Journal of Neuro-Oncology

, Volume 97, Issue 3, pp 323–337

Activated EGFR signaling increases proliferation, survival, and migration and blocks neuronal differentiation in post-natal neural stem cells

Authors

    • Neurosurgical Laboratory for Translational Stem Cell Research, Department of Neurosurgery, Weill Cornell Brain Tumor CenterWeill Cornell Medical College of Cornell University
    • Department of Cell MorphologyCentro de Investigación Principe Felipe and RETICS-CIBERNED
  • Jennifer A. Moliterno
    • Neurosurgical Laboratory for Translational Stem Cell Research, Department of Neurosurgery, Weill Cornell Brain Tumor CenterWeill Cornell Medical College of Cornell University
  • Sebila Kratovac
    • Neurosurgical Laboratory for Translational Stem Cell Research, Department of Neurosurgery, Weill Cornell Brain Tumor CenterWeill Cornell Medical College of Cornell University
  • Gurpreet S. Kapoor
    • Department of Neurological SurgeryUniversity of Pennsylvania School of Medicine
  • Donald M. O’Rourke
    • Department of Neurological SurgeryUniversity of Pennsylvania School of Medicine
  • Eric C. Holland
    • Department of Neurological SurgeryMemorial Sloan-Kettering Cancer Center
  • Jose Manuel García-Verdugo
    • Department of Cell MorphologyCentro de Investigación Principe Felipe and RETICS-CIBERNED
  • Neeta S. Roy
    • Department of Neurology and NeuroscienceWeill Cornell Medical College of Cornell University
    • Neurosurgical Laboratory for Translational Stem Cell Research, Department of Neurosurgery, Weill Cornell Brain Tumor CenterWeill Cornell Medical College of Cornell University
    • Department of Neurological SurgeryWeill Cornell Medical College of Cornell University
Laboratory Investigation - Human/Animal Tissue

DOI: 10.1007/s11060-009-0035-x

Cite this article as:
Ayuso-Sacido, A., Moliterno, J.A., Kratovac, S. et al. J Neurooncol (2010) 97: 323. doi:10.1007/s11060-009-0035-x

Abstract

Recent evidence supports the notion that transformation of undifferentiated neural stem cell (NSC) precursors may contribute to the development of glioblastoma multiforme (GBM). The over-expression and mutation of the epidermal growth factor receptor (EGFR), along with other cellular pathway mutations, plays a significant role in GBM maintenance progression. Though EGFR signaling is important in determining neural cell fate and conferring astrocyte differentiation, there is a limited understanding of its role in NSC and tumor stem cell (TSC) biology. We hypothesized that EGFR expression and mutation in post-natal NSCs may contribute to cellular aggressiveness including enhanced cellular proliferation, survival and migration. Stable subclones of C17.2 murine NSCs were transfected to over-express either the wild-type EGFR (wtEGFR) or its most common mutated variant EGFRvIII. Activated EGFR signaling in these cells induced behaviors characteristic of GBM TSCs, including enhanced proliferation, survival and migration, even in the absence of EGF ligand. wtEGFR activation was also found to block neuronal differentiation and was associated with a dramatic increase in chemotaxis in the presence of EGF. EGFRvIII expression lead to an increase in NSC proliferation and survival, while it simultaneously blocked neuronal differentiation and promoted glial fate. Our findings suggest that activated EGFR signaling enhances the aggressiveness of NSCs. Understanding the regulatory mechanisms of NSCs may lend insight into deregulated mechanisms of GBM TSC invasion, proliferation, survival and resistance to current treatment modalities.

Keywords

EGFRNeural stem cellsGliomaBrain tumors

Introduction

The epidermal growth factor receptor (EGFR) is a tyrosine kinase which is activated by ligand binding, thereby inducing receptor dimerization and consequent autophosphorylation of key tyrosines residues [1]. EGFR signaling affects many of the cellular events involved with corticogenesis, including neural cell survival, proliferation, differentiation, and migration [2]. During the subventricular zone (SVZ) development, cortical progenitor cells divide in response to EGF ligand, but retain the ability to differentiate into neurons, astrocytes, and oligodendrocytes [35].

EGFR is frequently over-expressed and often mutated in certain glial tumors, such as glioblastoma multiforme (GBM). The most common mutant encountered in gliomas, EGFRvIII, is a truncated extracellular domain that transforms into a constitutively active ligand-independent kinase [68]. EGFR is important in the mediation of a variety of biological responses in gliomas via both autocrine and paracrine mechanisms [9]. The over-expression of either the wild-type EGFR (wtEGFR) or its variant, EGFRvIII, has been extensively studied in vitro using GBM cell lines [1013], among others. In a glioma mouse model, somatic expression of EGFRvIII and PTEN loss were important post-natal genetic changes which directly resulted in high-grade gliomas [14]. Using other mouse models, expression of EGFRvIII in the glial lineage was found to induce lesions that were strikingly similar to human gliomas [1518]. Interestingly, these lesions were found to occur more frequently within the subpopulation of progenitor cells [17, 19, 20]. With infusions of EGF, type C-cells, which express high levels of EGFR, become highly migratory with invasive characteristics [2126].

It has more recently been suggested that brain tumors may derive from the transformation of stem cell precursors located in the germinal regions [27, 28]. GBM often arise in or adjacent to the SVZ and contain cells of multiple lineages [29]. Type B astrocytes may be the cell-of-origin in some models of malignant gliomas [30]. Recent studies have also drawn attention to a slowly cycling, but highly tumorigenic, subpopulation of cells in human glioma with stem cell-like properties. These tumor stem cells (TSCs) invite comparisons to the relatively quiescent population of adult NSCs embedded within the SVZ.

EGFR signaling has been shown to contribute to both NSC behavior and to gliomagenesis, but the understanding of its role in NSCs remains limited. Activated EGFR signaling in NSCs may induce characteristics seen in TSCs, such as heightened proliferation, migration and survival. Using the well-characterized murine NSC line C17.2 derived from the external germinal layer of the postnatal cerebellum, we hypothesized that activated EGFR signaling in NSCs would lead to increased cellular proliferation, survival and chemotaxis, while blocking neuronal differentiation. Such findings would likely be important in understanding the regulatory mechanisms of NSCs, which in turn, may lend insight into deregulated mechanisms of TSC invasion, proliferation, survival and resistance to current treatment modalities.

Materials and methods

GBM samples and cell culture

All protocols were approved by the Institutional Review Board (IRB) at Weill Medical College of Cornell University. Human GBM samples were collected for comparison of baseline wtEGFR and EGFRvIII expression levels. Fresh tumor tissue was obtained from patients during surgical resection of human GBM. Mouse C17.2 NSCs were grown in M2 media (DMEM with 10% fetal bovine serum, 5% horse serum and 5% pen/strep) or differentiation media (DMEM-F12 (Gibco) supplemented with 1% N2 (Gibco), 2 mM glutamine, 0.6% (W/v) glucose, 0.02 mg/ml Insulin (Sigma) and 15 mM HEPES (Gibco)) [31]. 293T cells were grown in DMEM with 10% fetal bovine serum and 5% pen/strep. Cell lines were maintained at 37°C in 80% air/5% CO2.

Vector construction

wtEGFR and EGFRvIII cDNAs (a gift from Don M. O′Rourke, MD, Philadelphia, PA) were PCR amplified and blunt-end ligated into pLIRE, a plasmid derived from pIRES2-EGFP plasmid (BD Biosciences-Clontech, Palo Alto, CA). wtEGFR was HA-tagged to distinguish between mouse and human EGFR. An empty vector cassette was made by restriction enzyme digest of EGFRvIII cDNA from the plasmid using HindIII. The pLIREiresEGFP plasmid was column purified and relegated. The plasmids were later sequenced to confirm the orientation of the inserts as well as the absence of mutations. Retroviruses were generated by co-transfecting the backbone carrying the gene of interest (pLIRE-wtEGFR, pLIRE-ΔEGFR and empty pLIRE), along with p-Gag-Pol and pVSVg into 293T cells. Transfection was performed using Calcium Phosphate and viral particles were collected after 48 h. The viral particles were purified by ultra-centrifuging on a SW-28 rotor at 25,000 rpm for 90 min at 4°C and resuspended in PBS. These viruses were used to infect mouse C17.2 NSCs.

Development of stable infected cell lines and maintenance

After infection with the retrovirus vectors, NSCs were sorted by green fluorescent protein (GFP) expression using a FACSVantage (Becton–Dickinson Vantage cell sorter, Hospital for Special Surgeries, NY). One cell per well were plated by FACS-automatic cell deposition in 96-wells plates and allowed to expand clonally with the media being replaced every other day. The three clones with the highest GFP selection, as confirmed by immunocytochemistry, were used for all experiments. The level of GFP correlated with the level of transgene expression. However, with each successive passage, diminishing levels of either GFP or transgene were observed. Methylation at LTRs is commonly associated with retroviral silencing [32] and this could be a factor in the decrease of wtEGFR or EGFRvIII over time. To minimize variations in the transgene expression with subsequent passages, cells from passage 4 to 7 where the percentage of GFP expressing cells, assessed by flow cytometry, was at least 80%. For maintenance, NSCs were passaged when they reached 80% confluence and media was changed every 2 days. For differentiation assays, 3000 cells per well were plated into 24 well/plates.

RT/QRT-PCR

For RT-PCR, total RNA was isolated by using TRIzol (Invitrogen). One μg of RNA was used for cDNA synthesis. These samples were amplified with specific primers targeting wtEGFR or EGFRvIII [33] and the Paq5000 polymerase (Stratagene), in a Mastercycler (Eppendorf). QRT-PCR was run in a LightCycler 480 Instrument (Roche). RNA Isolation was performed using Microkit (QIAGEN) following the manufacture′s recommendation. Reactions for QRT-PCR were carried out by using the following pair of primers: wtEGFR (Fw 5′-GAG GTG GTC CTT GGG AAT TTG-3′, and Rev 5′-GCA ATG AGG ACA TAA CCA GCC-3′), ΔEGFR (Fw 5′-GTC GGG CTC TGG AGG AAA AG-3′, and Rev 5′-GCC GTC TTC CTC CAT CTC ATA G-3′). Primers specifically designed to target proliferation, apoptosis and differentiation markers as well as the two housekeeping genes used to normalize the samples are listed as supplemental data (Table 1). All designed primers showed an amplification product of the expected size for all conditions (Supplemental Fig. 3). For each experiment, controls were performed in which reverse transcriptase was omitted from the cDNA reaction mixture and template DNA was omitted from the PCR mixture.

Oxidative stress measurement

Oxidative stress was assessed from intracellular levels of reactive oxygen species like superoxide anion using hydroethidine (HE) (Invitrogen) and membrane potential using tetramethylrhodamine (TMRM) (Invitrogen). Cell suspensions (106 cells/ml in PBS) were stained during 15 min at 37°C in the dark using 5 μM HE or 75 nM of TMRM. Flow cytometry analyses were performed using an EPICS flow cytometer (Beckman-Coulter, San Jose, CA, USA).

Immunocytochemistry

Cells were fixed in 4% PFA paraformaldehyde at room temperature for 20 min. After three rinses in PBS, cells were incubated for 15 min in PBS containing 1% normal goat serum, 0.01% (w/v) saponin and 0.9% (w/v) NaCl. Cells were blocked in PBS containing 5% normal goat serum, 0.05% (w/v) saponin and 0.9% (w/v) NaCl. The following primary antibodies were used: Nestin (1/5000, PRB-315C; Covance), Tuj1 (1/500, S-435P; Covance), BrDu (1/100, ab7384; Abcan), EGFR (1/500, 1005SC-03; Santa Cruz), GFP (1/1000, Molecular Probe), EGFRvIII (1/500, mouse monoclonar L8A4; a gift from Dr. Bigner). Incubation with these antibodies was performed at 4°C overnight. Cells were rinsed three times in PBS and incubated with fluorescent secondary antibodies. For the detection of the primary antibodies, Alexa Fluor secondary detection anti IgG (Molecular Probe) diluted 1/500 was used. All secondary antibodies were incubated for 1 h at room temperature, rinsed three times in PBS and counterstained with DAPI. Negative controls with the secondary antibodies were carried out in all cases. Cells were observed with a Leica DMI4000 B inverted microscope, and pictures were taken with a Leica DFC340FX camera.

Western blot

Cells were collected and homogenized in RIPA modified buffer: 20 mM Tris (pH 7.5), 150 mM NaCl, 10 mM EDTA, 1% Triton X-100, NaF 0.5 M, proteinase inhibitor (SIGMA), and phosphatase inhibitor (SIGMA). Protein concentration was assessed by using BCA Protein Assay Kit (Pierce). Equal amount of cell extracts was resolved in SDS-PAGE. The following primary antibodies were used: EGFR (1/500, 1005SC-03; Santa Cruz), pPLCγ1 (Tyr 775) (1:1000, ECM Biosciences), pStat3 (1:1000, Cell Signaling), pAKT (1:1000, Cell Signaling) 1:500; and pERK (phospho p44/42) (1:500, Cell Signaling), PLCγ1 (1:1000, Cell Signaling), Stat3 (1:1000, Cell Signaling), AKT (1:1000, Cell Signaling), ERK (p44/42) (1:1000, Cell Signaling). The following secondary detection was used: goat anti-rabbit (1:5000, Pierce). Bands were visualized using an ECL kit (Amersham Biscience).

MTT viability assay

Cells were plated in 24 well plates at 3,000 cells per well. The following day, cells were processed or starved for 1 or 2 days. MTT [3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium Bromide] was given to the cells for 4 h. Cells were lysed and kept at 37°C overnight. Viability was assessed by taking optical density readings at 560 nm as this reading is directly proportional to viable cells. The experiments were repeated three times.

Cell cycle analysis

Cells were collected at the end of treatment by trypsinization (0.25% Trypsin, GIBCO) for 5 min at 37°C, washed twice with PBS + 2%FBS and fixed in 70% ethanol for 1 h at 4°C. Afterwards, cells were washed in PBS and incubated with RNase-Free DNase (40 μg/ml) and propidium iodide (50 μg/ml) for 1 h. Samples were analyzed for DNA content by using a FACScan Flow Cytometry (Becton Dickenson, San Jose, Ca, USA). Each experiment was repeated three times.

Apoptosis assay

AnnexinV/PI (BD Biosciences) staining assay was used to detect early and late apoptotic cells following the manufacturer’s recommendations. After incubation in different conditions, cells were collected by trypsinization (0.25% Trypsin, GIBCO) for 5 min 37°C, harvested and washed two times with PBS + 2%FBS. Cells were resuspended in 400 µl of Annexin buffer solution. Annexin V-APC/PI (BD Biosciences) staining solution was added and incubated for 15 min, and cells were analyzed in a FACScan flow cytometry analyzer (Becton Dickenson, San Jose, Ca, USA). A total of 10,000 cells were analyzed per sample, and the experiments were repeated three times.

Transmigration assay

100.000 C17.2 cells were placed onto 8 μm pore size inserts and the inserts placed in a 12-well plate (BD Bioscience). Cells were incubated at 37°C for 24 h. For the experiments in serum free media, membranes were coated with low concentrations of MatrigelTM (BD Biosciences) to allow cells to attach without clogging the pores. Afterwards, cells in both upper and lower chambers were trypsinized and counted using a hemocytometer.

Differentiation assay

Differentiation assay was performed as previously described [34]. Briefly, cells were grown in M2 media until 90% confluence. M2 media was exchanged for differentiation media (see Supplemental Data) and cells were grown under differentiation conditions for 7 days. Media was changed every day, cells were fixed and differentiation markers were assayed.

Statistical analysis

Data are expressed as means ± SEM (standard error of the mean). Unpaired Student t-tests were performed and significance was defined by P-values equal to or less than 0.05.

Results

wtEGFR, and EGFRvIII can be stably expressed in C17.2 cells

Human wtEGFR, and EGFRvIII cDNAs were cloned under the control of the CMV promoter. Orientation was confirmed by sequencing, and no mutations were detected (Fig. 1a). Retroviruses were generated carrying wtEGFR, EGFRvIII or an empty vector control and parental NSCs were infected. Single Cells were FACS-automatic sorted based on GFP expression. Ten percent of the colonies survived after 1 week. Clones with the highest transgene expression were selected (Fig. 1b).
https://static-content.springer.com/image/art%3A10.1007%2Fs11060-009-0035-x/MediaObjects/11060_2009_35_Fig1_HTML.gif
Fig. 1

C17.2 NSCs Overexpressing wtEGFR and EGFRvIII. a wtEGFR and ΔEGFR DNAs were cloned into the retroviral backbone pLIRE-Ires-EGFP plasmid. b Retrovirus with wtEGFR or ΔEGFR cDNAs were used to infect C17.2 NSCs. Individual clones were FACS sorted and those with higher GFP expression were selected. c RT-PCR with wtEGFR or ΔEGFR specific primers revealed the presence of wtEGFR and ΔEGFR mRNA expression in the selected C17.2wtEGFR and C17.2ΔEGFR clones respectively. (1) C17.2 parental cDNA; (2) C17.2wtEGFR cDNA; (3) C17.2ΔEGFR cDNA; (4) Empty vector; (5) wtEGFR positive control (pLIREwtEGFR); (6) ΔEGFR positive control (pLIREΔEGFR) and (7) negative control (DNA−). d Protein from C17.2 parental, C17.2wtEGFR and C17.2ΔEGFR cells was isolated and probed with specific HA and ΔEGFR antibodies. EGFR and ΔEGFR were detected in C17.2wtEGFR and C17.2ΔEGFR clones respectively. e Comparative analysis of EGFR and ΔEGFR over-expression in human GBM tissue samples and EGFR/ΔEGFR transgenic C17.2 cells. The diagram shows that wtEGFR and ΔEGFR expression levels in C17.2wtEGFR and C17.2ΔEGFR respectively fell into the range of expression for these genes in human GBM. f HA and ΔEGFR specific antibodies were used to detect protein. Both EGFR and ΔEGFR were located in the nucleus and in the cytoplasmic membrane. Please note EGFRvIII and ΔEGFR are used interchangeably and refer to the most common EGFR variant.

RT-PCR was performed to detect the presence of wtEGFR or EGFRvIII in our C17.2 subclones (Fig. 1c). Using western blots, wtEGFR and EGFRvIII protein was detected in C17.2wtEGFR NSCs and EGFRvIII clones, respectively. No significant amount of protein was found in C17.2 parental NSCs or in NSCs infected with the empty vector controls (Fig. 1d). Immunocytochemistry with antibodies targeting the wtEGFR or human EGFRvIII proteins, specifically using anti HA-tag to detect wtEGFR and the human antibody L8A4 to detect EGFRvIII, revealed levels of GFP that correlated with the levels of either wtEGFR or EGFRvIII protein. As expected, wtEGFR and EGFRvIII protein was present around the nucleus and in the cell membrane (Fig. 1f).

To determine whether the levels of either wtEGFR or EGFRvIII protein expression in the NSC lines were physiologically relevant to human GBM, RNA was isolated from six human GBMs and compared with tumor wtEGFR and EGFRvIII levels in our established NSC lines. Three human GBMs over-expressed wtEGFR and two GBMs also expressed detectable EGFRvIII mRNA levels. Using QRT-PCR, we compared the EGFR and EGFRvIII mRNA expression levels in all four C17.2 clones (parental, empty vector, wtEGFR and EGFRvIII over-expressing) with those found in our human GBM specimens. The level of either wtEGFR or EGFRvIII mRNA in C17.2wtEGFR and C17.2 EGFRvIII cell lines was similar to levels of EGFR and EGFRvIII mRNA detected in human GBM tumors. wtEGFR or EGFRvIII mRNA expression in C17.2empty and parental cell lines was not observed (Fig. 1e).

Activation of EGFR signaling accelerates proliferation of NSCs

We hypothesized that the over-expression of wtEGFR or EGFRvIII conferred a proliferative advantage to NSCs. To test this hypothesis, MTT (3-(4,5 dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide) Cell Proliferation Assays were carried out in order to give a quantitative, convenient method for evaluating our NSCs response to external factors such as the addition of EGF ligand. In the absence of ligand and after 4 days in culture under serum-free conditions, we found a significant increase in NSCs over-expressing EGFRvIII and wtEGFR as compared to parental controls (Fig. 2a). The effect of adding EGF was notable after 2 days in culture with an increase in viable cells in all three clones.
https://static-content.springer.com/image/art%3A10.1007%2Fs11060-009-0035-x/MediaObjects/11060_2009_35_Fig2_HTML.gif
Fig. 2

GFP does not affect the biology of C17.2 NSCs. a MTT assay to compare proliferation of C17.2 parental and C17.2 empty cells. Both cell lines show similar proliferation patterns. Cells were plated in 24 well/plates. 24 h later cells were processed or starved for 1, 2, 3, 4 and 5 days. Proliferation patterns of wtEGFR, EGFRvIII and Parental are imported from experiments shown on Fig. 4 for comparative means. b wtEGFR and EGFRvIII signaling provides a growth advantage to C17.2 NSCs. Cells were plated in 24 well/plates. 24 h later cells were processed or starved for 1, 2, 3, 4 and 5 days. C17.2 parental, C17.2wtEGFR and C17.2EGFRvIII cellular proliferation in serum-free media was analyzed using MTT assay (see “Materials and methods”). AG1478 is diluted in DMSO. To check for any influence of DMSO in the result, controls with DMSO were included in the experiment. The experiments were repeated three times. Results are expressed as mean ± SEM. Please note EGFRvIII and ΔEGFR are used interchangeably and refer to the most common EGFR variant

To investigate whether EGFR pathway activation was causally responsible for the proliferative advantage in NSCs over-expressing wtEGFR or EGFRvIII, we repeated the MTT assay in the presence of the small molecule EGFR tyrosine kinase inhibitor, AG1478 (Fig. 2b). The addition of AG1478 returned the proliferation level of all subclones to below background levels. These results demonstrate that in the absence of ligand, there is a significant proliferative advantage in NCS over-expressing the constitutively active EGFR mutant EGFRvIII and, to lesser extent, wtEGFR.

In order to ensure that GFP expression alone (empty vector) would not influence the proliferation of C17.2 cells, our MTT assay was also done using C17.2 empty vector controls (GFP expressing cells) and these results were compared with those results obtained for C17.2 parental cells (no GFP expression). After 5 days in culture, we did not find a significant difference in terms of proliferation (Fig. 2a) due to GFP expression alone.

To assess whether GFP expression posed any toxicity to C17.2 cells, our NSCs subclones were assayed for increased oxidative stress by using the fluorescent probe hydroethidine (HE) technique for monitoring the intracellular superoxide production and for TMRM to determine the level of mitochondrial membrane depolarization. The level of fluorescence corresponding both to HE (C17.2 parental: 39,3 ± 22,2 FAV; C17.2empty: 38,3 ± 22,2 FAV) and to TMRM (C17.2parental: 26,5 ± 13 FAV; C17.2empty: 34,4 ± 10 FAV) was not statistically different (See Supplemental Data Fig. 1). These results suggest that GFP expression alone does not cause any NSC toxicity.

Based on these observations, C17.2parental cells were used as an experimental control to evaluate the effect of activated EGFR signaling on C17.2 NSC biology.

Downstream mediators of EGFR activation in NSCs

To test the hypothesis that over-expression of EGFRvIII or wtEGFR induced the activation of particular downstream pathways in NSCs, western blots were performed in the absence or presence of EGF ligand. In C17.2ΔEGFR NSCs, constitutive upregulation of pPLCγ, pERK and pAKT was observed, but not of pSTAT3. With the addition of EGF, both untransfected C17.2 parental cell line and transfected C17.2EGFRvIII NSCs demonstrated similar phosphorylation patterns which included a time dependant decrease in pSTAT3, pPLCγ, pERK and pAKT (Fig. 3b). In contrast, the addition of EGF in C17.2wtEGFR NSCs produced a sustained activation of pSTAT3, pPLCγ, pERK and pAKT. A constitutive upregulation of pPLCγ was consistently found in C17.2wtEGFR irrespective of the presence of EGF ligand.
https://static-content.springer.com/image/art%3A10.1007%2Fs11060-009-0035-x/MediaObjects/11060_2009_35_Fig3_HTML.gif
Fig. 3

Activation of EGFR downstream pathways in C17.2 cells over-expressing EGFR and EGFRvIII. a Schematic diagram showing the EGFR downstream pathways. b C17.2 cells and were cultured until 60–80% density. Afterwards, cells were serum starved for 24 h, and 10 ng/ml of EGF was added. Whole cell lysates were collected after 0, 10, 30 and 60 min, resolved by SDS-PAGE, and immunoblotted with phosphospecific antibodies. The membranes were stripped and re-probed with non-phosphorylated antibodies to measure total levels of proteins. Equal amount of protein was loaded. c C17.2 cells were cultured until 60–80% density. Afterwards, cells were serum starved for 24 h. Cultures were pretreated for 2 h with increasing concentration of AG1478 followed by 15 min incubation with 10 ng/ml of EGF. Please note EGFRvIII and ΔEGFR are used interchangeably and refer to the most common EGFR variant

To better delineate the effect of EGFR activation on these pathways, NSCs were treated with increasing concentrations of the EGFR inhibitor, AG1478, in the presence of EGF ligand. The addition of AG1478 was associated with inhibited phosphorylation of pSTAT3, pPLCγ, pERK and pAKT in all groups of cells. Slow dephosphorylation of pSTAT3, pPLCγ, pERK and pAKT was encountered as the concentration of AG1478 increased in C17.2parental and EGFRvIII cells (Fig. 3c). In C17.2wtEGFR cells, the dephosphorylation only occurred with a higher concentration of the AG1478. pSTAT3 reached baseline levels with inhibitor concentrations higher than 5 μM. pPLCγ was inhibited with low AG1478 concentration (0.1 μM) in C17.2 parental and in EGFRvIII cells and above 0.25 μM in wtEGFR cells. pAKT was inhibited at 2 μM and, pERK was dephosphorylated above 2 μM in C17.2 parental and EGFRvIII cells and 5 μM in C17.2wtEGFR cells. AG1478 inhibited phosphorylation of STAT3, PLCγ, ERK and AKT at 5 μM due to EGFR and/or EGFRvIII activation in all NSCs.

To better understand the role of activated EGFR signaling on cell cycle progression, we quantified S-phase progression by quantifying BrdU incorporating in NSCs over-expressing wtEGFR or EGFRvIII. In the absence of EGF ligand, wtEGFR and EGFRvIII activation increased the number of NSCs that went through the S-phase as evidenced by BrdU incorporation (see Supplemental Data Fig. 2). In order to evaluate whether an increased percentage of cells entering S-phase was related to induction of re-entry and/or an acceleration of cell cycle, NSCs were fixed, stained with Propidium Iodide (PI) and DNA content was analyzed by flow cytometry (FACS) (Fig. 4a). Consistent with the BrdU incorporation data, in serum containing media, a significantly higher percentage (64%, SEM ± 1%, n = 3) of C17.2EGFRvIII NSCs were in the S-phase and G2/M as compared with parental controls (58%, SEM ± 1%, n = 3). In addition, C17.2wtEGFR NSCs also showed a higher percentage (61%, SEM ± 1.5%, n = 3) of cells in S-phase and G2/M, but this was not statistically significant. In both NSC subclones, the percentage of cells in G1 was lower than in controls suggesting an acceleration of G1 phase. After 24 h in serum-free media, all cells experienced arrest in G1. However, both C17.2EGFRvIII and C17.2wtEGFR, NSCs showed a significantly higher percentage of cells in S-phase, and reduced number in G1 as compared with parental controls. After 48 h in the same conditions, the differences were not significant. Theses results suggest that EGFRvIII, and to a lesser extent wtEGFR, induce proliferation of postnatal NSCs by accelerating G1 of the cell cycle.
https://static-content.springer.com/image/art%3A10.1007%2Fs11060-009-0035-x/MediaObjects/11060_2009_35_Fig4_HTML.gif
Fig. 4

EGFR activation accelerates G1 Phase and increases NSC survival in C17.2EGFRvIII and C17.2wtEGFR. a The cell cycle distribution of C17.2 parental, C17.2wtEGFR and C17.2EGFRvIII cells is shown. Cells were grown under different conditions for 24 and 48 h. Cells were at 80% confluence when harvested. The results represent the average of three different plates. 10 ng/ml of EGF and 5 μM of AG1478 were added when necessary. b Apoptosis assay carried out by staining with annexinV/PI. Cells were grown for 48 h under specific conditions. Afterwards, they were collected and analyzed for apoptotic cells (see “Materials and methods”). EGFRvIII and to a lesser extent wtEGFR increased survival in NSCs. Experiments were repeated three times and are expressed as mean ± SEM. * P < 0.05, ** P < 0.005. c Semi-quantitative PCR showing gene expression levels for cell cycle and apoptotic markers. P27 and bad and bax are downregulated in NSC overexpressing EGFR or EGFrvIII. d p27, Bad and Bax gene expression analysis with or without AG1478 assayed by QRT-PCR. All the data were analysed by using the 2−ΔΔCT method. Two housekeeping genes (GAPDH and β-Actin) were used to normalize the data. All data are expressed as increased/decreased folds (relative Units). Keyword: M2, media M2; S, serum free; h, hour. Experiments were repeated three times and are expressed as mean ± SEM. ** P < 0.05; A-AG1478. Please note EGFRvIII and ΔEGFR are used interchangeably and refer to the most common EGFR variant

The effect of adding EGF ligand with or without an EGFR inhibitor on cell cycle progression was then assessed. After 24 h of exposure to EGF in serum-free media, a significant reduction was observed in the percentage of cells in G1 phase as compared with cells grown in the absence of EGF. After 48 h of ligand exposure, the percentage of C17.2wtEGFR NSCs in S phase increased, while the percentage in G1 decreased. The addition of AG1478 resulted in G1 arrest in NSCs from all three groups.

To better understand the mechanism by which either EGFR or EGFRvIII accelerates cell cycle progression, we checked for the expression of the cell cycle regulatory protein p21, p16 and p27. We did not observe significant changes in p21 and p16 gene expression. However, the expression of p27 was significantly downregulated in both C17.2 wtEGFR and EGFRvIII-over-expressing NSCs. This was reversed by the addition of 5 µM AG1478 suggesting that cell cycle regulation in NSCs overexpressing EGFR or EGFRvIII may occur through the downregulation of p27 (Fig. 4c).

EGFR activation enhances survival in NSCs

We then assessed whether the increase in cell number of C17.2EGFRvIII and wtEGFR NSCs, in addition to proliferation, were also a consequence of increased survival as a result of EGFR activation. NSCs were cultured and stained with annexin V and PI, and the number of cells that incorporated PI or stained for annexin V was determined (Fig. 4b). In serum-containing media there were no significant differences in NSC survival between the three groups. In serum-free media, however, 50% (SEM ± 2.5) of C17.2 parental cells entered early or late apoptosis. In contrast, wtEGFR over-expression increased NSC survival by 70% (SEM ± 1) and EGFRvIII over-expression by 85% (SEM ± 1) (Fig. 4b). The addition of EGF ligand did not significantly affect the percentage of surviving cells, whereas the addition of AG1478 significantly increased the number of apoptotic cells in the subclones. These results suggest that activated EGFR signaling improves survival and prevents apoptosis of post-natal NSCs. The reduction of surviving cells in the presence of AG1478 further implies that the survival advantage conferred to NSCs is mediated by activated EGFR signaling.

Additionally we checked for the expression of the proapoptosis markers Bax and Bad. After 48 h of growth in serum free media, we observed an increase of Bax and Bad gene expression in the parental C17.2 NSCs as compared to C17.2wtEGFR or EGFRvIII NSCs (Fig. 4c). When grown in the presence of the EGFR inhibitor 5 µM AG1478, Bad gene expression levels were significantly increased suggesting that the increased survival of NSCs overexpressing EGFR or EGFRvIII might be conferred through downregulation of the Bad pro-apoptotic molecule (Fig. 4d).

wtEGFR but not EGFRvIII signaling induces chemotaxis in C17.2 NSCs

In an effort to determine whether activated EGFR signaling increases the motility of postnatal NSCs, transmigration assays were performed under different growth conditions (Fig. 5a). Chemotaxis experiments involved adding serum-containing media to both the lower and the upper chambers of a transwell assay. Under these conditions, a low percentage of C17.2parental (16%, SEM ± 4, n = 3) or EGFRvIII (13%, SEM ± 4, n = 3) NSCs transmigrated. A significantly higher (36%, SEM ± 6, n = 3) percentage of C17.2wtEGFR NSCs transmigrated in the assay. Addition of EGF ligand to both the upper and lower chambers did not alter the number of transmigrated cells. Creation of an EGF gradient by adding EGF only to the lower chamber resulted in a significant increase in the percentage of transmigration by C17.2wtEGFR (62%, SEM ± 6, n = 3) NSCs, while a consistently similar percentage of C17.2parental (31%, SEM ± 4, n = 3) and EGFRvIII (36%, SEM ± 8, n = 3) NSCs transmigrated. The addition of AG1478 to the lower chamber reduced the percentage of transmigrated wtEGFR NSCs to numbers lower than those observed with just media. These findings suggest EGF may play a role in inducing chemo-attraction in postnatal NSCs, with over-expression of wtEGFR significantly increasing this chemotaxis (Fig. 5b).
https://static-content.springer.com/image/art%3A10.1007%2Fs11060-009-0035-x/MediaObjects/11060_2009_35_Fig5_HTML.gif
Fig. 5

wtEGFR but not EGFRvIII signaling increases NSCs chemotaxis. a A schematic representation of transmigration assay. Cells were collected, counted and re-suspended in the appropriated media. 1 × 105 cells were plated per well and growth for 24 h at 37°C. Cells in the upper and lower chambers were collected by trypsinization and counted. b Cells grown in FBS-containing media on uncoated membranes. c Cells grown in serum-free media on matrigel-coated membrane. wtEGFR increased directed migration of NSCs in the presence of EGFR ligand. The experiments were repeated at least three times and the results are expressed as mean ± SEM. * P < 0.05, ** P < 0.005. Please note EGFRvIII and ΔEGFR are used interchangeably and refer to the most common EGFR variant

The same experiments were done with serum free media in order to demonstrate that the chemotactic effect observed was not simply the result of the presence of growth factors in the serum. A higher percentage of transmigrated cells for both C17.2EGFRvIII and wtEGFR NSCs were observed when compared to parental controls. Similar to the previous findings, the generation of an EGF gradient resulted in a significant increase in the percentage of transmigrated cells. The addition of AG1478 reduced the percentage of transmigrated cells bellow the number observed with media alone (Fig. 5c). These additional findings support that the directional chemotaxis observed may be a function of EGFR activation. In addition, the higher percentage of transmigration exhibited by C17.2EGFRvIII NSCs in the absence of EGF or EGF gradient suggests that EGFRvIII over-expression might induce an increased random motility in post-natal C17.2 NSCs.

EGFR and EGFRvIII over-expression prevents differentiation of NSCs

To evaluate the effect of wtEGFR or EGFRvIII over-expression on the differentiation potential of NSCs, cells were maintained in differentiation medium for 7 days and then assessed for nestin (undifferentiated stem or progenitor cell), Tuj1 (neuron), and A2B5 (glial progenitors) (Fig. 6). All undifferentiated NSCs were positive for nestin. After differentiation, nestin was detected in 26–31% of C17.2 parental, wtEGFR and EGFRvIII over-expressing NSCs. Over 25 ± 4% of C17.2parental cells differentiated into neurons (Tuj1+), while 20 ± 7% differentiated into glial progenitor cells (GPCs) (A2B5+). In contrast, no Tuj1 expressing cells were observed in either C17.2wtEGFR or C17.2EGFRvIII NSCs. Instead, 45 ± 12% of C17.2wtEGFR NSCs and 26 ± 7% of C17.2EGFRvIII NSCs expressed A2B5 (Fig. 6a, b). These results suggest a role of EGFR activation in blocking neuronal differentiation of post-natal NSCs.
https://static-content.springer.com/image/art%3A10.1007%2Fs11060-009-0035-x/MediaObjects/11060_2009_35_Fig6_HTML.gif
Fig. 6

wtEGFR and EGFRvIII signaling blocks neuronal differentiation in NSCs. Cells were cultured until 80% density and serum-starved for 7 days. Media was changed every other day. After 7 days, cells were fixed and stained with specific markers. a Before adding differentiation media 100% of the cells were Nestin+. b After 7 days in starvation media, the percentage of Nestin + cells dropped to 26–31%. Tuj1 + cells were detected in C17.2 parental cells, but not in C17.wtEGFR or C17.2EGFRvIII cells. Most C17.2wtEGFR cells were A2B5 positive after differentiation, while no more than 20–26% A2B5 positive cells were detected in C17.2parental and C17.2EGFRvIII NSCs. c Tuj1 gene expression analysis was done by QRT-PCR. All the data were analysed by using the 2−ΔΔCT method. Two housekeeping genes (GAPDH and β-Actin) were used to normalize the data. All data are expressed as increased/decreased folds (relative Units). The expression level of the neuronal Tuj1 gene was significally higher in parental C17.2 NSCs as compared with C17.2 wtEGFR (Fig. 6d) and EGFRvII (Fig. 6e) over expressing NSCs. Growing the cells in the EGFR inhibitor AG1478 increased the levels of Tuj1 in both C17.2 wtEGFR NSCs and C17.2EGFRvIII NSCs suggesting a role for EGFR in the expression of neuronal markers. The experiments were repeated three times. Keywords: M2, media M2; S, Serum free; h, hour; *, P < 0.005; **, P < 0.05; A-AG1478. Please note EGFRvIII and ΔEGFR are used interchangeably and refer to the most common EGFR variant

We then determined whether the expression levels of neuronal differentiation markers would be reversed by EGFR inhibition. At baseline, the expression level of the neuronal Tuj1 gene was significally higher in parental C17.2 NSCs as compared with C17.2 wtEGFR and EGFRvII over expressing NSCs at all time points (Fig. 6c). Growing the cells in 5 µM AG1478 increased the levels of Tuj1 in C17.2wtEGFR/vIII NSCs up 24 h (Fig. 6d, e).

Discussion

Variations of EGFR, whether in the form of over-expression, amplification or mutation, are significant factors for glioma progression, with important implications in tumor cell proliferation, migration, differentiation and resistance to therapy. Our laboratory has previously expressed EGFRvIII in C17.2 NSCs and found that the constitutive kinase activity of EGFRvIII sustains an immature cell phenotype and enhances NSC migration [35]. Though complementary, these previous studies, however, did not provide insight into the tumorigenic implications of the amplification and/or mutation of EGFR [35]. Recent findings have shown that brain tumors contain undifferentiated neural precursors reminiscent of NSCs and that these cells can be isolated from the brain tumor mass. Several studies suggest that these GBM TSCs bear remarkable similarity to normal NSCs in that they have the NSC characteristics of self-renewal, multipotency and generation of distinct progeny [36].

We sought to better delineate the role of wtEGFR and EGFRvIII in modulating the aggressive, characteristics of NSCs by developing stable subclones of C17.2 murine NSCs that over-express wtEGFR or EGFRvIII. The C17.2 cell line was immortalized from a neural progenitor cell [37, 38] with a retrovirus carrying v-myc. Therefore, the cell line itself is not a “normal” NSC in that it has been transformed by v-myc. However, we believe that it is a suitable cell line to test our hypotheses as the cell line derives from a postnatal germinal zone stem or progenitor cell [39, 40]. In order to control for the expression of v-myc, all control experiments were done using either parental cell or empty vector controls that also expressed v-myc.

We found that activated EGFR signaling in immortalized NSCs was associated with the activation of those downstream mediators widely implicated in GBM tumor progression. Higher constitutive levels of phosphorylated mediators downstream of EGFR have been previously reported in mice fibroblasts [41] and in GBM cell lines [42]. It is possible that the high density of the EGF receptor on the NSC membrane might confer a low level of dimerization and activation even in the absence of ligand. Specific inhibition of activated EGFR completely down regulated the phosphorylation of these intermediates implicating EGFR signaling is significantly important to this pathway activation. Further studies will be needed to determine whether the activation of AKT, ERK and PLCγ pathways in C17.2EGFRvIII NSCs is secondary to the direct activation by EGFRvIII or through dimerization and transactivation of other endogenous EGFR family members. Taken together, these data suggest that cell signaling pathways commonly implicated in glioma progression are important in postnatal NSCs that have activated EGFR signals. Therefore, therapies that target these pathways may be important in eliminating TSCs in GBM or in inhibiting the recruitment of endogenous NSCs to the brain tumor mass.

We found that activated EGFR signaling in NSCs was associated with increased NSC proliferation. EGFRvIII over-expression has been previously shown to significantly promote proliferation in several cell lines [12, 4244] in serum-free media, while wtEGFR over-expression had a lesser effect on proliferation under the same conditions [42]. However, wtEGFR also increased NSC proliferation when compared to parental controls, albeit to a lesser extent than EGFRvIII. This significant increase in NSC proliferation mediated by EGFRvIII and wtEGFR may have been due to the acceleration of the G1 phase or the induction of survival signals and we show that this cell cycle regulation occurs through downregulation of p27, which is reversed in the presence of the EGFR inhibitor AG1478. The consequential reduction in the percentage of C17.2EGFRvIII NSCs in the G1 phase after 24 h in serum-free media and increased cell survival after 48 h were significant compared to C17.2parental cells. While the addition of EGF had the same effect on C17.2EGFRvIII and C17.2parental cells, it significantly decreased the percentage of C17.2wtEGFR in the G1 phase after 48 h, with no significant effect on survival. These results are consistent with those found in other cell lines [10, 43, 44], whereby activated EGFR signaling was also associated with a decrease in the percentage of apoptotic cells and increased cell survival. We show that this survival advantange may be, mediated via downregulation of the pro-apoptotic proteins Bad and Bax. If activated EGFR signaling enhances the survival of NSCs, this may have important implications for TSC survival following radiation and chemotherapy and may be important in the recruitment of endogenous NSCs into the brain tumor mass.

EGFR signaling has been shown to play a very important role in the motility of both neural and non-neural cells. Increased migration and invasion is likely a central feature of GBM TSCs and may be important in the recruitment of endogenous NSCs into the tumor mass. EGFRvIII increased random motility in fibroblasts (NR6) and in glioma cell lines (U87). After the addition of EGF, NR6 cells with low levels of wtEGFR showed increased random motility [41]. EGFR-mediated chemotaxis of NSCs has been previously reported in experiments in vivo and in vitro [45]. In the absence of EGF, we observed a higher percentage of transmigrating C17.2wtEGFR cells when compared to C17.2parental cells. C17.2EGFRvIII NSCs also showed a higher percentage of transmigrating cells in the absence of EGF. The higher percentage of C17.2wtEGFR transmigrating cells observed was irrespective of the presence of serum in the media. While the addition of EGF to both chambers did not alter the percentage of transmigrated cells, it did increase the total number of cells within the chambers. Furthermore, the formation of an EGF gradient strongly increased the chemotaxis of C17.2wtEGFR NSCs. These findings suggest that activated EGFR signaling in NSCs could induce migration toward sources of EGFR ligand. Such an implication is relevant if EGFR signaling plays a role in TSC dispersal or in the recruitment of endogenous NSCs to a brain tumor mass.

EGFR signaling has been shown to play an important role in determining neural cell fate [46]. EGF infusion into the lateral ventricle of the SVZ has been shown to increase the number of newborn glia and decrease the number of newborn neurons [47]. Activated EGFR signaling confers to progenitor cells the ability to differentiate into astrocytes [48]. Downstream of EGFR, it has been demonstrated that neuronal terminal differentiation was inhibited by activated STAT3 in Lif-deficient mice [49]. STAT3 knockdown in C17.2 cells drove them to neuronal differentiation without generating glial cells [50]. Our immunocytochemistry results showed that C17.2wtEGFR and C17.2vIII EGFR failed to differentiate into neurons but were successful in differentiating into glial progenitors. By gene expression analysis we observed little expression of the neuronal marker Tuj1 in wtEGFR and EGFRvIII over-expressing C17.2 NSCs when compared with parental C17.2NSCs. Growing the cells in the presence of the EGFR inhibitor increased the levels of Tuj1 gene expression in C17.2wtEGFR and EGFRvIII NSCs up to 24 h suggesting an EGFR mediated role in the expression of neuronal genes.

Conclusion

Whether malignant gliomas derive from the transformation of undifferentiated NSCs remains to be determined. NSCs may be the cell-of-origin of GBM tumor-initiating cells or may play a role in recruiting progenitors to the tumor mass [27]. EGFR over-expression and mutation, in collaboration with other mutations, is known to play an important role in the maintenance and progression of malignant gliomas. We demonstrate that activated EGFR signaling in post-natal NSCs induces aggressive characteristics, including enhanced proliferation, survival and migration. Over-expression of the most common mutant, EGFRvIII, also increased NSC proliferation and survival while at the same time it blocked neuronal differentiation and promoted a glial fate. wtEGFR over-expression increased proliferation and survival in the absence of ligand, with the effect becoming even more profound with EGF. wtEGFR over-expression also blocked neuronal differentiation in NSCs and caused a dramatic increase in chemotaxis in the presence of an EGF gradient. Since evidence suggests that GBM TSCs bear remarkable similarity to NSCs, this knowledge of NSC biology may be important to understanding the deregulated pathways in GBM TSCs. It is possible that in GBM, EGFRvIII over-expression in TSCs increases proliferation and survival and modulates TSC fate, irrespective of exposure to EGF ligand. Alternatively, wtEGFR over-expression may increase TSC proliferation when the cells are exposed to EGF, and increase TSC chemotaxis in the presence of a ligand gradient, thus allowing for TSCs to spread throughout the brain parenchyma. Evaluating patients’ TSC subpopulation for activated EGFR signaling may lend further insight into the role of EGFR in TSC biology. Small molecular therapies, targeted to the TSC subpopulation, may effectively inhibit EGFR mediated TSC proliferation, survival and migration and may improve anti-cancer strategies in patients with GBM.

Acknowledgements

We are especially grateful to Dr. Glickstein, Dr. Musatov, Dr. Falcon, and Dr. Heck for their technical advice and FACS core facility at CIPF. Angel Ayuso-Sacido is a recipient of a Postdoctoral Fellowships from Fulbright/MEC program, RETIC, Ministerio de Educacion y Ciencia, Spain. This work was supported in part by grants from The American Brain Tumor Association (JAB), the American Association of Neurological Surgeons (JAB), the Starr Foundation Stem Cell Award (JAB) and K08 CA130985-01A1 (JAB) from the NIH/NCI.

Supplementary material

11060_2009_35_MOESM1_ESM.pdf (636 kb)
Supplementary material 1 (PDF 635 kb)

Copyright information

© Springer Science+Business Media, LLC. 2009