Sorafenib selectively depletes human glioblastoma tumor-initiating cells from primary cultures

Sorafenib exerts in vitro antiproliferative effects in GBM TICs

During preliminary experiments, cell concentration ensuring an exponential proliferation for the entire duration of treatment (up to 4 d) was determined for each GBM TIC culture, as we previously reported.²¹ From these experiments we identified 2,500, 3,000, 25,000, 12,000, 5,000 and 15,000 cells/well for GBM 3, 23, 24, 17, 19 and 6, respectively, as the optimal cell number for seeding each culture in a 96-wells plate (data not shown).

To evaluate whether sorafenib exerts antiproliferative effects on GBM TICs in vitro, cultures derived from all six GBMs were treated with increasing concentrations of sorafenib (from 5 to 20 µM) for different time intervals (48 and 72 h). Viability of cells was measured at each time point by MTT assay, and results were expressed as percentage of treated cells vs. control (vehicle treated) cells. As summarized in Figure 1A, sorafenib induced a dose- and time-dependent reduction of cell number in all GBM TIC cultures. Sorafenib half-maximal inhibitory concentration (IC₅₀), measured at 48 h from dose-response curves (data not shown), was very similar in all the cultures: 15 µM for GBM 3, 23, 24 and 17; 12 µM for GBM 19 and 8 µM for GBM 6.

Sorafenib inhibition of GBM TIC DNA synthesis

To investigate whether sorafenib antiproliferative activity also involves a reduction of GBM TIC DNA synthesis, we measured the effects of this drug on BrdU incorporation after 24 h of treatment. As compared with vehicle-treated cells, the percentage of BrdU-positive cells was significantly decreased (Fig. 1B) from 20.1% to 0.7% in GBM 3 (p < 0.01), from 34.8% to 5.3% in GBM 23 (p < 0.01), from 15.5% to 0.8% in GBM 24 (p < 0.05), from 10.3% to 0.6% in GBM 17 (p < 0.05), from 20.2% to 2.7% in GBM 19 (p < 0.05) and from 13.2% to 0.8% in GBM 6 (p < 0.01).

These results clearly indicated that sorafenib induces growth arrest in all GBM TICs in the first 24 h of treatment, preventing the entering the S phase.

Sorafenib induces apoptosis in the vast majority of GBM TICs cultures

The contribution of cell death to the antiproliferative effects of sorafenib in GBM TICs was measured after 24 and 48 h of treatment with drug concentrations corresponding to the respective IC₅₀. To accurately assess TIC viability, apoptosis was analyzed using annexin-V binding assay, in conjunction with propidium iodide (PI) staining to detect necrotic cells (Fig. 2). Data are reported as percentage of overall cell death, including PI-positive necrotic cells, early apoptotic cells (annexin-V-positive) and late apoptotic cells (labeled with both PI and annexin-V). The response to sorafenib treatment widely differed among cultures: TICs from GBM 3, 23, 24 and 17 showed a considerable increase of overall cell death compared with control, already significant at 24 h and reaching maximal values at 48 h of treatment (Fig. 2): from 8.8% to 44.7% in GBM 3, from 7.9% to 68.4% in GBM 23, from 10.1% to 74.2% in GBM 24 and from 20.1% to 72.2% in GBM 17. GBM 19 showed a lower cell death induction that became statistically significant only after 48 h (from 8.6% to 35.6%). Conversely, no evidence of cell death induction was detected in GBM 6, suggesting that, in this case, cell death and apoptosis do not contribute to the antiproliferative effect of sorafenib. The contribution of the apoptotic process to cell growth arrest exerted by sorafenib in GBM TICs is highlighted in Figure 2 by bars including the percentage of apoptotic fraction in each culture. Importantly, the apoptotic component in all TIC cultures (with exception of GBM 6 and partially GBM 19) represented a high percentage of the total cell death reaching values largely greater than 50%.

Intracellular mechanisms mediating sorafenib pro-apoptotic effects

To define the intracellular mechanisms by which sorafenib affects human GBM TIC survival in vitro, we analyzed the activation of intracellular pathways related to GBM development and, in particular, STAT-3, MEK/ERK1/2 and Akt pathways. The analysis was performed by western blot in control cells and in cells treated for 24 h with sorafenib, using concentrations corresponding to 0.5X, 1X and 2X-fold the calculated IC₅₀ for each GBM TIC culture. Sorafenib-induced modulation of STAT-3 phosphorylation, one of the downstream molecules involved in GBM invasion and proliferation, was not strictly correlated with antiproliferative or pro-apoptotic effects observed in GBM TIC cultures, STAT-3 activation being variably modulated or unaffected (Fig. 3). Conversely, as expected according to the known mechanism of action of sorafenib, the treatment caused a significant reduction of the activation of MEK and ERK1/2 as well as of Akt in the responsive cultures (GBM 3, 23, 24 and 17) (Fig. 3; Fig. S1 for the densitometric quantification of the immunoreactive bands). This effect was obtained only for Akt at the highest concentration tested (2X IC₅₀) in the partially responsive culture GBM 19, and completely absent in GBM 6, which we’ve demonstrated to be insensitive to apoptotic effects of the drug (Fig. 3).

Since Mcl-1, an anti-apoptotic member of the Bcl-2 family, was reported to play a key role in pro-apoptotic effects of sorafenib,³³ we investigated its expression in GBM TICs. The activity of sorafenib on MEK/ERK and Akt activation was paralleled by the dramatic downregulation of Mcl-1, in sorafenib-responsive TICs, starting from the lower concentration used (0.5X IC₅₀) and proportionally increasing up to the 2X IC₅₀. In the GBM 19 TICs that showed low induction of the apoptotic fraction after exposure to sorafenib, we observed a slight reduction of Mcl-1 only at the highest concentration tested, while Mcl-1 expression was unaffected in GBM 6, which did not activate the apoptotic process in response to sorafenib (Fig. 3; Fig. S1).

Sorafenib impairs the clonogenic potential in vitro and the tumorigenicity in vivo of sensitive GBM TICs

In order to understand whether sorafenib preferentially reduces the viability of the stem cells subset, or whether it also might affect the differentiated subpopulation present in GBM cultures, the clonogenic potential of cells surviving 48 h treatment with sorafenib was evaluated in three GBM TIC cultures showing different levels of sensitivity to the pro-apoptotic effect of sorafenib (namely, GBM 3, 19 and 6). The same number of viable cells was plated for control and sorafenib-treated conditions. Clonogenic assays identify single cell-derived colonies grown above an arbitrarily designated cut-off size in a given time: committed cells with a limited residual proliferative potential give rise to small colonies, while stem-like cells with self-renewal capacity are able to originate large colonies. The cut-off chosen was 100 cells after 2 weeks of culture (see “Materials and Methods”). As shown in Figure 4A, sorafenib treatment induced a significant reduction of the clonogenic potential of GBM 3 and GBM 19: from 67.9% to 26.8%, and from 38.9% to 11.9%, respectively. In contrast, sorafenib did not reduce clonogenicity of GBM 6 TICs (Fig. 4A). Correspondingly, a significant reduction of the clonogenic potential after treatment was confirmed in GBM 3 and GBM 19 by MTT colorimetric assays performed three weeks after cell seeding, using an absorbance value of 0.400 as a threshold (data not shown).

These results suggest a selective activity of sorafenib on the clonogenic potential, an in vitro test to evaluate cell stemness. To confirm the depletion of the stem cells pool by sorafenib, the expression of differentiation (glial fibrillary acidic protein, GFAP and microtubule-associated protein 2, MAP2) or stemness (nestin, Olig2 and Sox2) markers was evaluated in treated and untreated cells. The levels of basal expression of GFAP, astroglial differentiation marker and MAP2, neuronal differentiation marker, considerably differed among the GBM TICs used in this study (Fig. 4B, upper panels). With the only exception of GBM 6, whose GFAP- and MAP2-positive cell percentages were not affected by sorafenib (from 13.6% to 13.2% of GFAP-positive cells and from 54.2% to 55.7% of MAP2-positive cells in untreated and treated cells, respectively), all other cultures showed a significant increment of the fraction of GFAP-positive cells: from 21.1% to 44.4% in GBM 3, from 21.0% to 37.9% in GBM 23, from 36.8% to 61.6% in GBM 24, from 27.1% to 40.6% in GBM 17 and from 27.3% to 56.2% in GBM 19. The cell populations which survived the treatment also included a consistent fraction of MAP2-positive cells that generally was at least 2-fold greater in sorafenib-treated cells as compared with controls (Fig. 4B, upper panel).

The analysis of stem-specific markers evidenced that, with the exception of GBM 6, all cultures significantly reduced the percentage of nestin- and Olig2-expressing cells after 48-h exposure to sorafenib (Fig. 4B, middle panels). In particular, the amount of sorafenib-dependent decrease was about 35% (min -25% for GBM 19/max -47% for GBM 23) for nestin positivity, and more marked (average -67%; min -53% for GBM 19/max -79% for GBM 23) as far as Olig2 expression. GBM 6 showed no reduction in nestin-expressing cell content and a low decrease in Olig2-positive cells (from 72.9% to 52.2%) that did not reach statistical significance (Fig. 4B, middle panels). Furthermore, the expression levels of Sox2, a marker for normal neural cells also linked to self-renewal properties of GBM stem cells, were evaluated by western blot analysis. After 48 h of treatment, sorafenib decreased Sox2 protein levels in all GBM cultures. GBM 3, 17 and 23 cells showed the highest reduction of Sox2 expression after sorafenib treatment in comparison with vehicle-treated control cells (-70%, -74% and -75%, respectively). GBM 6, 19 and 24 displayed a lower, but still significant decrease in Sox2 expression (-42%, -56% and -38%, respectively) (Fig. 4B, lower panel).

Since GBM TICs that survived sorafenib treatment are enriched in cells expressing differentiation markers and depleted in markers associated with stemness, we evaluated whether this drug might have selective effects toward TIC subpopulation. Thus the effects of sorafenib were tested on GBM 3, 19 and 6 cultures after they were induced to differentiate for two weeks in a medium containing 10% of fetal bovine serum. After this period, cells were treated with sorafenib 20 µM (the highest concentration tested on GBM cultures) for 48 h. Cell proliferation was measured by MTT assay, and results are expressed as percentage of cells detected after treatment vs. control (vehicle-treated). Differentiated GBM cells showed a significantly lower sensitivity to sorafenib than the corresponding undifferentiated cultures (Fig. 4C). In particular, we observed 72% of cells in differentiated vs. 50% in undifferentiated GBM 3, 97% in differentiated vs. 39% in undifferentiated GBM 19, and 88% in differentiated vs. 35% in undifferentiated GBM 6 (Fig. 4C). The reduced in vitro sensitivity of GBM differentiated cells, together with both enrichment in differentiation markers and inhibition of the clonogenic potential of GBM TICs after treatment with sorafenib, strongly suggests a selective action of sorafenib on the stem (undifferentiated) compartment of GBM cultures.

To confirm the in vitro results, seven mice were orthotopically transplanted with 100,000 GBM 3 cells, survived to 48 h treatment with sorafenib or left untreated. Six out of seven (85%) NOD/SCID mice intracranially injected with untreated GBM 3 cells developed a tumor mass, while only two out of seven (28%) mice injected with in vitro sorafenib-treated GBM 3 cells formed a tumor (Fig. 4D).

Isolation of human glioblastoma tumor-initiating cells

Tumor specimens from GBM resections obtained after patients’ informed consent and Institutional Ethical Committee approval from the Neurosurgery Department (IRCCS-AOU San Martino-IST) were histologically classified according to the WHO classification as astrocytic glioma grade IV. Tumors from six patients, three males and three females, with an average age of 53.8 y (41–70) entered this study and were coded as GBM 3, 23, 24, 17, 19 and 6. Four out of six tumor samples (GBM 3, 23, 17 and 6) derived from primary GBMs, GBM 19 from a secondary GBM and one sample (GBM 24) was from a recurrence after initial radio- and chemotherapy. After surgery, tumor samples were immediately processed for isolation of TICs as previously described.^21,45 Briefly, single cells were obtained by mechanical dissociation and plated in Dulbecco’s modified Eagle medium (DMEM)/F12 and Neurobasal (Gibco-Life Technologies, #31331 and #21103), supplemented with B27 (Gibco-Life Technologies, #12587–010), 2 mM L-glutamine (Euroclone, #3000D, penicillin-streptomycin (Euroclone, #ECB1001D), recombinant human basic FGF (10 ng/ml; Peprotech, #100–18B) and recombinant human EGF (20 ng/ml; Peprotech, #AF-100–15). All GBM cultures gave rise to neurospheres and were also able to grew as a monolayer on Matrigel (BD, #354234)-coated flask without losing their spherogenic and tumorigenic properties. To induce differentiation, GBM TICs were seeded and maintained for 2 wk in DMEM/F12 medium supplemented with 2 mM L-glutamine, penicillin-streptomycin and 10% of fetal bovin serum (FBS; Euroclone, #ECS0180L).

Intracranial tumor assays

Non-obese diabetic severe combined immunodeficient (NOD-SCID) mice were used to evaluate the tumorigenicity of cell cultures previously treated in vitro with sorafenib. Animals were housed in pathogenic-free conditions, and they were handled in agreement with guidelines conforming to current Italian regulations for the protection of animals used for scientific purposes (D.lvo 27/01/1992, no.116). Procedures were approved by the Ethical Committee for Animal Experimentation of the National Institute of Cancer Research and by the Italian Ministry of Health and were performed according to the National Regulation on Animal Research Resources. Intracranial inoculations were performed on adult NOD-SCID mice (6–8 weeks) anesthetized with i.m. ketamine and xylazine. Animals were positioned into a stereotactic apparatus (Model 900; David Kopf Instruments), and a hole was made using a 21-gauge needle, 2 mm lateral and 1 mm anterior from the intersection of the coronal and sagittal sutures (bregma). Up to 5 µl of cell suspension, containing 100,000 trypan blue-negative GBM TICs, were injected using a Hamilton syringe (series 7000; Sigma-Aldrich, #Z120359) at the depth of 3 mm. After six months, all mice were sacrificed by CO₂ asphyxiation, and brains were cryopreserved. Analysis/identification of tumors was performed on 10 μm coronal sections stained with hematoxylin-eosin, as previously described.⁴⁵

Cell proliferation assay

To test sorafenib effects on cell proliferation, MTT assays were performed.⁵³ Sorafenib, purchased from US Biological (#S5343–15), was reconstituted in dimethyl sulfoxide (DMSO) at the concentration of 10 mM. GBM TICs were plated in Matrigel coated 96-well microplates. The number of cells of each culture was adjusted to a density that allowed an exponential growth for the duration of the assay (72 h) ranging from 2,500 to 25,000 cells/well. After exposure to sorafenib at increasing concentrations (from 5 to 20 μM) for different time intervals (0, 48, 72 h), cells were incubated for 4 h in MTT solution (2 mg/ml in PBS; Sigma-Aldrich, #1001089589). After removing MTT, 150 µl of DMSO were added in each well to dissolve formazan crystals, and the absorbance was determined at 540-nm wavelength using a plate reader. Dose-response curves were generated, and IC₅₀ values were calculated by Graph Pad Prism 5 software.

BrdU incorporation assays

Effects of sorafenib on DNA synthesis were evaluated by 5-bromo2'-deoxy-uridine (BrdU, Sigma-Aldrich, #B5002) incorporation assay. Briefly, after a treatment of 24 h, cells were incubated with 10 µM BrdU for 30 min at 37°C and 5% CO₂. Cells were harvested and fixed with ethanol for 30 min. Following fixation cells were treated with PureLink RNaseA (1 mg/ml; Invitrogen, #12091–021) and incubated in 0.1 N HCl for 10 min on ice. After a denaturation step (90°C for 20 min then 2 min on ice), cells were incubated with mouse anti-BrdU antibody (1:100; DAKO, #M0744) for 30 min and with FITC-conjugated goat anti-mouse IgG (1%; Jackson ImmunoResearch, #115–095–003) for 20 min at room temperature. Subsequently cells were stained with 10 µg/ml propidium iodide and analyzed by flow cytometry.

Apoptosis/cell death

To assess the effect of sorafenib on cell survival, treated and control cultures were double stained with Annexin V-FITC and propidium iodide (20 µg/ml) (Annexin V – FITC Apoptosis Detection kit; eBioscience, #BMS500FI), following the manifacture guidelines. Each sample was analyzed by flow cytometry with a FACScalibur (BD Bioscience) recording 10,000 events per sample.

Clonogenic assay

To verify the effect of sorafenib on stemness of GBM TIC cultures, the ability of single cells to form a colony was tested. Untreated or sorafenib-treated (48 h, using IC₅₀ concentration for each culture) cells were seeded out in 96-well microplates with a density of 0.8 alive cell/well. The day after plating, cells were observed under light transmitted microscope, and wells containing no alive cells or more than one cell were excluded. Cells were maintained in complete medium for a period of time of 2 wk to allow the growth of colonies. Then the dimension of each colony was determined counting the cells in each well by microscope inspection. A cut off of 100 cells was chosen to discriminate between clusters derived from committed cells from clones originated from clonogenic cells. This value was chosen, because it showed the better correlation with the tumorigenic ability of GBM TICs. After three weeks, cell content of clones was measured by MTT assay, as previously described. An absorbance threshold value of 0.4 was used to discriminate true clones from clusters.

Immunofluorescence

To determine the expression of different markers of stemness and differentiation in GBM cultures, untreated or sorafenib-treated cells were plated onto matrigel-coated glass coverslips and then fixed with 4% paraformaldehyde for 15 min at r.t. Cells were incubated with primary antibodies as follows: mouse anti-human nestin (1:10.000; Abcam, #ab6329), rabbit anti-Olig2 (1:500; Millipore, #AB9610), rabbit anti-GFAP (1:10:000; DAKO, #Z0334) and mouse anti-MAP2 (1:1.000; Chemicon International, #MAB378). Immunocomplexes were detected with secondary fluorescent antibodies as DyLight 488 goat anti-mouse IgG and DyLight 459-goat anti-Rabbit IgG (Jackson ImmunoResearch, #111–485–003 and #111–505–003). Cells were counterstained with Hoechst 33342 dye (Sigma-Aldrich, #14533) to identify all nuclei. Images were acquired by automated Zeiss AxioImager M2 equipped with an Axiocam MRM (Zeiss). Results are showed as percentage of stained cells from randomly selected fields.

Western blot

Cells were cultured for 24 h in complete medium before being treated. Cells were lysed and total proteins were isolated and measured as standard methods.⁵⁴ Primary antibodies (all from Cell Signaling Technology): against phospho-Akt (Ser-473, #9271), phospho-ERK1/2 (#9101), phospho-STAT-3 (#9134), phospho-MEK (#9121), Mcl-1 (#5453), total Akt (#9272), total ERK1/2 (#9102), total MEK (#9122) and α-tubulin (Sigma Aldrich, #T5168) and Sox2 (Millipore, #AB5603). Secondary ECL mouse (#GEHNA931) and rabbit (#GEHNA934) HRP-linked whole antibodies were from GE Healthcare. Immunocomplexes were detected using chemiluminescence system according to manufacturer's instructions (Immobilon, Millipore, #WBKLS0500). Densitometric analysis of resulting bands was performed on a Chemi-Doc system by the Image Lab Software (Bio-Rad Laboratories).

Statistical analysis

Experiments were repeated at least three times and statistical significance determined by two-tailed t test. P values lower than 0.05 were considered statistically significant.