Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2011 Nov 1.
Published in final edited form as: Int J Cancer. 2010 Nov 1;127(9):2063–2075. doi: 10.1002/ijc.25249

Glioblastoma cells incorporate into tumor vasculature and contribute to vascular radioresistance

Candice A Shaifer 1, Jianhua Huang 2, P Charles Lin 2,3,4
PMCID: PMC2932815  NIHMSID: NIHMS218583  PMID: 20162571

Abstract

Glioblastoma multiforme (GBM) remains the most devastating neoplasm of the central nervous system and has a dismal prognosis. Ionizing radiation represents an effective therapy for GBM, but radiotherapy remains only palliative because of radioresistance. In this study, we demonstrate that glioma cells participate in tumor vascularization and contribute to vascular radioresistance. Using a 3-dimensional coculture system, we observed an intimate interaction of glioma cells with endothelial cells whereby endothelial cells form vascular structures, followed by the recruitment and vascular patterning of glioma cells. In addition, tumor cells stabilize the vascular structures and render them radioresistant. Blocking initial endothelial vascular formation with endothelial specific inhibitors prevented tumor cells from forming any structures. However, these inhibitors exhibited minimum effects on vascular structures formed by tumor cells, due to the absence of the targeted receptors on tumor cells. Consistent with the in vitro findings, we show that glioma cells form perfused blood vessels in xenograft tumor models. Together, these data suggest that glioma cells mimic endothelial cells and incorporate into tumor vasculature, which may contribute to radioresistance observed in GBM. Therefore, interventions aimed at the glioma vasculature should take into consideration the chimeric nature of the tumor vasculature.

Keywords: glioma, endothelial cell, cell-cell interaction, radioresistance, vascular mimicry, VEGF

Introduction

GBM, a grade IV tumor, is a malignant astrocytoma. Astrocytomas are classified as either diffuse or localized based on how they interact with their immediate microenvironment. Diffuse astrocytomas are invasive at the peritumoral edge and are able to metastasize to distant sites. On the other hand, localized astrocytomas exhibit limited invasiveness and a restricted pattern of growth 1. According to the World Health Organization, GBM is characterized by the presence of microvascular proliferation and necrosis, which indicates a close interaction between glioma cells and vascular endothelial cells 1.

Tumor angiogenesis is essential for tumor growth and progression. Angiogenesis is controlled by a variety of growth factors and their cognate receptors. Among them, two families of receptor tyrosine kinases, VEGF receptor (VEGFR) and Tie2, play a direct role, evidenced by the fact that these receptors are mainly expressed in vascular endothelium. Glioblastomas are known to express Tie2 and its ligands angiopoietin-1 (Ang 1) and angiopoietin-2 (Ang 2), as well as VEGFR1, VEGFR2 and VEGF 2–5, with expression levels of Ang1, Ang2 and Tie2 correlating with malignancy of glioblastoma 6, 7. In human glioblastoma tissues, Tie2 expression is restricted to blood vessels, and Tie2 expression and phosphorylation is increased in GBM blood vessels compared to blood vessels in low grade astrocytomas and normal brain tissue 6.

Ionizing radiation represents an effective therapy for GBM, but radiotherapy remains only palliative because of radioresistance 8, and long-term survival of GBM patients is rare (median survival ranges from 12–15 months) compared to patients with low grade tumors (median survival ranges from 10–15 years). Interestingly, experimental studies suggest that glioblastoma blood vessels are also resistant to radiotherapy, showing no response to doses of radiation used in clinical settings 9–11. In contrast, non-glioma tumor lines showed vascular regression in response to clinical doses of radiotherapy in both syngeneic and xenograft models 9–11. However, the molecular mechanism of glioblastoma vascular radioresistance is unclear. One way to approach vascular radioresistance is to examine the interaction between the glioma cells and their supporting vasculature.

Here we report that human glioblastoma cells form perfusable blood vessels. We observed an intimate interaction between tumor cells and endothelial cells in a 3-D culture. Endothelial cells form vascular structures which tumor cells mimic thereby creating a mosaic vasculature. The tumor cells stabilize the structure and render vascular radioresistance. This unique characteristic may be particularly important in understanding the survival mechanisms of tumors and their associated vasculature that are non-responsive to standard treatment modalities, and warrants development of novel approaches to therapeutic intervention.

Materials and Methods

Mice and tumor models

Athymic Nude-FOXN1nu (nu/nu) male mice (age 5–6 weeks) were purchased from the Harlan lab (Indianapolis, IN). The mice were housed in pathogen-free units at the Vanderbilt University School of Medicine in compliance with Institutional Animal Care and Use Committee (IACUC) regulations. For the subcutaneous tumor model, 1×106 U251MG cells were implanted in the flank of nude mice (n=5). For cranial tumor implantation, U251MG-EGFP cells (7.2×105) or U87MG-GFP cells (1×106) in 5 μL PBS were stereotactically injected into the striatum of mice (n=12 and n=15, respectively). All mice injected with tumor cells developed tumors. For illuminating tumor blood vessels, 0.5 mg WGA lectin tetramethylrhodamine (Molecular Probes, Eugene, OR) in 150 μl PBS was injected via carotid artery. Circulation of lectin was terminated after 10 minutes. Brain tumor tissues were imaged using a FV-100 Olympus confocal microscope. Linescan spectrum of vasculature in the confocal image was generated using the Metamorph program.

Cell Culture and Reagents

The human umbilical vein endothelial cells (HUVEC) were purchased and maintained in EBM supplemented with 5% FBS, hydrocortisone, hFGF-B, VEGF, R3-IGF-I, ascorbic acid, hEGF and GA-1000 (Lonza Cellgro, Walkersville, MD). The human U251MG, U87MG, T98G, and D54 glioma cell lines were maintained in high glucose DMEM (Mediatech, Inc., Herndon, VA) supplemented with 10% FBS and 1% penicillin/streptomycin. The human U87MG-GFP glioma cell line was a kind gift from Dr. Jacques Galipeau (Lady Davis Institute, Montreal, Quebec, Canada). U251MG-EGFP stable clones were generated using the pEGFP-N3 vector (BD Biosciences) and selected with G418. Cells were cloned in the Flow Cytometry and Cell Sorting Facility in the VUMC Veteran Affairs Hospital.

VEGFR2 kinase small molecule inhibitor, SU5416 12, and VEGFR-2 antibody were purchased from EMD Bioscience (San Diego, CA) and Cell Signaling Technology, respectively. Recombinant soluble Tie2 receptor, ExTek, and adenoviral vector directing the expression of ExTek (AdExTek) were generated as described 9, 13. Antibodies against human and murine CD31 were purchased from PharMingen.

Three-dimensional organotypic coculture

HUVECs were labeled with PKH26-Red dye (Sigma, St. Louis, MO) and combined (4:1 ratio) with T98G-GFP, D54-GFP or U251MG-EGFP glioma cells and seeded on growth factor reduced MatrigelR (BD Bioscience, Bedford, MA) in EGM. The cells were fixed on Matrigel using 4% paraformaldehyde 24, 48, 72 hours after seeding, stained with DAPI and imaged.

For vascular regression experiments, cells were treated with increasing doses (0–10 μg/mL) of ExTek protein to neutralize Tie2 activation 9, 13. For each group, the number of vascular branch points was counted and averaged. For inhibition of Tie2 activation, PKH26-HUVECs were infected with either AdExTek or Adβ-gal as a control adenovirus (MOI 10) 12. For neutralization of VEGFR2 function, PKH26-HUVECs were pre-treated with 5 μm SU5416 12 or DMSO control for 30 minutes, then mixed with EGFP-U251MG glioma cells (4:1 ratio) on Matrigel in EBM supplemented with 2% FBS containing SU5416. Vascular structures were imaged at 24 and 48 h.

To determine whether the vascular structures generated in coculture form lumens, HUVECs were embedded in Matrigel at a density of 1×105 cells in a Boyden Chamber. U251MG-EGFP tumor cells (2.5×104) were seeded on top of the Matrigel layer in EGM-2. Another mixture of HUVECs embedded in Matrigel was plated over the tumor cell monolayer in each chamber and allowed to polymerize. The cultures were fed with EGM-2 every 2 days for 6 days. Digital photographs were captured using an Olympus IX70 inverted microscope.

Conditioned medium

HUVECs on the 3rd–5th passage were cultured to 60–70% confluency. The endothelial cells were then washed with PBS and cultured in serum-free EBM for 24 hours.

Histological analyses

Subcutaneous tumor tissues were harvested 17 days after implantation and processed for immunohistochemical analysis. For detecting endothelium from mouse and human origin, antibodies specific against either mouse CD31 (1:100) or human CD31 (1:50) were used (PharMingen). Microvascular density was assessed in 5 random fields at 200× magnification. Human CD31-PAS (periodic acid-Schiff, stains basement membrane) dual staining was performed according to standard protocols and counterstained with Mayer’s hematoxylin. The hematoxylin and eosin slides were evaluated by a neuropathologist and imaged on an Olympus BX51 microscope. For lectin perfused brain tumor tissue, mice were perfused with 4% paraformaldehyde. Brains were removed, further fixed with 4% paraformaldehyde and embedded in optimal cutting temperature (OCT) compound. To visualize colocalization of lectin staining and GFP frozen sections (20 μm) were fixed in ice-cold acetone and imaged using confocal microscopy.

Western blot

Cell lysates were separated by 4–12% SDS/PAGE, transferred to Protran nitrocellulose membranes (Perkin Elmer) blocked, and incubated with anti-Tie2 (1:100), anti-VEGFR2 (1:1,000), or β-tubulin (1:1,000) followed by anti-mouse or anti-mouse rabbit secondary antibodies (1:20,000).

RT-PCR

Total RNA was extracted using RNeasy Mini Kit (QIAGEN) and first-strand cDNA synthesis was generated using the iScript cDNA Synthesis kit as per the manufacturer’s protocol (BioRad). The RT-PCR primer sets with their 5’-3’ sequences are listed below. cDNA amplification with human VEGFR2 primers was performed as described 14.

Tie2- TAAATTTGACCTGGCAACCA; TAGCACCGAAGTCAAGTTGC

VEGFR2- GTGACCAACATGGAGTCGTG; CCAGAGATTCCATGCCACTT

EphA2- AGACGCTGAAAGCCGGCTAC; GAGCCGGATAGACACGCGG

VE-cadherin- CCGGCGCCAAAAGAGGA; CTGGTTTTCCTTCAGCTGGAAGTGGT

MMP1- ATTGGAGCAGCAAGAGGC; GTCCACATCTGCTCTTGGC

MMP2- TGGCAGTGCAATACCTGAAC; CAAGGTCCATAGCTCATCGTC

Nodal- CCTTCCTGAGCCAACAAGAG; AGGTGACCTGGGACAAAGTG

CD133- TACCAAGGACAAGGCGTTCAC; CAGTCGTGGTTTGGCGTTGTA

GAPDH- GTCAGTGGTGGACCTGACCT; AGGGGTCTACATGGCAACTG

β-actin- GACAACGGCTCCGGCATGTGTGC; TGGCTGGGGTGTTGAAGGTC

Radiation treatment

Cultured cells were irradiated at indicated single doses with a Mark I 137Cesium irradiator system at a dose rate of 1.841 Gy/minute.

Statistical analysis

Results were analyzed using GraphPad Prism software by a two-way ANOVA. A two-tailed Student’s t-Test was used to analyze statistical differences between the human and mouse CD31 data. p< 0.05 was considered statistically significant.

Results

Glioma cells incorporate into vascular structures in vitro

Based on the findings that glioma vasculature in experimental models is resistant to radiotherapy, we studied the interaction of tumor cells with endothelial cells using a 3-dimensional organotypic coculture system. Therefore, we cocultured EGFP stably transfected tumor cells (U251MG-EGFP) with endothelial cells labeled with a red vital dye, PHK26, on Matrigel. This approach allows us to distinguish tumor cells from vascular endothelial cells while studying their interaction. We found that within 24 hours post-seeding, endothelial cells first exhibit fibroblast-like morphology and form vascular branches. Very interestingly, U251MG-EGFP tumor cells follow afterwards. They become elongated, interact closely with the preexisting vascular structures, and contribute to the morphology of the vascular structures creating a mosaic vascular network. By 48 hours, majority of the endothelial cells in the structure were mimicked by glioma cells resulting in a vascular network largely comprised of U251MG-EGFP tumor cells and endothelial cells (Figure 1A). This observation was further confirmed with in-gel DAPI staining of the vascular structures present (Figure 1A). The data shows that DAPI corresponds to the nuclei of the EGFP-U251MG glioma cells and endothelial cells.

Figure 1. Glioma cells incorporate into vascular networks in vitro.

Figure 1

Figure 1

Open in a new tab

EGFP-U251MG cells were cocultured with PHK26 labeled HUVECs (1:4 ratio) in Matrigel. Vascular network formation was imaged at various time points as indicated under bright and fluorescent field under microscopy (Panel A). For illumination of nuclei, the gel was fixed in 4% paraformaldehyde, followed by staining with DAPI (Panel A). EGFP-expressing human T98G and D54 glioma cells were cocultured with PHK26 labeled HUVECs (1:4 ratio) on Matrigel in EGM-2 medium. The images were taken 24, 48 and 72 hours after coculture (Panel B and C). Each experiment was performed in triplicate and repeated twice. EGFP-U251MG tumor cells were cultured between two Matrigel layers embedded with HUVECs for four days (Panel D, left) (100×). Cultures were sectioned and stained with H&E (Panel D, right panel). Arrows point to the luminal structures (400×). Representative images are shown from three independent experiments performed in duplicate.

A similar observation was achieved in two other human glioma cell lines, T98G and D54 (Figure 1B and 1C). These tumor cells incorporated into vascular structures formed initially by endothelial cells and created a mosaic network. Furthermore, we observed that endothelial cells seeded alone on Matrigel form vascular networks within 24 hours, which regressed thereafter. However, vascular networks formed in the coculture remained stable for weeks, maintained lumens and exhibited a distinct morphology (Figure 1D). These findings support that glioma cells stabilize vascular networks.

Tumor cells mimic endothelial cell vascular patterning

Based on the observation that endothelial cells form vascular networks first, followed by incorporation of tumor cells, we postulated that glioma cells mimic endothelial cell patterning. To test this hypothesis, we used a Tie2 specific inhibitor, ExTek that prevents the ligand, angiopoietins, from binding to the endogenous Tie2 receptor on cell surface 13. We showed that Tie2 is present in endothelial cells, but absent in glioma cell lines (Figure 2A). Next, we infected endothelial cells with an adenoviral vector expressing ExTek for inhibition of Tie2 signaling, and placed them in coculture with tumor cells. We observed limited vascular network formation within 24 hours after infection and the networks gradually regressed by 48 hours. As expected the U251MG-derived vessel-like structures were inhibited when Tie2 signaling within endothelial cells was abrogated (Figure 2B). In contrast, endothelial cells treated with control vector formed vascular networks with incorporation of tumor cells.

Figure 2. Blocking endothelial cells from forming vascular networks prevented tumor cells from forming any structures.

Figure 2

Open in a new tab

Tie2 expression on HUVEC, U251MG, T98G and D54 cells was examined by Western blot. Beta-tubulin was used as an internal control (Panel A). HUVECs were infected with AdExTek or control β-galactosidase at MOI=10, respectively. The cells were then labeled with PHK26 vital dye, mixed with U251MG-EGFP cells (4:1 ratio) and seeded on Matrigel. Vascular networks were imaged under bright and fluorescent microscopy 24 and 48 hours after coculture. Representative images are shown (Panel B) (100×). The expression of VEGFR2 in HUVECs and U251MG cells was analyzed by Western blot (Panel C). PKH26 labeled HUVECs were treated with 5μM SU5416 for 30 minutes prior to mixing with EGFP-U251MG glioma cells (4:1 ratio). The cells were then seeded on Matrigel and cultured in EBM supplemented with 2% serum and 5 μM SU5416. Vascular structures were imaged at 24 and 48h under microscopy (Panel D) (100×). Each experiment was performed in triplicate and repeated twice.

Consistently, we showed that VEGFR2, another endothelial specific receptor, is present in vascular endothelial cells, but absent from glioma cells (Figure 2C). Pre-treating endothelial cells with SU5416, a specific VEGFR2 kinase inhibitor at 5 μM 12, 15, impaired vascular structure formation by the endothelial cells, which subsequently prevented the formation of vascular structures by tumor cells (Figure 2D). Together, these findings indicate an intimate interaction between endothelial cells and glioma cells. In coculture, endothelial cells initiate the patterning of tumor cells and together the endothelial and glioma cells create a stable, mosaic vascular network.

Neutralization of Tie2 activity has no effect on vascular structures formed by tumor cells

Since we observed that U251MG glioma cells incorporate into vascular structures initially formed by endothelial cells and create mosaic vascular structures (Figure 1A), we next determined whether the tumor cells acquire the endothelial-specific Tie2 receptor expression and responds to anti angiogenic therapy under the coculture condition. Increasing concentrations of ExTek protein was added to the culture medium at the time of cell seeding. Endothelial cells and tumor cells were cocultured on Matrigel for 24 hours in EBM containing increasing concentrations of ExTek protein to block Tie2 activation16. Interestingly, we did not find any significant effects on U251MG-derived vascular structures when compared to control treated groups although the treatment inhibited vascular network formation by endothelial cells (Figure 3A and 3B). Consistently, we did not detect Tie2 and VEGFR2 protein expression in tumor cells cultured under the endothelial cell conditioned media on Matrigel (Figure 3C). Semi quantitative RT-PCR confirmed absence of Tie2 and VEGFR2 mRNA in the tumor cells (Figure 3D). These findings show molecular differences between vascular structures formed by glioma cells and those formed by endothelial cells. This suggests that targeting endothelial specific angiogenic pathways is less likely to be effective in vascular structures formed by tumor cells.

Figure 3. Vascular structures formed by U251MG cells are resistant to anti-angiogenic treatment.

Figure 3

Open in a new tab

HUVECs (4×104) or U251MG-EGFP (8×104) cells were cultured on Matrigel in serum-free EBM (100×) (Panel A). Increasing concentrations of ExTek protein was added to the culture medium at the time of cell seeding. Micrographs were obtained 24h after the treatment (Panel A). The number of vascular branch points was counted in ten randomly selected fields under microscopy (100×) 1 day after the treatment (Panel B). Each experiment was performed in triplicate and repeated twice, *p<0.01. The expression of Tie2 and VEGFR2 was analyzed in HUVECs cultured in EGM, and U251MG cells cultured in endothelial cell-conditioned media for 24 hours by Western blot (Panel C) and semi quantitative RT-PCR (Panel D). Beta-tubulin and actin were used as internal controls, respectively.

Characterization of vascular mimicry-related genes in glioma cells

To characterize gene expression associated with vascular mimicry, we performed a series of semi-quantitative RT-PCR experiments on tumor cells cultured in DMEM (Figure 4A) and endothelial conditioned media on Matrigel (Figure 4B). As shown in Figure 4A, we found that EphA2 and Nodal were highly expressed in HUVEC, U251MG, T98G and D54 cell lines whereas expression was absent in U87MG cell line. Expression of the endothelial-specific marker VE-cadherin and the hematopoietic/endothelial progenitor cell marker, CD133 was absent in all glioma cell lines. MMP2 mRNA was detected in all cell lines however; MMP1 was only detected in HUVECs and the T98G cell line. The astrocyte marker, glial fibrillary acid protein (GFAP) was expressed in glioma cell lines with highest expression in U87MG cells, medium level in U251MG and D54 cell lines and minimal in the T98G cell line.

Figure 4. Characterization of vascular mimicry-related genes in glioma cells.

Figure 4

Open in a new tab

HUVECs, U251MG, U87MG, T98G and D54 cells were cultured in their native media (Panel A) or endothelial conditioned media for 24 hours (Panel B). RNA was obtained and semi-quantitative RT-PCR was performed. GAPDH was used as an internal control. Representative images were shown. The experiments were repeated twice.

Interestingly, under endothelial cell conditions there is a general reduction of astrocyte specific marker (GFAP), accompanied with an induction of endothelial progenitor marker (CD133) in all the glioma cell lines except U87MG (Figure 4B). We also observed an induction of VE-cadherin expression in U251MG cells (Figure 4B). These data suggest that glioma cells can indeed trade their identity by downregulating the astrocyte-specific gene, GFAP, to acquire endothelial and endothelial progenitor gene expression. It is worth to note that we did not found a significant reduction of GFAP in U87MG cells, nor did we see an increase of CD133 under endothelial conditions (Figure 4B). Furthermore, there is no significant change in EphA2, MMP2 and Nodal between the groups.

Vascular networks formed in the presence of glioma cells are resistant to radiotherapy

Ionizing radiation represents an effective therapy for GBM, but radiotherapy remains only palliative because of radioresistance 8. Interestingly, experimental data showed that glioblastoma blood vessels are also resistant to radiotherapy 9–11. Based on our observation that glioma cells incorporate into vascular structures initially formed by endothelial cells and stabilize the vascular structures, we therefore investigated whether the presence of tumor cells in these networks render radioresistance. Endothelial cells were cultured alone or in combination with U251MG glioma cells on Matrigel. Upon the formation of vascular networks after 9 hours, we irradiated the cells with a single dose of radiation. We observed a time dependent destruction of vascular networks formed by endothelial cells. In contrast, the networks formed by tumor cells and endothelial cells were resistant to the therapy (Figure 5). This finding provides a potential molecular explanation for the radioresistant tumor blood vessels associated with malignant glioma in animal models.

Figure 5. Incorporation of tumor cells into vascular structure renders vascular radioresistance.

Figure 5

Open in a new tab

PHK26 labeled HUVECs were cultured alone on Matrigel for 9 hours to allow vascular networks to develop, followed by irradiation at 0 and 6 Gy. Vessel structures were imaged under microscopy 1, 2 and 3 days post radiation (Panel A). PHK26 labeled HUVECs were cocultured with U251MG-EGFP glioma cells for 9 hours on Matrigel, followed by irradiation at 0 and 6Gy. Vessel structures were imaged 1, 2, and 3 days post radiation (Panel B). Vascular branch points were counted in 90 randomly selected fields under microscopy 1 day (Panel C) and 3 days (panel D) post radiation, **p< 0.0001.

Malignant glioma cells participate in tumor vascular formation in vivo

To expand these in vitro studies, we examined whether glioma vascular mimicry occurs in vivo. First, U251MG glioma cells were subcutaneously implanted in the flank of nude mice for 17 days, tumor tissues were processed and analyzed by immunohistochemistry for vessel density using antibodies specific for mouse CD31 to detect host-derived endothelium, and specific for human CD31 to detect tumor cell contribution to the vasculature (Figure 6A and 6B). We confirmed that the anti mouse CD31 antibody did not cross react with human tissues (Figure 6C), and the anti human antibody with mouse samples (Figure 6D). Consistent with the in vitro data, we detected human CD31 positive tumor vasculature (Figure 6A and 6E), indicative of tumor cell contribution into the vasculature. Furthermore, immunohistochemical dual staining using human CD31 and PAS (periodic acid-Schiff, stains basement membrane) confirms the positive reaction for CD31 on the luminal side of vessels and a PAS positive reaction on the vessel wall in U251MG and U87MG xenografts, which is indicative of vascular mimicry (Figure 7A–7D). Hematoxylin and eosin staining illustrates tumor cell invasion into normal brain and red blood cells in vessel lumens (Figure 7E–7H).

Figure 6. Malignant glioma cells incorporate into perfused tumor vasculature in vivo.

Figure 6

Open in a new tab

1×106 U251MG-EGFP human glioma cells were subcutaneously injected in athymic FOXN1nu/nu mice. Tumor tissues were harvested 17 days after tumor implantation, processed and stained with specific antibodies against human CD31 or mouse CD31 (Panel A and B) (200×). Staining controls were human brain tissue for mouse CD31 antibody (Panel C) and mouse brain tissue for human CD31 antibody (Panel D). The number of mouse CD31 and human CD31 positive vessels was counted in 5 random selected fields at x100 magnification (Panel E). n= 5 mice per group.

Figure 7. Histological evaluation of tumor sections.

Figure 7

Open in a new tab

U251MG-EGFP (Panel A and B) or U87MG (Panel C and D) tumors were harvested from stereotactically injected mice five weeks and two weeks, respectively after tumor implantation. Tumor tissues were processed, sectioned and stained for human CD31 (brown color), PAS (pink color) and counterstained with hematoxylin. White arrows point to human CD31 positive vasculature surrounded by PAS positive basement membrane. Magnifications: A, C = 200X; B, D =400X. Adjacent U251MG (Panel E and F) and U87MG (Panel G and H) tumor tissue sections were stained with hematoxylin and eosin. White arrows points to tumor invasion into the normal brain. Br= brain, Tu= tumor. Magnifications: E,G =100X; F,H = 200X.

In addition, we orthotopically implanted U251MG-EGFP and U87-GFP cells into mouse brains. Prior to harvesting tumor tissues, we intravenously injected WGA-lectin rhodamine into the mice to illuminate the perfused tumor vasculature in order to evaluate the functionality of glioma cell-derived blood vessels. Analysis of U251MG and U87MG tumor tissue sections by confocal microscopy and linescan analysis showed incorporation of tumor cells in perfused blood vessels and revealed a mosaic nature of glioma blood vessels, as indicated by the colocalization of perfused lectin-rhodamine with U251MG-GFP glioma cells (Figure 8A and 8B) and U87MG-GFP glioma cells (Figure 8C and 8D). Together with the in vitro data, these findings suggest that glioma cells participate in tumor blood vessel formation, which stabilizes tumor vasculature and likely renders tumor vascular radioresistance.

Figure 8. Malignant glioma cells incorporate into perfused tumor vasculature in vivo.

Figure 8

Figure 8

Open in a new tab

U251MG-EGFP cells (7.2×105) were stereotactically injected into the striatum of mice (n=12 mice). Five weeks after tumor implantation, 0.5 mg WGA lectin tetramethylrhodamine in 150 ul PBS was injected via the carotid artery. Brain tumor tissues were harvested, sectioned, and directly imaged using a FV-100 Olympus confocal microscope (Panel A). Arrows indicate incorporation of GFP positive tumor cells into perfused vasculature. Lines in panel A indicate a 2-dimensional optical cross section in the tumor. Linescan spectrum of mosaic vasculature from Panel A (Panel B). U87MG cells (1×106) were stereotactically injected into the striatum of mice (n=15 mice). Two weeks after tumor implantation, 0.5 mg WGA lectin tetramethylrhodamine in 150 ul PBS was injected via the carotid artery. Brain tumor tissues were harvested, sectioned, and directly imaged using a FV-100 Olympus confocal microscope (Panel C). Arrows indicate incorporation of GFP positive tumor cells into perfused vasculature. Lines in panel C indicate a 2-dimensional optical cross section in the tumor. Linescan spectrum of mosaic vasculature from Panel C (Panel D).

Discussion

In this study, we report that malignant glioma cells participate in tumor vascular formation in vitro and in vivo. Interestingly, we found that endothelial cells form vascular networks first, followed by incorporation of tumor cells into vascular structures. Accordingly, inhibition of endothelial cells from forming the initial vascular networks by targeting VEGFR2 and Tie2, two endothelial specific tyrosine kinases, blocked tumor cells from forming any structures. In contrast, inhibition of Tie2 had no effect on vascular structures that had been formed mainly by tumor cells. Further analysis indicates these tumor cells did not acquire VEGFR2 or Tie2 expression under angiogenic conditions despite their ability to form vascular networks in the presence of endothelial cells in 3-D Matrigel cultures. This finding illustrates molecular differences between vascular structures formed by glioma cells and structures formed by endothelial cells. It provides an additional explanation of the vascular abnormality observed in tumor blood vessels. Future interventions aimed at the glioma vasculature should take this property into consideration. In addition, we found that the presence of glioma cells stabilizes vascular structures, which remained for weeks and exhibited a tortuous morphology and enlarged lumens, compared to vascular structures formed by endothelial cells alone that only last for a few days. Incorporation of glioma cells in the vascular networks also renders vascular radioresistance in vitro.

Importantly, we detected tumor cell incorporation into vasculature in vivo. Nonresectable GBM is a uniformly fatal disease. Although ionization radiation remains a major approach to control the disease, but radiotherapy remains only palliative because of radioresistance 8. Disease progression within the field of irradiation occurs despite wide margins and high dose. Notably, studies using both syngeneic and xenograft models have demonstrated that tumor blood vessels in GBM are also resistant to radiotherapy 9–11. However, the underlying molecular mechanism is unclear. Our study demonstrates that incorporation of glioma tumor cells into the vascular structure stabilizes the structure and contributes to vascular radioresistance, providing a molecular explanation for this phenotype.

The generation of perfusable vascular channels by genetically deregulated, aggressive tumor cells was termed “vasculogenic mimicry” to emphasize their de novo generation without the participation of endothelial cells and being independent of angiogenesis. It was reported that the formation of these networks in vitro resembled that of embryonic vascular structures, and they were associated with the distinctly patterned ECM-rich networks that are observed in patients with aggressive tumors 17. Vasculogenic mimicry has been reported in ovarian 18, malignant melanoma 17, and prostate 19 cancers. The presence of loops or patterned networks within human uveal melanomas had a significant correlation with death 17. We show that malignant glioma cells have the potential to mimic vascular endothelial cells, incorporate into vascular networks in vitro and forms ECM rich and perfusable tumor blood vessels in vivo. At the molecular level, these glioma cells lose GFAP expression, a specific marker for astrocytes, and acquire CD133 expression, a marker for endothelial progenitors. U251MG cells also acquire VE-Cadherin expression under this condition.

Studies on the molecular profile of aggressive tumors suggest that there is a deregulation of the tumor-specific phenotype which is accompanied by simultaneous transdifferentiation of tumor cells into other cell types, such as endothelial cells. Many of the genes associated with vasculogenic mimicry in aggressive tumors are involved in angiogenesis such as CD31 17 and vascular endothelial-cadherin (VE-cadherin) 20 and it has been shown that tumor cells express endothelial specific genes during vascular mimicry and they acquire endothelial properties. However, we failed to detect any expression of Tie2 and VEGFR2, two receptors preferentially expressed on vascular endothelium and important in angiogenesis 21, 22, in U251MG tumor cells despite a clear incorporation of glioma cells into vascular structures. One difference between our studies and others may rely on the different tumor lines that were used in each study. Accordingly, neutralizing Tie2 and VEGFR2 function has no effect on U251MG-derived vascular structures. This result illustrates molecular differences of tumor blood vessels, which depends on the composition of vascular structures. It is conceivable that targeting these angiogenic pathways will have limited response in tumor blood vessels that are formed by tumor cells.

In conclusion, this study illustrates that malignant glioma cells display plasticity and have the ability to mimic vascular endothelial cells to form vascular structures in vitro and CD31+ blood vessels in vivo. Incorporation of tumor cells into vascular networks renders vascular radioresistance. This observation could have important implications in GBM therapy therefore future therapies aimed at the glioma vasculature should take into consideration the mosaic nature of the tumor blood vessels.

Acknowledgments

We thank Drs. Dennis Hallahan, Jin Chen and Miss Kimberly Boelte at Vanderbilt University Medical Center, and Dr. Marilyn Thompson at Belmont University, for critical reading the manuscript and helpful discussions. We thank Dr. Sam Wells and Mr. Sean Schaffer at the Vanderbilt Cell Imaging Shared Resources, Dr. Lianli Ma at the Vanderbilt Mouse Metabolic Phenotyping Center, Dr. Angela M. Boutte, Mrs. Youmin Zheng, Dr. Ty Abel, the VA FACS Core, the HTA Core and the Vanderbilt Immunohistochemistry Core for technical assistance. This work is supported in part by grants from NIH (CA108856, NS45888 and AR053718) to PCL and a training grant and fellowship to CS (5T32GM08554 and F31GM78856).

Footnotes

Impact : This study demonstrates that malignant glioma cells participate in tumor vascularization and contribute to vascular radioresistance in mouse models. Therefore, interventions aimed at the glioma vasculature should take into consideration the chimeric nature of the tumor vasculature.

References

  • 1.Kleihues P, Louis DN, Scheithauer BW, Rorke LB, Reifenberger G, Burger PC, Cavenee WK. The WHO classification of tumors of the nervous system. J Neuropathol Exp Neurol. 2002;61:215–25. doi: 10.1093/jnen/61.3.215. discussion 26-9. [DOI] [PubMed] [Google Scholar]
  • 2.Hu B, Guo P, Fang Q, Tao HQ, Wang D, Nagane M, Huang HJ, Gunji Y, Nishikawa R, Alitalo K, Cavenee WK, Cheng SY. Angiopoietin-2 induces human glioma invasion through the activation of matrix metalloprotease-2. Proc Natl Acad Sci U S A. 2003;100:8904–9. doi: 10.1073/pnas.1533394100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Koga K, Todaka T, Morioka M, Hamada J, Kai Y, Yano S, Okamura A, Takakura N, Suda T, Ushio Y. Expression of angiopoietin-2 in human glioma cells and its role for angiogenesis. Cancer Res. 2001;61:6248–54. [PubMed] [Google Scholar]
  • 4.Puputti M, Tynninen O, Sihto H, Blom T, Maenpaa H, Isola J, Paetau A, Joensuu H, Nupponen NN. Amplification of KIT, PDGFRA, VEGFR2, and EGFR in gliomas. Mol Cancer Res. 2006;4:927–34. doi: 10.1158/1541-7786.MCR-06-0085. [DOI] [PubMed] [Google Scholar]
  • 5.Du R, Petritsch C, Lu K, Liu P, Haller A, Ganss R, Song H, Vandenberg S, Bergers G. Matrix metalloproteinase-2 regulates vascular patterning and growth affecting tumor cell survival and invasion in GBM. Neuro Oncol. 2008;10:254–64. doi: 10.1215/15228517-2008-001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Stratmann A, Risau W, Plate KH. Cell type-specific expression of angiopoietin-1 and angiopoietin-2 suggests a role in glioblastoma angiogenesis. Am J Pathol. 1998;153:1459–66. doi: 10.1016/S0002-9440(10)65733-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Zadeh G, Qian B, Okhowat A, Sabha N, Kontos CD, Guha A. Targeting the Tie2/Tek receptor in astrocytomas. Am J Pathol. 2004;164:467–76. doi: 10.1016/S0002-9440(10)63137-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Bao S, Wu Q, McLendon RE, Hao Y, Shi Q, Hjelmeland AB, Dewhirst MW, Bigner DD, Rich JN. Glioma stem cells promote radioresistance by preferential activation of the DNA damage response. Nature. 2006;444:756–60. doi: 10.1038/nature05236. [DOI] [PubMed] [Google Scholar]
  • 9.Geng L, Donnelly E, McMahon G, Lin PC, Sierra-Rivera E, Oshinka H, Hallahan DE. Inhibition of vascular endothelial growth factor receptor signaling leads to reversal of tumor resistance to radiotherapy. Cancer Res. 2001;61:2413–9. [PubMed] [Google Scholar]
  • 10.Sarkaria JN, Carlson BL, Schroeder MA, Grogan P, Brown PD, Giannini C, Ballman KV, Kitange GJ, Guha A, Pandita A, James CD. Use of an orthotopic xenograft model for assessing the effect of epidermal growth factor receptor amplification on glioblastoma radiation response. Clin Cancer Res. 2006;12:2264–71. doi: 10.1158/1078-0432.CCR-05-2510. [DOI] [PubMed] [Google Scholar]
  • 11.Kozin SV, Boucher Y, Hicklin DJ, Bohlen P, Jain RK, Suit HD. Vascular endothelial growth factor receptor-2-blocking antibody potentiates radiation-induced long-term control of human tumor xenografts. Cancer Res. 2001;61:39–44. [PubMed] [Google Scholar]
  • 12.Mendel DB, Schreck RE, West DC, Li G, Strawn LM, Tanciongco SS, Vasile S, Shawver LK, Cherrington JM. The angiogenesis inhibitor SU5416 has long-lasting effects on vascular endothelial growth factor receptor phosphorylation and function. Clin Cancer Res. 2000;6:4848–58. [PubMed] [Google Scholar]
  • 13.Lin P, Polverini P, Dewhirst M, Shan S, Rao PS, Peters K. Inhibition of tumor angiogenesis using a soluble receptor establishes a role for Tie2 in pathologic vascular growth. J Clin Invest. 1997;100:2072–8. doi: 10.1172/JCI119740. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Dias S, Hattori K, Zhu Z, Heissig B, Choy M, Lane W, Wu Y, Chadburn A, Hyjek E, Gill M, Hicklin DJ, Witte L, et al. Autocrine stimulation of VEGFR-2 activates human leukemic cell growth and migration. J Clin Invest. 2000;106:511–21. doi: 10.1172/JCI8978. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Fong TA, Shawver LK, Sun L, Tang C, App H, Powell TJ, Kim YH, Schreck R, Wang X, Risau W, Ullrich A, Hirth KP, et al. SU5416 is a potent and selective inhibitor of the vascular endothelial growth factor receptor (Flk-1/KDR) that inhibits tyrosine kinase catalysis, tumor vascularization, and growth of multiple tumor types. Cancer Res. 1999;59:99–106. [PubMed] [Google Scholar]
  • 16.Lin P, Buxton JA, Acheson A, Radziejewski C, Maisonpierre PC, Yancopoulos GD, Channon KM, Hale LP, Dewhirst MW, George SE, Peters KG. Antiangiogenic gene therapy targeting the endothelium-specific receptor tyrosine kinase Tie2. Proc Natl Acad Sci U S A. 1998;95:8829–34. doi: 10.1073/pnas.95.15.8829. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Maniotis AJ, Folberg R, Hess A, Seftor EA, Gardner LM, Pe'er J, Trent JM, Meltzer PS, Hendrix MJ. Vascular channel formation by human melanoma cells in vivo and in vitro: vasculogenic mimicry. Am J Pathol. 1999;155:739–52. doi: 10.1016/S0002-9440(10)65173-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Sood AK, Seftor EA, Fletcher MS, Gardner LM, Heidger PM, Buller RE, Seftor RE, Hendrix MJ. Molecular determinants of ovarian cancer plasticity. Am J Pathol. 2001;158:1279–88. doi: 10.1016/S0002-9440(10)64079-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Liu C, Huang H, Donate F, Dickinson C, Santucci R, El-Sheikh A, Vessella R, Edgington TS. Prostate-specific membrane antigen directed selective thrombotic infarction of tumors. Cancer Res. 2002;62:5470–5. [PubMed] [Google Scholar]
  • 20.Hendrix MJ, Seftor EA, Meltzer PS, Gardner LM, Hess AR, Kirschmann DA, Schatteman GC, Seftor RE. Expression and functional significance of VE-cadherin in aggressive human melanoma cells: role in vasculogenic mimicry. Proc Natl Acad Sci U S A. 2001;98:8018–23. doi: 10.1073/pnas.131209798. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Visconti RP, Richardson CD, Sato TN. Orchestration of angiogenesis and arteriovenous contribution by angiopoietins and vascular endothelial growth factor (VEGF) Proc Natl Acad Sci U S A. 2002;99:8219–24. doi: 10.1073/pnas.122109599. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Suri C, Jones PF, Patan S, Bartunkova S, Maisonpierre PC, Davis S, Sato TN, Yancopoulos GD. Requisite role of angiopoietin-1, a ligand for the TIE2 receptor, during embryonic angiogenesis. Cell. 1996;87:1171–80. doi: 10.1016/s0092-8674(00)81813-9. [DOI] [PubMed] [Google Scholar]

RESOURCES