Microglial Stimulation of Glioblastoma Invasion Involves Epidermal Growth Factor Receptor (EGFR) and Colony Stimulating Factor 1 Receptor (CSF-1R) Signaling
- 7 Downloads
Glioblastoma multiforme is a deadly cancer for which current treatment options are limited. The ability of glioblastoma tumor cells to infiltrate the surrounding brain parenchyma critically limits the effectiveness of current treatments. We investigated how microglia, the resident macrophages of the brain, stimulate glioblastoma cell invasion. We first examined the ability of normal microglia from C57Bl/6J mice to stimulate GL261 glioblastoma cell invasion in vitro. We found that microglia stimulate the invasion of GL261 glioblastoma cells by approximately eightfold in an in vitro invasion assay. Pharmacological inhibition of epidermal growth factor receptor (EGFR) strongly inhibited microglia-stimulated invasion. Furthermore, blockade of colony stimulating factor 1 receptor (CSF-1R) signaling using ribonucleic acid (RNA) interference or pharmacological inhibitors completely inhibited microglial enhancement of glioblastoma invasion. GL261 cells were found to constitutively secrete CSF-1, the levels of which were unaffected by epidermal growth factor (EGF) stimulation, EGFR inhibition or coculture with microglia. CSF-1 only stimulated microglia invasion, whereas EGF only stimulated glioblastoma cell migration, demonstrating a synergistic interaction between these two cell types. Finally, using PLX3397 (a CSF-1R inhibitor that can cross the blood-brain barrier) in live animals, we discovered that blockade of CSF-1R signaling in vivo reduced the number of tumor-associated microglia and glioblastoma invasion. These data indicate that glioblastoma and microglia interactions mediated by EGF and CSF-1 can enhance glioblastoma invasion and demonstrate the possibility of inhibiting glioblastoma invasion by targeting glioblastoma-associated microglia via inhibition of the CSF-1R.
Glioblastoma multiforme (grade IV) remains one of the most difficult cancers to treat, with a median survival rate of 12 months after diagnosis (1,2). This result is primarily because high-grade glioblastomas are highly locally invasive and, as a result, a relatively large population of invading cells is left behind after surgical resection (3). Therefore, targeting the motile properties of this tumor is an attractive strategy for therapeutic intervention. Although progress has been made in elucidating the genetic alterations and oncogenic signaling pathways responsible for glioblastoma, the molecular mechanisms of glioblastoma cell invasion are still largely undefined.
It is becoming increasingly clear in the field of cancer biology that the tumor microenvironment, which includes stromal cells such as fibroblasts and cells of the immune system, plays a critical role in disease progression (4). Tumor-associated macrophages are an important class of cells in the microenvironment that can potentiate carcinoma cell invasion and metastasis (5, 6, 7, 8).
There is emerging evidence that the resident macrophages of the central nervous system, microglia, may play a similar role in aiding glioblastoma progression. Glioblastoma tumors contain substantial numbers of microglia and/or macrophages, especially at the invasive border (9). Brain slices cultured ex vivo and depleted of microglia using clodronate liposomes are refractory to glioblastoma cell invasion (10,11). Microglial-secreted transforming growth factor (TGF)-β can act as a chemoattractant for glioblastoma cells in in vitro invasion assays (12). Importantly, recent studies have reported that in vivo ablation of microglia results in a significant reduction in glioblastoma tumor size (13,14). Interestingly, there also is genetic evidence that microglia promote the growth of optic glioma in Nf1+/− mice (15). However, the glioblastoma-produced factors that influence microglial behavior remain undefined. In this study, we demonstrate a role for epidermal growth factor receptor (EGFR) and colony stimulating factor 1 receptor (CSF-1R) signaling during microglia-stimulated glioblastoma invasion.
Materials and Methods
Cell Culture and Reagents
The cells used in this study were GL261 murine glioblastomas (obtained from the National Cancer Institute, Frederick, MD, USA), U251 human glioblastomas (American Tissue Culture Collection, Manassas, VA, USA) and normal murine microglia originally isolated from C57Bl/6J mice, as previously described by Dobrenis (16). Primary microglia cultures were generated from high-density mixed cell-type cultures of neonatal neocortex by differential adhesion methods producing highly purified (>99%) microglial populations, as assessed by cell-type specific markers including F4/80 (13). To further maximize and ensure purity for experiments, the isolated cells were subcultured an additional three times with stringent selective adhesion on nontissue culture-treated “suspension cell” plates (Sarstedt) to further limit nonmicroglial cells. All cultures were maintained in Macrophage-SFM (MSFM; Life Technologies Corporation, Grand Island, NY, USA) with 10% fetal calf serum. Microglia were supplemented with 10 ng/mL recombinant mouse granulocyte macrophage-colony-stimulating factor (GM-CSF) (R&D Systems, Minneapolis, MN, USA). All cells were cultured in a humidified incubator containing 5% CO2 at 37°C. Recombinant human colony stimulating factor 1 (CSF-1) was a gift from Chiron Corporation. Iressa was obtained from AstraZeneca. A novel CSF-1R receptor inhibitor, 4-cyano-1H-pyrrole-2-carboxylic acid [4-(4-methyl-piperazin-1-yl)-2-(4-methyl-piperidin-1-yl)-phenyl]-amide (hereafter referred to as “JnJ”), provided by Johnson & Johnson Pharmaceutical Research & Development (Johnson & Johnson, New Brunswick, NJ, USA), was used at 10 nmol/L (17). Two additional novel CSF-1R inhibitors, PLX3397 and PLX5622, were provided by Plexxikon. Small interfering RNA (siRNA) si-GENOME duplexes targeting mouse CSF-1R were acquired from Dharmacon, and microglia were transfected with 2 µL Dharmafect Reagent #1 with 20 nmol/L si-GENOME siRNA against murine CSF-1R (Dharmacon/Thermoscientific). Microglia were seeded 24 h before transfection in a six-well plate at 70% confluency. The siRNA and Dharmafect mixture was added to 1.6 mL complete microglia growth media (MSFM) with 10% fetal bovine serum and 10 ng/mL GM-CSF) and added to the cells. Cells were incubated with transfection mix for 72 h before the experiment.
GL261 and microglial (MG) cells were stained with cell tracker green (5-chloromethylfluorescein diacetate (green) [CMFDA]) and with cell tracker red (CMTPX; Invitrogen), respectively, and then cocultured on Matrigel-coated invasion chambers (BD, Franklin Lakes, NJ, USA). For most assays, to maintain constant cell numbers, cells were plated at a density per invasion chamber of 100,000 labeled GL261 cells with an additional 100,000 unlabeled GL261 cells or with 50,000 unlabeled GL261 and 50,000 MG cells in MSFM with 0.3% bovine serum albumin (BSA) supplemented with 10 ng/mL GM-CSF. Invasion chambers were incubated for 48 h, after which they were fixed in 3.7% paraformaldehyde in phosphate-buffered saline (PBS). Imaging of the cells on the bottom of the filter was performed with a Bio-Rad Radiance 2000 Confocal microscope. The extent of invasion was quantified by counting the number of glioblastoma cells (green) that were on the underside of the filter in at least seven 20x fields.
GL261 cells were plated on 8-µm pore transwells (Costar) that were lightly precoated with approximately 100 ng/mL Matrigel for 2 h at room temperature. GL261 cells were plated at a density of 100,000 cells in a 100-µL volume of MSFM with 0.3% BSA on the top of the chamber, whereas the bottom contained media alone, 10 nmol/L epidermal growth factor (EGF) or 50 ng/mL CSF-1 and was incubated for 4 h. Transwells were then fixed in 3.7% paraformaldehyde and stained with 0.2% crystal violet. The extent of chemotaxis was quantified by counting the number of cells on the underside of the filter.
Enzyme-Linked Immunosorbent Assays
Conditioned media were collected from GL261 cells alone, stimulated with 10 nmol/L EGF or cocultured with microglia overnight in MSFM + 0.3% BSA and 10 ng/mL GM-CSF. Supernatants were harvested, centrifuged at 144g to remove any cellular material and then subjected to enzyme-linked immunosorbent assay (ELISA) for mouse CSF-1 (MMC00) performed according to the Quantikine protocol (R&D Systems).
Cell cultures of microglia starved overnight in MSFM 0.3% BSA and microglia stimulated with CSF-1 in the presence of various inhibitors were lysed directly into 1x sample buffer (2% sodium dodecyl sulfate [SDS], 10% glycerol, 62.5 mmol/L Tris) containing β-mercaptoethanol and loaded onto 10% SDS-polyacrylamide gel electrophoresis (PAGE) gels. The proteins were resolved and transferred to a polyvinylidene fluoride (PVDF) membrane and blotted using an anti-phospho CSF-1R Y723 (#3155; Cell Signaling Technology) or anti-phospho ERK antibody (#9101; Cell Signaling Technology), followed by secondary blot using goat anti-rabbit antibody conjugated to IR800 in Licor Blocking Buffer (Licor). For detecting EGF, blotting was carried out using anti-EGF (sc-1342; Santa Cruz Biotechnology, Santa Cruz, CA, USA) at a concentration of 1:100 followed by secondary donkey anti-goat antibody conjugated to IR800 (Licor). For detecting CSF-1R, blotting was carried out using a rabbit antibody that recognizes the C-terminus of the CSF-1R (C-15 ). Blots were scanned on the Odyssey system. Blots were stained with anti-total ERK (#137F5; Cell Signaling Technology) and actin (A5441; Sigma-Aldrich) for loading control.
Microglia were plated on glass-bottomed MatTek dishes and cultured in MSFM supplemented with GM-CSF for 48 h. Cells were then fixed in 3.7% paraformaldehyde in PBS for 15 min and blocked with 5% fetal calf serum in PBS for 1 h. Primary antibody staining was carried out by using either control goat IgG or anti-EGF (M-20) (Santa Cruz Biotechnology; sc-1342) at a concentration of 4 µg/mL overnight at 4°C. After washing with PBS, cells were stained with secondary anti-goat antibody conjugated to Cy3. Images were acquired by use of a Bio-Rad Radiance 2000 Confocal microscope.
Sulforhodamine B Proliferation Assays
The effect of the inhibitors on proliferation/survival was assessed by use of sulforhodamine B (SRB) proliferation assays as described by Skehan et al. (19). GL261 cells (20,000) and microglia (10,000) were plated on 96-well plates and fixed by using 10% trichloroacetic acid (Sigma Aldrich) after incubation with dimethyl sulfoxide (DMSO), 5 µmol/L Iressa or 10 nmol/L JnJ for 48 h. After fixation, 0.2% SRB reagent in 1% acetic acid was used to stain the cells and was solubilized with 10 mmol/L unbuffered Tris base solution for 30 min. The absorbance was measured at 490 nm.
Intracranial Injection of Glioma Cells
All procedures involving mice were conducted in accordance with the National Institutes of Health regulations concerning the use and care of experimental animals. The study of mice was approved by the Albert Einstein College of Medicine Animal Use Committee. For intracranial injection, C57BL/6J mice (10–12 wks old; Jackson) were anesthetized with isofluorane. A hole was made 1 mm lateral and 2 mm anterior from the intersection of the coronal and sagittal sutures (bregma). The 2 × 104 GL261 cells were injected using a Hamilton syringe series 7000 at a depth of approximately 1 mm in a volume of 0.2 µL in the cortex. Mice were allowed to recover 24 h after injection and then fed either control chow or chow containing PLX3397 (Research Diets). Two weeks after injection, animals were killed and their brains were fixed in formalin for 72 h, paraffin embedded and sectioned. Microglia were visualized by immuno-histochemistry with an Iba1 antibody (Wako; #019-19741). Invasion was measured by counting the number of tumor cells, visualized with anti-Ki67 antibody (Vector; #VP-R451), that were found outside the tumor border. The total number of cells found outside of the tumor border in each field was counted and normalized to the total length of the border (in mm) for that field.
Fluorescent In Situ Hybridization
Fluorescent probes directed against murine CSF-1 were obtained from Stellaris (Novato, CA, USA), and the procedure was carried out according to the manufacturer’s protocol. Briefly, paraffin-embedded sections containing GL261 tumor were deparaffinized, rehydrated in PBS and incubated overnight in formamide, 2× sodium chloride-sodium citrate and 2.5 µmol/L CSF-1 probe at 37°C. Samples were then washed, stained with DAPI for 10 min, sealed with a cover slip in mounting medium (1% n-propyl gal-late, 100 mmol/L Tris, pH 8.0, and 50% glycerol) and visualized by use of a confocal fluorescence microscope.
All supplementary materials are available online at www.molmed.org .
Microglia Are Able to Stimulate Glioblastoma Cell Invasion
Microglia-Stimulated Glioblastoma Invasion Depends on EGFR Signaling
Work from our laboratory and collaborators has shown that a paracrine loop exists between breast carcinoma cells and tumor-associated macrophages that involves EGF and CSF-1 (or M-CSF) (7). Blockade of either EGF/EGF receptor (EGFR) or CSF-1/CSF-1 receptor (CSF-1R) signaling was sufficient to block macrophage-mediated breast carcinoma cell invasion both in vitro and in vivo (7,8). We therefore investigated the role of EGFR and CSF-1R in mediating the stimulatory effect of microglia-stimulated glioblastoma invasion.
Microglia Stimulation of Glioblastoma Invasion Depends on CSF-1R Signaling
Using the concentrations of CSF-1R inhibitors that were effective at inhibiting CSF-1-induced CSF-1R and ERK phosphorylation, we next tested their effects on microglia-stimulated glioma invasion. A total of 10 nmol/L JnJ or 1 µmol/L of either PLX compound completely blocked microglia-stimulated glioma invasion (Figure 3B). Basal GL261 invasion was unaffected by any of the inhibitors (data not shown). None of these inhibitors at the concentrations used in the invasion assay had any influence on proliferation rate or survival of either cell type under the duration time of these experiments (Figures 3C, D). As a complementary approach to pharmacological inhibition, we examined the role of CSF-1R by RNA interference. We delivered siRNA duplexes that target CSF-1R by transient transfection into microglia. We achieved at least 50% knockdown of CSF-1R at the protein level (Figure 3E). When cocultured with GL261 cells, CSF-1R-depleted microglia were unable to stimulate glioblastoma invasion (Figure 3F). These data demonstrate that the activity of the CSF-1R receptor in microglia is necessary for microglial stimulation of glioma invasion.
Glioblastoma Cells and Microglia Generate Ligands for CSF-1R and EGFR, Respectively
Glioblastoma and Microglia Respond to EGF and CSF-1, Respectively
Blockade of CSF-1R Inhibits Glioblastoma Invasion In Vivo
The results from this study provide new insight into the mechanisms of glioblastoma invasion. Coculturing of glioblastoma cells with proportions of microglia similar to what has been observed in human patients results in a dramatic increase in glioblastoma invasion. Microglia-stimulated glioblastoma invasion was strongly inhibited by blockade of either EGFR or CSF-1R, implicating these receptors in mediating communication between these two cell types during invasion. We propose that the constitutive production of CSF-1 by glioblastoma cells can attract and stimulate microglia from the surrounding parenchyma. As microglia approach the glioblastoma cells in response to the CSF-1, microglia in turn stimulate glioblastoma cell invasion via, at least in part, EGFR activation.
In some breast cancer models, EGFR and CSF-1R stimulate production of the other receptor’s ligands (7). In our studies, we find no evidence of stimulated secretion of the ligands by coculture or by stimulation with EGF or CSF-1. We conclude that production of CSF-1 by glioblastoma is necessary for the generation of interactions between microglia and glioblastoma cells that stimulate glioblastoma cell invasion. We observe that EGFR is necessary for microglia stimulation of glioblastoma invasion and that microglia express EGF. The most straightforward interpretation for these data is that microglia activate EGFR on glioblastoma. The EGF precursor is a membrane-spanning protein of a 150–180 kDa that is expressed on the plasma membrane in an orientation that allows cleavage and release of EGF (20). Because the precursor itself has EGF activity (21), the surface expression of EGF raises the possibility that microglia can activate EGFR signaling on glioblastoma cells in a juxtacrine fashion. Alternatively, it is possible that proteases required for EGF processing are present on glioblastoma cells and EGFR activation is triggered by EGF release only when glioblastoma and microglia cells are in close proximity. Any EGF released in this fashion might be rapidly taken up by EGFR on the glioblastoma surface and endocytosed. This result perhaps explains our inability to detect EGF in the supernatant of cocultures using ELISA (data not shown). Alternatively, other ligands for EGFR, such as heparin-binding (HB)-EGF or TGF-α, might be expressed by microglia and required for the stimulation of glioblastoma invasion through activation of EGFR. The expression of these additional EGFR ligands by microglia is currently being investigated. Finally, it is possible that microglia induce glioblastoma cells to activate EGFR by an autocrine mechanism. For example, proteases expressed on the surface of microglia may process EGFR ligand precursors on glioblastoma cells once the two cell types are in proximity.
Several studies have observed an effect on GL261 tumor growth in vivo when macrophages/microglia were ablated using the CD11b-TK system (13,14). Although treatment with PLX3397 substantially reduces the number of microglia / macrophages associated with the tumor to a level that is similar to that of gancyclovir-treated tumors in CD11-TK mice (Figure 6), we did not observe any obvious difference in tumor size between control and PLX3397-treated animals. Thus, it is possible the microglia/macrophages that remain in the CD11-TK mice are also affected by the gancyclovir treatment and are unable to promote tumor growth. We also note that our investigation examines the effect of blocking CSF-1R signaling, which represents a more subtle treatment than eliminating these cells. Indeed, it is possible that blockade of CSF-1R signaling, while interfering with the ability of microglia to promote cell invasion into normal brain, has little or no effect on their support of tumor proliferation. Further studies will need to be carried out to elucidate the mechanisms of microglia-dependent invasion versus growth.
Consistent with our observations, it has been reported that glioblastoma specimens from human patients express CSF-1 and that expression of CSF-1 correlates with the degree of glioblastoma progression (22). This study also demonstrated that CSF-1 expression correlated with the increased presence of macrophages bearing markers that are associated with the so-called “M2 phenotype,” which are believed to be one of the critical subpopulations of macrophages involved in tumor progression (6). In addition, a recent study used the Sleeping Beauty transposon mutagenesis technique in the brains of whole animals to screen for novel genes associated with glioma oncogenesis (23). It was discovered that >60% of the highgrade astrocytoma occurrence was a result of transposon-mediated enhanced expression of the Csf-1 gene. Using PLX3397, a small molecule inhibitor of CSF-1R that crosses the blood-brain barrier, we show that interfering with CSF-1R signaling in vivo prevents glioblastoma cells from invading into the parenchyma. Recent results from the Coussens laboratory show that inhibition of CSF-1R in a breast carcinoma animal model synergizes with paclitaxel in preventing metastasis and improving animal survival (22). It will be interesting to determine whether PLX3397 alone, or in combination with standard regimens currently used for treating glioblastoma, is able to extend survival of tumor-bearing animals. Clinical trials involving PLX3397 in treating glioblastoma, lymphoma, leukemia and breast carcinoma are currently underway or are in planning. Our results provide a mechanistic explanation for the involvement of CSF-1 in glioblastoma progression and indicate that inhibition of CSF-1R signaling could provide a novel approach to limiting glioblastoma invasion.
ER Stanley is a consultant for Plexxikon on CSF-1/CSF-1R biology and a consultant for Amgen for therapies targeting the CSF-1R. BL West is an employee of Plexxikon.
The study was funded by AACR-Amgen, Inc., Fellowship in Clinical/Translational Cancer Research (09-40-11 to SJ Coniglio); the Dana Foundation (SJ Coniglio, MH Symons and JE Segall), Association for International Cancer Research (AICR; SJ Coniglio and JE Segall), National Institutes of Health (PO1 CA 100324 to JE Segall and ER Stanley and P30 CA13330 and RO1 CA25604 to ER Stanley); and Voices Against Brain Cancer Foundation (MH Symons).
We thank Carl Manthey for helpful discussions on the use of JnJ, AstraZeneca and Johnson & Johnson for reagents, Plexxikon for PLX3397 and PLX5562, and the Segall, Condeelis, Cox and Hodgson labs for helpful comments. JE Segall is the Betty and Sheldon Feinberg Senior Faculty Scholar in Cancer Research.
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, and provide a link to the Creative Commons license. You do not have permission under this license to share adapted material derived from this article or parts of it.
The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
To view a copy of this license, visit (http://creativecommons.org/licenses/by-nc-nd/4.0/)