The anti-vascular endothelial growth factor receptor-1 monoclonal antibody D16F7 inhibits invasiveness of human glioblastoma and glioblastoma stem cells

We enrolled 42 adults [mean age 60.51 (34–79), 27 males/15 females], who underwent surgery for primary GBM at the Institute of Neurosurgery, “Università Cattolica del Sacro Cuore” (Rome, Italy), from March 2005 to September 2011. Diagnosis of GBM was established on histological examination according to the WHO classification (grade IV) of tumors of the nervous system. All patients provided written consent to use their specimens for research and the research proposal was approved by the university Ethical Committee. Tissues were fixed in 4% paraformaldehyde in 0.1 M phosphate buffer pH 7.6 at 4 °C overnight. Tissues were rehydrated in graded ethanol solutions, xylene and finally embedded in Paraplast Plus (Tyco/Healthcare, Mansfield, MA). Sections, 3–4 μm thick, were deparaffinized and incubated in 10 mM citrate buffer, pH 6.0, dry heated for 10 min each to unmask antigen sites, cooled and washed in phosphate-buffered saline (PBS). Endogenous peroxidase activity was inhibited by rinsing the slides in 3% hydrogen peroxide for 5 min. Nonspecific binding was blocked by 5 min incubation with the Super Block Solution (ScyTek Laboratories, UT). After washing in PBS, sections were incubated for 10 min at room temperature with rabbit anti-Human Flt-1/VEGFR-1 polyclonal antibody (1:50; Spring Bioscience, Pleasanton, CA). The immunostaining conditions were standardized using human placenta as positive control (data not shown). Sections were washed extensively with PBS and subsequently processed using the Ultra Tek Anti-Polyvalent kit (ScyTek Laboratories). Finally sections were treated with 3,3′-diaminobenzidine as chromogen, contrasted with hematoxylin and mounted [26]. Two blinded examiners evaluated staining of human tumor specimens. For each specimen, the number of VEGFR-1 positive cells in total 50 cells was counted.

The human GBM cell lines A172, U87, LN18, T98G and U373 were from American Type Culture Collection (ATCC, Manassas, VA). Cells were maintained in DMEM (Sigma-Aldrich, St. Louis, MO) supplemented with 10% fetal bovine serum (FBS, Sigma-Aldrich), 2 mM L-glutamine, 100 units/ml penicillin, and 100 μg/ml streptomycin sulfate, at 37 °C in a 5% CO₂ humidified atmosphere.

GSCs were isolated from 18 surgical samples of adult patients who had undergone craniotomy at the Institute of Neurosurgery, “Università Cattolica del Sacro Cuore” (Rome, Italy). Prior to surgery all patients provided written informed consent according to the Declaration of Helsinki and the research proposal was approved by the university Ethical Committee. In regard to GSCs origin, the diagnosis of GBM was established on histological examination according to the WHO classification (grade IV) of tumors of the nervous system. Tumor samples were subjected to mechanical dissociation. The resulting cell suspension was cultured in a serum-free medium supplemented with 20 ng/ml epidermal growth factor (EGF) and 10 ng/ml FGF-2 (PeproTech, Rocky Hill, NJ). Generation of GSCs was defined by the following criteria: in vitro formation of primary neurospheres expressing stem cell markers such as CD133, SOX2, Musashi-1 and nestin, capacity of self-renew, ability to co-express astrocytic as well as neuronal phenotypic markers after serum-induced differentiation in vitro [27‐29]. GSCs were characterized by immunofluorescence analysis as previously described [30]. All the GSC lines tested in this study were positive for SOX2, Musashi-1 and nestin, whereas they expressed different levels of CD133 (data not shown).

P3, EGFRwt⁺, and EGFRvIII⁺ GSC lines were previously described [31]. Cells were cultured in neurobasal medium (NBM) supplemented with 2 mM GlutaMAX and 1× B-27 (Life Technologies, Carlsbad, CA), 1× penicillin/streptomycin (Lonza, Basel, Switzerland), 1 U/ml heparin (Sigma-Aldrich) and 20 ng/ml FGF-2 (hereafter referred to as complete NBM).

The human umbilical vascular endothelial cells (HUVEC) were isolated from freshly delivered umbilical cords as previously described [32] and cultured in EGM-2.

The human GR-Mel and M14 melanoma cell lines, used as positive and negative controls for VEGFR-1 or VEGFR-2 transcripts, were obtained and cultured as previously described [33].

Human GBM cell lines were authenticated by STR profiling (BMR genomics, Padova, Italy) and GSCs lines were periodically tested for the expression of phenotypic markers [in P3-derived cells, EGFR amplification or mutation; in GSCs, the above described markers].

Cell clones were obtained by limiting dilution from U87 cells and one clone was transfected with the pBLAS49.2 or pBLAS49.2/VEGFR-1 plasmids. The pBLAS49.2/VEGFR-1 construct was obtained by cloning of VEGFR-1 cDNA from pcDNA3/VEGFR-1 plasmid (a generous gift of Dr. K. Ballmer-Hofer, PSI, Zurich) into pBLAS49.2 vector (InvivoGen, San Diego, CA). Transfection was performed using lipofectamine 2000 (Invitrogen, Camarillo, CA), as described by the manufacturer, and transfected cells were selected in blasticidine (Invitrogen) containing culture medium. Antibiotic resistant clones were isolated by ring cloning and U87 clones maintained in the presence of 2.5 μg/ml blasticidine. VEGFR-1 expressing subclones were identified by RT-PCR and Western blotting.

Quantification of membrane VEGFR-1 and VEGFR-2 transcripts was performed by quantitative real-time reverse transcriptase –polymerase chain reaction (qRT-PCR) according to the dual-labeled fluorigenic probe method and using an ABI Prism 7000 sequence detector (PerkinElmer, Groningen, the Netherlands), as previously described [34]. Expression levels were calculated by the relative standard curve method. Primers used were as follows: VEGFR-1, forward 5′-ACCGAATGCCACCTCCATG-3′ and reverse 5′-AGGCCTTGGGTTTGCTGTC-3′; VEGFR-2, forward 5′-GTCTATGCCATTCCTCCCCC-3′ and reverse 5′-GAGACAGCTTGGCTGGGCT-3′. For each sample, the level of VEGFR-1 or VEGFR-2 transcripts was normalized to that of 18S RNA (TaqMan® Gene Expression Assay, Applied Biosystems, Foster City, CA) and referred to the values of the VEGFR-1 and VEGFR-2 negative M14 cell line, to which the arbitrary value of 1 was assigned.

In VEGFR-1-transfected cells detection of VEGFR-1 transcript was confirmed by RT-PCR analysis. The cDNA preparation followed by PCR amplification to evaluate VEGFR-1 expression was performed as previously described [35], utilizing an annealing temperature of 58 °C and the following primers: human VEGFR-1, forward primer 5′-CTCCTGAGTACTCTACTCCT-3′, reverse primer 5′-GAGTACAGGACCACCGAGTT-3′ (640 bp fragment); human glyceraldehyde-3-phosphate dehydrogenase (GAPDH), forward primer 5′-TCCCATCACCATCTTCCA-3′, reverse primer 5′-CATCACGCCACAGTTTCC-3′ (380 bp fragment).

Conditioned media from GBM cells were obtained by incubating semi-confluent cultures for 24 h in 0.1% BSA/DMEM medium without FBS. These conditions did not significantly affect cell viability. Supernatants were concentrated at least 10-fold in Centriplus concentrators (Amicon, Beverly, MA). Cells were detached from the flasks with PBS/EDTA. Cytokine secretion values were normalized by the total number of cells.

Quantification of the amount of VEGF-A and PlGF in the conditioned medium was performed using goat anti-VEGF-A or anti-PlGF IgGs (R&D Systems, Abingdon, UK), at a concentration of 10 μg/ml in PBS, to coat Maxisorp Nunc immunoplates (Nunc, Roskilde, Denmark). Detection of the cytokines was performed with biotinylated goat anti-VEGF or anti-PlGF IgGs (0.4 μg/ml; R&D Systems) followed by incubation with streptavidin alkaline phosphatase conjugate (1:10,000) (Roche, Monza, Italy) and alkaline phosphatase reaction. Optical density at 405 nm was measured in a 3550-UV Microplate reader (Bio-Rad, Hercules, CA).

Proteins were run in 10% SDS-polyacrylamide gels and transferred to supported nitrocellulose membranes by standard techniques. Immunodetection was performed using the following antibodies: mouse monoclonal anti-VEGFR-1 (clone D2, 1:500; Santa Cruz Biotechnology, Santa Cruz, CA); mouse monoclonal anti-EGF receptor (EGFR) (528, 1:1000; Santa Cruz Biotechnology); mouse monoclonal antibody anti-EGFRvIII (L8A4; 1:1000; Absolute Antibody, Oxford, UK); rabbit polyclonal anti-phosphorylated VEGFR-1 at tyrosine 1213 (1:500; R&D Systems); rabbit polyclonal anti-Erk1&2 (1:1000; Genetex, Irvine, CA); rabbit polyclonal anti-phospho-Erk1&2 (Thr/Tyr185/187, 1:1000; Invitrogen); or rabbit polyclonal anti-β-actin (1:10,000; Sigma Aldrich) primary antibodies. Anti-mouse or anti-rabbit Ig/Horseradish peroxidase secondary antibodies and ECL Western blotting detection reagents from GE Healthcare (Milan, Italy) were used to identify the proteins of interest.

In vitro migration assay was performed using Boyden chambers equipped with 8 μm pore diameter polycarbonate filters (Nuclepore, Whatman Incorporated, Clifton, NJ) coated with 5 μg/ml gelatin (Sigma-Aldrich), as previously described [36, 37]. Treatment with D16F7 was carried out by incubating the cells in the presence of the indicated mAb concentrations in a rotating wheel for 30 min at room temperature. Cells (2 × 10⁵/chamber) were then loaded in the upper compartment of Boyden chambers and migration assay, toward stimuli (50 ng/ml VEGF-A or PlGF) present in the lower compartment, was done in the absence or in the presence of D16F7 mAb or, in selected experiments, of an equivalent amount of a species- and isotype-matched control antibody (mouse IgG1, R&D Systems) for 18 h. Migrated cells, attached to the lower side of the filters, were stained with crystal violet counted in triplicate samples for a total of 12 high power (200× magnification) microscopic fields.

For spheroid invasion assay, tumor cells (25,000–30,000 cells/ml) were suspended in DMEM-1640 containing 10% FBS (for GBM cell lines) or in complete NBM (for P3-derived GSC lines), supplemented with methyl cellulose (0.24% final concentration; Sigma-Aldrich), seeded in 96-well round bottom cell culture plates (100 μl/well; Corning® Costar® Ultra-Low attachment multi-well, Sigma-Aldrich) and centrifuged at 3000 rpm for 90 min [31]. Plates were then incubated for 24 h under standard culture conditions (5% CO₂, at 37 °C) to allow spheroid formation. Spheroids were collected, embedded individually in 100 μl of matrigel (reduced growth factor basement membrane matrix, Pathclear, Cultrex, Gaithersburg, MD) in 0.1% BSA/DMEM or NBM medium, with or without VEGF-A or PlGF (50 ng/ml) and/or D16F7 mAb, and plated in each well of a 96-well flat bottom plate, previously coated with 50 μl of matrigel. Five to ten replicates were set up for each experimental group. After matrigel solidification at 37 °C, 100 μl of invasion medium, with or without VEGF-A, PlGF or EGF (50 ng/ml), were added and plates incubated at 37 °C for up to 72 h. Spheroids were visualized and photographed using a Nikon Eclipse TS100 microscope in conjunction with a Nikon DS-Fi1 high resolution camera (Melville, NY). Measurements were performed using Adobe Photoshop CS6 software. Relative invasion area was defined as area of spheroids (in mm²) at each time point minus area on day 0.

Preliminary experiments on U87 and U87-MF24 cell migration in response to PlGF and in the presence of graded concentrations of D16F7 indicated that the IC₅₀ values were 1.54 ± 0.22 μg/ml and 2.49 ± 0.56 μg/ml, respectively. Therefore, based on these results we selected the mAb concentrations to be tested in the functional assays.

Cell proliferation was evaluated in 96-well plates using the tetrazolium compound MTS [3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl) 2-(4-sulphophenyl)-2H–tetrazolium, inner salt] from Promega (Madison, WI), as previously described [38]. Briefly, increasing numbers of GBM cells, suspended in complete medium containing graded concentrations of D16F7 up to 20 μg/ml of D16F7 or control antibody or without antibodies, were dispensed into flat-bottom 96-well plates and grown at 37 °C in a 5% CO₂ humidified atmosphere. Six replica wells were used for every condition. After 3 days, 20 μl of MTS solution were added to each well and cells were incubated at 37 °C for 2 h. Absorbance was read at 490 nm (reference wavelength 655 nm) using a 3550-UV Microplate reader (Bio-Rad).

Statistical analysis of the differences between pairs of groups was performed by Student’s t test. For multiple comparisons ANOVA analysis, followed by Bonferroni’s post-test, was used. Statistical significance was determined at α = 0.05 level. Differences were considered statistically significant when p < 0.05.

To investigate the relevance of VEGFR-1 in GBM, we initially investigated the expression of VEGFR-1 by immunohistochemistry in tissue specimens obtained from 42 adult GBM patients (Table 1). Most of GBM tissue samples showed a significant VEGFR-1 immunoreactivity. VEGFR-1 staining was observed in association with GBM cells as well as with endothelial cells (Table 1 and Additional file 1: Figure S1 and Additional file 2: Figure S2) in accordance with previous studies [39, 40].

With the purpose of analyzing the effect of the anti-VEGFR-1 D16F7 mAb in GBM models, a set of human GBM cell lines was characterized for the expression of membrane VEGFRs and production of VEGFR-1 ligands. The VEGFR-1 mRNA, evaluated by qRT-PCR, was detected in five out of six cell lines (Fig. 1a, upper panel), even though at lower levels compared with the positive control HUVEC. In the same analysis, VEGFR-2 mRNA was observed in all cell lines, although in most of them at a lower level with respect to VEGFR-1 (Fig. 1a, lower panel). ELISA analysis of PlGF and VEGF-A secretion in culture supernatants collected from the different GBM cell lines revealed that: a) PlGF was produced by most of the cell lines tested and was nearly undetectable in T98G cells (Fig. 1b, upper panel), and b) all cell lines secreted substantial amounts of VEGF-A (Fig. 1b, lower panel).

Activation of VEGFR-1 is involved in endothelial and monocytic cell migration, and PlGF has a role in controlling cell motility and invasiveness of cultured cancer cells [3]. In this regard, we recently demonstrated that the anti-VEGFR-1 D16F7 mAb inhibits human melanoma chemotaxis in response to PlGF [15]. The influence of D16F7 on GBM migratory response to VEGF-A and PlGF was tested using VEGFR-1-positive U87 and LN18 cells in Boyden chambers containing gelatin coated filters. VEGF-A and PlGF stimulated substantial chemotaxis of U87 cells, and pre-incubation with D16F7 markedly reduced ligand-induced migration (Fig. 1c). On the other hand, a murine IgG1 control did not affect migration of U87 cells upon stimulation of VEGFR-1 (Additional file 3: Figure S3), in accordance with a previous study with melanoma cells [15]. These data indicated a prevalent role for VEGFR-1 in the promotion of GBM cell migration compared to VEGFR-2, even when VEGF-A was used as stimulus. The anti-VEGFR-1 mAb also inhibited the migratory response of LN18 cells (data not shown). The influence of D16F7 on VEGF-A and PlGF-induced chemotaxis of other VEGFR-1 positive cell lines (i.e., A172 and T98G) could not be tested in this assay because cells failed to adhere to gelatin-coated filters. Therefore, the ability of D16F7 to reduce ECM invasion of A172 cells was tested by a spheroid invasion assay. The VEGFR-1-selective ligand PlGF stimulated matrigel-embedded spheroids to markedly invade the surrounding matrix in a time-dependent manner, and D16F7 significantly reduced GBM cell invasiveness (Fig. 1d). Notably, D16F7 at the concentrations tested in the chemotaxis and invasion assays did not inhibit proliferation in vitro in any of the GBM cell lines tested (data not shown).

GBMs harbor a subset of GBM stem-like cells (GSCs) which are radioresistant and chemoresistant, participate in tumor neovascularization, and promote tumor recurrence [30, 41]. To assess whether VEGFR-1 plays a role in the aggressive behavior of this GBM cell subset, 18 patient-derived GSC lines (Table 2) were analyzed for VEGFR-1 and VEGFR-2 expression by qRT-PCR. VEGFR-1 expression was quite heterogeneous, while all GSC lines were weakly positive for VEGFR-2 (Fig. 2a). VEGFR-1 expression was verified by Western blotting (Fig. 2b) in GSCs expressing the greatest levels of VEGFR-1 transcript. VEGFR-1 positive GSC lines #213, #169 and #74 migrated in response to both VEGF-A and PlGF, and treatment with D16F7 reverted ligand-induced chemotaxis (Fig. 2c and d).

Mutation/amplification of EGFR has been reported in a wide proportion of GBMs. In fact, a large-scale sequencing study indicated that 57% of GBM patient samples contain mutation, rearrangement, altered splicing, and/or focal amplification of EGFR and that various mutations often co-occur with EGFR rearrangement and/or amplification [42]. In this context, we next investigated the influence of D16F7 mAb in a patient-derived GSC line (P3) modified to over-express wild-type EGFR (EGFRwt⁺) or mutant EGFR (ligand binding domain-deficient EGFRvIII⁺) (Fig. 3a). These GSCs are characterized by invasive or angiogenic in vivo behavior depending on whether EGFRwt (more invasive) or EGFRvIII (more angiogenic) is overexpressed [31]. Analysis of P3 cells and the two P3-derived cell lines by qRT-PCR revealed that all of them expressed VEGFR-1 as well as VEGFR-2 (Fig. 3b). VEGFR-1 level was significantly higher in EGFRvIII⁺ than in P3 and EGFRwt⁺ cells.

Strikingly, PlGF stimulated ECM invasion in all cell lines in a time-dependent manner while, as expected, only P3 and more significantly EGFRwt⁺ cells responded to EGF (Fig. 3c, Additional file 4: Figure S4, Additional file 5: Figure S5 and Additional file 6: Figure S6). Consistently, treatment with D16F7 mAb inhibited invasion induced by PlGF, but it did not affect ECM invasion in response to EGF (Fig. 3c, Additional file 4: Figure S4, Additional file 5: Figure S5 and Additional file 6: Figure S6).

These results suggest that D16F7 may inhibit the aggressive behavior of GSCs expressing mutated EGFR.

Despite the presence of a conserved kinase domain containing an ATP binding site in VEGFR-1, ligand stimulation of the receptor results in only minor tyrosine phosphorylation in vitro and in vivo [43]. Indeed, the weak kinase activity of the receptor has posed challenges to studying features of VEGFR-1 signal transduction. To investigate the ability of D16F7 to affect signal transduction in GBM cells, we over-expressed the human VEGFR-1 membrane form in U87 expressing little endogenous VEGFR-1 protein. Cells transfected with control or VEGFR-1 cDNA-containing vectors were subsequently analyzed by RT-PCR (Fig. 4a) and Western blotting (Fig. 4b). For further experiments to assess the influence of D16F7 on signal transduction, the clone exhibiting the greatest VEGFR-1 protein expression was tested (i.e., U87-MF24). The Tyr 1213 residue in VEGFR-1 is regarded as one of the major auto-phosphorylation sites responsible for activation of intracellular signaling pathways [44, 45]. Accordingly, we investigated whether receptor stimulation by its ligands (VEGF-A and PlGF) resulted in Tyr 1213 phosphorylation in transfected cells and whether D16F7 could inhibit this effect. Exposure of U87-MF24 (Fig. 4c) cells to VEGF-A or PlGF for 10 min induced robust VEGFR-1 kinase activity, as indicated by a marked increase of protein phosphorylation at Tyr 1213, and D16F7 reduced receptor auto-phosphorylation in a dose-dependent manner. VEGFR-1 activation by VEGF-A or PlGF in normal cells (e.g., endothelial cells, fibroblasts, monocytes) stimulates Erk1/2 of the MAPK signaling pathway [3, 16], which, in turn, promotes tumor cell invasion and migration [46]. Accordingly, we investigated whether VEGF-A- or PlGF-induced receptor auto-phosphorylation at Tyr 1213 in U87-MF24 cells was accompanied by Erk1/2 phosphorylation, and whether D16F7 could reduce Erk1/2 activation. Results revealed that D16F7 markedly inhibited Erk1/2 phosphorylation stimulated by both ligands (Fig. 4c). Quantitative results are summarized as percentage inhibition of VEGFR-1 or Erk1/2 phosphorylation achieved after treatment with D16F7, calculated across three independent experiments (Fig. 4d).

VEGFR-1 over-expression in U87-MF24 cells highly stimulated ECM invasion triggered by PlGF and inhibition of PlGF-induced signaling by D16F7 resulted in abrogation of ECM invasion (Fig. 4e).