Sunitinib, an inhibitor of kinases, including VEGFR and platelet-derived growth factor receptor (PDGFR), efficiently induces apoptosis in vitro in glioblastoma (GBM) cells, but does not show any survival benefit in vivo. One detrimental aspect of current in vitro models is that they do not take into account the contribution of extrinsic factors to the cellular response to drug treatment. Here, we studied the effects of substrate properties including elasticity, dimensionality, and matrix composition on the response of GBM stem-like cells (GSC) to chemotherapeutic agents. Thirty-seven cell cultures, including GSCs, parenchymal GBM cells, and GBM cell lines, were treated with nine antitumor compounds. Contrary to the expected chemoresistance of GSCs, these cells were more sensitive to most agents than GBM parenchymal cells or GBM cell lines cultured on flat (two-dimensional; 2D) plastic or collagen-coated surfaces. However, GSCs cultured in collagen-based three-dimensional (3D) environments increased their resistance, particularly to receptor tyrosine kinase inhibitors, such as sunitinib, BIBF1120, and imatinib. Differences in substrate rigidity or matrix components did not modify the response of GSCs to the inhibitors. Moreover, the MEK–ERK and PI3K–Akt pathways, but not PDGFR, mediate at least in part, this dimensionality-dependent chemoresistance. These findings suggest that survival of GSCs on 2D substrates, but not in a 3D environment, relies on kinases that can be efficiently targeted by sunitinib-like inhibitors. Overall, our data may help explain the lack of correlation between in vitro and in vivo models used to study the therapeutic potential of kinase inhibitors, and provide a rationale for developing more robust drug screening models. Mol Cancer Ther; 13(6); 1664–72. ©2014 AACR.

Several strategies have emerged to attempt to inhibit chemoresistance, but the fact remains that resistance is a problem for every effective anticancer drug (1). One of the most challenging problems faced by cancer researchers is the lack of correlation between in vitro cell lines and animal tumor models and human in vivo tumors (2). In line with this, sunitinib, a direct inhibitor of the tyrosine-kinase activities of VEGFR, platelet-derived growth factor receptor (PDGFR), and other related kinases (3), has been shown to induce apoptosis in vitro in glioblastoma (GBM) cells. However, it did not show any survival benefit in GBM xenograft models and its clinical success against GBM is not clear yet (4). A number of cell culture platforms have been developed in recent years to improve predictivity and further to elucidate the mechanisms governing the differing responses observed in vitro versus in vivo. Despite their importance for drug testing, in vitro methods are beset by pitfalls and inherent limitations (5). However, given that it is practically impossible to test large quantities of new anticancer agents in vivo, cell culture conditions that mimic the niche of tumor cells need to be further investigated. It is anticipated that the development of in vitro systems representative of tumor microenvironment will lead to an increase in the quality of and reduction in the timelines and costs associated with screening drugs, and enhancement in efficacy information for regulatory decisions. Current in vitro models do not take into account the contribution of different extrinsic factors on the responsiveness of the cells to drug treatment (6). It has been shown for breast cancer cell lines, that culture within a three-dimensional (3D) environment decreased taxol-induced cell death (6). In addition, several lines of evidence indicate that extracellular matrix-derived biomechanical properties, including extracellular matrix components and matrix rigidity, regulate proliferation and migration of GBM cells (7, 8). GBM is a highly aggressive tumor that invades the surrounding tissue and leads to regrowth of a recurrent tumor, which is not significantly altered by radiation or chemotherapy (9). The identification of GBM stem-like cells (GSC), which are considered to be the cause of tumor formation and recurrence, provided a promising cellular target for more effective therapies against GBM (10). Although cancer stem-like cells are predicted to be more resistant to chemotherapy and radiation therapy than the other surrounding cancer cells, the susceptibility of GSCs to chemotherapeutic drugs is controversial as the existing literature presents conflicting experimental data (11). This is likely because the in vitro models used do not recapitulate extrinsic factors of the niche where GSCs reside, which are needed for these cells to maintain their stem cell features (12). Thus, a systematic study of the contribution of different environmental properties to the chemoresponse of GSCs is yet to come. That study will help develop robust in vitro models of GSCs and will be of clinical importance for the discovery of effective treatments against this deadly disease. In this work, we have analyzed the effects of biophysical and biochemical properties of the substrate on the response of GSCs to an array of chemotherapeutic agents used in the clinic or under clinical trials in patients with cancer.

Antitumor compounds and kinase inhibitors

The following compounds were used for treatment of GBM cells at the indicated concentrations. Carmustine, cisplatin, roscovitine (Sigma-Aldrich), bortezomib (Millennium Pharmaceuticals), flavopiridol (Santa Cruz Biotechnology), sorafenib, sunitinib (Toronto Research Chemicals), ABT-737, BIBF1120 (vargatef), Gleevec (imatinib; Selleck Chemicals), obatoclax (Cayman Chemical), PD98059, LY294002 (Calbiochem), and AG490 (Sigma-Aldrich).

Cell lines

Human GBM cell lines DBTRG, T98, U87, A172 (from American Type Culture Collection), and 8MGBA, 42MGBA, DKMG, GAMG, GMS10 (from DSMZ) were obtained from commercial providers within the last 2 years and no authentication was performed by the authors. Cells were cultured in Dulbecco's Modified Eagle Medium (DMEM; Life Technologies) in the presence of 10% (v/v) FBS (Lonza). Cell lines were seeded at 30,000 cells per cm2 for cytotoxicity analysis.

Primary cell cultures

Cell cultures from tissue specimens used in this study (GBM and mesenchymal stem-like cells) have been previously characterized and described by our group (8, 13, 14). The stem-like origin of the cells was tested by analyzing the patterns of differentiation just before their use (13, 14). GSCs were maintained in serum-free DMEM/F12 medium (Life Technologies) containing 20 ng/mL EGF, 20 ng/mL basic fibroblast growth factor (both from Sigma-Aldrich), 2 μg/mL heparin, and B-27 (20 μL/mL of medium; Invitrogen) and plated at a density of 3 × 106 live cells/60-mm plate. Primary neurospheres were enzimatically dissociated (acutase; Sigma-Aldrich) every 4 to 5 days to facilitate cell growth. To promote differentiation, neurospheres were cultured in the same medium but in the presence of 10% (v/v) FBS for 4 days (14). All primary cells were used at low (<20) passage number and seeded at 30,000 cells per cm2 for cytotoxicity analysis.

Cell culture on 3D matrices

Bovine skin collagen I (PureCol 5005-B, Nutacon BV, Leimuiden) was prepared at 1.7 mg/mL in HEPES, pH 7.3, buffer with serum- and phenol-free DMEM as described (8) and allowed to polymerize at 37°C for 2 hours. For glycation of collagen, collagen stock solutions were mixed with 0.5 mol/L D-ribose (Amresco) to a final concentration of 0, 50, and 200 mmol/L ribose in 0.1% (v/v) sterile acetic acid and incubated at 4°C for 5 days, as described (15). To study matrix composition, 1.7 mg/mL collagen I and 2.4 mg/mL of low- (29 kDa) or high- (1,540 kDa) molecular-weight hyaluronic acid (Sigma-Aldrich) dissolved in DMEM (pH 7.4) were mixed to weight ratios of 1:0, 1:2, and 1:8. The viscous solutions were incubated at 37°C overnight to allow the natural gelatin of the structure. Following polymerization, cells were seeded on top of matrix in medium, allowed to penetrate for 48 hours (more than 90% of cells within the matrix), and treated with the indicated compounds. Gel stiffness was measured using a rheometer (Anton Paar) and quantified in Pascal units (Pa).

Cell culture on 2D substrates

Collagen I solution (1.7 mg/mL) was added to culture plates and incubated for 16 hours at 4°C. Non-attached collagen was removed by washing before cell incubation. Polyacrylamide matrices of different rigidities were prepared by mixing acrylamide and bisacrylamide at different ratios in the presence of 10 mg/mL NHS cross-liker (Sigma-Aldrich) as previously described (7). All hydrogels were covalently functionalized with collagen (Sigma-Aldrich) at a nominal surface density of 1.4 μg/cm2. A rheometer (Anton Paar) was used to measure the macroscopic elastic shear modulus of each gel.

Immunofluorescence analysis

Cells were assayed for the expression of glial fibrillary acidic protein (GFAP) and Ki-67 by fluorescence microscopy as described previously (14). Briefly, cells were fixed in 3.7% (v/v) formaldehyde and permeabilized with 0.5% (v/v) Triton X-100. For immunostaining, cells were incubated overnight with 10 μg/mL rabbit anti-GFAP (DAKO) or anti-Ki-67 at a 1:250 dilution (Thermo Scientific). Then, Texas red-conjugated goat anti-rabbit secondary antibodies (Jackson ImmunoResearch) were used for fluorescence detection.

Cell nuclei were stained with 0.5 μg/mL Hoechst fluorochrome (Life Technologies) for 15 minutes.

When indicated, GSCs were stained with 5 μmol/L CFDA SE Cell Tracer, also called CFSE (Invitrogen) for 15 minutes and incorporated into a 3D collagen matrix as described above. Then, blocking antibodies against β1-integrin (1 μg/mL; BD Bioscience) were added to the matrix and 24 hours later, fluorescently labeled cells were visualized under a confocal microscope (Nikon A1R). Images were taken with a ×40 objective.

Cell death and apoptosis analyses

Quantitative analysis of cell viability was performed by using the Alamar blue bioassay (Life Technologies). To quantify apoptosis, cells were incubated with annexin V/7-ADD according to the manufacturer's instructions (Annexin red kit; Millipore). Samples were analyzed on a FACScan cytometer (BD Bioscience) or immunofluorescence. When indicated, apoptosis was determined by analyzing the digestion of PARP, a substrate of caspases, as previously described (16).

Gene silencing

Short hairpin RNA (shRNA)-containing plasmid (1 μg per 106 cells) specific for PDGFR-α or a control plasmid containing a scramble sequence (Origene) was transfected into GSCs by using a neural stem cell nucleofector kit as recommended by the manufacturer (Lonza). Following 24 hours of transfection, cells were seeded in a collagen matrix as already described.

Western blot analysis

Total protein extracts were separated on a 10% (w/v) SDS-polyacrylamide gel and transferred to nitrocellulose. Blots were blocked with 5% (w/v) bovine serum albumin and incubated overnight with antibodies against Stat3, pStat3, α-tubulin, PARP (all from Santa Cruz Biotechnologies), Akt, pAkt, extracellular signal-regulated kinase (ERK), or pERK (all from Cell Signaling Technology) at 1:500 or 1:250 dilutions, followed by incubation with IRDye-labeled anti-mouse or anti-rabbit antibodies at a 1:15000 dilution for 1 hours (LI-COR Biosciences). Bound antibody was detected by using the Odyssey Infrared Imaging System (LI-COR Biosciences. Gels were cropped with Adobe Photoshop software for clarity and conciseness purposes.

Quantitative RT-PCR

Total RNA was extracted by using a TRI Reagent solution (Sigma-Aldrich). To assess the expression of individual genes, a cDNA was generated and amplified by using primers for human GFAP and β-actin (8). Quantitative real-time PCR (RT-PCR) was performed in a 7000 Sequence Detection System (Life Technologies) as described (8).

Statistical analysis

All statistics were calculated with the SPSS statistical package (version 13.0). Data are presented as mean ± SD. Differences between groups were tested for statistical significance using the unpaired two-tailed Student t test. The significance level was set at P < 0.05.

Response of GBM cell lines to different chemotherapeutic agents

We first studied the sensitivity of nine different GBM cell lines, cultured on plastic surfaces, to genotoxic drugs (carmustine and cisplatin), inhibitors of antiapoptotic Bcl-2 family members (ABT-737 and obatoclax), cyclin-dependent kinase (CDK) inhibitors (roscovitine and flavopiridol), multikinase inhibitors (sorafenib and sunitinib), and the proteasome inhibitor bortezomib. Kinase and proteasome inhibitors were used at two concentrations either alone or in combination with the antiapoptotic inhibitors to enhance their killing efficacy (Fig. 1A). Carmustine and cisplatin were not lethal for almost any cell line even at concentrations of 30 μmol/L (cisplatin) and 120 μmol/L (carmustine), and similar results were obtained when cells were treated with Bcl-2 inhibitors obatoclax, a pan-Bcl inhibitor, and ABT-737 that inhibits Bcl-2, Bcl-x, and Bcl-w. Targeted therapies showed heterogeneity in treatment efficacy against GBM cell lines. Among the CDK inhibitors, 1 μmol/L flavopiridol was more effective than 15 μmol/L roscovitine. However, roscovitine drastically increased its killing activity when combined with ABT-737 or obatoclax in DKMG, GAMG, and GMS10 cells. The multikinase inhibitor sorafenib promoted a low level of cell death even at concentrations of 30 μmol/L (six of nine cell lines showed a viability higher than 50%), and did not significantly increased its cell death activity in the presence of pan-Bcl inhibitors. On the contrary, another receptor tyrosine kinase inhibitor, sunitinib, at 30 μmol/L was very effective as a single agent (six of nine cell lines showed a viability in the range of 0%–25%) and significantly increased its capacity to induce cell death in combination with obatoclax. Finally, bortezomib showed to be very effective as a single agent in GAMG and GMS10 cells and moderately increased its effectiveness in the presence of pan-Bcl inhibitors. From these data, we concluded that sunitinib was the most effective treatment when used as a single agent. We also determined that U87, GAMG, T98, and 42MGBA formed neurospheres when cultured in the absence of serum (Fig. 1B), similar to those formed by GBM stem-like cells. T98 and 42MGBA, which were two of the most resistant GBM cell lines, were cultured under serum-free conditions and treated with single agents (Fig. 1C). Cisplatin and ABT-737 were not effective in either cell line. However, the sensitivity to roscovitine, sorafenib, and sunitinib was clearly increased in the absence of serum in both cell lines. Taking into consideration that chemoresistance is a main feature of cancer stem-like cells, our results suggest that stem-like cell culture conditions in standard plastic surfaces may not be a reliable in vitro model to test compounds against this cell population. We also confirmed that sunitinib induced apoptotic cell death, as determined by measuring the proportion of Annexin V(+) and 7-Aminoactinomycin D (7-AAD)(−) cells following 12 hours of treatment (Supplementary Fig. S1A). Almost 90% of DKMG cells were apoptotic when treated with 10 μmol/L sunitinib. As apoptosis is mostly regulated by Bcl-2 family members, we analyzed the expression of antiapoptotic proteins Mcl-1, Bcl-2, and Bcl-x in all cell lines. Bcl-2 showed the biggest differences in expression among cell lines. However, there was not a clear correlation between the levels of Bcl-2 and the sensitivity to sunitinib in all cell lines (Supplementary Fig. S1B).

Figure 1.

Response of GBM cell lines to chemotherapy. A, GBM cell lines were cultured on plastic in the presence of different compounds and 48 hours later cell viability was determined by using the Alamar bioassay. A grey scale was used to represent cell viability. B, T98 cells maintained in the standard medium (with serum) or under stem-like cell conditions (serum-free). Note the presence of neurospheres in the absence of serum, a feature of GBM stem cells in culture. Scale bar, 50 μm. C, two chemoresistant GBM cell lines were incubated with or without serum and cell viability was analyzed 48 hours later as described above. Values represent the mean ± SD of three independent experiments. *, consistent significant differences (P < 0.01) between serum-supplemented and serum-free conditions. CIS, cisplatin; CAR, carmustine; ABT, ABT-737; OBA, obatoclax; ROS, roscovitine; FLA, flavopiridol; SOR, sorafenib; SUN, sunitinib; BOR, bortezomib.

Figure 1.

Response of GBM cell lines to chemotherapy. A, GBM cell lines were cultured on plastic in the presence of different compounds and 48 hours later cell viability was determined by using the Alamar bioassay. A grey scale was used to represent cell viability. B, T98 cells maintained in the standard medium (with serum) or under stem-like cell conditions (serum-free). Note the presence of neurospheres in the absence of serum, a feature of GBM stem cells in culture. Scale bar, 50 μm. C, two chemoresistant GBM cell lines were incubated with or without serum and cell viability was analyzed 48 hours later as described above. Values represent the mean ± SD of three independent experiments. *, consistent significant differences (P < 0.01) between serum-supplemented and serum-free conditions. CIS, cisplatin; CAR, carmustine; ABT, ABT-737; OBA, obatoclax; ROS, roscovitine; FLA, flavopiridol; SOR, sorafenib; SUN, sunitinib; BOR, bortezomib.

Close modal

GBM stem-like cells are sensitive to chemotherapeutic agents in a 2D environment

GSCs are considered to be responsible for tumor regrowth, in part, due to their resistance to chemotherapy. We studied the response of thirteen samples of GSCs to the same array of compounds used in cell lines, cultured on flat plastic surfaces. Interestingly, most inhibitors of target proteins (roscovitine, flavopiridol, sunitinib, and bortezomib) were lethal for GSCs as single agents (Fig. 2). Sunitinib alone promoted an almost complete death of the tumor cell population (only one culture showed a viability higher than 25%). Another multikinase inhibitor, sorafenib, was not effective when used alone but killed most cells (cell viability of 11 cultures below 25%) when it was combined with the pan-Bcl inhibitor obatoclax. On the contrary, carmustine, a nitrosourea used in the treatment of GBM, promoted a weak cell death response in 11 GSC cultures. We also studied mesenchymal stem cells (MSC) from bone marrow (two samples) and cells isolated from the bulk of GBM tumors (tumor parenchymal cells; 13 samples) and found that both cell types were clearly more resistant than GSCs to the same treatments (Fig. 2). Moreover, the most effective antitumor agent against GSCs, sunitinib, induced a heterogeneous response in tumor parenchymal cells, but only 1 of 13 samples showed a strong cell death response (below 25%).

Figure 2.

Response of stem-like and parenchymal GBM cells to chemotherapy. GSCs, parenchymal GBM cells and MSCs were treated with different compounds and cell viability was analyzed (see legend of Fig. 1).

Figure 2.

Response of stem-like and parenchymal GBM cells to chemotherapy. GSCs, parenchymal GBM cells and MSCs were treated with different compounds and cell viability was analyzed (see legend of Fig. 1).

Close modal

Biophysical features of the substrate in the response of GSCs to sunitinib

Very recently, it has been shown that 3D culture platforms induced alterations of markers of malignancy in a GBM cell line (17). We studied the decoupled effects of substrate properties such as elasticity, dimensionality, and matrix composition to the response of GSCs to chemotherapeutic agents using two-dimensional (2D) and 3D platforms (Fig. 3). Significant differences in viability between cells cultured in 2D and 3D substrates were obtained following treatment with cisplatin, flavopiridol, and sunitinib (Fig. 4A). Notably, we observed the highest difference in response to sunitinib. This multikinase inhibitor decreased cell viability to less than 5% on collagen-coated surfaces, whereas the proportion of viable cells in 3D collagen matrices was about 50%. Sunitinib-induced cell death was due to apoptosis as determined by analyzing the cleavage of PARP (Fig. 4B). Consistently, PARP cleavage was more evident in cells cultured on a 2D substrate than in cells incubated in a 3D matrix. To analyze whether this dimensionality-dependent response to sunitinib was specific for GSCs, we treated sunitinib-sensitive and sunitinib-resistant GBM parenchymal cells with increasing concentrations of the kinase inhibitor in 2D and 3D environments (Fig. 4C). In sensitive cells, viability decreased at similar rates both in 2D and 3D substrates, whereas in resistant cells a 3D environment provided a significant protection against sunitinib compared with 2D conditions only at a concentration three times higher than that used in GSCs. These results suggest that a 3D environment is more critical for the survival of GSCs than for the viability of GBM tumor cells. Nevertheless, the comparison of 2D with 3D models may be influenced by differences in stiffness. Thus, we studied the influence of substrate elasticity on the response of GSCs to sunitinib. As a 2D environment, we used a nonpermeable collagen-coated polyacrylamide hydrogel of different stiffness, from 1 kPa (approximately the elastic modulus of brain tissue) to 55 kPa (more than four times stiffer than skeletal muscle), and found that sunitinib was equally effective at killing cells at any of the elasticity coefficients (Fig. 5A). A 3D environment of different elastic conditions was created by glycation of collagen, which allows up to a 4-fold increase in the stiffness of collagen matrices without significant modification to the collagen architecture (15). Again, no significant differences in cell viability were observed between the different elasticity coefficients (Fig. 5B). Finally, to analyze the contribution of substrate composition, we used matrices made of collagen and hyaluronic acid at different ratios. As shown in Fig. 5C, even collagen: hyaluronic acid ratios of 1:8 did not significantly modify sunitinib-mediated cell death in comparison with collagen-only matrices. The same result was obtained using either low- or high-molecular-weight hyaluronic acid. Therefore, our data indicate that substrate stiffness and changes in weight ratios of collagen and hyaluronic acid do not modify the response of GSCs to sunitinib.

Figure 3.

Schematic representation of the different substrates used to culture GSCs. To evaluate dimensionality, cells were either cultured on collagen-coated plastic surfaces (2D) or embedded in a collagen matrix (3D). For stiffness, collagen-coated polyacrylamide gels (2D) or ribose-modified collagen matrices (3D) were used. For matrix composition, collagen and hyaluronic acid (HA) were mixed at different ratios.

Figure 3.

Schematic representation of the different substrates used to culture GSCs. To evaluate dimensionality, cells were either cultured on collagen-coated plastic surfaces (2D) or embedded in a collagen matrix (3D). For stiffness, collagen-coated polyacrylamide gels (2D) or ribose-modified collagen matrices (3D) were used. For matrix composition, collagen and hyaluronic acid (HA) were mixed at different ratios.

Close modal
Figure 4.

3D collagen matrices provide a survival advantage to GSCs treated with sunitinib. A, two GSC cultures were incubated on a collagen-coated surface (2D) or in a collagen matrix (3D) and then treated with the indicated compounds. Cell viability was determined by Alamar Blue 48 hours after treatment. B, apoptosis of sunitinib-treated GSCs was assessed by PARP cleavage products. α-Tubulin was included to assure equal loading. C, GBM parenchymal cells were treated with increasing concentrations of sunitinib and cell viability was determined. Histograms represent the mean ± SD of three independent experiments. *, significant differences (P < 0.01) between 2D and 3D conditions.

Figure 4.

3D collagen matrices provide a survival advantage to GSCs treated with sunitinib. A, two GSC cultures were incubated on a collagen-coated surface (2D) or in a collagen matrix (3D) and then treated with the indicated compounds. Cell viability was determined by Alamar Blue 48 hours after treatment. B, apoptosis of sunitinib-treated GSCs was assessed by PARP cleavage products. α-Tubulin was included to assure equal loading. C, GBM parenchymal cells were treated with increasing concentrations of sunitinib and cell viability was determined. Histograms represent the mean ± SD of three independent experiments. *, significant differences (P < 0.01) between 2D and 3D conditions.

Close modal
Figure 5.

Effect of stiffening and matrix composition on the response of GSCs to sunitinib. Two GSC cultures were incubated on collagen-coated polyacrylamide gels (A), or in ribose-cross-linked collagen matrices (B) of different rigidity, and in a matrix of collagen mixed with low-molecular-weight hyaluronic acid at different molar ratios (C). In all cases, cell viability was analyzed 48 hours after treatment. Histograms represent the mean ± SD of three independent experiments.

Figure 5.

Effect of stiffening and matrix composition on the response of GSCs to sunitinib. Two GSC cultures were incubated on collagen-coated polyacrylamide gels (A), or in ribose-cross-linked collagen matrices (B) of different rigidity, and in a matrix of collagen mixed with low-molecular-weight hyaluronic acid at different molar ratios (C). In all cases, cell viability was analyzed 48 hours after treatment. Histograms represent the mean ± SD of three independent experiments.

Close modal

MEK–ERK and PI3K–Akt pathways contribute to preserve the resistance of GSCs to sunitinib in 3D collagen matrices

To see whether dimensionality was able to modify biologic features other than cell survival, we cultured GSCs in 2D or 3D collagen substrates and determined the proliferation rate. Following 4 days of incubation, the number of proliferating cells was similar in both conditions as determined by Ki-67 immunofluorescence staining (Supplementary Fig. S2A and S2B) and by counting the number of cells over time (Supplementary Fig. S2C). We also showed that the differentiation pattern of GSCs was similar in both 2D and 3D conditions as determined by immunofluorescence staining of the astrocytic marker GFAP (Supplementary Fig. S2D) and quantitative analysis of the mRNA levels of GFAP (Supplementary Fig. S2E).

Next, we aimed at determining how GSCs sense their physical environment under 3D conditions. The major receptors of extracellular matrix are integrins. The β1-integrin subfamily mediates cell interactions with different extracellular components, including collagen, and it has been shown to regulate chemoresistance of cancer cells (18). Cells were incubated in 3D matrices in the presence or in the absence of a β1-integrin neutralizing antibody. As expected, treated cells lost their elongated shape and adopted a rounded morphology characteristic of cells poorly attached to the substrate (Fig. 6A). However, blockade of β1-integrin did not significantly modify the cell viability of GSCs after treatment with sunitinib (Fig. 6B).

Figure 6.

MEK and PI3K inhibitors revert the protective effect of a 3D environment. A, GSCs were cultured in a 3D collagen matrix in the presence of a blocking anti-integrin β1 antibody or an irrelevant control antibody. After 24 hours of treatment, CFSE-labeled cell attachment to the matrix was assessed by confocal microscopy. Scale bar, 10 μm. B, viability of cells treated with sunitinib in the presence or in the absence of anti-integrin β1. C, Ab, control antibody. C, phosphorylation levels of Akt, ERK, and Stat3 in GSCs treated with PI3K (LY294002), MEK (PD98059), and JAK2 (AG490) inhibitors were analyzed by Western blotting. D, the indicated inhibitors were added to GSC cultures in the presence of sunitinib, and cell viability was assessed 48 hours later. *, significant differences (P < 0.01) between treatment with sunitinib alone and sunitinib plus the inhibitor. E, GSCs were cultured with sunitinib or two other multikinase inhibitors, BIBF1120 (BIB) and imatinib (IMA), and cell viability was determined following 48 hours of treatment. *, significant differences (P < 0.01) between 2D and 3D conditions. Histograms represent the mean ± SD of three independent experiments.

Figure 6.

MEK and PI3K inhibitors revert the protective effect of a 3D environment. A, GSCs were cultured in a 3D collagen matrix in the presence of a blocking anti-integrin β1 antibody or an irrelevant control antibody. After 24 hours of treatment, CFSE-labeled cell attachment to the matrix was assessed by confocal microscopy. Scale bar, 10 μm. B, viability of cells treated with sunitinib in the presence or in the absence of anti-integrin β1. C, Ab, control antibody. C, phosphorylation levels of Akt, ERK, and Stat3 in GSCs treated with PI3K (LY294002), MEK (PD98059), and JAK2 (AG490) inhibitors were analyzed by Western blotting. D, the indicated inhibitors were added to GSC cultures in the presence of sunitinib, and cell viability was assessed 48 hours later. *, significant differences (P < 0.01) between treatment with sunitinib alone and sunitinib plus the inhibitor. E, GSCs were cultured with sunitinib or two other multikinase inhibitors, BIBF1120 (BIB) and imatinib (IMA), and cell viability was determined following 48 hours of treatment. *, significant differences (P < 0.01) between 2D and 3D conditions. Histograms represent the mean ± SD of three independent experiments.

Close modal

Activation of ERK1/2 has been shown to be associated with resistance to cytotoxic agents in cells exposed to collagen (19). We treated cells with 1 μmol/L PD98059, an inhibitor of MAP–ERK kinase (MEK)-1/2 which is an upstream activator of ERK1/2. Treatment with the inhibitor significantly decreased the phosphorylation of both ERK proteins although pERK1 showed a stronger (about 8-fold) reduction (Fig. 6C). Interestingly, although MEK1/2 inhibitor did not affect viability of GSCs, it drastically decreased (more than 3-fold) cell viability when cotreated with sunitinib (Fig. 6D). Adhesion to collagen may also activate the PI3K–Akt pathway (20). Thus, we inhibited the pathway by treating cells with 1 μmol/L LY294002 as assessed by downregulation of phosphorylated Akt (Fig. 6C), and found a clear reduction of cell viability only when combined with sunitinib (Fig. 6D). However, incubation of cells in the presence of 20 μmol/L AG490, an inhibitor of Stat3, which is a signal transducer relevant in many cancer models, did not modify the response to sunitinib (Fig. 6C and D). Treatment of GSCs with any of these inhibitors alone did not produce a significant reduction of cell viability (85%–100% viable cells) either in 2D or 3D systems (Fig. 6D and data not shown). Other inhibitors that share some kinase targets with sunitinib such as BIBF1120 (Vargatef) and imatinib (Gleevec; ref. 21), were also less effective at inducing cell death in 3D than in 2D substrates (Fig. 6E). All three inhibitors target PDGFR. Silencing with specific shRNAs reduced the expression of PDGFR-α (about 40%) but did not modify the levels of other related genes (Supplementary Fig. S3A). However, this reduction did not promote an increase in cell death (Supplementary Fig. S3B and S3C).

Cancer stem cells play a major role in sustaining tumor growth. Thus, targeting this rare cell population has become a challenge for researchers. However, the in vitro conditions used in most studies do not allow a good correlation between in vitro and in vivo models of chemotherapy response (4, 22, 23). Among the different compounds used in this work, sunitinib was the most effective against GBM cell lines and tumor parenchymal cells isolated from patients with GBM. This is in agreement with previous studies showing that sunitinib was the only receptor tyrosine kinase inhibitor, among eleven compounds, that could induce apoptosis in GBM cells (4). GSCs cultured as neurospheres maintain genomic and phenotypic changes of the primary tumor better than traditional cell lines (24). Interestingly, the cell death response of GSC-containing neurospheres to most compounds tested, including sunitinib, was clearly stronger than that of parenchymal cells, which does not fit with the chemoresistance that characterizes GSCs (11). To shed light on these contradictory data, we analyzed different properties of the in vitro microenvironment. This experimental approach is based on the relevance that the environment plays in regulating migration, proliferation, and apoptosis of cancer stem cells (8, 12). It has been shown that mechanical rigidity of the extracellular matrix contributes to key GBM cell properties, including migration and proliferation (7). However, differences in the stiffness of collagen-based substrates did not promote changes in response to sunitinib. Microenvironment rigidity has been associated with the response of cancer cells to certain apoptotic stimuli (25, 26), which may rely on signaling pathways that regulate actomyosin contractility. However, GSCs are undifferentiated cells with poor adhesion capabilities to substrates. Thus, survival signals are not likely to come from the contractility machinery that needs focal adhesions to initiate the process. Matrix composition may modulate the response to chemotherapy (27) through the interaction between cancer cells and the extracellular matrix molecules. Our data indicate that collagen either alone or in combination with hyaluronic acid, a major component of the brain parenchyma, did not modify the cell death response of GSCs to sunitinib. Although we cannot rule out that other extracellular matrix proteins could have a modifier effect on the chemosensitivity of GSCs, we have confirmed that blockade of integrin-β1, that plays a pivotal role in cell adhesion to many different types of extracellular matrix molecules, had no effect on the viability of sunitinib-treated GSCs. A third important external parameter that may modify cell fate is the dimensionality of the substrate where cells are cultured. 3D in contrast with 2D cell cultures, better recapitulate the features of the in vivo environment (28–31). Most of these works use cancer cell lines or primary tumor cells but not primary cancer stem-like cells, which have biologic properties different from other cancer cells, including the response to chemotherapy. In line with this, we found a highly significant protection against sunitinib and other similar inhibitors when GSCs were cultured in 3D collagen matrices. The effect of dimensionality on chemoresponsiveness could be partly explained in the tumor by the high cell density within the tissue, which may reduce drug penetration. However, cell density did not modify the observed response to sunitinib in our 3D cell system (data not shown). Alternatively, the spatial organization of a cell within a 3D microenvironment could modulate cell fate decisions such as survival, as previously suggested (32). Other multikinase inhibitors such as vargatef and imatinib gave similar results to sunitinib. All three inhibitors have a common kinase target, PDGFR. It has been shown that blockade of PDGFR impairs the ability of embryonic stem cells to assemble 3D extracellular matrix, most likely due to decreased expression of matrix proteins including collagen and fibronectin (33). However, we show that downregulation of PDGFR does not mimic the cell death effect of sunitinib, suggesting that different kinases are needed or alternatively that downstream mediators of cell viability are activated in a 3D environment, likely through focal adhesions proteins. Consistent with this, our data showed that inhibition of the Akt and ERK pathways, which are activated by focal adhesions (34, 35), reverts the protective effect of a 3D environment against sunitinib.

Our work thoroughly analyzed the influence of dimensionality along with other environmental features on the resistance of primary cancer stem-like cells to sunitinib and sunitinib-like multikinase inhibitors. A number of previous works have shown that 3D environments may modulate the response to chemotherapy (36–38). However, they mostly use epithelial tumor cells, but not primary cancer stem cells, which form a distinct cell population with particular properties not shared by other tumor cells, including their tumorigenic potential. Taken together, we show that a culture system that mimics a 3D environment surrounding primary GSCs recapitulates, at least in part, the chemoresistance of GSCs to tyrosine kinase inhibitors. We also provide a simple and robust in vitro model for the screening of multikinase inhibitors, against GSCs and shed light on the molecular pathways that connect a extrinsic factor such as the 3D environment to the machinery of cell survival in GSCs.

No potential conflicts of interest were disclosed.

Conception and design: J.L. Fernandez-Luna

Development of methodology: G. Fernandez-Fuente, P. Mollinedo, L. Grande, J.L. Fernandez-Luna

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Vazquez-Barquero, J.L. Fernandez-Luna

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): G. Fernandez-Fuente, J.L. Fernandez-Luna

Writing, review, and or revision of the manuscript: J.L. Fernandez-Luna

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P. Mollinedo, L. Grande

Study supervision: J.L. Fernandez-Luna

This work was supported by the Instituto de Salud Carlos III (ISCIII), Spanish Ministry of Science and Innovation grant PI10/02002, and program Red Temática de Investigación Cooperativa en Cáncer (RTICC; grants RD06/0020/0074 and RD12/0036/0022) and from the Instituto de Formación e Investigación Marqués de Valdecilla (IFIMAV) grant API2011-04 (to J.L. Fernandez-Luna).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Lonning
PE
. 
Molecular basis for therapy resistance
.
Mol Oncol
2010
;
4
:
284
300
.
2.
Ledford
H
. 
Translational research: 4 ways to fix the clinical trial
.
Nature
2011
;
477
:
526
8
.
3.
Karaman
MW
,
Herrgard
S
,
Treiber
DK
,
Gallant
P
,
Atteridge
CE
,
Campbell
BT
, et al
A quantitative analysis of kinase inhibitor selectivity
.
Nat Biotechnol
2008
;
26
:
127
32
.
4.
Joshi
AD
,
Loilome
W
,
Siu
IM
,
Tyler
B
,
Gallia
GL
,
Riggins
GJ
. 
Evaluation of tyrosine kinase inhibitor combinations for glioblastoma therapy
.
PLoS ONE
2012
;
7
:
e44372
.
5.
Zips
D
,
Thames
HD
,
Baumann
M
. 
New anticancer agents: in vitro and in vivo evaluation
.
In Vivo
2005
;
19
:
1
7
.
6.
Hakanson
M
,
Textor
M
,
Charnley
M
. 
Engineered 3D environments to elucidate the effect of environmental parameters on drug response in cancer
.
Integr Biol
2011
;
3
:
31
8
.
7.
Ulrich
TA
,
de Juan Pardo
EM
,
Kumar
S
. 
The mechanical rigidity of the extracellular matrix regulates the structure, motility, and proliferation of glioma cells
.
Cancer Res
2009
;
69
:
4167
74
.
8.
Ruiz-Ontanon
P
,
Orgaz
JL
,
Aldaz
B
,
Elosegui-Artola
A
,
Martino
J
,
Berciano
MT
, et al
Cellular plasticity confers migratory and invasive advantages to a population of glioblastoma-initiating cells that infiltrate peritumoral tissue
.
Stem Cells
2013
;
31
:
1075
85
.
9.
Giese
A
,
Bjerkvig
R
,
Berens
ME
,
Westphal
M
. 
Cost of migration: invasion of malignant gliomas and implications for treatment
.
J Clin Oncol
2003
;
21
:
1624
36
.
10.
Singh
SK
,
Hawkins
C
,
Clarke
ID
,
Squire
JA
,
Bayani
J
,
Hide
T
, et al
Identification of human brain tumour initiating cells
.
Nature
2004
;
432
:
396
401
.
11.
Beier
D
,
Schulz
JB
,
Beier
CP
. 
Chemoresistance of glioblastoma cancer stem cells–much more complex than expected
.
Mol Cancer
2011
;
10
:
128
.
12.
Borovski
T
,
De Sousa
EMF
,
Vermeulen
L
,
Medema
JP
. 
Cancer stem cell niche: the place to be
.
Cancer Res
2011
;
71
:
634
9
.
13.
Fontalba
A
,
Gutierrez
O
,
Fernandez-Luna
JL
. 
NLRP2, an inhibitor of the NF-kappaB pathway, is transcriptionally activated by NF-kappaB and exhibits a nonfunctional allelic variant
.
J Immunol
2007
;
179
:
8519
24
.
14.
Nogueira
L
,
Ruiz-Ontanon
P
,
Vazquez-Barquero
A
,
Lafarga
M
,
Berciano
MT
,
Aldaz
B
, et al
Blockade of the NFkappaB pathway drives differentiating glioblastoma-initiating cells into senescence both in vitro and in vivo
.
Oncogene
2011
;
30
:
3537
48
.
15.
Mason
BN
,
Starchenko
A
,
Williams
RM
,
Bonassar
LJ
,
Reinhart-King
CA
. 
Tuning three-dimensional collagen matrix stiffness independently of collagen concentration modulates endothelial cell behavior
.
Acta Biomater
2013
;
9
:
4635
44
.
16.
Grande
L
,
Bretones
G
,
Rosa-Garrido
M
,
Garrido-Martin
EM
,
Hernandez
T
,
Fraile
S
, et al
Transcription factors Sp1 and p73 control the expression of the proapoptotic protein NOXA in the response of testicular embryonal carcinoma cells to cisplatin
.
J Biol Chem
2012
;
287
:
26495
505
.
17.
Pedron
S
,
Becka
E
,
Harley
BA
. 
Regulation of glioma cell phenotype in 3D matrices by hyaluronic acid
.
Biomaterials
2013
;
34
:
7408
17
.
18.
El Azreq
MA
,
Naci
D
,
Aoudjit
F
. 
Collagen/beta1 integrin signaling up-regulates the ABCC1/MRP-1 transporter in an ERK/MAPK-dependent manner
.
Mol Biol Cell
2012
;
23
:
3473
84
.
19.
Bourbonnais
E
,
Raymond
VA
,
Ethier
C
,
Nguyen
BN
,
El-Leil
MS
,
Meloche
S
, et al
Liver fibrosis protects mice from acute hepatocellular injury
.
Gastroenterology
2012
;
142
:
130
9
e4.
20.
Cordes
N
,
Seidler
J
,
Durzok
R
,
Geinitz
H
,
Brakebusch
C
. 
beta1-integrin-mediated signaling essentially contributes to cell survival after radiation-induced genotoxic injury
.
Oncogene
2006
;
25
:
1378
90
.
21.
Gori
B
,
Ricciardi
S
,
Fulvi
A
,
Intagliata
S
,
Del Signore
E
,
de Marinis
F
. 
New antiangiogenics in non-small cell lung cancer treatment: Vargatef (BIBF 1120) and beyond
.
Ther Clin Risk Manag
2011
;
7
:
429
40
.
22.
Fischbach
C
,
Chen
R
,
Matsumoto
T
,
Schmelzle
T
,
Brugge
JS
,
Polverini
PJ
, et al
Engineering tumors with 3D scaffolds
.
Nat Methods
2007
;
4
:
855
60
.
23.
Grantab
R
,
Sivananthan
S
,
Tannock
IF
. 
The penetration of anticancer drugs through tumor tissue as a function of cellular adhesion and packing density of tumor cells
.
Cancer Res
2006
;
66
:
1033
9
.
24.
Lee
J
,
Kotliarova
S
,
Kotliarov
Y
,
Li
A
,
Su
Q
,
Donin
NM
, et al
Tumor stem cells derived from glioblastomas cultured in bFGF and EGF more closely mirror the phenotype and genotype of primary tumors than do serum-cultured cell lines
.
Cancer Cell
2006
;
9
:
391
403
.
25.
Targosz-Korecka
M
,
Biedron
R
,
Szczygiel
AM
,
Brzezinka
G
,
Szczerbinski
J
,
Zuk
A
. 
Stiffness changes of tumor HEp2 cells correlates with the inhibition and release of TRAIL-induced apoptosis pathways
.
J Mol Recognit
2012
;
25
:
299
308
.
26.
Leight
JL
,
Wozniak
MA
,
Chen
S
,
Lynch
ML
,
Chen
CS
. 
Matrix rigidity regulates a switch between TGF-beta1-induced apoptosis and epithelial-mesenchymal transition
.
Mol Biol Cell
2012
;
23
:
781
91
.
27.
Hakanson
M
,
Kobel
S
,
Lutolf
MP
,
Textor
M
,
Cukierman
E
,
Charnley
M
. 
Controlled breast cancer microarrays for the deconvolution of cellular multilayering and density effects upon drug responses
.
PLoS ONE
2012
;
7
:
e40141
.
28.
Griffith
LG
,
Swartz
MA
. 
Capturing complex 3D tissue physiology in vitro
.
Nat Rev Mol Cell Biol
2006
;
7
:
211
24
.
29.
Chen
L
,
Xiao
Z
,
Meng
Y
,
Zhao
Y
,
Han
J
,
Su
G
, et al
The enhancement of cancer stem cell properties of MCF-7 cells in 3D collagen scaffolds for modeling of cancer and anti-cancer drugs
.
Biomaterials
2012
;
33
:
1437
44
.
30.
Gurski
LA
,
Jha
AK
,
Zhang
C
,
Jia
X
,
Farach-Carson
MC
. 
Hyaluronic acid-based hydrogels as 3D matrices for in vitro evaluation of chemotherapeutic drugs using poorly adherent prostate cancer cells
.
Biomaterials
2009
;
30
:
6076
85
.
31.
Xu
X
,
Gurski
LA
,
Zhang
C
,
Harrington
DA
,
Farach-Carson
MC
,
Jia
X
. 
Recreating the tumor microenvironment in a bilayer, hyaluronic acid hydrogel construct for the growth of prostate cancer spheroids
.
Biomaterials
2012
;
33
:
9049
60
.
32.
Zahir
N
,
Weaver
VM
. 
Death in the third dimension: apoptosis regulation and tissue architecture
.
Curr Opin Genet Dev
2004
;
14
:
71
80
.
33.
Hewitt
KJ
,
Shamis
Y
,
Knight
E
,
Smith
A
,
Maione
A
,
Alt-Holland
A
, et al
PDGFRbeta expression and function in fibroblasts derived from pluripotent cells is linked to DNA demethylation
.
J Cell Sci
2012
;
125
:
2276
87
.
34.
Thomas
SL
,
Alam
R
,
Lemke
N
,
Schultz
LR
,
Gutierrez
JA
,
Rempel
SA
. 
PTEN augments SPARC suppression of proliferation and inhibits SPARC-induced migration by suppressing SHC-RAF-ERK and AKT signaling
.
Neuro Oncol
2010
;
12
:
941
55
.
35.
Genua
M
,
Xu
SQ
,
Buraschi
S
,
Peiper
SC
,
Gomella
LG
,
Belfiore
A
, et al
Proline-rich tyrosine kinase 2 (Pyk2) regulates IGF-I-induced cell motility and invasion of urothelial carcinoma cells
.
PLoS ONE
2012
;
7
:
e40148
.
36.
Weaver
VM
,
Lelievre
S
,
Lakins
JN
,
Chrenek
MA
,
Jones
JC
,
Giancotti
F
, et al
beta4 integrin-dependent formation of polarized three-dimensional architecture confers resistance to apoptosis in normal and malignant mammary epithelium
.
Cancer Cell
2002
;
2
:
205
16
.
37.
Loessner
D
,
Stok
KS
,
Lutolf
MP
,
Hutmacher
DW
,
Clements
JA
,
Rizzi
SC
. 
Bioengineered 3D platform to explore cell-ECM interactions and drug resistance of epithelial ovarian cancer cells
.
Biomaterials
2010
;
31
:
8494
506
.
38.
Dangi-Garimella
S
,
Sahai
V
,
Ebine
K
,
Kumar
K
,
Munshi
HG
. 
Three-dimensional collagen I promotes gemcitabine resistance in vitro in pancreatic cancer cells through HMGA2-dependent histone acetyltransferase expression
.
PLoS ONE
2013
;
8
:
e64566
.