2) are found in a subset of gliomas. Among the many phenotypic differences between mutant and wild-type IDH1/2 gliomas, the most salient is that IDH1/2 mutant glioma patients demonstrate markedly improved survival compared with IDH1/2 wild-type glioma patients. To address the mechanism underlying the superior clinical outcome of IDH1/2 mutant glioma patients, we investigated whether overexpression of the IDH1^R132H protein could affect response to therapy in the context of an isogenic glioma cell background. Stable clonal U87MG and U373MG cell lines overexpressing IDH1^WT and IDH1^R132H were generated, as well as U87MG cell lines overexpressing IDH2^WT and IDH2^R172K. In vitro experiments were conducted to characterize baseline growth and migration and response to radiation and temozolomide. In addition, reactive oxygen species (ROS) levels were measured under various conditions. U87MG-IDH1^R132H cells, U373MG-IDH1^R132H cells, and U87MG-IDH2^R172K cells demonstrated increased sensitivity to radiation but not to temozolomide. Radiosensitization of U87MG-IDH1^R132H cells was accompanied by increased apoptosis and accentuated ROS generation, and this effect was abrogated by the presence of the ROS scavenger N-acetyl-cysteine. Interestingly, U87MG-IDH1^R132H cells also displayed decreased growth at higher cell density and in soft agar, as well as decreased migration. Overexpression of IDH1^R132H and IDH2^R172K mutant protein in glioblastoma cells resulted in increased radiation sensitivity and altered ROS metabolism and suppression of growth and migration in vitro. These findings provide insight into possible mechanisms contributing to the improved outcomes observed in patients with IDH1/2 mutant gliomas.

Diffuse gliomas are the most common type of primary adult brain tumor—with an annual incidence of ∼20 000 cases in the United States—and pose a significant health problem, as the vast majority remain incurable.^1,2 Of these, glioblastoma multiforme (GBM; World Health Organization grade IV) is the most common and represents one of the most devastating human malignancies. Currently, the primary treatments for gliomas include surgical resection followed by radiotherapy (RT) and/or temozolomide (TMZ).

In 2008, a genome-wide somatic mutational analysis of glioblastomas revealed a subset with an isocitrate dehydrogenase 1 (IDH1) mutation at codon R132 (IDH1^R132MUT).³ Subsequently, the IDH1 mutation was found in the majority of lower-grade and anaplastic diffuse gliomas.^4,5 The IDH1 gene is located on chromosome 2 and encodes IDH1. IDH1 catalyzes the conversion of isocitrate to alpha-ketoglutarate (α-KG) and generates nicotine adenine disphosphonucleotide (NADPH) in the cytoplasm and peroxisomes.⁶ IDH1 is involved in lipid metabolism, glucose sensing, and defense against oxidative stress.^7,8 At a much lower frequency, mutations in the functionally similar IDH2 gene have also been found in glioma patients. In addition to gliomas, the IDH1/2 mutation is found in a fraction of adult acute myelogenous leukemia (AML) patients.^9,10 The most common IDH1 mutation is R132H (∼90%) with R132C and R132S found less commonly; the most common IDH2 mutation is R172K.⁸ Monoclonal antibodies have been derived for R132H^11,12 and R132S.¹³

Since the identification of these mutations, there has been an explosion of interest in understanding how IDH1/2 mutation and the resulting alterations in cellular metabolism may contribute to tumor formation and behavior. Zhao and coworkers¹⁴ showed that the heterozygous IDH1 mutation dominantly inhibited the enzyme's activity, which led to the decreased formation of α-KG. The authors proposed that this contributed to gliomagenesis via increased expression of hypoxia-inducible factor 1α. Alternatively, Dang et al¹⁵ demonstrated that IDH1^R132MUT protein possessed a novel enzyme activity, leading to the production of 2-hydroxyglutarate (HG). Elevated 2-HG has also been found in AML cells carrying IDH1 or IDH2 mutations.^16,17 Recently, Koivunen et al¹⁸ demonstrated that the accumulation of 2-HG promoted transformation of human normal astrocytes, and multiple studies have shown that IDH1 mutant gliomas are associated with cytosine–phosphate–guanine (CpG) island hypermethylation.¹⁹

Patients with IDH1^R132MUT gliomas have also been shown to have improved survival.^5,20–22 The mechanism responsible for this improved survival is unclear, but it is possible that IDH1 mutant gliomas have a less aggressive phenotype and/or may respond better to standard therapies. One recent study showed that IDH1 or IDH2 mutations predicted response to TMZ and longer survival in low-grade gliomas.²¹ Another study showed that the mutation conferred improved prognosis independent of whether PCV (procarbazine, CCNU, vincristine) was given in addition to radiation.²³ In that study the impact of IDH mutations on untreated tumors was limited. In AML, in contrast, the presence of the IDH1 mutation is an unfavorable prognostic factor.^10,24

To evaluate the pathological consequence of the IDH1^R132MUT mutation and how it might explain the improved survival seen in patients with IDH1^R132MUT gliomas, we generated clonal U87MG glioma and U373MG glioma cell lines overexpressing the R132H mutant protein (IDH1^R132H). Compared with control cells and cells overexpressing IDH1^WT, IDH1^R132H cells showed increased sensitivity to radiation in both U87MG cells and U373MG cells. This effect was also seen in U87MG cells overexpressing IDH2^R172K, the most common glioma-associated IDH2 mutation. Examining the mechanism of radiosensitization, we found that U87MG-IDH1^R132H cells also showed increased apoptosis and a heightened oxidative response to radiation. Interestingly, we also detected decreased proliferation at high cell density and decreased migration of U87MG-IDH1^R132H cells. These findings suggest that the improved survival seen in patients with IDH1^R132MUT tumors may result, in part, from direct effects of the IDH1^R132MUT protein on proliferation, migration, and sensitivity to radiation.

Materials and Methods

Cell Culture

U87MG and U373MG glioma cells (obtained from Dr. Paul Mischel at UCLA, and originally from American Type Culture Collection) were maintained in Dulbecco's modified Eagle's medium (DMEM)/F12 medium supplemented with 10% fetal bovine serum (FBS; Invitrogen), 2 mm glutamine, and 100 units/mL of penicillin/streptomycin. Cells were dissociated with enzyme-free cell dissociation solution (Millipore) and cultured at 37°C and 5% CO₂ in a 90% humidified tissue incubator. TMZ was obtained from the National Cancer Institute/National Institutes of Health (NCI/NIH) Developmental Therapeutics Program and dissolved in dimethyl sulfoxide (DMSO). N-acetyl cysteine (NAC; Sigma) was dissolved in phosphate buffered saline (PBS) (pH 7.4).

Generation of Constructs

Full-length human wild-type IDH1 and IDH2 coding sequences were amplified from U87MG cells. The cDNA was fused in-frame with a FLAG tag at the C-terminus by the adapter primers (for IDH1: forward primer with the XhoI site: 5′-TACTCGAGATGTCCAAAAAAATCAGTGG-3′, reverse primer with the EcoRI site and FLAG tag: 5′-ACGAATTCTTACTTGTCATCGTCATCCTTGTAATCCATAAGTTTGGCCTG-3′; for IDH2: 5′-TACTCGAGATGGCCGGCTACCTGCG-3′, reverse primer with the EcoRI site and FLAG tag: 5′-ACGAATTCTTACTTGTCATCGTCATCCTTGTAATCCATCTGCCTGCCCAG-3′) and inserted into the XhoI- and EcoRI-linearized pLPCX vector. The IDH1^R132H and IDH2^R172K alterations were generated in pLPCX-IDH1-WT-FLAG and pLPCX-IDH2-WT-FLAG using the QuikChange method (Stratagene) with the following primer sequence: 5′R132H (5′-ACCTATCATCATAGGTCATCATGCTTATGGG-3′) and 3′R132H (5′-TGACCTATGATGATAGGTTTTACCCATCCAC-3′); 5′R172K (CCCATCACCATTGGCAAGCACGCCCATGG-3′) and 3′R172K (5′-TTGCCAATGGTGATGGGCTTGGTCCAGC-3′).

Stable Overexpression of IDH1^WT, IDH1^R132H, IDH2^WT, and IDH2^R172K Constructs in Glioma Cells

Using the FuGene HD transfection system (Roche), Plat-A cells were transfected with pLPCX (vector only), pLPCX-IDH1-WT-FLAG, pLPCX-IDH1-ΔR132H-FLAG, pLPCX-IDH2-WT-FLAG, and pLPCX-IDH2-ΔR172K-FLAG constructs. U87MG or U373MG cells were infected with retrovirus produced by Plat-A package cells in medium containing 8 μg/mL polybrene. Cells were then selected with puromycin (0.7 μg/mL for U87MG and 0.5 μg/mL for U373MG) after 48 h. For each construct, >10 single clonal cell lines were derived from the selected populations by limited dilution, and 2–4 representative cell lines per construct were selected for further experiments. All clones were maintained in culture media with puromycin. Protein expression was confirmed by western blot and immunofluorescence. For U87MG-control cells, U87MG-IDH1^WT cells, and U87MG-IDH1^R132H cells, 4 clones of each construct were selected based on expression level and were designated clones 1, 2, 3, and 4. Clones 1–4 were used in an assay by MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) (Sigma), and clones 1 and 2 were used in all other experiments. For U373MG-control cells, U373MG-IDH1^WT cells, U373MG-IDH1^R132H cells, U87MG-IDH2^WT cells, and U87MG-IDH2^R172K cells, 2 clones of each construct were selected based on expression level and were designated clones 1 and 2, which were used in all other experiments. All results represented the mean ± SEM for 4 or 2 clones. All clones for each construct had similar behavior in all experiments.

Immunofluorescence

Cells were grown on glass slides, fixed in 100% methanol at −20°C for 15 min, and stained with mouse monoclonal anti-FLAG M2 antibody (Sigma), goat anti-human IDH1 (Santa Cruz Biotech), mouse anti-human R132H mutant IDH1 (IDH1^R132H) antibody (Dianova),¹¹ fluorescein isothiocynate–conjugated goat anti-mouse immunoglobulin (Ig) G (Invitrogen), and Dy649-conjugated donkey anti-goat IgG (Jackson ImmunoResearch). Nuclei were counterstained with propidium iodide or 4′,6-diamidino-2-phenylindole (DAPI). Slides were mounted and imaged on an Olympus IX71 confocal microscope.

Western Blot

Total protein was isolated using radioimmunoprecipitation assay lysis buffer containing protease inhibitor, separated on 12% sodium dodecyl sulfate–polyacrylamide gel, and transferred to nitrocellulose membranes. Western blot was performed with mouse monoclonal anti-FLAG M2 antibody (Sigma), goat anti-human IDH1 (Santa Cruz Biotech), mouse anti-human IDH1^R132H antibody (Dianova), mouse anti-human α-tubulin (Sigma), horse radish peroxidase (HRP)–conjugated goat anti-mouse IgG (Jackson ImmunoResearch), HRP-conjugated rabbit anti-goat IgG (Santa Cruz Biotech), and the SuperSignal West Pico Kit (Thermo Scientific).

2-HG Determination by Chromatography and Mass Spectrometry

Extracellular concentrations of 2-HG were measured using high-performance liquid chromatography (HPLC) or gas chromatography (GC) followed by mass spectrometry (MS). For HPLC-MS determination of extracellular 2-HG, 50-μL aliquots of cell media were transferred to sample tubes, and 250 µL of acetonitrile was added for sample extraction. An aliquot of the supernatant (50 µL) was diluted with water (250 µL) and injected (10 µL) onto a reverse-phase HPLC column (Phenomenex Synergi Polar-RP, 4.6 × 150 mm, 4-μm particle size) equilibrated in 80% of eluent A (water/formic acid 100/0.1, volume/volume [v/v]) and 20% of eluent B (acetonitrile/formic acid, 100/0.1, v/v). The chromatographic separation was achieved with gradient elution of 20%–70% eluent B in 4 min. The effluent from the column was directed to an electrospray ionization (ESI) source connected to a triple quadrupole mass spectrometer (API 5500 QTrap, Sciex; −5 kV ESI capillary voltage, 500°C source temperature, 80 m dwell time per channel, 5-m interchannel delay, unit mass resolution on both Q1 and Q3) operating in the negative-ion selected reaction monitoring mode, in which the intensity of the m/z 147→129 transition of 2-HG was monitored.

MTT Cell Viability Assay

The 96-well plates were coated with 0.1% gelatin before seeding cells. Cell viability was determined by MTT assay as previously described.²⁵ Absorbance at 535 nm was measured on a Wallac Victor² 1420 plate reader (PerkinElmer) with a background reference filter at 660 nm.

Colony Formation Assay

Cell survival after radiation and TMZ treatment was assessed by colony formation assay. One thousand cells were seeded onto 60-mm dishes and incubated overnight. For radiation treatment, 7.43–7.49 Gy/min of irradiation was delivered to the cells on a rotating platform using a Mark-I Irradiator. TMZ treatment was performed as previously described.²⁵ The cell culture medium was replaced with fresh medium containing TMZ (NCI/NIH Developmental Therapeutics Program) or vehicle only (DMSO) as control. Cells were incubated after irradiation or TMZ treatment for 14 days and then fixed with methanol and stained with 0.25% crystal violet. Colonies containing >30 cells were counted under a dissecting microscope. The results are reported as a percentage of the colonies in untreated cultures of each corresponding clone.

Apoptosis Analysis by Flow Cytometry

Apoptotic cells were stained with YO-PRO-1 (Invitrogen) per the manufacturer's instructions and assayed with flow cytometry. Briefly, cells were treated with radiation, cultured for 48 h, and then stained with 1 µM YO-PRO-1 and 10 µg/mL propidium iodide for 30 min before being harvested in trypsin and analyzed with flow cytometry.

Reactive Oxygen Species Measurement

Cellular ROS levels were quantitated using the oxidant-sensitive fluorescent dye 2′,7′-dichlorfluorescein-diacetate (DCFH-DA; Sigma) and a Wallac Victor² 1420 plate reader (PerkinElmer) as previously described.²⁶ Briefly, cells were incubated with 10 μM DCFH-DA for 30 min at 37°C and then washed twice by using PBS before radiation or tert-butyl hydroperoxide (tert-BHP) exposure. Dichlorodihydrofluorescein (DCF) was measured 90 min (for radiation) or 10 min (for tert-BHP) after exposure by using the plate reader at excitation and emission wavelengths of 485 nm and 535 nm, respectively.

Soft Agar Growth Assays

Growth in soft agar was measured by colony assay as previously described.²⁷ Briefly, 2-mL underlayers of 0.6% agar medium were prepared in 6-well plates by combining equal volumes of 1.2% Noble agar (Fisher) and DMEM with 20% FBS. Cells (2 × 10³) were plated in 0.3% agar medium and cultured for 2–3 weeks. Colonies were then photographed, and colonies larger than 50 μm in diameter were counted in 5 random microscopic fields and tabulated.

Monolayer Cell Migration Scratch Assay

Cells were plated in 6-well plates coated with 0.1% gelatin to create a confluent monolayer. Scratch assay was performed as previously described.²⁸ The scratch wound was observed using contrast microscopy (Olympus, IX41), and images were taken 0 h, 3 h, 6 h, 9 h, 12 h, and 24 h after the initial scratch. Scratch wound distance was quantitated using Adobe Photoshop software.

Transwell Migration Assay

Cell migration assays were performed essentially as described previously.^29,30 Briefly, transwell plates (8-μm pore size) were coated with 100 μL gelatin (1 mg/mL) on the bottom side of the membrane and dried at room temperature. The transwell plates were assembled in a 24-well plate, and 1 × 10⁵ cells were suspended in 100 μL FBS free culture media and added to each upper chamber. After a 5-h incubation period, the transwell plates and the nonmigrating cells in the top wells were removed. The migrating cells remaining on the bottom side of the transwell plates were fixed and stained with 0.25% crystal violet. Quantification of cell migration was done by calculating the area of migrated cells on the bottom chamber and was expressed as a percent of control.

Statistical Analyses

Statistical analysis was performed with GraphPad Prism 5 software. Statistical significance was determined using Student's t-test. Data are expressed as mean ± SEM, and the level of significance is set at P < .05, P < .01, and P < .001.

Results

Stable Expression of Wild-type and Mutant FLAG-tagged IDH1 and IDH2 Constructs in Glioma Cells

To determine the effects of overexpression of human glioma-associated IDH mutant proteins within an isogenic glioma cell context, we generated carboxy-terminus FLAG-tagged IDH1^WT, IDH1^R132H, IDH2^WT, IDH2^R172K, and control (vector only) constructs, stably expressed them in U87MG cells, and isolated clonal cell lines (U87MG-control, U87MG-IDH1^WT, U87MG-IDH1^R132H, U87MG-IDH2^WT, and U87MG-IDH2^R172K) (Fig. 1). The same FLAG-tagged IDH1^WT and IDH1^R132H constructs were also stably expressed and single-cell cloned in parallel with vector only in U373MG cells (U373MG-control, U373MG-IDH1^WT, and U373MG-IDH1^R132H) (Fig. 1A, Fig. S1). Western blot analysis with anti-FLAG, anti-IDH1, and mutant-specific anti-IDH1^R132H antibodies confirmed the expression of the appropriate IDH1 protein products in the various cell lines. Similar confirmation of the IDH2 protein product expression was achieved by western blot analysis using an anti-FLAG antibody (Fig. 1B, Fig. S1A). By immunofluorescence, the steady-state cellular localization of the IDH1^R132H mutant protein appeared indistinguishable from that of the IDH1^WT protein, indicating that the mutations did not significantly alter the targeting of the IDH1 protein; similarly, the localization of the mutant IDH2^R172K protein appeared identical to that of the IDH2^WT protein (Fig. 1C and 1D). Using densitometric analysis of western blot, we found that for the selected clones, IDH1^WT was overexpressed ∼3–4-fold over endogenous IDH1, and IDH1^R132H was overexpressed by ∼5–6-fold in U87MG cells (Fig. S1B). The level of overexpression of IDH1^WT or IDH1^R132H was ∼2–3-fold more than endogenous IDH1 in U373MG cells (Fig. S1C).

Fig. 1.

Establishment of glioma stable cell lines overexpressing IDH1^WT, IDH1^R132H, IDH2^WT, and IDH2^R172K. (A) Schematic diagram of the IDH1-FLAG and IDH2-FLAG fusion protein amino acid sequence. The FLAG tag was linked to the C-terminus of IDH1 and IDH2. Amino acid residue 132 for IDH1 and 172 for IDH2 are indicated with (). (B) Representative western blot for FLAG-tagged IDH1^WT and IDH1^R132H or IDH2^WT and IDH2^R172K in stably transfected U87MG cells (2 representative independent clones for each construct, C1 and C2, are shown) using anti-FLAG, anti- IDH1^WT, and anti-IDH1^R132H antibodies. (C) Immunofluorescence showed the diffuse cytoplasmic distribution of IDH1^WT and IDH1^R132H proteins stained by anti-FLAG antibody (green), anti-IDH1^WT antibody (red), and anti-IDH1^R132H (green). Nuclei were counterstained with DAPI (blue). 40×, bar = 20μm (D) Immunofluorescence shows the punctate cytoplasmic distribution of IDH2^WT and IDH2^R172K proteins stained by anti-FLAG antibody (green), and nuclei were counterstained with propidium iodide (PI) (red), 40×, bar = 20μm. (E) ∼70-fold increase in 2-HG level was found in the conditioned medium of U87MG-IDH1^R132H cells by HPLC-MS. Data were normalized by cell number and expressed as fold change relative to U87MG-control (mean ± SEM, n =3). **P < .01 compared with U87MG-control and U87MG-IDH1^WT cells. (F) ∼140-fold increase in 2-HG level was found in the conditioned medium of U87MG-IDH2^R172K cells by HPLC-MS. Data were normalized by cell number and expressed as mean fold change relative to U87MG-control (mean ± SEM, n =3). **P < .01 compared with U87MG-control and U87MG-IDH2^WT cells.

Consistent with prior reports,¹⁵ we found, using HPLC-MS, that U87MG-IDH1^R132H cells generated a concentration increase in 2-HG of ∼70-fold in conditioned medium compared with U87MG-control and U87MG-IDH1^WT cells (Fig. 1E). These results were independently confirmed using GC-MS (Fig. S2A). Using HPLC-MS, similarly increased 2-HG levels were seen in the conditioned medium from U373MG-IDH1^R132H (Fig. S2B) and U87MG-IDH2^R172K cells (Fig. 1F).

IDH1^R132H and IDH2^R172K Overexpression Increases Sensitivity to Radiation Treatment

Given the clinical observation that patients with IDH1/2 mutant tumors have prolonged survival, we hypothesized that one contributing factor could be that IDH1^R132H tumors are more sensitive to treatment. To test this hypothesis, we treated the isogenic U87MG-control, U87MG-IDH1^WT, and U87MG-IDH1^R132H clonal cell lines with radiation, a common treatment modality for glioma. Following escalating doses of radiation, MTT cell viability (Fig. 2A and 2B) and colony formation assays (Fig. 2C) showed that the U87MG-IDH1^R132H cells exhibited dose- and time-dependent increased loss of viability compared with U87MG-control and U87MG-IDH1^WT cell lines. To determine whether this effect could be observed in another isogenic glioma cell context, we performed similar experiments on U373MG cells, which notably differ from U87MG cells in harboring mutant p53 and phosphatase and tensin homolog (PTEN). In both the MTT assay (Fig. 2A and 2B) and the colony forming assay (Fig. 2C), we found that U373MG-IDH1^R132H cells also demonstrated increased sensitivity to radiation compared with U373MG-control and U373MG-IDH1^WT cells, albeit to a slightly less degree. This may reflect lower levels of overexpression of IDH1^R132H in U373MG cells compared with U87MG cells (Fig. S1A and S1C). To determine whether overexpression of IDH2^R172K, the most common glioma-associated IDH2 mutation, was also associated with radiosensitization, we tested U87MG-control, U87MG-IDH2^WT, and U87MG-IDH2^R172K cells. In both MTT (Fig. 2A and 2B) and colony forming assays (Fig. 2C), we observed that U87MG-IDH2^R172K cells also showed increased sensitization to radiation similar to that seen for U87MG-IDH1^R132H cells.

Fig. 2.

Overexpression of IDH1^R132H or IDH2^R172K increased the sensitivity of glioma cells to radiation treatment. (A) Cells were treated with various doses of radiation and incubated for 48 h for U87MG cells or 72 h for U373MG cells. Viability was quantitated with MTT assay and expressed as mean percentage of untreated control cells (mean ± SEM, n =3). *P < .05, **P < .01 compared with control, IDH1^WT or IDH2^WT cells. (B) Cells were treated with 25 Gy of radiation for different incubation times. Viability was quantitated with MTT assay and expressed as mean percentage of untreated control cells (mean ± SEM, n = 3). *P < .05, **P < .01 compared with control, IDH1^WT or IDH2^WT cells. (C) Equal number of cells was seeded into 60-mm dishes and treated with different doses of radiation 24 h later. After 14 days, the cells were fixed and stained with 0.25% crystal violet. Mean clonogenic survival of cells after radiation was plotted on a logarithmic scale (mean ± SEM, n =3). *P < .05, **P <.01 compared with control, IDH1^WT or IDH2^WT cells.

IDH1^R132H and IDH2^R172K Overexpression Does Not Affect TMZ Sensitivity, Whereas IDH1^WT and IDH2^WT Overexpression Appears Protective Against TMZ

Since the alkylating agent, TMZ, is also commonly used in the treatment of gliomas, we sought to determine whether overexpression of IDH1^R132H altered sensitivity to TMZ. After exposing U87MG-control, U87MG-IDH1^WT, and U87MG-IDH1^R132H cells to varying concentrations of TMZ, we found, using MTT assays, that there were no differences in viability among the 3 cell lines, in either a dose- (Fig. 3A) or a time-dependent manner (Fig. 3B). Similar lack of differences was seen among U373MG-control, U373MG-IDH1^WT, and U373MG-IDH1^R132H (Fig. 3A and 3B). When measured by clonogenic assay, we also found that TMZ sensitivity was similar between U87MG-control and U87MG-IDH1^R132H cells. Interestingly, however, U87MG-IDH1^WT cells showed increased survival percentage when treated with TMZ, compared with U87MG-control and U87MG-IDH1^R132H cells, suggesting a protective effect of overexpressed IDH1^WT protein against TMZ cytotoxicity (Fig. 3C). We observed this same pattern in U373MG cells, in which U373MG-IDH1^WT cells demonstrated decreased TMZ sensitivity by clonogenic assay compared with either U373MG-IDH1^R132H or U373MG-control cells (Fig. 3C). In addition, when we performed the same experiments on U87MG-control, U87MG-IDH2^WT, and U87MG-IDH2^R172K cells, we similarly observed no difference among any of the cell lines on MTT assay (Fig. 3A and 3B), whereas the colony forming assay also showed decreased TMZ sensitivity of the U87MG-IDH2^WT cells (Fig. 3C).

Fig. 3.

IDH1^R132H and IDH2^R172K overexpression did not affect TMZ sensitivity. (A) Cells were treated with various doses of TMZ for 72 h. Viability was quantitated with MTT assay and expressed as mean percentage of untreated control cells (mean ± SEM, n =3). (B) Cells were treated with 400 μM of TMZ for different periods of time. Viability was quantitated with MTT assay and expressed as mean percentage of untreated control cells (mean ± SEM, n =3). (C) Clonogenic survival of cells after TMZ treatment was plotted on a logarithmic scale (mean ± SEM, n =3). *P < .05, **P < .01 compared with control, IDH1^R132H or IDH2^R172K cells.

U87MG-IDH^R132H Cells Show Significant Increase in Apoptosis Following Radiation Treatment

Using the U87MG-control, U87MG-IDH1^WT, and U87MG-IDH1^R132H cells, we investigated molecular mechanisms mediating the effect of overexpression of IDH^R132H protein on increasing radiation sensitivity. To determine whether increased radiation sensitivity was accompanied by increased apoptotic index, we assessed the apoptotic index in the various U87MG cell lines using the fluorescent dye YO-PRO-1. Following radiation, measurement of YO-PRO-1 staining by flow cytometry showed that U87MG-IDH1^R132H cells had an increased apoptotic index compared with U87MG-control and U87MG-IDH1^WT cells (Fig. 4). In comparison, all of the various U87MG cells demonstrated similar increases in apoptotic index upon treatment with paclitaxel, used as a positive control (Fig. S3).

Fig. 4.

U87MG-IDH^R132H cells showed significant increase in apoptosis following radiation treatment. (A) Apoptosis was assessed by YO-PRO-1 and propidium iodide (PI) staining after 48 h radiation treatment (0 Gy or 25 Gy). Harvested cells were evaluated by flow cytometry. Column plot summarized 3 independent experiments, each performed with triplicates (mean ± SEM, n =3). **P < .01 compared with U87MG-control and U87MG-IDH1^WT cells. (B) Apoptosis % was normalized to U87MG-control cells treated with 0 Gy and expressed as mean apoptotic index (mean ± SEM, n =3). **P< .01 compared with U87MG-control and U87MG-IDH1^WT cells.

U87MG-IDH1^R132H Cells Show Increased ROS Levels Upon Radiation Treatment

Given that IDH1 functions to generate NADPH, we hypothesized that impaired NADPH production resulting in increased susceptibility to oxidative stress might contribute to the increased radiation sensitivity in U87MG-IDH1^R132H cells. To investigate whether IDH1^R132H alters ROS metabolism, we measured levels of cellular oxidants using the fluorescent dye DCFH-DA, which is retained in cells following de-esterification and becomes fluorescent when oxidized. We found that basal DCF fluorescence level was lower in U87MG-IDH1^R132H cells compared with U87MG-control and U87MG-IDH1^WT cells by using the plate reader (Fig. 5A) and fluorescence-activated cell sorting (Fig. S4). However, after radiation treatment, the U87MG-IDH1^R132H cells showed an accentuated elevation in ROS level compared with both U87MG-control and U87MG-IDH1^WT cells, which had little increase in DCF signal after radiation treatment (Fig. 5B). Similar oxidant response was shown when cells were treated with tert-BHP, a mimic of lipid hydroperoxides, with increased levels detected in U87MG-IDH1^R132H cells compared with either U87MG-control or U87MG-IDH1^WT cells (Fig. S5).

Fig. 5.

U87MG-IDH^R132H cells showed increased oxidative response to radiation. Cells were incubated with the oxidant sensitive fluorescent dye DCFH-DA for 30 min before exposure to radiation, and ROS level was measured 90 min later by the plate reader. (A) The baseline ROS levels were measured by DCF assay and expressed as mean fold change relative to U87MG-control cells (mean ± SEM, n = 3). *P < .05, **P< .01 compared with U87MG-control and U87MG-IDH1^WT cells. (B) The ROS level after radiation treatment was measured by DCF assay and expressed as mean fold change relative to untreated U87MG-control, U87MG-IDH1^WT, and U87MG-IDH1^R132H cells (mean ± SEM, n =3). *P < .05 compared with U87MG-control and U87MG-IDH1^WT cells.

N-Acetyl Cysteine Protects U87MG-IDH1^R132H Cells From Radiation

To determine whether addition of an ROS scavenger can protect the U87MG-IDH^R132H cells from radiation, we pre-incubated cells with 5 μM NAC for 4 h before exposure to different doses of radiation and determined cell viability 48 h later by MTT assay. The choice of the low dose of 5 μM NAC incubation was determined as a concentration with minimal effects on viability of the various cell lines (Fig. S6). We observed that the increased radiosensitivity of U87MG-IDH1^R132H was abrogated in the presence of 5 μM NAC. In contrast, this dose of NAC did not affect the radiosensitivity of either U87MG-IDH1^WT or U87MG-control cells (Fig. 6).

Fig. 6.

ROS scavenger protected the U87MG-IDH^R132H cells from radiation. Cells were pretreated with NAC for 4 h before exposure to different doses of radiation. Viability was determined 48 h later by MTT assay and expressed as mean percentage of untreated U87MG-control cells (mean ± SEM, n= 3). *P < .05, **P < 0.01 compared with U87MG-control and U87MG-IDH1^WT cells.

U87MG-IDH1^R132H Cells Show Decreased Cell Proliferation at High Growth Density, Growth in Soft Agar, and Glioma Migration

During routine cell culture, we observed that the U87MG-IDH1^R132H cells appeared more cuboidal than U87MG-IDH1^WT cells and that the mutant cells appeared to grow more slowly in culture. At higher cell densities, U87MG-IDH1^R132H cells maintained a monolayer, whereas the U87MG-control and U87MG-IDH1^WT cells formed multilayered cell clusters (Fig. 7A). The MTT proliferation assays showed that U87MG-IDH1^R132H cells grew more slowly than U87MG-control and U87MG-IDH1^WT cells at higher cell density (Fig. 7B). These findings were confirmed by direct cell counting (Fig. S7). We measured colony formation of the cells in soft agar (non-adherent) for the various cell lines and found that the U87MG-IDH1^R132H cells had significantly decreased colony formation ability as well as smaller colony size than U87MG-control and U87MG-IDH1^WT cells (Fig. 7C). Based on the altered growth characteristics demonstrated by the U87MG-IDH1^R132H cells, we sought to investigate whether the IDH1^R132H protein could suppress cell migration using a wound-healing assay to study directional cell migration in vitro.²⁸ As seen qualitatively in Fig. 7D, U87MG-IDH1^R132H cells showed decreased migration. Quantitative assessment of these assays clearly show decreased cell migration in the U87MG-IDH1^R132H cells compared with U87MG-control and U87MG-IDH1^WT cells (Fig. 7E). Similar results showing decreased migration of U87MG-IDH1^R132H cells were obtained with the transwell migration assay (Fig. 7F).

Fig. 7.

U87MG-IDH^R132H protein overexpression decreased cell proliferation, growth in soft agar, and migration. (A) Phase contrast microscopy pictures of U87MG-control, U87MG-IDH1^WT cells, and U87MG-IDH1^R132H cells in culture. Clusters are found at higher densities U87MG-control, U87MG-IDH1^WT cells (indicated by white arrows). 4×, bar = 100 μm (B) Cell proliferation was determined by MTT assay and expressed as mean absorbance value at 535 nm with reference at 660 nm (mean ± SEM, n =3). ***P < .001 compared with U87MG-control and U87MG-IDH1^WT cells. (C) Soft agar colony formation assay was performed by counting individual colonies (>50 µm) in 5 random microscopic fields, and representative micrographs are shown. The quantification of colony numbers are shown as mean ± SEM (n =3). 4×, bar = 100 μm, **P < .01 compared with U87MG-control and U87MG-IDH1^WT cells. (D) and (E) Monolayer cell migration scratch assay. Representative phase micrographs of U87MG-control, U87MG-IDH1^WT, and U87MG-IDH^R132H cells at 0 h, 3 h, 6 h, 9 h, 12 h, and 24 h after monolayer wounding are shown in D. Monolayer scratch wound distances were quantitated in (E) and expressed as mean ± SEM (n = 3). 4×, bar = 100 μm, *P < .05 compared with U87MG-control and U87MG-IDH1^WT cells. (F) Cell migration quantitated with a transwell migration assay. U87MG-control, U87MG- IDH1^WT, and U87MG-IDH1^R132H cells were plated in the transwell chamber for 5 h before migrated cells were fixed, stained, and counted. Top: representative micrographs of migrated cells. Bottom: migrated cell number is quantitated and expressed as mean percentage of U87MG-control (mean ± SEM, n =3). 4×, bar = 100 μm, **P < .01 compared with U87MG-control and U87MG-IDH1^WT cells.

To determine whether overexpression of IDH1^R132H had similar effects in U373MG cells, we compared proliferation and migration among U373MG-control, U373MG-IDH1^WT, and U373MG-IDH1^R132H. In contrast to what we found with U87MG cells, we did not observe decreased growth (Fig. S8A) and migration in U373MG-IDH1^R132H (Fig. S8B), suggesting that these effects are cell context dependent. The slightly lower lever of overexpression of IDH1^R132H provides another possible explanation for the lack of effect on U373MG cells (Fig. S1B and S1C).

Discussion

Using an isogenic U87MG model system, we demonstrated that overexpression of the IDH1^R132H mutant protein enhanced the cytotoxic effect of radiation exposure, possibly through altered ROS metabolism. There was also similar radiosensitization in cells overexpressing IDH2^R172K. These findings are consistent with improved survival in patients with IDH1^R132H or IDH2^R172K mutant tumors because radiation is the most common therapeutic treatment for glioma patients. For example, in a recent study by the European Organisation for Research and Treatment of Cancer (EORTC), patients with tumors harboring IDH1 or IDH2 mutations showed similarly improved outcomes whether they received RT or RT/PCV.²³ From these data, the authors concluded that IDH mutation is a prognostic marker but not predictive of PCV response. Since all patients received RT, an interpretation of these data is that IDH1^R132H predicts improved response to RT. In addition, our data showed that IDH1^R132H protein did not sensitize U87MG cells to TMZ and that, interestingly, overexpression of IDH1^WT increased resistance of U87MG cells to TMZ. Thus far, retrospective clinical studies have reported conflicting results regarding TMZ response in IDH1 mutant gliomas. For example, Houillier and coworkers²¹ found that IDH1^R132MUT predicted response to TMZ in low-grade gliomas, while Dubbkink et al²² found that IDH1^R132MUT predicted overall survival but not response to TMZ. Definitive clinical evidence on whether IDH1^R132MUT predicts and/or influences response to RT or TMZ may be derived from the results of an ongoing randomized prospective trial that compares RT-only with TMZ-only treatments (the EORTC 22033-26033 trial).³¹

In terms of the mechanism of enhanced radiosensitivity in the U87MG-IDH1^R132H cell lines, data from the current study suggest that mutant IDH1 may result in increased ROS levels and a heightened oxidative response to radiation exposure in the U87MG-IDH1^R132H cells. It is possible that the observed increased oxidative stress in IDH1 mutant cells is due to decreased buffering of ROS from lower available levels of NADPH. Previous studies have shown that IDH-mediated NADPH production and overall NADPH levels are reduced in IDH1^R132H glioblastomas in situ.²⁰ Similarly, it has been previously shown that mitochondrial IDH-regulated radiation induced apoptosis in vitro, as demonstrated by increased radiation sensitivity in cells treated with mitochondrial IDH small interfering (si)RNA.³² In another study, this effect was also shown in vivo based on increased radiosensitivity in mice fed oxalomalate, an inhibitor of IDH1.²⁹ Interestingly, we found that the basal (untreated) ROS levels were lower in the U87MG-IDH1^R132H cell lines. While it is possible that this lower basal DCFH-DA staining may somehow represent an artifact, the apparent decreased level of ROS could result from lower basal levels of oxidant generation from slower cellular metabolism. The lower basal ROS level may also contribute to the decreased proliferation observed in U87MG-IDH1^R132H cells.^33–36 Although supported by the current study, the mechanism(s) by which IDH1^R132H overexpression results in increased sensitivity to radiation are unclear and warrant further studies.

Several points regarding our in vitro model must be considered that may limit the clinical relevance of the current results. First, the “overexpression” model used here has an approximately 5–6-fold increased mutant protein expression compared with endogenous IDH1^WT expression (Fig. S1A and S1B). In patient glioma samples, IDH1^R132MUT is heterozygous, which would suggest a 1:1 ratio of IDH1^WT: IDH1^R132MUT gene dosage. Presently, “naturally occurring” cell lines derived from a patient tumor harboring the IDH1^R132MUT have been elusive, with only one reported cell line derived from an oligodendroglioma.³⁷ Existing studies have all utilized an overexpression system similar to that used here.^14,15,38 Further experiments using cell lines with varying levels of overexpression may be useful to estimate the effects of gene dosage. Second, the isogenic U87MG cell model represents a specific genetic background: the salient molecular features of the U87MG cells include IDH1^WT,³⁹ methylated MGMT,⁴⁰ p53^WT,⁴¹ EGFR^WT,⁴² and PTEN null.⁴³ This raises the possibility that our results are specific to the U87MG genetic background. In order to investigate whether radiosensitization was specific to U87MG cells, we performed similar experiments in U373MG cells that had a background of IDH1^WT,³⁹ methylated MGMT,⁴⁰ p53^MUT,⁴¹ EGFR^WT,⁴⁴ and PTEN null⁴³ and found similar radiosensitization by overexpression of IDH1^R132H in U373MG cells. These results support the notion that radiosensitization of glioma cells by overexpression of IDH1^R132H may apply to the varied genetic backgrounds seen in the clinic.

In addition, similar to other reported results,⁴⁵ we found that U87MG-IDH1^R132H cells demonstrated reduced baseline growth in soft agar, migration, and growth at higher cell density. Similar to our results in U87MG cells, it has been shown that siRNA knockdown of endogenous IDH1^WT results in lower growth in vitro in SF188 cells, a pediatric glioblastoma cell line that is p53^MUT and PTEN^WT.⁴⁶ However, we did not observe these effects in the context of U373MG cells, suggesting that these effects are cell context dependent, although there is a possibility that U373MG cells did not show these effects on basal growth because of lower dosage of IDH1^R132H overexpression than in the U87MG cells. Overall, these data suggest that lowered basal growth and migration might be other mechanisms for improved outcomes, albeit in only selected IDH1 mutant tumors.

In summary, our results indicate that overexpression of the IDH1^R132H protein can sensitize glioma cells to radiation and, in the context of U87MG cells, exerts suppression of growth and migration. These in vitro results are consistent with the clinical observation that patients with IDH1^R132H tumors have improved survival. If validated in patients, these findings may have broader implications regarding the possibility of therapeutically targeting IDH1-mediated pathways. Additional in vitro and in vivo studies are needed to further elucidate the pathophysiology and clinical implication of IDH1 mutation.

Supplementary Material

Supplementary material is available online at Neuro-Oncology (http://neuro-oncology.oxfordjournals.org/).

Funding

This work was supported by the National Cancer Institute at the National Institutes of Health (K08CA124479 to A.L.); University of California/Cancer Research Coordinating Committee (to A. L.); and American Brain Tumor Association Basic Research Fellowship (to S.L.); American Association of Neurological Surgeons Neurosurgery Research and Education Foundation Fellowship sponsored by the Section on Tumors (to A.P.C.); Congress of Neurological Surgeons Wilder Penfield Fellowship (to A. P. C.).