Dasatinib-induced autophagy is enhanced in combination with temozolomide in glioma

Glioblastoma is the most common and malignant astrocytic glioma, carrying the WHO designation of grade IV out of four grades (1, 2). Glioblastoma is histologically defined by its marked vascular proliferation and necrosis and shows high cellularity, mitotic activity, and aggressive invasion into normal brain. Molecularly, these de novo neoplasms are characterized by epidermal growth factor (EGF) receptor (EGFR), platelet-derived growth factor receptor, vascular endothelial growth factor receptor, and cyclin-dependent kinase 4 mutations and amplifications; additionally, they exhibit a high frequency of PTEN dysfunction (3). Despite maximal tumor resection, radiation, and chemotherapy, the high degree of infiltration of glioblastoma leads to 100% recurrence and an average life expectancy of only 12 to 14 months (4–6).

BMS-354825 (dasatinib) is an ATP-competitive small-molecule inhibitor effective in combating malignancies characterized by drug-resistant mutants of BCR-ABL, KIT, and EGFR by interfering with aberrantly activated phosphorylation sites (7–10). At low nanomolar concentrations, dasatinib inhibits progression of chronic myelogenous leukemia, pancreatic adenocarcinoma, colon cancer, prostate cancer, non-small cell lung cancer, and breast malignancies (10–15). In glioblastoma, we aimed to target the SRC nonreceptor tyrosine kinase. SRC family kinases include eight homologous members (SRC, YES, FYN, LYN, LCK, HCK, FGR, and BLK) that are implicated in numerous oncogenic pathways (14, 16–18). SRC-type family members exhibit two primary phosphorylation sites: an activating phosphorylation at Tyr⁴¹⁹ and a deactivating phosphorylation at Tyr⁵²⁷ (19, 20). Oncogenesis has been found to occur when inactive and active forms of SRC kinases are in disequilibrium. This imbalance may take place either when SRC itself is mutated or when proteins that regulate SRC, such as the activating agents PTPα and SHP1 and the deactivating agents CSK and CHK, are abnormal or overexpressed (21). Several SRC family members have been shown to be activated in glioblastoma, including the malignant phenotype-promoting members LYN and FYN (22, 23).

SRC mediates many critical proteins that when dysregulated contribute highly to proliferation, invasion, and vasculogenesis through interconnected signaling cascades. These proteins include focal adhesion kinase (FAK), which colocalizes with p130CAS, integrin α_vβ₃, and paxillin to form focal adhesions, and AKT, which is a downstream effector of the phosphatidylinositol 3-kinase (PI3K) pathway (24–28). In addition, SRC appears to activate the mitogen-activated protein kinase (p44/42 MAPK) in cooperation with Grb2/PI3K regulation via activation by platelet-derived growth factor receptor, ultimately instigating tumorigenic transcription and gene expression (29, 30).

We studied dasatinib as a known inhibitor of SRC activation in glioblastoma with emphasis on inhibiting the SRC-activating phosphorylation at Tyr⁴¹⁹, a site that has been implicated in the autophosphorylation of FAK at Tyr³⁹⁷ (28). We show the inhibitory effect of dasatinib on tumor proliferation in multiple glioma cell lines via the blockage of SRC, AKT, and ribosomal protein S6 (rpS6) activation. Dasatinib also inhibited glioma invasion in a Matrigel invasion assay by preventing FAK autophosphorylation and focal adhesion formation. Furthermore, we show that dasatinib has the capacity to initiate type II (autophagic) programmed cell death without inducing apoptosis. Finally, we show that dasatinib and temozolomide combine synergistically in the inhibition of cell proliferation.

To assess the effect of dasatinib in human gliomas, we used the LN18, U251, U87/EGFR (U87 cells stably expressing wild-type EGFR), and U87/EGFR PTEN (U87 cells stably expressing both wild-type EGFR and functional PTEN; both U87 cell lines were a generous gift of Dr. Paul Mischel, University of California at Los Angeles) human glioblastoma cell lines. The LN18 cell line with stable lentiviral vector-mediated knockdown of PTEN (LN18 shPTEN) was a gift from Dr. Ta-Jen Liu (The University of Texas M. D. Anderson Cancer Center). All cell lines were maintained as monolayer cultures in DMEM/F-12 supplemented with 10% fetal bovine serum (Life Technologies) and 1.0% penicillin/streptomycin and were incubated in an environment of 10% CO₂/90% air at 37°C.

Dasatinib was dissolved in DMSO (Sigma) to a concentration of 20 mmol/L. The solution was stored at −20°C and further diluted as needed to an appropriate final concentration in serum-containing DMEM/F-12 (Life Technologies). Chemotherapeutic sources: temozolomide (Schering-Plough), irinotecan (Pharmacia & Upjohn), and carboplatin (Sigma). 3-Methyladenine (3-MA; Sigma), an inhibitor of autophagosome formation, was dissolved in medium containing 10% fetal bovine serum.

The antiproliferative activity of dasatinib on cells growing in culture was quantified using a sulforhodamine B (SRB)-based assay (31). Cells were seeded into 96-well plates at 3 × 10³ per well and allowed to adhere overnight in medium containing 10% fetal bovine serum. The cells were maintained in standard culture conditions after treatment with dasatinib, and cell growth was measured by fixing the cells with 10% TCA and subsequent staining with SRB. The wells were read at 530 and 620 nm using spectrophotometric analysis. Drug interactions were determined using the CalcuSyn Software (Biosoft). This program uses the median effect equation to determine the median dose effect, Dm, which is analogous to the IC₅₀. Synergy or antagonism [quantified by the combination index (CI)] is based on the multiple drug effect equation where CI = 1 indicates an additive effect, CI < 1 indicates synergy, and CI > 1 implies antagonism (32).

Invasion of glioma cells in vitro was measured by the invasion of cells through Matrigel-coated 24-well Transwell inserts. Cells (1 × 10⁶) were treated with 50, 100, and 200 nmol/L dasatinib for 24 h in 10-cm plates and washed with serum-free medium. Cells (2.5 × 10⁵) were then transferred to each Transwell in duplicate, and the cells were allowed to invade for 24 h at 37°C. Invading cells were stained with 0.1% crystal violet in methanol and quantified by dissolving the Transwell inserts in a 2.0% sodium deoxycholate solution, which was subsequently analyzed spectrophotometrically at 595 nm.

U251, U87/EGFR, and LN18 cells were plated on glass slides in 6-well culture plates at a concentration of 2 × 10⁵ per well for 24 h and subsequently treated with dasatinib for an additional 24 h. Thereafter, the cells were fixed with a 3.7% formaldehyde solution in PBS and then permeabilized with 1.0% Triton X-100 in PBS. Paxillin was visualized by incubating first with mouse anti-paxillin monoclonal antibody (BD Biosciences) and then with Texas red-conjugated goat antibody to mouse IgG (Invitrogen). Actin filaments were stained with phalloidin conjugated with Alexa Fluor 488 (Invitrogen). Nuclei were stained with 5.0 ng/mL Hoechst 33258 for 15 min at room temperature. Pictures were taken under a Zeiss Axioskop 40 microscope equipped with the appropriate filters and aid of AxioVision 4.4 software.

To determine dasatinib-mediated cell signal changes, we evaluated the effect of dasatinib on phosphorylation of SRC, FAK, MAPK, AKT, and rpS6 in U251 human glioma cells. Cells were plated onto 10-cm plates at a concentration of 2 × 10⁶ and then incubated for 24 h. The next day, cells were treated for the indicated length of time with varying concentrations of dasatinib diluted in fresh DMEM. To assess the activity of dasatinib in cell lines with and without expression of the tumor suppressor PTEN, the effect of dasatinib was evaluated in the context of inhibiting EGF-stimulated tyrosine phosphorylation of EGFR, SRC, FAK, MAPK, AKT, and rpS6. Cells (2 × 10⁶) were seeded onto 10-cm plates in DMEM, serum starved for 24 h, and then treated with various concentrations of dasatinib in serum-free medium for 8 h. The cells were further treated with 50 ng/mL human recombinant EGF (Upstate Biotechnology) for 15 min. Protein (30 μg) was electrophoresed on a 10% SDS-PAGE gel and transferred to an Immuno-Blot polyvinylidene difluoride membrane (Bio-Rad). Membranes were then probed with primary antibodies: phospho-EGFR (Tyr¹¹⁷³), phospho-SRC (Tyr⁴¹⁹), phospho-SRC (Tyr⁵²⁷), phospho-MAPK (Thr²⁰²/Tyr²⁰⁴), total MAPK, phospho-AKT (Thr³⁰⁸, Ser⁴⁷³), total AKT, and phospho-S6 (Ser²³⁵/Ser²³⁶, Ser²⁴⁰/Ser²⁴⁴; Cell Signaling); PTEN, total SRC, and total FAK (Santa Cruz Biotechnology); phospho-FAK (Tyr³⁹⁷; Zymed Invitrogen); EGFR Ab4 (Calbiochem); and β-actin (Sigma-Aldrich).

Glioma cells were plated at a density of 5 × 10⁵ per well in 10-cm plates and maintained in 10% fetal bovine serum medium overnight. The following day, the cells were treated with varying concentrations of dasatinib and temozolomide in 10% DMEM for the indicated length of time. Cells were harvested, washed with 1× PBS, and suspended in 100% ethanol at −20°C overnight. Cells were spun out of fix at 1,500 rpm for 5 min, and the pellets were stained with 1× propidium iodide and followed by treatment with 7 μg/mL RNase A. DNA content was then analyzed on a FACScan flow cytometer.

Apoptosis was quantified using the Annexin V-FLUOS Apoptosis kit (Roche) according to the manufacturer's instructions. Treated U87/EGFR, LN18, and U251 glioma cells were incubated in standard culture conditions with and without drug treatment for 72 h before staining and subsequent flow cytometric analysis. We also analyzed cells for evidence of apoptosis using a terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling kit per the manufacturer's instructions (Calbiochem).

To detect the formation of acidic vesicular organelles, cells were treated and then stained with 1.0 μg/mL acridine orange for 15 min. Cells were then trypsinized, collected in PBS, and analyzed using the FACScan flow cytometer.

U87 and U251 cells were transfected with a GFP-light chain 3 (LC3) vector (in pEGFP-C1) kindly provided by Dr. Oliver Bogler (The University of Texas M. D. Anderson Cancer Center) using the FuGENE 6 transfection reagent (Roche) according to the manufacturer's instructions. At 48 h, cells were plated on glass slides and treated with drug. After exposure to temozolomide, dasatinib, or the combination of these agents for 72 h, cells were fixed in 1% paraformaldehyde and analyzed by fluorescence microscopy.

Mean and SD were calculated for the biological effect of dasatinib in the different assays. Statistical significance between groups was compared using the Student's t test. One-sided tests were considered significant at a level of 0.05. All data were analyzed using the GraphPad Instat3 software (GraphPad Software).

The ability of dasatinib to inhibit glioma cell proliferation in vitro was evaluated with the aid of a SRB assay (31). LN18, U251, and U87/EGFR cells were treated with a wide range of dasatinib concentrations (0, 1.0, 10, 100, 500, and 1,000 nmol/L) to capture its IC₅₀ of antitumor activity. Cell viability was assessed after 72 h, and the results of SRB staining are shown (Fig. 1A). Dasatinib showed significant (P < 0.05) antiproliferative activity at low nanomolar concentrations in U251 and LN18 glioma cells, yielding an IC₅₀ at 210 and 420 nmol/L, respectively. U87 cells stably expressing wild-type EGFR were found to be more resistant to dasatinib and exhibited an IC₅₀ of 1.5 μmol/L.

To explore the mechanisms mediating the growth-inhibitory effect of dasatinib, we tested its ability to inhibit kinases important in mitogenic signaling and survival pathways, including SRC, p44/42 MAPK, AKT, and rpS6. U251 was selected as a representative cell line due to its IC₅₀ and mutant p53 and PTEN-deficient status. Cells were treated with 200 nmol/L dasatinib for 2, 4, 8, 24, and 48 h. Attached and floating cells were collected for protein and Western blotting analysis (Fig. 1B) and revealed a decrease in SRC Y419 phosphorylation at 8 h. After an initial decrease, there was a time-dependent increase in SRC Y527 phosphorylation reflecting an accumulation of SRC protein with phosphorylation of its negative regulatory tyrosine residue (33). Slight decreases in both serine and threonine phosphorylation of AKT were seen as early as 2 h, an effect that became more apparent by 8 h. Phosphorylation of the AKT Thr³⁰⁸ residue, however, exhibited some reactivation at 24 and 48 h, whereas Ser⁴⁷³ phosphorylation was not reinstated but did not reach the same level of attenuation as threonine phosphorylation. Similarly, phosphorylation of S6 was decreased by dasatinib treatment, with Ser²³⁵/Ser²³⁶ showing a more rapid response and subsequent recovery at 48 h, whereas Ser²⁴⁰/Ser²⁴⁴ phosphorylation declined more slowly but was sustained. The inhibition of S6 phosphorylation is important as a downstream indicator of transcriptional inhibition. In contrast to SRC, AKT, and S6, the phosphorylation of p44/42 MAPK showed an increase over the course of 24 h.

We further pursued a possible link between PTEN expression and sensitivity to dasatinib by analyzing the expression of endogenous PTEN in U251 and the syngeneic cell line pairs of U87/EGFR and U87/EGFR PTEN (courtesy of Dr. Paul Mischel; characterized previously; ref. 34) and LN18 and LN18 shPTEN. The Western blot shown in Fig. 2A shows a moderate and high level of PTEN expression in LN18 and U87/EGFR PTEN, respectively, compared with the diminished PTEN levels in LN18 shPTEN, U251, and U87/EGFR cells. With these data, it is possible to make a preliminary association between the PTEN status of a cell line and the effect of dasatinib (Fig. 2B). LN18, with an IC₂₅ of 22 nmol/L and an IC₅₀ of 420 nmol/L, is more sensitive to dasatinib in comparison with LN18 shPTEN, which was 3- to 5-fold more resistant with IC₂₅ and IC₅₀ values of 100 nmol/L and >1.5 μmol/L, respectively. The PTEN-negative U251 is less resistant than LN18, exhibiting an IC₂₅ of 9 nmol/L and an IC₅₀ of 210 nmol/L (Fig. 1A). U87/EGFR, the more resistant PTEN-negative cell line studied, displayed an IC₂₅ of 62 nmol/L and an IC₅₀ of 1.5 μmol/L. U87/EGFR became >2.5- to 8-fold more sensitive to dasatinib when PTEN was reintroduced (IC₂₅, 24 nmol/L; IC₅₀, 180 nmol/L) compared with U87/EGFR. The disparity of dasatinib sensitivity between U87/EGFR and U251 may be explained by the abundance of EGFR in the former cell line and genetic heterogeneity and may consequently not be a reflection of PTEN levels alone.

To determine whether mitogenic signaling via AKT was affected by PTEN expression status, we stimulated U87/EGFR and U87/EGFR PTEN cells with human recombinant EGF. A dose slightly lower than each cell line's respective IC₅₀ was used during the treatment phase: 1.0 μmol/L for U87/EGFR and 100 nmol/L for U87/EGFR PTEN. Western blotting (Fig. 2C) shows that dasatinib was able to maintain an inhibitory effect vis-à-vis AKT phosphorylation and transcription-activating rpS6 phosphorylation at Ser²⁴⁰/Ser²⁴⁴ (but not Ser²³⁵/Ser²³⁶) in serum-free medium and when EGFR was activated. Additionally, the same inhibitory effect was achieved with lower concentrations of dasatinib in PTEN-containing cells (U87/EGFR PTEN) as in PTEN-deficient cells (U87/EGFR PTEN) treated with a higher dose. Western blot analysis was done three times with similar results.

To determine the effect of dasatinib-induced SRC inhibition on glioma migration, we conducted a Matrigel Transwell migration assay. LN18 and U251 cells were treated with 0, 50, 100, and 200 nmol/L dasatinib for 24 h in serum-containing medium before being allowed to invade through Matrigel-coated membranes in serum-free medium. No change in cell viability was seen following 24 h treatment with dasatinib according to a previously conducted SRB assay. Dasatinib suppressed the migration of glioma cells across the Matrigel-coated Transwells: quantification of the stained membranes exposes a dose-dependent decrease in invasion that begins at treatment with 50 nmol/L dasatinib. Indeed, a ∼50% decrease in invasion was observed at 50 nmol/L in both cell lines, with LN18 being slightly less sensitive (Fig. 3A).

Because SRC affects autophosphorylation of FAK at Tyr³⁹⁷, promoting the localization of FAK, paxillin, and α_vβ₃ integrin to focal adhesions, immunofluorescence of paxillin was carried out to visualize the extent of focal adhesion formation in U251 glioma cells treated with 200 nmol/L dasatinib. Untreated cells (Fig. 3B) displayed focal adhesions staining positive for both actin and paxillin. Treatment with dasatinib significantly diminished the presence of these podia, and paxillin was observed to be diffuse throughout the cytoplasm; additionally, the dasatinib-treated cells adopted a more elongated morphology. Quantification of focal adhesions in untreated and treated U251 and LN18 cells (Fig. 3C) further underscores the extent of focal adhesion disruption in glioma cells.

In addition to SRC, which plays an important role in migration and invasion, activation of FAK Y397 is also integral to focal adhesion formation and cellular movement. LN18 and U251 glioma cells were treated with dasatinib for 24 h and cell lysate was obtained and analyzed by Western blotting. Figure 3D displays a dose-dependent decrease in FAK autophosphorylation beginning at 100 nmol/L in both cell lines. As suggested by the invasion assay, LN18 cells displayed more persistent FAK phosphorylation than U251 cells. In these glioma cells, inhibition of Transwell migration appears to be mediated to a large extent by inhibition of SRC, further augmented at higher dasatinib concentrations by concurrent inhibition of FAK.

Because dasatinib was found to be effective against cell proliferation, we appraised the effect of dasatinib on the cell cycle (Fig. 4A). U87/EGFR and LN18 cells were treated with 0, 100, 200, and 500 nmol/L dasatinib for 24, 48, and 72 h. Fluorescence-activated cell sorting analysis showed minimal increase in the proportion of cells in the G₁ phase at 72 h (<10%), and the G₂-M- and S-phase cell numbers were similar between treated and untreated cells (see Supplementary Fig. S1 for tabulated results).¹

Temozolomide is an alkylating agent used as the standard-of-care treatment of glioblastoma and is known to induce G₂ cell cycle arrest. To examine the effect of combining dasatinib with temozolomide, we performed fluorescence-activated cell sorting analysis on LN18, U251, and U87/EGFR cell lines treated for 24, 48, and 72 h with 200 nmol/L dasatinib and the varying concentrations of temozolomide to examine the extent of G₁ and G₂-M cell cycle change. Figure 4B shows that dasatinib induced minimal G₁ accumulation but temozolomide monotherapy induced a significant G₂-M arrest. The combination-treated cells exhibited a more equal accumulation in both G₁ and G₂-M. Additionally, a negligible decrease in the S-phase population was apparent. Little to no sub-G₁ cells were present compared with the total cell number.

Because little or no increase in the sub-G₁ cell fraction was observed, we evaluated the effect of these agents alone and in combination on the induction of apoptosis. A lack of apoptosis was confirmed by the absence of Annexin V staining (Fig. 4C), which was corroborated by a lack of terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling positivity in treated cells (Fig. 4D). This prompted the notion that dasatinib may trigger cell death by an alternate mechanism such as autophagy.

Considering the absence of apoptotic cell populations, we evaluated the ability of dasatinib to induce autophagic cell death. In addition, as temozolomide also induces autophagy (35), we studied these agents in combination to determine whether this possibly shared mechanism of action resulted in synergistic antitumor activity. U87/EGFR and U251 cells were treated with 200 nmol/L dasatinib and 50 and 5 μmol/L temozolomide, respectively, for 48 and 72 h. Following treatment, the cells were stained with 1.0 mg/mL acridine orange, which detects the formation of acidic cellular vesicles, a hallmark of autophagy (but not exclusively so; refs. 35, 36; Fig. 5A). U87/EGFR cells showed an increased percentage of cells stained with acridine orange when treated for 48 h with 200 nmol/L dasatinib and 50 μmol/L temozolomide as monotherapies; this effect was intensified when the two drugs were used in tandem. However, U87/EGFR cells were not as sensitive as U251 cells, which displayed approximately twice as much response when treated with dasatinib alone. This may be because the IC₅₀ of dasatinib for U251 is much lower than that of U87/EGFR. Additionally, U87/EGFR cells were more sensitive to temozolomide than U251 cells; this may be due to the higher dose of temozolomide (50 μmol/L) needed to compensate for the relatively low dose of dasatinib. Nevertheless, an additive effect of the two drugs was revealed by increased acridine orange staining in both cell lines. The addition of 2.0 mmol/L 3-MA, a PI3K inhibitor that disrupts the formation of autophagosomes (37), after 24 h of dasatinib and temozolomide treatment partially inhibited the increase in acridine orange staining. The same inhibition by 3-MA was seen when cells were treated for 72 h (data not shown).

To further investigate the possible autophagic effect, we analyzed cells treated with dasatinib and temozolomide for the presence and processing of microtubule-associated protein 1 LC3. LC3 is the first known mammalian protein that specifically associates with the autophagosomal membrane. An increase in expression of LC3-I and processing of LC3-I to the lower molecular weight form LC3-II is observed in autophagic (but not apoptotic) cells (38). GFP-LC3 has recently been shown to correlate with the induction of autophagy (38) and we used epifluorescence to further substantiate the idea that dasatinib induces glioma cell autophagy. In Fig. 5B, we show that both dasatinib and temozolomide are individually able to increase in the number and intensity of punctate LC3 fluorescence, representing autophagic vacuoles. Combination treatment with dasatinib and temozolomide further increased the number and intensity of punctuate LC3 fluorescence staining.

Western blotting of U87/EGFR cell lysates (Fig. 5B) showed an increase in LC3-I levels 24 and 48 h after treatment with dasatinib and temozolomide alone. There was an increase in processing of LC3-I to the lower molecular weight form LC3-II in U87/EGFR treated with dasatinib when comparing 24 with 48 h. Expression of LC3-I was elevated when dasatinib and temozolomide were applied to cells in combination for 48 h, more so than when cells were treated only with dasatinib or temozolomide. This enhancement of activity was confirmed by densitometry using Image J software (NIH; Fig. 5B). In contrast, U251 cells did not display a LC3-I signal increase until 48 h when treated with dasatinib alone, and temozolomide induced only a slight increase in LC3-I levels at 24 and 48 h. Regardless, LC3-I expression was more elevated when U251 cells were subjected to combination therapy than when treated individually with dasatinib or temozolomide. Conjointly, the expression of LC3 observed in both cell lines in response to treatment with dasatinib and temozolomide was successfully attenuated with the addition of 2.0 mmol/L 3-MA 24 h into treatment consistent with abrogation of autophagy. Several processes may have affected the accumulation of LC3-II in the U251 cell line experiments including the observation that LC3-II is degraded in cells undergoing autophagy (39) and that LC3-II may be converted back to LC3-I in late stages of autophagy (38). The combination of data obtained from acridine orange staining, LC3 epifluorescence, and LC3 Western blot analyses showing induction of autophagy and subsequent inhibition of autophagy after treatment with 3-MA emphasizes the ability of dasatinib to induce autophagy alone and in combination with temozolomide.

In view of the shared mechanism of action and the apparent additive effect at the molecular level between dasatinib and temozolomide, we examined the combination effect on cellular proliferation using a 72 h SRB growth assay (Fig. 5C). LN18, U251, and U87/EGFR cells were treated with a base concentration of 200 nmol/L dasatinib and varying concentrations of temozolomide ranging from 5 to 250 μmol/L. The result of the combination of dasatinib and temozolomide revealed synergistic cell growth inhibition (P < 0.05). Synergy was quantified by calculating a CI value for each data point, with values < 1 indicating synergism. With the exception of the 5 and 10 μmol/L combinations in LN18, all values obtained from the CI calculation indicated synergy. More dramatic, the IC₅₀ of temozolomide in combination with 200 nmol/L dasatinib was markedly lower than the IC₅₀ of temozolomide alone. The resultant IC₅₀ values of temozolomide for LN18, U87/EGFR, and U251 were ∼25.0, 50.0, and <5.0 μmol/L, respectively. These IC₅₀ values represent a dramatic increase in sensitivity to temozolomide when combined with dasatinib than when used alone.

In addition to temozolomide, we tested the ability of two additional cytotoxic chemotherapy agents to augment the antiproliferative effect of dasatinib in glioma. The SRB assay was done using a fixed dose of dasatinib (200 nmol/L) in combination with increasing doses of carboplatin (2.5-20 μg/mL) and irinotecan (2.5-20 μg/mL), agents commonly used in patients with glioblastoma. Although there was a reduction in the IC₅₀ and IC₇₅ for both U87/EGFR and U251 cells lines for both combination treatments, only U251 cells treated with dasatinib and irinotecan showed a >5-fold increase in sensitivity compared with irinotecan treatment alone (Fig. 5D). Thus, combining dasatinib with chemotherapy agents exhibiting proapoptotic mechanisms of action (carboplatin and irinotecan) may be not as effective in inhibiting glioma proliferation compared with combining dasatinib and temozolomide.

The invasive and proliferative nature of glioblastoma has been one of the impetuses for the testing of targeted therapy in this disease. Dasatinib (BMS-354825) is an orally bioavailable tyrosine kinase inhibitor, which shows exceptionally promiscuous inhibitive activity in multiple endothelial and solid tumors (7). In this report, we show that dasatinib is effective in glioblastoma by inhibiting tyrosine kinases such as SRC and EGFR, leading to the inhibition of proproliferative serine/threonine phosphorylation of AKT and rpS6, downstream indicators of PI3K and p70S6K activation, respectively. Additionally, we examined dasatinib sensitivity as a function of PTEN status and found that cells expressing functional PTEN were more sensitive than PTEN-deficient cells. In vitro proliferation and invasion of glioma cells were greatly reduced by treatment with dasatinib, with effective doses reaching low nanomolar, therapeutic concentrations. No apoptosis was observed; however, treatment with dasatinib induced considerable levels of autophagy in glioma cells. This type II programmed cell death was further magnified by combining dasatinib with low micromolar concentrations of temozolomide, resulting in a significant augmentation of sensitivity to temozolomide. Less significant inhibition of cell proliferation was observed when dasatinib was combined with other cytotoxic chemotherapies such as carboplatin and irinotecan.

The PI3K pathway, which is up-regulated by PTEN deficiency and amplification of EGFR, has been heavily implicated in the genesis and progression of glioblastoma. AKT is a critical downstream mediator of PI3K, and phospho-PI3K and phospho-AKT levels correlate with more aggressive gliomas (40). A similar relationship among phospho-PI3K, p70S6K, and rpS6 has been described (37). High levels of these activated proteins have exhibited an inverse association with cleaved caspase-3 detection, intimating the suppression of apoptosis via PI3K. We found that AKT and rpS6 phosphorylation in U251 was decreased beginning at 8 and 4 h, respectively, when treated with the IC₅₀ of dasatinib. This same inhibitory effect was observed when doses below the IC₅₀ were employed for a period of 24 h, indicating the potency of dasatinib with respect to inhibiting the PI3K pathway. We also examined the activity of dasatinib against cells with wild-type and dysfunctional PTEN and in cells with a knockdown of PTEN (LN18 shPTEN). Although growth assays revealed that PTEN-competent cells were more sensitive to dasatinib than PTEN-deficient cells, the inhibitor continued to decrease activated SRC, AKT, and rpS6 in these cells at low concentrations. This effect was sustained even when cells were stimulated with EGF. Furthermore, activation of EGFR itself was curtailed, signifying either direct inhibition of EGFR phosphorylation or indirect inhibition through decreased cyclic SRC-EGFR phosphorylation (24). The increased sensitivity of LN18 and U87/EGFR PTEN cells may indicate an additive effect of PTEN phosphatase activity vis-à-vis AKT phosphorylation and indirect inhibition by dasatinib through SRC. It is also possible that PTEN is managing the activity of FAK, a known regulator of the cell cycle, whereas dasatinib simultaneously inhibits autophosphorylation of FAK through inhibition of SRC (41, 42).

The role of SRC in invasion and the targeting of SRC as a mechanism of inhibiting cellular invasion have been described in glioblastoma (18, 21, 22, 28, 43–45). Dasatinib abrogated Transwell migration in vitro in a dose-dependent manner at low nanomolar concentrations that effected little or no alteration in FAK activation, suggesting that this inhibition of cell migration is mainly due to SRC inhibition with additional augmentation by FAK inactivation at higher dasatinib concentrations. Our findings corroborate invasion studies conducted in various other malignancies (10, 12). Moreover, immunofluorescence studies indicated that dasatinib was able to attenuate colocalization of paxillin and actin to focal adhesions. Interestingly, PTEN-positive LN18 cells were slightly more resistant to dasatinib in the context of invasion than PTEN-defunct U251 cells, a finding that was intimated by both immunofluorescence and analysis of autophosphorylated FAK levels. This suggests that, although SRC is able to mediate invasion through FAK, most of the aforementioned PTEN-dasatinib additive effect may occur through the PI3K/AKT/p70S6K pathway in glioblastoma instead of through FAK.

Cell cycle analysis was done to assay for the presence of a sub-G₁ cell population. Very little G₁ arrest and no sub-G₁ population were observed in glioma cells treated with dasatinib. We confirmed the absence of apoptosis with Annexin V and terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling analyses. This finding is contrary to those of Johnson et al. (46) and Shor et al. (12), which detail the ability of dasatinib to induce apoptosis in squamous cell carcinoma and bone sarcoma cells, respectively. Intrigued, we investigated the possibility of dasatinib inducing type II programmed cell death or autophagy, a process through which intracellular vesicles sequester proteins and organelles for cellular recycling (47). By staining cells with acridine orange and examining LC3 epifluorescence and protein processing, we found that nanomolar concentrations of dasatinib significantly increase autophagic cell death in glioblastoma, an effect that has not been reported previously. Dasatinib-induced autophagy in glioma cells may offer a possible explanation for the increase in phospho-MAPK levels seen over the course of 48 h, because a similar increase in MAPK activity has been reported in HT-29 colon cancer cells (48). Unfortunately, many mechanisms of autophagy remain to be elucidated, truncating our understanding of the mechanisms by which dasatinib induces autophagy.

The alkylating agent temozolomide is known to induce G₂-M arrest and remains the front-line standard-of-care treatment for patients with glioblastoma. Temozolomide induces autophagy in glioma cells at doses >100 μmol/L (35). When combined with dasatinib, autophagy was induced at significantly lower doses of temozolomide. These two drugs were highly additive in combination: dasatinib significantly decreased the amount of temozolomide needed to achieve a 50% decrease in glioma cell proliferation and displayed an additive effect in combination with temozolomide with respect to induction of autophagy. Surprisingly, dasatinib combined with carboplatin or irinotecan did not lead to a significant decrease in cell proliferation or a dramatic decrease in the chemotherapy dose required to reach the IC₅₀, suggesting that perhaps using two agents that induce autophagy increases the therapeutic effect.

Advances in targeted molecular therapies are serving to more efficiently mitigate tumorigenic invasion, proliferation, and survival. A dasatinib monotherapy phase II trial for patients with recurrent glioblastoma (RTOG-0627) is ongoing, and we presented compelling data for the screening of tumors for SRC, EGFR, FAK, AKT, and PTEN expression to potentially predict response to this agent. Because some tumors and cell types have been shown to be refractory to temozolomide monotherapy, combination therapy with dasatinib may be a viable clinical option for resistant tumors (49, 50). Dasatinib, characterized here as an anti-invasive cytostatic agent in glioblastoma, may also be successfully combined with radiation therapy or antiangiogenic therapy, which has been recently shown to increase tumor invasion (51). We have initiated a study evaluating the efficacy of dasatinib in combination with temozolomide and radiation therapy in patients with newly diagnosed glioblastoma, and tumor screening will be critical in this study in determining the most important molecular predictors of outcome.

No potential conflicts of interest were disclosed.

We thank Dr. William Bornmann for generously providing the dasatinib used in these studies.