Cell-Cell Contact Between Microvascular Endothelial

DIRECT CONTACT WITH PERIVASCULAR TUMOR CELLS ENHANCES INTEGRIN

αVβ3 SIGNALING AND MIGRATION OF ENDOTHELIAL CELLS

A dissertation submitted

to Kent State University in partial

fulfillment of the requirements for the

degree of Doctor of Philosophy

Monica Elizabeth Burgett

August 2016

Dissertation written by

Monica Elizabeth Burgett

B.S., Eastern Michigan University, 2008

Ph.D., Kent State University, 2016

Approved by

Candece L. Gladson, M. D. , Chair, Doctoral Dissertation Committee

Gail Fraizer, Ph.D. , Members, Doctoral Dissertation Committee

Derek Damron, Ph.D. ______

Matthew Summers, Ph.D.______

Songping Huang, Ph.D.______

Accepted by

Ernest J. Freeman, Ph.D. , Chair, School of Biomedical Sciences

James L. Blank, Ph.D. , Dean, College of Arts and Sciences

TABLE OF CONTENTS……………………………………...………………………... iii

LIST OF FIGURES…………………………………………………………...….……… v

ACKNOWLEDGEMENTS…………………………………………………………….. vii

CHAPTERS:

I. INTRODUCTION……...……………….……………………………...……. 1

Malignant Glioma.….…………...…………………………………………… 1

Angiogenesis in Malignant Glioma..…..…………………………….………. 4

The Role of Cell-adhesion Molecules in Tumor Angiogenesis ...…….…… 10

Integrins and Downstream Signaling..………..……….……………..…….. 10

L1 Cell Adhesion Molecule….…………………………….…….…..…….. 16

Cell-cell Interaction and Communication…...... ……………….....………... 17

The Perivascular Niche and its Regulation of Tumor Angiogenesis…….…. 20

Significance and Aims………………………….………...... …………. 22

II. MATERIALS AND METHODS…...... ……..……………………….... 26

III. RESULTS…...……………..…..…………………………….…………..…. 33

IV. DISCUSSION………………………..…………………………...………… 68

Discussion and Conclusions…………………………………...…………… 68

Future Directions and Perspectives………………………………………… 73

REFERENCES……………………………………………………………………...…...77

APPENDICES

A. Supplemental Data……………………………………………………………… 98

iii

B. List of Abbreviations….……………………………..……………...…………. 104

iv LIST OF FIGURES

Figure 1.Model of an integrin-initiated pro-migratory signal transduction pathway……….…...13

Figure 2. Integrin αvβ3 on ECs and L1CAM on CSCs mediate the direct contact of ECs and CSCs………………………………………………...……34

Figure 3. ECs express β3; integrin αvβ3 is likely the major integrin mediating CSC - EC adhesion…………………………………………..………35

Figure 4. An RGD peptide and β3 or L1CAM downregulation inhibits CSC-EC adhesion……………………………………………………………….36

Figure 5. EC and CSC interaction in mouse brain slice organotypic culture……………….…...38

Figure 6. CSCs from GBM promote network formation of brain ECs…...... 39

Figure 7. The direct interaction of ECs and CSCs activates ECs…………………….…………40

Figure 8. The direct interaction of ECs and CSCs promotes activation of integrin αvβ3……………………………………………………………………….…...41

Figure 9. The direct interaction of ECs and CSCs promotes phenotypic changes in ECs………………………………………………………………..…....……42

Figure 10. The direct interaction of ECs and CSCs promotes activation of p130CAS ECs………………………………………………………………………………….…....43

Figure 11. p130CAS co-localizes with the EC membrane in ECs co-cultured with CSCs……………………………………………………………...…....44

Figure 12. The direct interaction of ECs and CSCs activates ERK and JNK in ECs in an RGD-peptide-dependent manner…………………………...... …46

Figure 13. CSCs increase the directional 2D motility of ECs through secreted factors. EC migration is mediated by an α6 integrin……………………………………….….....47

Figure 14. CSCs increase the directional 2D motility of ECs through secreted factors………....49

v Figure 15. CSCs increase the motility of ECs through direct contact………………………..….51

Figure 16. The direct interaction of ECs and CSCs promotes EC migration towards bFGF…………………………………………………………………………..52

Figure 17. EC migration on laminin is not mediated by integrin αvβ3; the direct interaction of ECs and CSCs promotes EC migration towards bFGF………54

Figure 18. BMX, FAK, and p130CAS are required for enhanced EC migration…………..……56

Figure 19. BMX is necessary for p130CAS activation in ECs contacting CSCs……………..…58

Figure 20. Direct contact with CSCs increases phospho-BMX in ECs…………………..……...59

Figure 21. Integrin αvβ3 is necessary for BMX activation……………………………….……..60

Figure 22. Xenograft orthotopic model of GBM: Experimental Schematic………….………….61

Figure 23. LN-308 GBM cells express L1CAM, bind to ECs, and promote EC migration in an L1CAM-dependent manner………………………………63

Figure 24. Administration of the RGD-peptide significantly reduces the proximity of Sox2-positive tumor cells to ECs ECs in an intracerebral xenograft model of GBM………………………………………………………………..64

Figure 26. Model of proposed integrin αvβ3-mediated pro-migratory signaling on ECs when binding L1CAM on CSCs………………………………………67

Figure 27. Specificity of rabbit affinity purified anti-phospho-p130CAS (Y234) antibody……………………………………………………………………………98

Figure 28. Controls for the cell-cell adhesion assay and 2P-LSM; and an association of ECs and CSCs after co-seeding on Matrigel®………………………………………100

ACKNOWLEDGMENTS

It gives me great pleasure to express my utmost appreciation towards my advisor, Dr.

Candece Gladson, for her kind acceptance, encouragement, and guidance for the past six years. I would also like to thank my committee members Dr. Gail Fraizer, Dr. Derek Damron, Dr.

Matthew Summers, and Dr. Songping Huang for their valuable time and participation in my academic journey. I additionally owe a special thanks to Judy Wearden and Dr. Robert Dorman for their advice and assistance throughout these years.

I consider it an honor to have worked alongside my lab colleagues—your training, advice, comic relief, and mentorship were instrumental in my success. Our enormously fascinating discussions will hardly be matched again.

I am indebted to my former college professors, for encouraging me to “aim high”, for their patience and tutoring, and for a continual assurance in my abilities more immense than I could instill in myself at such an early stage of my career.

Lastly, this dissertation would not have been possible without the company of my friends, particularly those who made my first year in college the most memorable and unforgettable; and of course, procuring this milestone consummation would have remained nothing more than an aspiration, had it not been for the infinite confidence and love from my family. I owe my sincere gratitude to all who were instrumental in my success.

vii

CHAPTER 1

INTRODUCTION

Malignant Glioma

Malignant gliomas are lethal primary brain tumors commonly characterized by an invasive, permeative phenotype (1). These tumors can arise in both children and adults; however, the incidence increases with age, occurring most often in adults over the age of 45 (2).

The World Health Organization (WHO) classification of malignant gliomas grades the malignancy of gliomas according to histological and specific genetic characteristics, such as hypercellularity, nuclear atypia, mitotic figures, gene amplification/deletion, evidence of angiogenesis, and the degree to which the disease responds to treatment. WHO grades gliomas on a scale of I to IV; individuals with grade I tumors can have a long survival due to the benign nature of the tumor and if the location of the tumor allows resection, however, grade IV tumors, otherwise known as glioblastomas (GBM), are remarkably challenging to treat, unresponsive to surgery, and are incurable due to their unbridled cellular and vascular proliferation, evasion of

1 apoptosis, aggressive invasion, and inclination for necrosis, which induces hypoxia. As a result, patients diagnosed with GBM experience tumor relapse and have a bleak prognosis, with a median survival of merely 12-18 months in spite of current therapeutics (1-4). As the most malignant form of gliomas, GBMs are subdivided into primary or secondary categories, dependent upon the path of disease development. Primary, or de novo, GBM is the more common of the two subtypes, and occurs most often in adults greater than 55 years of age. This form of GBM arises without evidence of the presence of a pre-existing lower grade glioma.

Conversely, secondary GBMs are more uncommon and tend to develop over a period of 5-10 years (1) in younger patients from previously diagnosed lower grade astrocytomas (grade II/III) due to accumulated genetic alterations. Both primary and secondary GBMs fulfill the aforementioned diagnostic criteria for grade IV classification, and as a result, are impossible to distinguish histologically regardless of the differences in development (1-3, 5).

Despite common clinical behavior, primary and secondary GBMs have a multitude of different genetic alterations and differences in gene expression that lead to disease formation.

Many of these genetic aberrations target signaling pathways affecting cellular proliferation, survival, angiogenesis, and invasion (1). For example, epidermal growth factor receptor

(EGFR) gene amplification and loss of phosphatase and tensin homolog (PTEN) has been elucidated to be a feature of most primary GBM, while the loss of tumor protein p53 (TP53) but no EGFR amplification is associated with the secondary GBM subtype (6). Analysis of differences in gene expression patterns and DNA copy number in adult GBM samples has suggested three molecular subclasses of GBM, namely, classical, mesenchymal, proneural, with a possibility of perhaps a neural subtype (5, 7). Amplification of EGFR and its overexpression was prominently observed in the classical GBM subtype, in addition to the enhanced expression

of Notch and Sonic Hedgehog signaling pathway components. The mesenchymal GBM subtype was characterized, for example, by neurofibromin (NF1) mutations and overexpression of genes in the NF-B pathway, coupled with a high amount of inflammation and necrosis. The highest rate of amplification of platelet derived growth factor receptor α (PDGFRA) and its overexpression occurred in the proneural GBM subtype, as well as mutations in TP53 and isocitrate dehydrogenase 1 (IDH), which rarely occur in the classical subtype. Interestingly, the proneural subtype is connected with GBM in younger patients and is the major classification of secondary GBMs. Finally, the neural GBM subtype was mainly described by expression of neuronal markers, such as neurofilament (NEFL) (5,7).

About half of GBM patients also harbor methylation of the MGMT (O6-methylguanine methyltransferase) promoter, which causes silencing of a DNA repair protein. As a result, the tumor has a reduced capability for restoring DNA in response to chemotherapy, e.g.,

Temozolomide, and confers a survival advantage to older patients with a methylated MGMT promoter (8-9). However, not every patient with MGMT promoter methylation responds to

Temozolomide therapy, which may reflect the effect of intratumoral heterogeneity of GBM, where there are specific areas in the tumor which bear different mutations and gene expression attributes. Examples of such heterogeneity include the alternative or co-amplification of different receptor tyrosine kinases such as EGFR, MET, and PDGFRA within a GBM tumor

(10), or different mutations in TP53 (11). All in all, GBM pathogenesis is partially a consequence of complex interactions between genetic and epigenetic variations and gene expression, which is not uniform among patients who suffer from the disease.

Angiogenesis in Malignant Glioma

A striking feature of GBM is sprouting angiogenesis, which is due in part to disregulated growth factor and cell-adhesion receptor signaling in tumor cells and endothelial cells (ECs).

This process of forming new blood vessels from pre-existing vessels contributes to the rapid infiltrative and proliferative phenotype of GBM tumors (3,12), as well as post-therapy recurrence

(4,12-13). A tumor may co-opt existing blood vessels or may arise in an avascular manner, growing until the angiogenic process has been initiated to further promote vessel formation in order to continue to support the growth and progression of the tumor via oxygen and nutrients.

Sprouting angiogenesis in tumors can very broadly be divided into four basic processes: 1) EC activation and vessel breakdown, 2) protease degradation of the EC basement membrane, 3) EC sprouting and migration to generate new blood vessels, and 4) lumen and tube formation (12-13)

. Sprouting angiogenesis is stimulated by a hypoxic tumor microenvironment, where blood vessels respond to pro-angiogenic growth factors that are upregulated and released by stromal and tumor cells in response to hypoxia. Some of the most potent pro-angiogenic growth factors secreted include VEGF-A (vascular endothelial growth factor A) and bFGF (basic fibroblast growth factor or FGF2), which activate ECs, resulting in heightened EC survival, protease secretion, proliferation, sprouting, and migration (13-16). VEGF-A binds and activates its receptors VEGFR1 and VEGFR2, but primarily VEGFR2, as it has more elevated tyrosine kinase and signaling activity. VEGF-A can stimulate EC mitotic activity and therefore proliferation by ligating VEGFR2, which in turn as been found to interact with, and phosphorylate, PLC-γ, leading to PKC-β, Raf-1, MEK1/2, and MAPK activation, prompting

DNA synthesis in primary ECs (17-18). Vessel breakdown is initiated by the upregulation of

angiopoietin-2 (Ang-2) by tumor cells and ECs, which acts as an antagonist towards its EC- specific receptor Tie2, causing disruption of EC junctions or contact with support cells, vessel destabilization, vessel sprouting, and increased sensitivity to VEGF-A. Ang-2 is also believed to inhibit the activity of angiopoietin-1 (Ang-1), a Tie2 receptor alternative ligand, that normally aids in maintaining vessel stabilization. As a result, in the presence of upregulated VEGF-A, which aids in EC survival and proliferation, the angiogenic response is established (13, 19-21).

Protease degradation of the extracellular matrix (ECM) and basement membrane (BM) assist in the local migration and invasion of ECs and glioma cells through the ECM. Proteolytic degradation of the ECM or BM is mediated by matrix metalloproteases (MMPs), serine proteases, such as urokinase-type PA (uPA) and plasmin, and cysteine proteases (22-23). These proteases disengage supportive pericytes from blood vessels and release bound pro-growth and pro-angiogenic factors in the ECM which further prompts the vascularization of ECs, in part by actuating signal-transduction cascades that promote motility. As compared to the low-grade versions of glioma, GBMs highly express MMPs—MMP-2 is notably upregulated in ECs of

GBM, as well as MMP-9 in proliferating ECs (22-25). Research has shown that MMP-9 is supplied by cells that are contiguous to the vasculature or by ECs themselves, and is required for initiating the “angiogenic swtich” (26). The interaction of SNB19 glioma cells with human brain microvascular ECs with the addition of a PKC activator (4-phorbol-12-myristate 13-acetate

(PMA)) enhanced EC network formation and increased MMP-9 production (27). SNB19 cells cultured with human dermal ECs in the presence of an tissue inhibitor of MMPs (TIMP) , a PKC inhibitor, or an anti-MMP9 antibody caused a decrease in tube-like structure formation (28). The majority of MMPs are secreted in an inactive form (pro-MMP) , but can be activated by proteinases such as plasmin or by integrins as in the case of integrin αvβ3 promoting activation

5 of pro-MMP-2 (29). In addition to activating MMPs, plasmin can also cleave uPA to its active form and contribute to ECM degradation (22-23, 30). uPA and its receptor (uPAR) also have important roles in neovascularization. Studies have indicated an augmentation of uPA and uPAR in GBM tissue in contrast with normal brain (28,30- 31). Downregulation of uPA or uPAR inhibited invasion of glioma cells through Matrigel®, which contains laminin, collagen IV, and proteoglycans (28). The involvement of uPAR in cell migration is evidenced by its known association with integrins, and activation of cytoskeletal proteins, resulting in the activation of

MEK-ERK and JAK-STAT pathways (32). In particular, uPAR was mostly detected at the leading edge of tumors, and was discovered to be coupled with integrin αvβ3 at focal contacts between U251MG GBM cells and the ECM (29, 33). This interaction of uPAR with integrin receptors can also indirectly escalate MMP activity (32). Ultimately, proteases and their cell- surface receptor interactions, including interactions with integrins, aid in establishing the foundation for cell migration, a key step in angiogenesis.

Cell adhesion to the ECM is imperative for EC motility. The adhestion, migration, and proliferation of ECs towards paracrine-acting pro-angiogenic factors is mediated by integrins in cooperation with growth factors/growth factor receptors (33). Cell surface-expressed integrins recognize and are activated by ligands typically localized in the extracellular environment. This activation, in turn, results in activation of cytoplasmic kinases, such as Src and focal adhesion kinase (FAK), as well as adaptor molecules, such as p130CAS/BCAR1 (Breast cancer anti- estrogen resistance protein 1) , described in more detail in the following section. This results in reorganization of the actin cytoskeleton and generation of the mechanical force needed to pull the cell forward (34-35). Motile ECs first sense stimuli, such as VEGF, through filopodia extensions, and subsequently become polarized by forming cytoplasmic protrusions of

6 lamellipodia. This is aided by re-organization of the cytoskeleton, promoted by active Cdc42 and

Rac (downstream of VEGFR2) , which target WASP and WAVE proteins, respectively (36-37).

Small focal adhesions formed by clustering integrins anchor the cell cytoskeleton to the ECM and convert protrusions into forward movement (38). Ensuing activation of RhoA and ERK pathways allows for contraction of stress fibers, while disassembly of focal adhesions at the rear end of the cell complete the forward movement (39-40). Migration diminishes when integrin clusters become larger and stabilize.

In conjunction with integrins themselves, constituents of the ECM and BM are also known to play a major role in EC migration and capillary formation. The aforementioned proteolysis of the BM and ECM exposes previously sequestered domains of collagens, for example collagen type IV, which was shown to play a functional role in angiogenesis, mediated by integrin αvβ3 binding (41). In GBM, an alternatively spliced form of fibronectin, which is limited to the ECM surrounding new blood vessels, ligates integrin αvβ5, and is thought to be an important interaction during the angiogenic process (42). Other pro-angiogenic ECM constituents found in gliomas include osteopontin and tenascin-C (42). Conversely, as more of the BM and ECM is degraded, endogenous molecular fragments, such as endostatin, that are resistant to proteases can act in an anti-angiogenic manner (42-43). In addition to exposure of

ECM components, proteolysis facilitates the creation of vascular guidance tunnels, through which ECs can migrate without further MMP activity (44).

The formation of tubes or vessels occurs via development of cable-like structures, stimulated, for example, by collagen I and β1 integrin interactions, such as α2β1. It has been demonstrated that the collagen receptor integrin α2β1 is up-regulated on brain microvascular endothelial cells in malignant astrocytomas (45-46). This interaction can induce activation of

Rho and Src, leading to induction of actin stress fibers (47). In addition, interaction of α2β1 on

ECs with collagen was shown to activate Rac1 and Cdc42, promoting EC tube and lumen formation, a requirement for blood vessel stabilization. Establishment of EC apical-basal polarity is needed for tube and lumen formation; Cdc42 activates and forms a complex with the downstream cell polarity proteins Par3 and Par6b, as well as atypical PKCζ, which regulates fusion of intracellular vacuoles and thus lumen formation (48-49). VE-cadherin phosphorylates atypical PKCζ to aid in junctional localization and activation of the Par3-Par6b-PKCζ polarity complex to further support vascular tube and lumen formation (50). It was also found that activation of FAK contributes to brain microvessel endothelial tube formation in a 3D collagen gel (46). Cytokines bound to the ECM, like intereukin-3 (IL-3) or stromal-derived factor-1 a

(SDF-1α) can regulate tube formation with pericytes (44). Morphogenesis of the vasculature is therefore profoundly influenced by growth factor and cell-surface integrin – ECM interactions.

New construction of the vascular basement membrane also corresponds with maturation and stabilization of blood vessels. Pericytes have been implicated in reinforcing new BM matrix formation around neovessels by synthesizing laminins, fibronectin, collagen type IV, and integrins, such as α6β1, which recognize the reassembled BM matrix (44, 51). However, it is known that the EC BM in tumors, including GBM, has irregularities, including unusual thickness, and weak connections to endothelial cells and pericytes. Pericytes themselves are disorganized, lack proper structure (are tortuous), are reduced in number, and are not well associated with the vasculature (51). Nevertheless, ECs recruit pericytes in gliomas by releasing platelet-derived growth factor (PDGF) (52-53). Interestingly, glioblastoma tumor cells with cancer stem-like properties (CSCs) have been suggested to transdifferentiate into pericytes through the actions of TGFβ (54). Overall, pericytes assist in depositing an EC BM to aid in

blood vessel stabilization, although the resultant vasculature has abnormalities in its BM, pericyte organization, and its own structure and function.

Additional mechanisms of vascularization include glioma cell co-option of previously existing vessels of the host, allowing the tumor to be vascularized earlier. Host defense mechanisms then cause vessel regression, resulting in a hypoxic tumor. However, tumors may overcome regression by inducing new angiogenesis by EC sprouting and migration, which is again facilitated by an increase of signals in the tumor microenvironment such as Ang-2, VEGF-

A, and Ephrin-B2 (55). Vasculogenesis, vascular mimicry, and transdifferentiation of CSCs to

ECs remain as yet other possible mechanisms of GBM vascularization, although the relative significance and contribution of these potential mechanisms has yet to be confirmed. In vasculogenesis, endothelial progenitor cells (EPCs) circulate and express CXCR4, and thus are recruited to, and differentiate at, sites of neovascularization, especially in hypoxic conditions due to the actions of SDF-1α, as well as VEGF and Ang-2 (56-57). Vascular mimicry may also serve to complement vascularization in GBM, through the generation of vessel-like channels by tumor cells. These channels lack ECs but are still functional and stable via the presence of glioma cells lining the structure externally. It was reported that these glioma cells shifted to a more stem cell- like phenotype as indicated by an increase in CD133 expression (58). It was also reported that glioma stem-like cells have increased expression of VEGFR2, which was fundamental for tubule formation in GBM xenograft tumors (59-60). The most recently described and controversial mechanism of vascularization is the idea of transdifferentiation of glioma cancer stem cells to

ECs. Reports have suggested that neurospheres or CD133+/CD105- GBM cells that were cultured in EC culture conditions generated CD31+ vessel-like structures with increased expression of

CD105, CD31, and VEGFR2 (61-62). However, the relevance of these briefly described alternate processes of vascularization in GBM must still be validated.

The Role of Cell-Adhesion Molecules in Tumor Angiogenesis

As annotated earlier, surface cell-adhesion molecules on ECs play a crucial role in the initiation of angiogenesis, namely by attaching to the ECM, by ligating pro-angiogenic growth factors that are soluble or immobilized by the ECM such as VEGF, HGF, SDF1α, and IL-3, and by cooperating with growth factor receptors or other cell surface adhesion proteins, all of which mediate processes such as migration, and induce signaling events that regulate EC proliferation, activation, and survival (63). The following section will focus on the integrin family and immunoglobulin superfamily, with special emphasis on integrin αvβ3 and L1CAM, as both have been implicated to play important roles in cell motility and have increased expression in GBM

(64-65).

Integrins and Downstream Signaling

Integrins are a large group of cell-adhesion proteins that function in cell motility, invasion, growth, survival and gene expression by forming contacts with the ECM and by transducing signals both into and out of the cell. This family of receptor heterodimers are made up of two subunits, an α subunit, and a β subunit that both span the lipid bilayer and associate noncovalently. A short cytoplasmic tail links these heterodimers to the actin cytoskeleton.

Several cell surface-expressed integrins on ECs recognize and are activated by diverse ligands typically localized in the extracellular environment, such as fibronectin, laminin, and collagen.

Integrins can additionaly recognize cell surface receptors like ICAM-1 or VCAM-1. Some integrins have several ligands, for example, the αvβ3 integrin recognizes the RGD peptide motif in multiple ligands (35). Notably, certain integrins, such as αvβ3, have increased expression in

GBM tumors, particularly on proliferating ECs, aiding in cell adhesion to the ECM, cytoskeletal reorganization, motility, and survival (65, 69). Integrin αvβ3 expression is additionally detected at the periphery of high-grade gliomas (65). Integrin αvβ3 also controls cell invasion through the

ECM by regulating expression and activation of MMPs, particularly MMP-2, which degrade

ECM components (67-68). Ligand binding promotes integrin receptor clustering that initiates focal adhesion complex assembly at the leading edge of the cell where actin stress fibers are connected to the ECM through recruited cytoskeletal and adaptor proteins, including talin, vinculin, paxillin, Src, and FAK (35, 67). Talins and kindlins bind to the β subunit of the integrin heterodimer, promoting an integrin conformational change and increasing integrin affinity for extracellular ligands in the ECM, which aids in establishing cell polarity and driving cell migration (69). Activation of recruited FAK via integrin ligation to ECM ligands was found to be necessary for EC tube formation by promoting EC migration (70-71). The interactions between activated integrins at focal adhesions with multiple scaffolding and signaling proteins drive biological processes such as migration, by regulating multiple signaling pathways, as highlighted below.

FAK is one of the important non-receptor tyrosine kinase proteins that mediate integrin- stimulated signaling, ultimately promoting turnover of focal adhesions during cell migration, and also affecting EC proliferation and survival. FAK is recruited to integrin clusters at focal

adhesions, and like talin, interacts with the integrin β subunit cytoplasmic tail and has binding sites for downstream signaling molecules (67). Within the carboxy-terminus of FAK is the focal adhesion targeting (FAT) domain, which contains binding sites for paxillin and talin, thereby linking integrins with the actin cytoskeleton. The FAT domain also interacts with and phosphorylates p190RhoGEF, thus activating and promoting RhoA activity (72). The C-terminal half of the FAK molecule contains two proline-rich regions, which allow SH3-containing proteins to bind, such as p130CAS. The N-terminal FERM domain (also known as band four point one, erzin, radixin, and moesin) is important in linking FAK to other lipids and proteins and transducing signals from activated growth factor receptors such as EGFR and PDGFR, which is implicated in stimulating cell motility (73). The binding of integrins to the ECM or growth factor receptor activation triggers FAK phosphorylation, through the association of the

N-terminal FERM domain of FAK to the β integrin subunit, changing the confirmation of FAK and exposing residue Y397 to phosphorylation. This creates a high-affinity SH2 binding site for molecules such as c-Src, the Shc adaptor protein, PLCγ, and PI3K. c-Src further heightens FAK activation by phosphorylating Y576 and Y577 in the central catalytic domain of FAK, which in turn increases Src autophosphoylation (73-74). FAK can also be phosphorylated at Y407, Y861 and Y925. Src phosphorylation of FAK at Y925 promotes SH2-mediated binding of Grb2.

FAK/Grb2 was found to signal to ERK2 (Fig. 1), resulting in enhanced VEGF expression, tumor growth, and neovascularization (75). Enhanced activation of the MAP kinase pathway in response to Grb2-Sos-Ras activation has been implicated in cell migration. The binding of Src to

FAK and subsequent activation was found to be promoted by a WISP-1/integrin αvβ3 interaction, ultimately regulating VEGF expression (76). Other signaling effectors activated by

FAK that participate in cell migration include p130CAS, PI3K and Ras, leading to enhanced

activation of the MAP kinase pathway, the recruitment and activation of JNK at focal contacts through phosphorylation of the JNK scaffold protein JSAP1 and the transduction of signals through the interaction of the adaptor protein p130CAS with FAK and multiple cytoplasmic signaling proteins (73).

Figure 1. Model of a pro-migratory signal transduction pathway initiated by integrin ligation. Activation of integrins by a RGD-containing ligand triggers FAK autophosphorylation at Tyr-397, creating a high-affinity SH2 binding site for c-Src, which in turn phosphorylates FAK at Tyr-925 promoting SH2-mediated binding of Grb2. Grb2 signals to ERK, enhancing cell migration. The binding of FAK and Src to p130CAS promotes phosphorylation of p130CAS in its substrate domain. The SH2 domain of the adaptor protein Crk binds to the phosphorylated substrate domain in p130CAS, and Rac1 is activated. Rac1 can target JNK, which phosphorylates paxillin at Ser178, further enhancing cell motility. EC = endothelial cell.

As mentioned above, integrin engagement promotes the activation of Src and FAK, and in fact Src and FAK co-activate each other. FAK binds to the SH3-containing amino-terminal of p130CAS. Activated c-Src binds to the carboxyl-terminal of p130CAS via its SH2 domain, and the FAK-Src-p130CAS complex promotes phosphorylation of p130CAS in its substrate domain

(Fig. 1) and recruitment of p130CAS to focal adhesions (74, 77). FAK overexpression was implicated in prompting p130CAS phosphorylation and cell migration in fibroblasts (78). The

SH2 domain of the adaptor protein Crk binds to the phosphorylated substrate domain in p130CAS, and Rac1 is activated (Fig. 1 ) from successive recruitment of the guanine nucleotide exchange factor DOCK180, influencing cytoskeletal reorganization. Rac1 can target JNK, which phosphorylates paxillin at Ser178 or manages gene transcription of proteases like MMP2 and

MMP9. In ECs, it was found that Cdc42 and RhoA were both activated downstream of p130CAS

(73-74, 79).

In some cells or conditions, regulation of cell migration through p130CAS is also promoted by p130CAS interaction with BMX/Etk. BMX (Bone Marrow Tyrosine Kinase Gene

In Chromosome X) is a member of the Bruton’s tyrosine kinase family and contains a pleckstrin homology (PH) domain, a SH2 domain, a SH3 domain, and a kinase domain. Although largely known for its function in lymphoid cells, it is becoming apparent that BMX is also expressed in other cell types, such as in ECs. In addition to FAK and c-Src, BMX phosphorylates p130CAS within its substrate domain, promoting the binding of BMX via its SH2 domain to p130CAS. It was discovered that BMX enhances p130CAS-Crk complex formation and was found to form a complex with p130CAS at membrane ruffles following integrin-mediated adhesion, enhancing the formation of lamellipodia and enhancing cell motility (80). The activity of BMX itself was observed to be dependent on FAK Y397 phosphorylation in ECs after integrin engagement, with

13 the possible involvement of c-Src. FAK participates in BMX activation by phosphorylating the

Y40 residue in the BMX PH domain, and also co-localizes with BMX at membrane ruffles via its FERM domain (81). This signaling pathway can aid in mediating integrin signaling in ECs and promote cell migration.

Integrin signaling frequently functions in cooperation with signals initiated by growth factor receptors. The MAP kinase pathway, for example, is most efficiently activated when integrins and growth factor receptors are concurrently stimulated (82). The incitement of angiogenic responses in ECs has been shown to be distinctly influenced by integrin αvβ3 in concert with growth factors, such as EGF, PDGF, and VEGF (83). The coordination of integrins with growth factor receptors can instigate a certain type of signal. For instance, integrin αvβ3 has been shown to interact directly with PDGFR and VEGFR2. Stimulation of ECs with VEGF increases c-Src interaction with phosphorylated β3 and PDGFR, inducing integrin-dependent cell migration. In addition, VEGF intensifies integrin αvβ3 activity in ECs by promoting an association of integrin αvβ3 with VEGFR2. Phosphorylation of the β3 subunit at Y747 and Y759 by c-Src at integrin clusters promotes the interaction between VEGFR2 and integrin αvβ3, further bolstering VEGFR2 phosphorylation, and ultimately regulates VEGF signaling in the processes of EC migration, as well as EC proliferation (84). It was notably found that VEGFR2 does not form a complex with the β1 or β5 integrin subunits (85). In addition to interacting with integrins, FAK also interacts with growth factor receptors through its N-terminus, and may function in integrin-growth factor receptor cross-activation. Integrin αvβ3 has also been suggested to interact with the growth factor receptors FGFR3 and Met, enhancing reciprocal phosphorylation. (84, 86). These interactions highlight the various ways integrins can mediate

15 biological processes by transducing intracellular signals, and how integrin activation can regulate integrin affinity for its ligands and its cell surface interacting partners.

L1 Cell Adhesion Molecule

L1CAM is a transmembrane neuronal adhesion receptor that regulates cell-cell and cell-

ECM interactions (64). Physiologically, L1CAM is involved in the development of neuronal tissue, but has also been detected in other tissues. It is part of the immunoglobulin superfamily of cell adhesion molecules (IgCAM) and its extracellular domain is made up of six Ig-like and five

FN type-III domains, with a transmembrane region and a conserved cytoplasmic tail. L1CAM functions in neural processes, but its expression has also been found to be correlated with advanced uterine tumors, melanoma, colon cancer, GBM, and its expression is increased in

CSCs from GBM (64, 87). In particular, L1CAM was discovered to have higher levels in CSCs than in normal neural progenitors, and in CD133+ CSCs isolated from GBM tumors, suppression of L1CAM blocked CSC tumorigenic capacity and CSC survival (88). L1CAM expression can be induced by TGF-β1, TNF-α, and IL-1β (89). L1CAM is able to bind in a homophilic manner to another L1CAM molecule on the same or on an adjacent cell, or in a heterophilic manner to other proteins, such as RGD (Arg-Gly-Asp )-peptide binding integrins through its RGD sequence in its sixth Ig domain (87, 90). When bound to ligands, L1CAM can enhance cell motility in normal and cancer cells by activating ERK through the adaptor protein

RanBPM, or through loss of adherens junctions, depending on the cell type. ERK activation can also lead to an inhibition of apoptosis via PI3K/AKT signaling. L1CAM can be cleaved by

matrix metalloproteases such as ADAM10 after the sixth Ig domain to release the ectodomain, which can, in addition to full-length L1CAM, bind to integrins, receptor tyrosine kinases, or an

L1CAM molecule on the same or neighboring cell, aiding cancer cell invasion (87, 90). GBM cells have been suggested to have an increased migratory capacity when stimulated with soluble

L1CAM (91), which may occur through regulation of FAK activation (92). Soluble L1CAM was also suggested to act as an angiogenic stimulus by ligating integrins on bovine aortic ECs, as evidenced by an increase in bovine aortic EC growth, Matrigel® invasion and tube formation.

The combination of soluble L1 and VEGFA165 promoted a three-fold increase in VEGFR2 activation in bovine aortic ECs, as compared to treatment with VEGFA165 alone (93).

Interestingly, as described earlier, integrin αvβ3 plays a role in the full activation of VEGFR2 in the presence of VEGF. Aside from the soluble ectodomain, the cleavage of L1CAM by

ADAM10 and γ-secretase was demonstrated to cause the C-terminal intracellular portion of

L1CAM to be translocated to the nucleus in HEK293 cells, affecting L1CAM-dependent gene expression of Cathepsin B and HoxA9, among other genes (94). As with integrin αvβ3, L1CAM has been shown to cooperate with growth factor receptors. L1CAM can interact with FGFR,

EGFR, and c-Met to induce a more motile phenotype; for example, a dimerized form of L1CAM can bind to FGFR in trans, which activates the receptor in a similar fashion as the canonical

FGF ligand (94-96).

Cell-cell interaction and communication

Tumors of various origins, including GBM, contains cell masses that are heterogeneous.

Various unique tumor microenvironments embody diversified cell populations which

17 communicate with each other through secreted factors but also through cell-cell adhesion interactions. These interactions between distinctive cells can provide molecular signals to influence cell functions like growth, invasion, and survival (64, 97-98). In terms of integrins mediating cell-cell interactions with cell adhesion molecules, it has been reported that integrin

αvβ3, αvβ5, αvβ1, α5β1, and α9β1 can interact with L1CAM (94, 99-101). Integrins αvβ3,

α5β1, and αvβ1 bind to the RGD site in the sixth Ig domain of L1CAM (87, 94, 100-101). It was also proposed that the third FN-like domain can support RGD-independent recognition by integrins αvβ5, α5β1 and α9β1 (101-102). Adhesion of human umbilical vein ECs (HUVECs) to a L1CAM-fusion protein including only the sixth Ig-domain (L1-Ig6) could be blocked by an anti- integrin αvβ3 neutralizing antibody in the presence of divalent cations. Wild-type HUVEC adhesion to L1-Ig6 in, for example, either Ca2+ or Mg2+ alone, relied wholly on integrin αvβ3.

Furthermore, the greatest adhesion of HuVECs resulted from conditions where all three cations--

Ca2+, Mg2+, and Mn2+, were present, and was significantly decreased by a blocking antibody to integrin αvβ3. There was no indication that there is an interaction between L1-Ig6 and integrin

αvβ5 (100, 102). The possibility exists that divalent cations may induce conformational changes in integrins, and the structural arrangement of L1CAM can alter its association with integrins.

Subsequent research found that the various conformations of the L1CAM ectodomain regulated its interactions with integrins. Phosphorylation events at the serine residue (S1181) or the threonine residue (T1172) of the cytoplasmic tail of L1CAM promote changes in the conformation of the L1CAM extracellular domain. Phosphorylation at T1172 by PKC promotes the folding of L1CAM into a folded, condensed monomer, while dephosphorylation of T1172 promotes extension of the extracellular domain. This extension of Ig domains 1-6 allows for the full potency of L1CAM homophilic binding. Integrin αvβ3 has been reported to bind to the open,

multimerized conformation, while integrin αvβ5 and/or integrin α5β1 bind to the folded configuration, likely to the FN3 domain. Immobilized integrin αvβ3 was not determined to be capable of interacting with the third FN domain in a condition that mimics the third FN domain present in the whole L1CAM molecule in the plane of the cell membrane. However, integrin

αvβ5 was found to readily bind the third FN domain. Binding of integrin αvβ3 to the expanded multimer of L1CAM further stabilizes the extension, and may also potentiate integrin activation and signaling, affecting processes like cell motility and survival (101-102). HUVECs stimulated with the extracellular domain of L1CAM (L1Ig1-6) showed higher levels of tyrosine phosphorylated integrin αvβ3 and activation of ERK1 & 2 . It was also determined that HUVECs stimulated with L1-Ig6 in the presence of an integrin αvβ3 blocking antibody experienced apoptosis (103). Of note, L1CAM has additional cis ligands, such as the signal transducer CD24, tetraspanin CD9, and the neural cell adhesion molecule NCAM (100) , and the potential remains for binding these ligands in a trans manner.

As previously stated, cleavage of the L1CAM extracellular domain can be mediated by matrix metalloproteases or ADAM family members and can also potentially modulate neighboring EC function. ADAM-mediated cleavage is prone to occur when L1CAM exists in the globular conformation, whereas serine proteolysis is more likely to occur in the stabilized extended multimer configuration (102). Ovarian cancer cells were demonstrated to shed

L1CAM through cleavage at the cell surface, or on the other hand, through intracellular cleavage and secretion in membrane vesicles, which constitutes another mode of cell-cell communication (104). Membrane vesicles (MVs) are released from a variety of different cell types, including tumor cells and endothelium. MVs not only contain cell-adhesion molecules, but also lipids, such as arachidonic acid and mRNA, which can promote an array of biological

19 reactions, such as EC activation and growth factor secretion (105). L1CAM in vesicles was reported to be functionally active and can promote chinese hamster ovary (CHO) cell migration

(106). It has also been suggested that MVs can be produced by GBM cells with stem-like properties (107). Other cell-cell communication mechanisms within the tumor microenvironment are cell-interactions which utilize cadherins, junctional adhesion molecules within tight junctions, or the use of gap junctions (108).

The Perivascular Niche and its Regulation of Tumor Angiogenesis

In various tissues, stem cells reside within specialized niches or microenvironments for protection and maintenance. The perivascular niche has a high density of blood vessels and represents an ideal microenvironment for the regulation of stem cell fate (109). Also, a vascular niche has been described for neural stem cells (NSC) within the adult brain, and there is a perivascular niche at the subventricular zone (SVZ) that includes a monolayer of ependymal cells and a niche at the dentate gyrus (DG) of the hippocampus. Within these niches, intimate communication occurs between ECs and NSCs, through soluble factors such as brain-derived neurotrophic factor (BDNF). Also, the expression of specific cell adhesion receptors, such as integrin α6β1 on NSCs, promotes NSC neurogenesis (109-111). Additional signals from other cells types and the ECM also may enhance NSC proliferation and function. Conversely, NSCs, through VEGF secretion, can aid in new vessel formation, suggesting a symbiotic bi-directional communication that sustains and activates both NSCs and ECs (112). In malignant tumors, a specialized perivascular microenvironment has been recently identified in which CSCs reside in

very close proximity to ECs (113). The basement membrane generated by blood vessels in the perivascular niche has been suggested to regulate CSC survival in part through integrin α6 expression on CSCs . Use of integrin α6 as a marker of CSCs indicated that approximately 60% of integrin α6-positive cells reside within 5-μm of blood vessels in GBM (114). It has been suggested that ECs maintain the CSC population, promoting the survival of CSCs through secreted soluble factors (e.g., nitric oxide) and laminin-α2 (114-116). Furthermore, ECs are thought to generate ligands that activate Notch receptors on CSCs, promoting CSC growth and self-renewal (117). Similar to NSCs, CSCs have markers such as Nestin and Sox2, and also secrete pro-angiogenic factors, but CSCs have genetic changes in molecular pathways different from NSCs, helping to enhance tumor progression, and may hold a heightened ability to promote angiogenesis (118). Whether gliomas are derived from NSCs that have ndergone genetic mutations permissive for tumor growth, or are derived from a differentiated cell type that has reverted back towards self-renewal has not yet been determined (110). The interaction between

CSCs and ECs within the perivascular niche in glioma and GBM has primarily been suggested to influence CSC biological processes and the CSC phenotype. Yet the secretion of VEGF and

SDF-1 by CSCs was shown to contribute to the migration and proliferation of ECs (14-15) .

Interestingly, the receptor of SDF-1, CXCR4, is elevated on CD133+ CSCs, and may further promote VEGF secretion by CSCs (119). As a result, while the perivascular niche may provide an ideal microenvironment for enriching the CSC population through paracrine interactions with

ECs, thus supporting tumor recurrence and progression, angiogenic capabilities of ECs may be also enhanced by intimate interactions with CSCs, contributing to angiogenesis that is a major hallmark of malignant gliomas (97). There is some evidence that CSCs may interact directly with

ECs; medulloblastoma cancer stem-like cells bind to ECs plated on Matrigel® and promote EC

network formation (113) and melanoma cell contact with ECs in serum-containing media promotes expression of genes that regulate cancer cell migration and tumor progression (97). In these studies, however, neither the mechanisms underlying the interactions nor the cell type expressing the upregulated genes were identified. Thus, the possibility that CSCs interact directly with ECs in the perivascular niche and affect EC behavior remains largely unexplored. The investigation of the role that EC-CSC bi-directional crosstalk plays in GBM pathobiology will lead to a better understanding of signaling mechanisms in the tumor microenvironment.

Significance and Aims

The standard therapy for GBM includes surgery, chemotherapy, and radiotherapy.

Despite these efforts, GBM remains as a non-curative and deadly disease (120). Because patients with gliomas typically have a poor prognosis due to the highly invasive and angiogenic features of the tumor, inhibiting the formation of new vasculature in gliomas through anti- angiogenic treatments has been highly pursued in order to curb tumor growth. Unfortunately, a number of anti-angiogenic treatments that target VEGF have encountered significant challenges, such as acquired resistance, rapid tumor recurrence, and an aggressive proinvasive adaptation of the tumor to the therapy (121-123). This study aims to elucidate new molecular processes to target for anti-angiogenic therapy.

Determining the precise mechanisms of communication between CSCs and ECs, and downstream signals will aid in developing better therapeutic interventions against the molecules promoting EC migration and angiogenesis in GBM patients. Brain tumor therapies have only just

22 begun to target molecular pathways in CSCs and components of the microenvironment. These resolved mechanisms will contribute to the current scientific understanding of how microenvironmental conditions, specifically the perivascular niche, regulate tumor vascularization, thereby improving scientific knowledge and current treatment of malignant gliomas. When the proposed aims are achieved, then anti-angiogenic treatments may potentially be directed at disrupting EC-CSC heterotypic interaction within the protective perivascular niche. This may have implications not only in reducing EC migration and subsequent angiogenesis, but also in a reduction of CSC survival and invasion in malignant glioma.The goal of this research is to decipher the molecular mechanisms involved in the promotion of angiogenesis that is caused by the interplay and crosstalk between ECs and CSCs within the perivascular niche in malignant glioma tumors. I hypothesized that CSCs not only promote EC migration through secreted factors, but also through a direct interaction. As noted, previous studies imply that cell-cell contact and L1CAM activation can augment integrin activation and signaling. As a result, L1CAM may directly promote processes typically affiliated with integrin- binding to the ECM, including cell motility. Hence, I hypothesized that the interaction between

L1CAM on CSCs derived from GBM or GBM cell lines and integrin αvβ3 on primary brain ECs can contribute to the attachment of CSCs to ECs, affect integrin αvβ3 signaling on ECs, and enhance chemotactic migration of ECs, thereby promoting angiogenesis.

I will dissect this hypothesis in the aims below:

Aim 1: Determine the mechanism of cell-cell interaction between CSCs and ECs and elucidate the downstream signals generated in ECs when in direct contact with CSCs.

a) Identify the cell-adhesion molecules required for interaction using blocking antibodies, peptides and downregulation studies in CSC-EC cell binding/adhesion assays.

b) Determine the proximity of CSCs to ECs in human GBM biopsies through double-label immunofluorescence, and the mechanism of interaction of CSCs with ECs on co-injection into mouse brain slices followed by 2-Photon Laser Scanning Microscopy (2P-LSM).

c) Examine the activity state of integrin αvβ3, L1CAM and downstream MAP kinase(s) by determining the phosphorylation state after contact of ECs and CSCs. Examine the mRNA levels of EC activation markers after contact of ECs and CSCs.

Aim 2: Determine the mechanisms by which CSCs isolated from GBM promote the migration of normal primary microvascular brain ECs.

a) Test EC motility with migration assays in a 2D monolayer assay and 3D transwell format.

Manually track EC paths in the 2D assay and determine cell velocity and center of mass. b) Identify the cell-adhesion molecules required using blocking antibodies, peptides, and siRNA. c) Compare the effects of conditioned media from EC-CSC co-culture (secreted factors) on EC migration to the physical interaction of CSCs with ECs (direct contact) on EC migration and phenotype.

d) Identify the MAP kinase pathway(s) or alternative downstream signaling effectors activated in ECs on interaction with or in proximity to CSCs using multi-label immunofluorescence during

24 migration, as well as the necessity of the effectors with small molecule inhibitors or downregulation techniques.

Aim 3: Determine whether cyclic RGD peptide (Cilengitide) inhibits the interaction of

CSCs and ECs in vivo in a mouse model of malignant glioma, thereby inhibiting angiogenesis.

a) Inject and propagate L1CAM-expressing GBM cells in the nude mouse brain followed by euthanasia, harvesting of the brains, and analysis of CSC proximity to integrin αvβ3+ vessels, angiogenesis, and blood vessel surface density. b) Propagate the L1CAM-expressing GBM cells in the nude mouse brain, and once the tumor is established, administer cyclic RGD peptide versus vehicle for 5 days followed by euthanasia, harvesting, and analysis of CSC proximity to integrin αvβ3+ vessels, angiogenesis, blood vessel surface density, and activation states of downstream signaling effectrors.

CHAPTER 2

MATERIALS AND METHODS

Cells

Primary human normal brain ECs were purchased (Cell Systems, Kirkland, WA), cultured in EGM medium with growth factors (SingleQuots; Lonza, Basal, Switzerland), used in the first eight passages and von Willebrand factor (vWf) expression repeatedly verified, as described (124). Primary human CSCs from four different GBM xenografts (08387, 3691, 3832,

4302) were isolated and propagated as neurospheres in complete neural basal medium (NBM)

(Life Technologies, Waltham, MA) without FBS. Re-authenticated LN-308 cells were received from Dr. Nicolas de Tribolet (Lausanne, Switzerland). Re-authentication was performed by DNA profiling of 8 highly polymorphic regions of Short Tandem Repeats using Nonaplex PCR at the

Leibniz Institute DSMZ – German Collection of Microorganisms and Cell Cultures before use in these experiments. ECs were labeled with PKH26-red-fluorescent cell linker (Sigma, St. Louis, MO). Human astrocytes were purchased (Cell Systems) and cultured in

DMEM with 10% FBS. Two-photon laser scanning microscopy (2P-LSM) of

fluorescent-labeled cells in mouse brain slices in organotypic culture and analysis of EC-CSC interaction using Imaris software was performed as described (125).

Inhibitors and downregulation studies

Peptides (Peptides International, Louisville, KY) and the Src inhibitor PP2 (EMD

Millipore) were purchased. Cyclic-RGD-peptide (Cilengitide) used in the mouse model was provided by Merck (Darmstadt, Germany). Small interfering ON-TARGETplus SMART pool

RNA (siRNA) oligonucleotides were purchased (Dharmacon, Lafayette, CO). Cells were transfected with oligonucleotides (HiPerfect, Qiagen, Hilden, Germany) and downregulation confirmed by immunoblotting (124, 126).

Antibodies

Antibodies were purchased: mAbs, anti-integrin αvβ3, anti-integrin β3(CD61), and anti- integrin α5β1 (Millipore, Billerica, MA), anti-human CD171 (L1CAM; BD Pharmingen, San

Diego, CA), anti-integrin αvβ5 (R&D Systems, Minneapolis, MN), anti-actin, anti-β-tubulin, antiphospho- ERK, and anti-phospho-JNK (Santa Cruz, Dallas, TX), and anti-L1CAM UJ127

(GeneTex, Irvine, CA); rabbit antibodies, anti-integrin β3 (Epitomics, Burlingame, CA), and anti-phospho-FAK (Y397), anti-total FAK, anti-phospho-BMX/Etk (Y40), anti-total BMX/Etk, and anti-total p130CAS (E1L9H) (Cell Signaling, Beverly, MA). Rabbit affinity purified phospho-specific anti-p130CAS (Y234) was generated using the following peptide: -

AQPEQDE[pY]DIPRHL, corresponding to residues 227-240 in human p130CAS, by 21st

Century Biochemicals, Inc (Marlboro, MA). (Fig.27 )

Immunofluorescence

GBM biopsy specimens or GBM xenograft sections were fixed in 4% paraformaldehyde and double-labeled for immunofluorescence using rabbit anti-von Willebrand Factor (vWF;

Millipore) and mAb anti-Sox2 (R & D Systems), or anti-pβ3 (759) and anti-vWf. ECs (2.5 x

104) and CSCs (2.5 x 104) were mixed and co-cultured on laminin-coated coverslips in NBM with 10 ng/mL bFGF (3 hrs, 37ºC, 5% CO2) and then multi-labeled for immunofluorescence as described previously (124). Confocal images were taken at 0.5μm steps (40X magnification) with a Leica-TCS-SP5II-AOBS confocal microscope (NDD detector).

Prior to qRT-PCR, the co-culture of ECs and CSCs was harvested with accutase, cells washed and sorted for CD31-positive and -negative cells. Both CD31-positive and CD31- negative cells were then seeded onto laminin-coated coverslips (20 µg/mL). CD31-positive cells were labeled with rabbit anti-vWF followed by anti-rabbit Alexa-594-conjugated secondary antibody, photographed and percent cells positive counted. CD31-negative cells were labeled with mAb anti-Sox2 and rabbit anti-vWf, followed by fluorescent-conjugated secondary antibodies. Cells that sorted as CD31-positive were 92% positive for vWf by immunofluorescence whereas cells that sorted as CD31-negative were 97% Sox2-positive. At least 5 fields were photographed at 20X magnification and quantified for vWf and Sox2 positivity.

Cell-cell adhesion assay

Unlabeled ECs (5x104) were plated on laminin (20 μg/ml) in replicates of five for each condition in a black 96-well microplate in adhesion assay buffer (140 mM NaCl, 5.4mM KCl,

5.56mM D-glucose, 10mM Hepes, pH 7.4, 1mM MgCl2, 100µM MnCl2, 1% BSA), and allowed to attach overnight (37°C, 5% CO2) and the EC monolayer was confluent prior to beginning the assay. Before seeding, the plate was blocked with 5% BSA (30 min). Primary human astrocytes were used as a control. Anti-αvβ3 and anti-L1CAM blocking antibodies (5.0 μg/ml) were incubated for 30 min with ECs or CSCs, respectively. Alternatively, cyclic-RGD or RAD peptides were incubated with both cells. GFP-CSCs (3x104) were plated over the ECs and allowed to adhere for 30 min then washed 3X with PBS and the remaining fluorescence was detected using a fluorimeter at 485nm absorption, 535nm emission. The number of CSCs remaining attached to ECs was calculated based on a CSC standard curve performed in the same experiment.

Matrigel® EC network formation

10 µL of Matrigel® was placed into the lower chamber of an angiogenesis μ-slide (Ibidi

LLC, Verona, WI), and allowed to polymerize for 1 hour. 50 µL of PKH26-red fluorescent-ECs

2X104 ECs, or 2X104 each of ECs and GFP-CSCs were mixed and seeded on the Matrigel® in complete NBM, and incubated at 37C, 5% CO2. Pictures from three independent assays were manually taken on a Zeiss Scope A1 microscope (W.E.L. Instrument Co., LLC, Mars, PA) at 5X

magnification at 2, 4, and 24 hours. CM from CSCs was prepared by seeding CSCs on laminin in complete NBM (72 h).

Motility assays

For the 2D-cell motility assay, cell culture inserts (Ibidi, Research Products International) were placed inside a 6-well plate coated with 20 μg/ml laminin (R&D Systems). PKH26- red fluorescent-ECs in EGM media (3x104) were plated on one side of the insert, and 3x104 GFP-

CSCs in complete NBM on the other side. Controls had ECs or CSCs on both sides.

Subsequently, the media was replaced with fresh complete NBM, the culture insert removed, and live-imaging of migration into the gap carried out every 15 min for 24 h using a Leica motorized-stage microscope (5% CO2, 37°C). The number of CSCs and ECs migrating into the gap was measured by counting each fluorophore in a fully automated fashion using customized software (Image I.Q., Cleveland, OH, USA), see description of software below). Manual

Tracking (Image J) was used to manually trace EC paths in 15 min increments for 24 h during migration into the 500 µm gap, as described (127). Raw data was imported into the Chemotaxis and Migration Tool 2.0 (Ibidi), which projected each of the 30 ECs trajectories on the XY plane and referred the trajectories to the same initial position that coincides with origin of the coordinate system for a qualitative analysis of EC motility (127). To evaluate the motility of ECs quantitatively, the average velocity (displacement/time in μm/min) of ECs was calculated for individual and total trajectories for the time indicated (127). Using Image J, the average velocity of ECs was calculated in the time before (< 7 h) and after contact with CSCs (7-24 h) and also without the presence of CSCs (ECs on both sides as a control) for < 7 h and 7-24 h. The spatial

averaged point of all cell endpoints (center of mass) over a given time for all the collective trajectories was calculated for the noted time frames. The point of origin marked the center of mass of all cells at the beginning of each experiment.

For analysis of number of CSCs and ECs migrating into the wound gap, an automated algorithm was generated by ImageIQ (Cleveland, OH) designed to batch process and analyze time-lapse, multi-channel stacks within ImagePro Plus 7.0 (MediaCybernetics, Silver Spring,

MD). Briefly, each channel stack for a given field (CSC, EC, and phase-contrast) was loaded.

Using a Euclidean Distance map filter on the first frame of the phase contrast stack, an initial wound region (timepoint 0) was established. For each subsequent frame, spectral enhancement filters were applied to each channel to segment cells from background. Cells were counted and summed within the initial wound region additionally evaluating any overlapping or touching cells (CSCs and ECs). Any empty regions not occupied by cells were analyzed to determine wound closure rates for a given field. Each quantified metric was exported to Excel on a frame by frame basis, and segmented cells were pseudo-colored and superimposed upon the original phase-contrast stack for visualization.

For the 3D migration assay (Transwell, Corning), 3-µm pore filters were coated on both sides with 20 µg/ml laminin and cells were allowed to migrate over 6 h. PKH26-red fluorescent-

ECs (3x104) were mixed with 3x104 GFP-CSCs and seeded in NBM plus 1% BSA, on top of the filter. The lower chamber contained either NBM without growth factors, NBM plus 10µg/mL bFGF, or 72 h-CSC-conditioned NBM. After removal of non-migrating cells, cells on the lower filter surface were washed and fixed (4% paraformaldehyde) (128). Migrating red-fluorescent-

ECs for each condition were photograph and counted in 5-10 fields with a 20X objective (Leica-

DFC425C-QImaging-Q15729-QCapturePro).

qRT-PCR

RNA was harvested using the Trizol method, reverse transcribed, and quantitative PCR performed using the Syber Green system and levels of mRNA normalized to GAPDH, as described previously (124). A 2-fold or greater change in normalized mRNA level relative to the control cells was required for interpretation as a significant change.

Animal studies

Animal experiments were carried out as described previously (129) with the approval of the Institutional Animal Care and Use Committee at the University Hospital Zurich (Zurich,

Switzerland). LN-308 cells were cultured in DMEM with 10% FBS, harvested, resuspended in

PBS, and 105 cells in 2-µl stereotactically injected into the striatum of 6-12 week-old anesthetized athymic CD1 nude mice (purchased from Charles Rivers Laboratories). Eight weeks after glioma cell injection, PBS (200µL) or Cilengtide (90 mg/kg) were injected on five successive days (n=5 / group); followed by euthanasia at 2 hrs post-treatment and brain harvest.

Statistics

All statistical analyses were performed/overseen by the biostatistician. The statistical test used is stated in the figure legend; a p value <0.05 was considered significant.

CHAPTER 3

RESULTS

Direct contact of ECs and CSCs involving cell-cell adhesion mediated by integrin αvβ3 on ECs and L1CAM on CSCs

Using a cell-cell adhesion assay, we found that the CSCs readily adhere/bind to ECs and that 5-fold more CSCs adhered/bound to ECs than to astrocytes (Fig. 2, Fig. 28A). We used laminin as the substrate in these experiments as CSCs from GBM retain their stem cell phenotype when plated on laminin in neural basal media (NBM) (130) and integrin α6β1 mediates brain EC attachment to laminin (124).

Integrin αvβ3, an RGD peptide-binding integrin that promotes EC adhesion, migration and survival [reviewed in (35)], is upregulated on tumor-associated ECs in GBM biopsies (65).

The expression of the cell adhesion molecule L1CAM that contains an RGD-peptide is increased on CSCs from GBM (88).

Figure 2. Integrin αvβ3 on ECs and L1CAM on CSCs mediate the direct contact of ECs and CSCs. CSC adhesion to an EC monolayer with blocking antibodies: unlabeled-ECs (50,000/well) were seeded in serum-free adhesion assay buffer on plates coated with 20 μg/mL laminin, allowed to attach overnight and then incubated with blocking antibody or control IgG for 30 min (replicates of five). GFP-CSCs- (08387) (30,000) incubated with blocking antibody, or IgG (30 min) were plated over the ECs, allowed to adhere (30 min), washed 3X with PBS and GFP-CSC fluorescence detected using a fluorometer (485nm absorption, 535 nm emission). Statistics: two-sided exact Wilcoxon rank-sum test. Data graphed as the mean+SEM.

Immunoblotting confirmed expression of the integrin β3 subunit on ECs and expression of L1CAM on CSCs (Fig.3A). Pre-incubation of ECs with a neutralizing antibody to integrin

αvβ3 or αvβ5 significantly reduced CSC adhesion to ECs (43% and 10%, respectively), but preincubation with a neutralizing antibody to α5β1 did not (Fig. 3B). As anti-integrin αvβ3 and anti-

L1CAM in combination did not further inhibit CSC adhesion to ECs, integrin αvβ3 is most likely the major integrin mediating this adhesion (Fig. 2).

A B

Figure 3. ECs express β3; integrin αvβ3 is likely the major integrin mediating CSC - EC adhesion. A. Immunoblotting of adherent ECs; and B. CSC adhesion to an EC monolayer with blocking antibodies: unlabeled-ECs (5x104/well) were seeded in serum-free adhesion assay buffer on plates coated with 20 μg/mL laminin, allowed to attach overnight and then incubated with blocking antibody, or control IgG for 30 min (replicates of five). GFP-CSCs- (08387) (3x104) incubated with blocking antibody, or IgG (30 min) were plated over the ECs, allowed to adhere (30 min), washed 3X with PBS and GFP-CSC fluorescence detected using a fluorometer (485nm absorption, 535 nm emission). Statistics: two-sided exact Wilcoxon rank-sum tests. Data graphed as the mean+SEM.

A cyclic-RGD-peptide significantly inhibited CSC adhesion to ECs in a concentration dependent manner whereas a control RAD-peptide did not (Fig. 4A). Moreover, downregulation of either the integrin β3 subunit on ECs or L1CAM on CSCs significantly inhibited CSC adhesion to ECs (Fig. 4B&C).

Figure 4. An RGD peptide and β3 or L1CAM downregulation inhibits CSC-EC adhesion. A. Unlabeled-ECs (5x104/well) were seeded in serum-free adhesion assay buffer on plates coated with 20 μg/mL laminin, allowed to attach overnight and then incubated with peptide for 30 min (replicates of five). GFP-CSCs-(08387) (3x104) incubated with peptide (30 min) were plated over the ECs, allowed to adhere (30 min), washed 3X with PBS and GFP- CSC fluorescence detected using a fluorometer (485nm absorption, 535 nm emission). B. Immunoblot confirms downregulation of β3 and L1CAM; C. CSC adhesion to an EC monolayer performed similarly after EC or CSC treatment with siRNA to β3 or L1CAM. Statistics: Line graph, two-way ANOVA. Bar graphs: two-sided exact Wilcoxon rank-sum tests. Data graphed as the mean+SEM. 36

We then co-injected red-fluorescent-ECs with GFP-CSCs into mouse brain slices in organotypic culture and used two-photon-laser-scanning microscopy (2P-LSM) to analyze the interaction of CSCs with ECs (125). Treatment of ECs with siRNA to integrin β3 resulted in a

38% reduction in the interaction with CSCs as compared to treatment with control siRNA

(Fig.5A). On downregulation of L1CAM in CSCs there was a significant reduction in the number of CSCs associated with ECs (Fig. 5B), with highly similar results when these experiments were repeated using CSCs isolated from a different GBM tumor (Fig. 5B). Neither downregulation of β3 in ECs nor downregulation of L1CAM in CSCs altered the number of ECs or CSCs detected in the brain slices at 24 h (Fig.28 B-D).

Figure 5. EC and CSC interaction in mouse brain slice organotypic culture. A-B. 2P-LSM of fluorescent-labeled cells in mouse brain slices in organotypic culture. 1:1 mix of ECs and CSCs-(08387 or 4302) (in DMEM with 10% FBS), or ECs alone, was injected into three replicate mouse brain slices and cultured (24 h). EC-CSC interaction was evaluated using 2P-LSM. Representative images and percent ECs in contact with CSCs are shown. Arrows denote EC contact with CSCs. Statistics: two-sided exact Wilcoxon rank-sum test. Data graphed as the mean+SEM..

CSCs from GBM promote network formation of brain ECs, activation of integrin αvβ3 and phenotypic changes in ECs

Upon co-seeding primary brain ECs with CSCs on Matrigel®, an interaction between

ECs and CSCs could be seen at 2 h (Fig. 28E). The number of EC segments/branches (network formation) was higher on co-seeding of CSCs with ECs than when ECs were seeded in CSC conditioned media (CM) (Fig. 6). Thus, in subsequent experiments we differentiated the effects of direct contact to those of soluble factors by comparing the effects of co-seeding to the effects of CM.

Figure 6. CSCs from GBM promote network formation of brain ECs. Redfluorescent ECs (20,000) alone or mixed with GFP-CSCs (20,000) were seeded onto Matrigel® in complete NBM and network/branch formation quantified as the number of segments/branches at 24 h. Controls included ECs (20,000) mixed with astrocytes (20,000). Statistics: two-sided exact Wilcoxon rank-sum test. Data graphed as the mean+SEM..

To determine whether the direct contact activates ECs, we analyzed mRNA levels of two markers of EC activation using qRT-PCR. The results indicated significantly higher E-selectin and VCAM-1 mRNA levels (9-fold and 34-fold, respectively) in ECs that were co-seeded with

CSCs than in ECs seeded in conditioned media obtained from co-seeded ECs and CSCs

(CM/EC+CSC), indicating that EC activation was not induced by a secreted factor at 3 h (Fig. 7).

Figure 7. The direct interaction of ECs and CSCs activates ECs. ECs seeded onto laminin in CM/EC+CSC (growthfactor- free NBM) or mixed directly with CSCs-(08387) (growth-factor-free NBM) for 3 h (37oC, 5% CO2), EC isolation by sorting for CD31 and qRT-PCR performed. mRNA levels of genes marking EC activation after normalization to GAPDH. Statistics: two-sided exact Wilcoxon rank-sum test; data plotted as mean+SEM.

Addition of recombinant L1CAM to ECs seeded alone activated integrin αvβ3, as assessed by staining for phospho-β3 (pY759) (84-85), confirming that integrin αvβ3 can be activated upon binding L1CAM (Fig. 8). The growth factor bFGF is upregulated in GBM; thus, to determine whether integrin αvβ3 on ECs is activated on binding L1CAM on CSCs, we co- seeded ECs and CSCs on laminin in the presence of bFGF (3 h) using Sox2 as a positive control for CSCs (131). A significant increase in integrin β3 phosphorylation was observed when ECs were co-seeded with CSCs (Fig. 8).

Figure 8. The direct interaction of ECs and CSCs promotes activation of integrin αvβ3. ECs in CM/EC+CSC (growth factor-free NBM + 10 ng/ml bFGF) or mixed directly with CSCs (growth factor-free NBM + bFGF) were seeded onto laminin (3 h). Immunofluorescence for phospho-β3 (Y759), marker of integrin αvβ3 activation; white arrows denote phospho-β3 and yellow arrows denote CSCs. Statistics: two-sided exact Wilcoxon rank-sum test; data plotted as mean+SEM.

Co-seeding of ECs with CSCs resulted in a phenotypic change in ECs (Fig. 9A). The number of ECs with stress fibers was significantly lower and the number with cortical actin was significantly higher as compared to ECs seeded in CM/EC+CSC (Fig. 9B). This suggested induction of a pro-migratory phenotype in ECs on co-seeding with CSCs.

Figure 9. The direct interaction of ECs and CSCs promotes phenotypic changes in ECs. A-B. ECs in CM/EC+CSC (growth factor-free NBM + 10 ng/ml bFGF) or mixed directly with CSCs (growth factor-free NBM + bFGF) were seeded onto laminin (3 h). Immunofluorescence for phalloidin-Alexa-594 (binds actin) and vWf (Alexa-647- yellow); arrows denote ECs with cortical actin or vWf expression.. Statistics: two- sided exact Wilcoxon rank-sum test; data plotted as mean+SEM.

Figure 10. The direct interaction of ECs and CSCs promotes activation of p130CAS in ECs. A-B. ECs in CM/EC+CSC (growth factor-free NBM + 10 ng/ml bFGF) or mixed directly with CSCs (growth factor-free NBM + bFGF) were seeded onto laminin (3 h). Immunofluorescence for phosphorylated p130CAS (phos-p130CAS Y234; Alexa-488, green) and vWf (magenta); the intensity of green signal calculated using FIJI (ImageJ) based on the sum of the mean gray level (intensity) output of all pixels of the area within the enclosed cell traced along its outer edge, for each confocal section taken at 0.5 μm steps comprising the Z-stack. Signal intensity was normalized by dividing the sum of the mean gray level by the area of the cell. White arrows, green phos-p130CAS; and yellow arrows, CSCs. Statistics: two-sided exact Wilcoxon rank-sum test; data plotted as mean+SEM.

As the phosphorylation of early downstream signaling effectors FAK and p130CAS is dynamic and depends on an organized actin cytoskeleton [reviewed in (35,79,132)], we quantitated the intensity of p130CAS phosphorylation (pY234) in confocal stacks of ECs. In ECs

co-seeded with CSCs, phospho-p130CAS-(pY234) was nearly 3-fold higher than in ECs seeded in CM/EC+CSC (Fig. 10A&B); however, there was no significant difference in total p130CAS in ECs co-seeded with CSCs versus seeded alone in CM/EC+CSC, based on cell harvest with

Accutase, sorting for CD31, cell lysis and blotting for p130CAS (Fig. 10B). FAK activity

(pY397) was similarly increased in ECs co-seeded with CSCs, as compared to ECs seeded in

CM/EC+CSC (data not shown). Double-labeling for phosphorylated p130CAS and actin

(Alexa-594-Phalloidin) followed by confocal microscopy showed focal co-localization at the membrane in ECs co-seeded with CSCs (Fig. 11)

Figure 11. p130CAS co-localizes with the EC membrane in ECs co-cultured with CSCs. ECs seeded alone or with CSCs. Immunofluorescence for phosphorylated p130CAS (Y234) (Alexa-488, green) and actin (Alexa-594-Phalloidin, red), followed by confocal microscopy. Sections comprising the Z-stack shown;white arrowheads-actin stress fibers; white arrows-co-localization (yellow) of phos-p130CAS and actin at membrane; and yellow arrows-CSCs.

The downstream effectors ERK and JNK can also promote cell migration (133). The activation of ERK and JNK was significantly higher in ECs co-seeded with CSCs from four different GBM tumors as compared to ECs seeded in CM/EC+CSC (Fig 12A&B), and this increase was blocked by addition of RGD peptide (Fig. 12C; Fig. 29A). Collectively, these results indicate that the direct interaction of CSCs with ECs promotes angiogenesis as indicated by network formation and a transition in ECs from a quiescent phenotype to an activated, migratory phenotype.

Figure 12. The direct interaction of ECs and CSCs activates ERK and JNK in ECs in an RGD-peptide-dependent manner. A-B. ECs seeded alone or with CSCs. Immunofluorescence for phospho-ERK or phospho-JNK (Alexa-594, red) and vWf (Alexa-647, magenta). White arrows denote vWf-positive ECs expressing pERK or pJNK; and yellow arrows denote GFP-CSCs-(08387). Immunofluorescence in ECs co- seeded with a second CSC isolate (4302) or in CM/EC+CSC; fluorescent intensity analyzed in 50 ECs and graphed as the fold-change. C. RGD- or RAD-peptide (50 μM) incubated with cells during seeding. Statistics: two-sided exact Wilcoxon rank-sum tests or two-sided Wilcoxon rank-sum test, data plotted as mean+SEM.

CSCs increase the motility of ECs through direct contact as well as secreted factors

We investigated the effect of CSCs on EC migration in a 2D cell motility assay in which

ECs and CSCs were seeded on laminin on either side of a 500-μm gap in NBM, and migration assessed over 24 h by live-imaging. EC migration was nearly completely inhibited by a neutralizing anti-integrin α6 antibody (Fig. 13A). As compared to CSCs or ECs on both sides of the gap, CSCs opposite ECs (on either side of the gap) significantly increased EC and CSC migration into the gap by 10- and 14-fold, respectively (Fig. 13B; Fig. 29B). The cells filled the gap over 24 h, whereas only 35% of the gap was filled when either CSCs or ECs were plated on both sides.

Figure 13. CSCs increase the directional 2D motility of ECs through secreted factors. EC migration is mediated by an α6 integrin. A-B. Cell culture inserts were coated with 20μg/mL laminin, and red-fluorescent-ECs (30,000, EGM media) were platedon one side of the insert, and GFP-CSCs-(08387) (30,000, complete NBM) on the other side. At 18 h, all media were replaced with complete NBM; the insert removed, and live-imaging performed q 15 min (24 h, 37oC, 5% CO2). Migration was measured by counting each fluorophore in an automated fashion (customized software). In certain experiments, blocking anti-integrin α6 antibody or IgG (5 μg/ml) was included in the complete NBM and nearly completed inhibited EC migration towards CSCs. Gap closure plotted. Statistics: two-sided exact Wilcoxon rank-sum tests.

Live video imaging showed that contact between ECs and CSCs first occurred after 7 h in this assay (Fig. 14A); thus, the effects of potential CSC-secreted factors in the absence of direct contact could be assessed during the first 7 h. On tracing the paths of 30 ECs, we found that in the absence of CSCs (-CSCs), ECs migrated randomly whereas when CSCs were present the majority of ECs displayed a preferential migration in the direction of the CSCs (negative X axis)

(Fig. 14B; Fig. 29C). During the first 7 h, there was no significant difference in the velocity of

ECs when plated opposite CSCs or ECs (Fig. 14C). Thus, potentially, CSC secreted factors promote the directional motility of ECs but do not alter the velocity.

Figure 14. CSCs increase the directional 2D motility of ECs through secreted factors. A. Examples of ECs contacting CSCs in the gap area at various time points after 7 h; white arrows-ECs, yellow arrows-CSCs, and black arrowhead-point of EC-CSC contact. B. Paths of 30 ECs were manually traced in the first 7 h before EC-CSC contact using Manual Tracker (Image J). Final destination of each EC denoted on the graph, mean overall displacement-Euclidean distance (all 30 ECs) shown with the arrow. C. Mean EC velocity (displacement/time-μm/min) graphed as a bar graph of the mean+SEM. Statistics: two-sided exact Wilcoxon rank-sum tests.

We then assessed the effects of direct contact of CSCs with ECs on EC migration by tracing trajectories during the 7-24 h time frame. We compared the trajectories of 30 ECs that had been observed to make contact with CSCs during the 7-24 h time frame with the trajectories of 30 ECs that never contacted CSCs but were plated opposite of CSCs, over the same time frame. We found that ECs which had been observed to contact CSCs exhibited a significantly greater displacement and velocity (21.3 μm and 0.298 μm/min, respectively), than the ECs that had never contacted CSCs (3.4 μm and 0.23 μm/min, respectively) (Fig. 15A&B; Fig. 29D). ECs that were seeded in the absence of CSCs continued to display random migration (Fig. 29E).

Figure 15. CSCs increase the motility of ECs through direct contact. A. Paths of 30 ECs were manually traced between 7-24 h when EC-CSC contact occurs using Manual Tracker (Image J). Final destination of each EC denoted on the graph, mean overall displacement-Euclidean distance (all 30 ECs) shown with the arrow. B. Mean EC velocity (displacement/time-μm/min) graphed as Box and Whisker plots of the mean+SEM. Statistics: two-sided exact Wilcoxon rank-sum test.

To assess the effect of EC-CSC contact on EC migration towards a chemo-attractant, we used a transwell assay with laminin-coated 3-μm pore filters to mimic the tight spaces in the brain extracellular matrix (134) (Fig. 16A). Red-fluorescent-ECs were mixed, co-seeded with

GFP-CSCs in the upper chamber and allowed to migrate for 6 h towards bFGF in NBM in the bottom chamber. Co-seeding of ECs with CSCs in the top chamber significantly increased EC migration as compared to migration of ECs seeded alone in the top chamber (Fig. 16B&C). The increase in EC migration when co-seeded with CSCs was not replicated by seeding ECs in

CM/EC+CSC, indicating the increase in EC migration was not due to a secreted factor(s) (Fig.

16D).

Src phosphorylates the substrate domain of p130CAS, which is necessary for p130CAS activation [reviewed in (35,79,132)]. Preincubation of ECs with the Src inhibitor 4-amino-5-(4- chlorophenyl)-7-(dimethylethyl)pyrazolo[3,4-α]pyrimidine (PP2) for 20 min followed by washing and seeding significantly inhibited EC migration towards bFGF whether the treated ECs were co-seeded with CSCs or seeded alone (Fig. 16B&C).

A B

C D

Figure 16. The direct interaction of ECs and CSCs promotes EC migration towards bFGF. A-D . Filters (3-μm pore) were coated on both sides with 20 μg/mL laminin. Redfluorescent- ECs were seeded alone or mixed with GFP-CSCs-(08387) and co-seeded in growth factor-free NBM with 1% BSA on top of the filter. Growth factor-free NBM with 10 ng/ml bFGF was placed in the bottom chamber, and the cells allowed to migrate (37°C, 5% CO2). At 6 h, cells were removed from the upper filter surface and cells on the lower filter surface were washed, fixed, photographed and counted. B-C. The Src inhibitor PP2 (1 μM) was incubated with ECs (20 min), followed by washing, and EC seeding alone or co-seeding with CSCs. Representative

52 photographs of the bottom of filter shown. D. ECs were seeded in CM/EC+CSC or co- seeded with CSCs. Statistics: two-sided exact Wilcoxon rank-sum tests, and data graphed as the mean+SEM.

Downregulation of β3 in ECs resulted in a significant decrease in migration of ECs co- seeded with CSCs, but had no significant effect on the migration of ECs that were seeded alone

(Fig. 17A). This was consistent with our observation that integrin αvβ3 is not mediating adhesion or migration of ECs on laminin. Downregulation of L1CAM on CSCs also resulted in a significant reduction in migration of ECs co-seeded with CSCs (Fig. 17B). Collectively, these data suggest that the interaction of CSCs with ECs activates signaling effectors in ECs thereby promoting migration towards bFGF that is not mediated by binding of integrin αvβ3 to the laminin substrate.

Figure 17. EC migration on laminin is not mediated by integrin αvβ3; the direct interaction of ECs and CSCs promotes EC migration towards bFGF. A-B. ECs were treated with integrin β3 or L1CAM specific siRNA or with control siRNA for 48 hours washed and seeded alone or with GFP-CSCs. Filters (3-μm pore) were coated on both sides with 20 μg/mL laminin. Red fluorescent- ECs were seeded alone or mixed with GFP-CSCs-(08387) and co-seeded in growth factor-free NBM with 1% BSA on top of the filter. Growth factor-free NBM with 10 ng/ml bFGF was placed in the bottom chamber, and the cells allowed to migrate (37°C, 5% CO2). At 6 h, cells were removed from the upper filter surface and cells on the lower filter surface were washed, fixed, photographed and counted. Data graphed as the mean+SEM.

Bone marrow tyrosine kinase on chromosome X (BMX) as well as FAK and p130CAS are required for enhanced migration of ECs towards bFGF when co-seeded with CSCs

To determine whether p130CAS was necessary for the enhanced migration of ECs when co-seeded with CSCs, we downregulated p130CAS in ECs (Fig 18A). This resulted in a significant inhibition of EC migration towards bFGF when co-seeded with CSCs or when seeded in CM/EC+CSC (Fig. 18C). Both FAK and BMX can phosphorylate p130CAS, although at different sites, and contribute to p130CAS activation (79-80). When FAK was downregulated in

ECs, we found a significant inhibition of EC migration whether ECs were co-seeded with CSCs or seeded in CM/EC+CSC (Fig. 18B&D). In contrast, when BMX was downregulated in ECs, we found a significant inhibition of EC migration when ECs were co-seeded with CSCs, but not when they were seeded alone in CM/EC+CSC (Fig. 18C&D). This indicates a differential requirement for BMX for the increased chemotactic migration of ECs when they are co-seeded with CSCs.

A B

Figure 18. BMX, FAK, and p130CAS are required for enhanced EC migration. A- C. ECs were treated with the indicated siRNA or control siRNA (48 h), detergent lysed and immunoblotted for the indicated antibodies. C-D. Filters (3-μm pore) were coated on both sides with 20 μg/mL laminin. Red fluorescent- ECs were seeded alone or mixed with GFP-CSCs-(08387) (3x104) and co-seeded in growth factor-free NBM with 1% BSA on top of the filter. Growth factor-free NBM with 10 ng/ml bFGF was placed in the bottom chamber, and the cells allowed to migrate (37°C, 5% CO2). At 6 h, cells were removed from the upper filter surface and cells on the lower filter surface were washed, fixed, photographed and counted. Data graphed as the mean+SEM.

We therefore determined whether BMX was necessary for p130CAS activation in ECs co-seeded with CSCs, by downregulating BMX in ECs that were then co-seeded with CSCs on laminin in NBM with bFGF (3 h). The downregulation of BMX significantly inhibited p130CAS phosphorylation in ECs co-seeded with CSCs but had no effect on p130CAS phosphorylation in

ECs seeded alone in CM/EC+CSC (Fig. 19A&B).

Figure 19. BMX is necessary for p130CAS activation in ECs contacting CSCs. A-B. ECs were seeded alone or with CSCs-(08387). ECs were treated with BMX siRNA or control siRNA (48 h), followed by immunofluorescence for phos-p130CAS (Y234) (Alexa-488, green) and vWf (Alexa-647, magenta, marker of ECs). Confocal sections (0.5-μm apart) comprising the Z-stack shown. White arrows denote phos-p130CAS and yellow arrows denote CSCs (vWf-negative). The intensity for phos-p130CAS was determined in 15 representative cells and graphed as the fold-change. Statistics: two- sided exact Wilcoxon rank-sum tests.

To determine whether BMX was activated in ECs on co-seeding with CSCs, phospho-

BMX was detected by immunoflorescence and fluorescent intensity quantitated as described for phospho-p130CAS. We found a significant increase (~3-fold) in phospho-BMX in ECs co- seeded with CSCs as compared to ECs seeded alone in CM/EC+CSC (Fig. 20A) although there was no significant difference in total BMX in ECs in the two conditions after Accutase harvest,

CD31 sorting, lysis and blotting for BMX (Fig. 20B).

Figure 20. Direct contact with CSCs increases phospho-BMX in ECs. A. ECs were seeded alone or with CSCs-(08387). ECs were treated with control siRNA (48 h), followed by seeding as above and immunofluorescence for phospho-BMX (pBMX) (Alexa-488 green) and vWf (Alexa-647, magenta, EC marker). Confocal sections (0.5- μm apart) comprising the Z-stack shown. White arrows denote pBMX, yellow arrows denote CSCs (vWf-negative) and red asterisks denote ECs in phase contrast images. B. Cells were harvested with Accutase, sorted for CD31, detergent lysed and immunoblotted for the indicated antibodies.

To further determine whether the increase in phospho-BMX was dependent on integrin

αvβ3 in ECs co-seeded with CSCs, we downregulated the β3 subunit in ECs. This resulted in a significant inhibition of BMX phosphorylation in the ECs with the downregulated β3 subunit when co-seeded with CSCs, but no effect on BMX phosphorylation in those seeded alone in

CM/EC+CSC (Fig. 20A; 21A&B). Taken together, these data indicate that BMX is necessary for 59

the increased p130CAS activation and EC migration towards bFGF in ECs co-seeded with CSCs on laminin, but is not necessary for p130CAS activation and EC migration when ECs are seeded alone.

Figure 21. Integrin αvβ3 is necessary for BMX activation. A-B. ECs were seeded alone or with CSCs-(08387). ECs were treated with integrin β3 siRNA (48 h), followed by seeding as above and immunofluorescence for phospho-BMX (pBMX) (Alexa-488 green) and vWf (Alexa-647, magenta, EC marker). Confocal sections (0.5-μm apart) comprising the Z-stack shown. White arrows denote pBMX, yellow arrows denote CSCs (vWf-negative). The intensity for pBMX was determined in 15 representative cells and graphed as the fold-change. Statistics: two-sided exact Wilcoxon rank-sum test.

RGD-peptide treatment reduces the proximity of Sox2-positive tumor cells to ECs and decreases phosphorylation of BMX and p130CAS in ECs in a xenograft model of GBM

Figure 22. Xenograft orthotopic model of GBM: Experimental Schematic. Two groups of nude mice (n=5) were injected with LN-308 cells (100,000), and after tumor establishment (55 days), administration of cyclic-RGD peptide (90mg/kg) or PBS was initiated for 5 days, followed by euthanasia at 2 hours post-treatment. The brains were harvested, fixed, frozen, sectioned, and stained.

Others have reported that cyclic-RGD-peptide (Cilengitide) treatment prolongs survival of scid mice bearing intracerebral GBM tumors, and of nude rats bearing intracerebral GBM tumors when administered with radiation (135-136). As we had found that an RGD-peptide blocks the interaction of integrin αvβ3 on ECs with L1CAM on CSCs, we examined the effects

of administration of the cyclic-RGD-peptide in anestablished orthotopic mouse model of GBM on the distance of Sox2-positive tumor cells from ECs and integrin αvβ3-mediated signaling in the ECs (Fig. 22).

LN-308 GBM cells were utilized as they express L1CAM when propagated as neurospheres in NBM, bind to ECs in the cell-cell adhesion assay in an L1CAM-dependent manner, and promote EC migration towards bFGF in an L1CAM-dependent manner (Fig. 23A-

D).

A B

C D

Figure 23. LN-308 GBM cells express L1CAM, bind to ECs, and promote EC migration in an L1CAM-dependent manner. A. LN-308 human GBM cells propagated in complete NBM in suspension were fluorescent-labeled for L1CAM (Alexa-594-red) ; B. allowed to adhere to a monolayer of ECs after downregulation of L1CAM; C. mixed with ECs, plated on laminin-coated filters and allowed to migrate to bFGF, or D. L1CAM was downregulated and the migration assay repeated. Statistics: two-sidedexact Wilcoxon rank- sum tests. Data graphed as the mean+SEM.

LN-308 cells were injected into the nude mouse brain and at day 55 administration of cyclic RGD-peptide was initiated, followed by euthanasia and brain harvest on day 60. We found a significant increase in the mean distance of Sox2-positive cells from ECs (Fig. 24A; 25A), and a significant decrease in the number of Sox2-positive cells within 25-μm of blood vessels (Fig.

24B; 25A)

A B

Figure 24. Administration of the RGD-peptide significantly reduces the proximity of Sox2-positive tumor cells to ECs in an intracerebral xenograft model of GBM. A-B. LN-308 cells (100,000) were injected into the nude mouse brain, tumor allowed to establish for 55 days, and mice were then administered cyclic- RGD-peptide (90 mg/kg) or PBS for 5 days (n=5 mice/group), followed by euthanasia at 2 h post- treatment and brain harvest. Sections were double-labeled for vWf (Alexa-594, red) and Sox2 followed by DAPI nuclear stain. The mean Sox2-positive tumor cell distance from vWf-positive ECs, and percent Sox2-positive tumor cells within 25-μm of a vWf- positive vessel from RGD-peptide-treated mouse tumors relative to vehicle-treated control tumors were determined. Statistics: two-sided exact Wilcoxon rank-sum tests. Data graphed as the mean+SEM.

We also found significant decreases in the percent of ECs expressing pβ3-(Y759), (Fig.

25A&B) in the intensity of phospho-BMX (BMX activation) (Fig 25. A&C), in phospho- p130CAS intensity in ECs (Fig. 25A&D) , and in vessel surface area in the tumors of RGD- peptide treated mice as compared to controls (Fig. 25A&E). Phospho-BMX staining of ECs in the xenograft tumors was detected in a population of vessels, consistent with BMX expression in vessels of arterial origin (137-138). These data suggest that ECs interact with tumor cells in the perivascular niche through an RGD-peptide-binding integrin and that this interaction promotes

BMX and p130CAS activation, thereby enhancing angiogenesis (Fig 26).

B C D E

Figure 25. Administration of the RGD-peptide significantly reduces the proximity of Sox2-positive tumor cells to ECs, and BMX and p130CAS activation in ECs in an intracerebral xenograft model of GBM. A-E. LN-308 cells (100,000) were injected into the nude mouse brain, tumor allowed to establish for 55 days, and mice were then administered cyclic- RGD-peptide (90 mg/kg) or PBS for 5 days (n=5 mice/group), followed by euthanasia at 2 h post-treatment and brain harvest. Sections were double-labeled for vWf (Alexa-594, red) and either Sox2, pβ3-(Y759), phospho- BMX, or phospho-p130CAS (Y234) (Alexa-488, green), followed by DAPI nuclear stain. The percent ECs positive for pβ3 (B), intensity of phospho-BMX in tumor- associated ECs (C), intensity of phospho-p130CAS in tumor-associated ECs (D), and vessel surface area in the tumor from RGD-peptide-treated mouse tumors relative to vehicle-treated control tumors (E) were determined. The intensity for phospho-BMX and phospho-p130CAS was calculated using FIJI (ImageJ) for vWf-positive blood vessels, and traced along its outer edge for each field taken at 40X magnification. Signal intensity was normalized by dividing the sum of the mean gray level by the area of the ECs in each blood vessel. Statistics: linear mixed regression model. Data graphed as the mean+SEM.

Figure 26. Model of proposed integrin αvβ3-mediated pro-migratory signaling on ECs when binding L1CAM on CSCs. ECs interact with tumor cells in the perivascular niche through an RGD-peptide-binding integrin αvβ3 with L1CAM. This interaction promotes FAK, Src, and BMX activation with subsequent p130CAS activation. BMX phosphorylates p130CAS after activation by FAK through phosphorylation of the Y40 residue in the BMX pleckstrin homology domain. Activation of BMX is required for EC migration stimulated by direct binding to CSCs, but not for EC migration stimulated by soluble factors. BMX and p130CAS activation following integrin-binding promotes membrane ruffles and enhances the formation of lamellipodia with cortical actin and cell motility. The promotion of paxillin phosphorylation by JNK can also lead to cytoskeletal changes, and the phosphorylation of myosin light chain kinase by ERK promotes EC migration. EC=brain endothelial cell. CSC= glioma cancer stem-like cell.

CHAPTER 4

DISCUSSION

Discussion and Conclusions

We describe for the first time a direct interaction of ECs with CSCs from GBM that is mediated through binding of integrin αvβ3 on ECs to the RGD-peptide in the extracellular domain of L1CAM on CSCs. This activates integrin αvβ3 on ECs and, in the presence of bFGF, results in significantly increased activation of BMX, FAK and p130CAS, as well as increased activation of the downstream effectors ERK and JNK and marked changes in the phenotype of the ECs. Importantly, this level of upstream and downstream effector activation from integrin

αvβ3 was not achieved by EC seeding in CM/EC+CSC with bFGF, indicating that it cannot be attributed to a secreted factor(s). Furthermore, the increase in ERK and JNK activation was

RGD-peptide-dependent suggesting a requirement for the direct interaction of ECs with CSCs rather than an effect due to integrin α6-mediated adhesion to laminin (which is RGD-peptide- independent) (reviewed in [35, 132]). In ECs co-seeded with CSCs, we found that integrin αvβ3

is necessary for BMX activation and BMX is necessary for p130CAS activation, whereas in ECs seeded in CM/EC+CSC, αvβ3 is not necessary for BMX activation and BMX is not necessary for p130CAS activation. BMX, a cytoplasmic non-receptor tyrosine kinase, can activate signaling pathways that promote migration and angiogenesis as well as other signaling pathways

(81, 137-138); thus, the increased BMX activity downstream of activated integrin αvβ3 in the

ECs co-seeded with CSCs could serve to amplify and diversify signaling.

The alterations in the actin cytoskeleton that occurred in ECs that contacted CSCs suggest a pro-migratory phenotype, and co-seeding ECs with CSCs on laminin significantly increased EC migration towards bFGF as compared to that observed when the ECs were seeded on laminin in CM/EC+CSC. An increase in EC velocity was also observed after contact with

CSCs in a monolayer 2D assay. This increased migration was dependent on integrin αvβ3 on

ECs and L1CAM on CSCs, supporting our hypothesis that the direct cell-cell contact promotes

EC migration towards bFGF. The possibility that the migration was mediated by integrin αvβ3 binding to the laminin substrate was ruled out by the failure of downregulation of β3 to significantly affect migration of the ECs seeded alone on laminin. Our studies did not determine the contribution of L1CAM shed from the cell surface of CSCs after three hours that could be deposited in the ECM or be a soluble component in CM and its effect on potentiating EC migration by αvβ3 ligation. These data still suggest, however, that interactions between L1CAM and integrin αvβ3 might potentially stimulate EC migration by inside-out signaling to increase the affinity of a non-RGD binding integrin, such as integrin α6, for the ECM, in our experiments a laminin substrate. Cross-talk between other integrins has been shown; ligand binding to integrin α5β1 stimulates α2β1-mediated adhesion to collagen I, and integrin αvβ3 regulates

α5β1-mediated migration towards fibronectin (139). Ligand binding to integrin α5β1 is

necessary for integrin αvβ3 internalization of vitronectin (140). We found that EC-CSC contact activates ECs based on increases in E-selectin and VCAM-1 mRNA; EC activation post-contact could potentially increase specific integrin ligand-binding affinities, although we have no evidence to support this.

We decided to focus on the signal transduction cascade directly modulated by integrin

αvβ3 ligation on ECs in the presence of CSCs. While some of the signals associated with migration of ECs induced by the direct interaction with CSCs and those induced by soluble factors in CM/EC+CSC are identical, there also were distinct differences including the requirement for BMX for cell-contact induced signaling of EC migration. Prior studies showing that integrin αvβ3 cooperation with bFGF promotes angiogenesis (132, 141-142) utilized models that do not take into account the effects of direct contact with perivascular CSCs or tumor cells on integrin αvβ3 signaling in the ECs. The current studies provide insights and also raise questions that are relevant to the ongoing development of therapeutic targeting. In terms of regulation of signaling, for example, integrin αvβ3 cooperation with bFGF in confluent monolayers of ECs has been shown to be due to a complex formed between integrin αvβ3, the tetraspanin CD9, and the junctional adhesion molecule-A (JAMA) that upon bFGF stimulation releases JAMA thereby regulating ERK activation and migration (143) but it is not known whether a similar complex is formed in the context of EC-CSC contact. Molecular complexes at the cell surface can integrate and transmit extracellular signals, which likely vary depending on the complex constituents. It was shown that CD9 can cooperate with integrin α6β1 to enhance neurite outgrowth of neurons on laminin (144). Interestingly, CD9 can also interact with L1CAM

(100), which may depend on L1CAM conformation. CD9 on the surface of neuronal cell bodies promotes Ca2+ entry into neural cells, which is a key step in integrin-dependent neurite outgrowth

(145). It was found that increasing levels of Ca2+ inhibited α5β1 and αvβ1 binding to an L1CAM peptide in the presence of Mn2+ in CHO cells (100). We did not include Ca2+ in our experiments, but there was no decrease in CSCs binding to ECs when inhibiting the α5β1 integrin with a neutralizing mAb in the presence of Mg2+ and Mn2+ , suggesting integrin α5β1 was not actively mediating binding. CD24, also known as signal transducer CD24, is a cell surface sialoglycoprotein, expressed on lymphocytes, neuroblasts, and granulocytes, as well as neural stem cells (146-147). The association of CD9 or CD24 with L1CAM in the plane of the

CSC membrane may also orchestrate the association of CSC- L1CAM with EC-integrin αvβ3

(100), but it is not known whether such a complex is formed in the context of EC-CSC contact.

The current studies focus on the effects on EC migration mediated by integrin αvβ3 after

EC-CSC contact; it is not known if this contact also promotes survival and, if so, whether it does so through a mechanism similar to that observed on binding to an extracellular substrate (148-

149). Others have reported that, during metastasis, melanoma cells undergo transendothelial cell migration through a mechanism involving an interaction between integrin αvβ3 on melanoma cells and L1CAM on ECs (150). It should be noted, however, that GBM tumors rarely metastasize outside of the brain (12).

We used primary normal brain microvascular ECs that did not express detectable levels of L1CAM; however, it is known that tumor-associated ECs harbor different attributes than normal ECs, and it has been demonstrated that the level of L1CAM is increased in the vasculature from other cancers as compared to normal tissue, notably upregulated by factors such as VEGF-A or IFN-γ (151-152). Expression of L1CAM in mouse lung ECs was shown to increase EC proliferation, migration, and tubulogenesis (151). While it remains a possibility that

L1CAM-expressing ECs coordinate EC behavior and vascular formation in tumors, our findings

have shown that exogenous L1CAM on the surface of CSCs also plays a major role in modulating EC migration, a critical process that guides angiogenesis.

Our in vivo animal studies indicate that treatment of an established orthotopic GBM xenograft tumor with cyclic-RGD-peptide significantly increased the mean distance of Sox2- positive tumor cells from ECs as compared to tumors treated with vehicle. We also observed significant decreases in the percent of ECs with integrin αvβ3 activation, in BMX activation and p130CAS phosphorylation in ECs, and in vessel surface area in tumors from RGD-peptide- treated mice. Clinical trial of the cyclic-RGD-peptide (Cilengitide) in combination with standard chemo-radiation for patients with newly diagnosed GBM showed improvement of survival in a subset of patients (153). Thus, our current data indicate previously unrecognized mechanisms of action that may affect therapeutic responsiveness to the RGD-peptide.

We showed that soluble recombinant L1CAM containing the RGD-peptide activated integrin αvβ3 on ECs. The possibility that soluble L1CAM shed from tumor cells can bind integrin αvβ3 on ECs and thereby interfere with the therapeutic effects of Cilengitide in cancer patients has not been explored. It is feasible that soluble L1CAM could decrease Cilengitide binding to integrin αvβ3, as both could compete for binding to the RGD recognition site located in integrin αvβ3. The therapeutic effects of Cilengtide in patients could be altered by the concentrations of soluble L1CAM and Cilengitide in circulation, the comparative binding affinity of both molecules to integrin αvβ3, as well as the degree to which divalent cations allosterically regulate L1CAM or Cilengitide binding to αvβ3.

In summary, we have identified a previously unrecognized integrin αvβ3/bFGF initiated signaling event in the perivascular niche in GBM, involving EC contact with CSCs or tumor

cells that requires BMX for activation of p130CAS and promotes chemotactic migration of ECs and thereby angiogenesis.

Future Directions and Perspectives

The studies detailed here suggest multiple avenues of new research regarding the role of cell-cell communication in the perivascular tumor microenvironment and have implications for the process of tumor vascularization in GBM and other cancers. In the most general sense, additional mechanisms of direct contact between ECs and CSCs have the potential to be elucidated, along with the ensuing downstream signaling cascades. In addition, all studies herein were conducted in conditions of normoxia, and thus studies into the role of hypoxia in this reciprocal crosstalk is worth initiating, as hypoxia is a major occurrence in GBM growth. For example, HUVECs that were activated with either TNF-α or exposed to hypoxia demonstrated an increase in CD24 expression (154), which, as noted above, is a ligand for L1CAM. Also, there is a likelihood of bi-directional signaling that occurs following EC-CSC contact. Our focus was on the impact of this contact on angiogenesis, and so we did not seek to determine the downstream pathways activated in CSCs following EC contact, which would certainly be of interest to those in stem cell-focused research. Specifically, one could focus on the downstream effectors that are involved in ligated-L1CAM signaling and the outcome of this signaling on gene expression of other proteins or the effect on CSC survival , migration, or proliferation in comparison to the effect of secreted factors on CSCs.

As noted, interactions between L1CAM on CSCs and integrin αvβ3 on ECs may potentially promote EC motility by inside-out signaling to increase the affinity of a non-RGD binding integrin, such as integrin α6β1, for the ECM. It was demonstrated that there is crosstalk between integrin αvβ3 and α6β1 during adhesion to the α4 laminin subunit, where a blocking antibody towards integrin αvβ3 abrogated α6β1 binding to α4 laminin (155). The α4 laminin subunit was found to have low levels of expression in mouse normal brain tissue as analyzed by northern blot (156). Stronger expression of α4 laminin correlated with increasing grades of glioma, with the highest levels in GBM. This was also true of the laminin β1 chain, which are the subunits comprising the laminin-8 isoform; notably integrin α6β1 is the receptor for laminin-8

(157). It could be of significance to ascertain the role of integrin αvβ3 and the molecular basis for any potential cross-regulation between integrin αvβ3 and a non-RGD-binding integrin, such as integrin α6β1, in mediating migration.

Another significant method of cell-cell crosstalk occurs via exosomes or microvesicles.

Research into the extent to which ECs/CSCs release microvesicles and the extent to which microvesicles activate EC/CSCs proteins or induce expression of cytokines and growth factors is of importance. Proteins or mRNA exogenously delivered from CSCs to ECs via exosomes could also induce epigenetic changes in ECs. In addition to L1CAM cell-surface ectodomain shedding by ADAM10 cleavage, L1CAM was shown to be present in membrane vesicles induced by the inhibitor of cholesterol synthesis methyl-β-cyclodextrin (MCD), and these L1CAM-positive vesicles prompted chinese hamster ovary cell haptotactic migration (158). The degree to which

CSCs release vesicles compared to non-stem tumor cells and any subsequent altercations to the tumor microenvironment, including possible pro-migratory effects on ECs, is a prospective study that would potentially provide further insight into the regulation of angiogenesis.

The relatively small beneficial effect of radiation therapy on patient with GBM is thought to be, in part, due to the radioresistance of CSCs (159). It is possible that the diversified components of the perivascular niche, including ECs, could modulate the CSC response to radiation; furthermore, it is possible that there may also be bi-directional regulation affecting the

EC response to radiation. It was observed that there was less apoptosis of ECs in co-culture with glioma cells than in monoculture after radiation treatment (160). This prompts the question of whether integrin αvβ3 activation following EC-CSC direct contact can control EC response to radiation, as integrin αvβ3 can signal for survival. In addition, upon EC contact with CSCs (ECs co-cultured with CSCs for 3 hours) we observed an increase in phosphorylated FAK compared to ECs cultured in CM/EC+CSC, and FAK can also signal for survival. An intriguing recent

Nature study found that ECs provide supportive local signals to surrounding tumor cells upon

DNA damage, but this behavior is ablated when there is deletion of FAK in ECs (161). This deletion of FAK in ECs was sufficient to raise tumor responsiveness to DNA-damaging treatments. Mechanistically, it was determined that deletion of FAK in ECs repressed NF-κB activation and the successive induction of cytokines during DNA-damaging chemotherapy (161).

It remains possible that CSCs which are only in direct contact with ECs in the perivascular niche enhance FAK activation in ECs, as we found, thereby enhancing EC cytokine production and resulting in a reduction in CSC sensitivity to radio- or chemotherapy and a promotion in CSC survival. This possibility warrants investigation. Therefore, disruption of the contact between

CSCs and ECs in the perivascular niche could in some manner synergize with current chemotherapies.

A new idea is emerging that suggests the promotion of a mature vasculature, rather than the eradication of the vasculature, could promote the delivery and efficacy of chemotherapy in

tumors. Anti-VEGF humanized monoclonal antibodies or VEGFR inhibitors currently prune the vasculature towards normalization, causing the restoration of vessel structure and oxygenation; however, anti-VEGF monoclonal antibody therapy is not optimized to sustain the integrity of tumor vessel structure and function, and the lack of stable normalization leads to further pruning, hypoxia, and tumor regrowth (162). Although our data suggests that CSCs promote angiogenesis, CSCs are also involved in tumor recurrence and resistance (162-163). It may be of interest to discover ways to normalize the vasculature while simultaneously reducing the bi- directional signaling that occurs between ECs and CSCs in the perivascular niche. A low concentration of Cilengitide, in combination with the Ca2+-channel blocker verapamil, was shown to increase vessel perfusion in tumors and the effectiveness of chemotherapeutic drugs

(164). If the administered dose of Cilengitide can be honed so that there is an increase in vessel perfusion, and a disruption in EC-CSC interaction, there may be better chemotherapeutic delivery, and also a decrease in cytokine production by ECs, which could further enhance tumor cell sensitivity to chemotherapeutics. Future studies can determine if promoting vessel normalization while targeting EC-CSC interaction in the GBM perivascular microenvironment is a relevant and successful therapeutic approach. It seems credible that the resolve to better understand the intricate nature of the brain tumor perivascular microenvironment and how it regulates angiogenesis will lead to the development of more effective therapeutic interventions for patients who suffer from GBM.

REFERENCES

1. Furnari FB, Fenton T, Bachoo RM, Mukasa A, Stommel JM, Stegh A, et al. Malignant

astrocytic glioma: genetics, biology, and paths to treatment. Genes Dev. 2007; 21: 2683-

2710.

2. Altieri R, Agnoletti A, Quattrucci F, Garbossa D, Calamo Specchia FM, Bozzaro M, et al.

Molecular biology of gliomas: present and future challenges. Transl Med UniSa. 2014;

10(7): 29-37.

3. Gladson CL, Prayson RA, Liu WM. The pathobiology of glioma tumors. Annu Rev Pathol.

2010; 5:33-50.

4. Ahluwalia MS, Gladson CL. Progress on antiangiogenic therapy for patients with malignant

glioma. J Oncol. 2010; 2010: 689018.

5. Sturm D, Bender S, Jones DTW, Lichter P, Grill J, Becher O, et al. Paediatric and adult

glioblastoma: multiform (epi)genomic culprits emerge. Nat Rev Cancer. 2014; 14(2): 92–

107.

6. Maher EA, Brennan C, Wen PY, Durso L, Ligon KL, Richardson A, et al. Marked genomic

differences characterize primary and secondary glioblastoma subtypes and identify two

distinct molecular and clinical secondary glioblastoma entities. Cancer Res. 2006;

66(23): 11502-13.

7. Verhaak RG, Hoadley KA, Purdom E, Wang V, Qi Y, Wilkerson MD, et al. Integrated

genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by

abnormalities in PDGFRA, IDH1, EGFR, and NF1. Cancer Cell. 2010; 17(1): 98.

8. Hegi ME, Diserens AC, Gorlia T, Hamou MF, de Tribolet N, Weller M, et al. MGMT gene

silencing and benefit from temozolomide in glioblastoma. N Engl J Med. 2005; 352(10):

997-1003.

9. Malmström A, Grønberg BH, Marosi C, Stupp R, Frappaz D, Schultz H, et al. Temozolomide

versus standard 6-week radiotherapy versus hypofractionated radiotherapy in patients

older than 60 years with glioblastoma: the Nordic randomized, phase 3 trial. Lancet

Oncol. 2012; 13: 916–926.

10. Snuderl M, Fazlollahi L, Le LP, Nitta M, Zhelyazkova BH, Davidson CJ, et al. Mosaic

amplification of multiple receptor tyrosine kinase genes in glioblastoma. Cancer Cell.

2011; 20: 810–817.

11. Ren ZP, Olofsson T, Qu M, Hesselager G, Soussi T, Kalimo H, et al. Molecular genetic

analysis of p53 intratumoral heterogeneity in human astrocytic brain tumors. J

Neuropathol Exp Neurol. 2007; 66(10): 944–954.

12. Onishi M, Ichikawa T, Kurozumi K, Date I. Angiogenesis and invasion in glioma. Brain

Tumor Pathol. 2011; 28(1): 13-24.

13. Hardee ME, Zagzag D. Mechanisms of glioma-associated neovascularization. Am J Pathol.

2012; 181(4): 1126-41.

14. Folkins C, Shaked Y, Man S, Tang T, Lee CR, Zhu Z, et al. Glioma tumor stem-like cells

promote tumor angiogenesis and vasculogenesis via vascular endothelial growth factor

and stromal-derived factor 1. Cancer Res. 2009; 69(18): 7243-51.

15. Bao S, Wu Q, Sathornsumetee S, Hao Y, Li Z, Hjelmeland AB, et al. Stem cell-like glioma

cells promote tumor angiogenesis through vascular endothelial growth factor. Cancer

Res. 2006; 66(16): 7843-8.

16. Presta M, Dell'Era P, Mitola S, Moroni E, Ronca R, Rusnati M. Fibroblast growth

factor/fibroblast growth factor receptor system in angiogenesis. Cytokine Growth Factor

Rev. 2005; 16(2): 159-78.

17. Takahashi T, Ueno H, Shibuya M. VEGF activates protein kinase C-dependent, but Ras-

independent Raf-MEK-MAP kinase pathway for DNA synthesis in primary endothelial

cells. Oncogene. 1999; 18(13): 2221-30.

18. Takahashi T, Yamaguchi S, Chida K, Shibuya M. A single autophosphorylation site on

KDR/Flk-1 is essential for VEGF-A-dependent activation of PLC-γ and DNA synthesis

in vascular endothelial cells. EMBO J. 2001; 20(11): 2768–2778.

19. Holash J, Maisonpierre PC, Compton D, Boland P, Alexander CR, Zagzag D, et al. Vessel

Cooption, Regression, and Growth in Tumors Mediated by Angiopoietins and VEGF.

Science. 1999; 284(5422): 1994-8.

20. Zadeh G, Koushan K, Pillo L, Shannon P, Guha A. Role of Ang1 and Its Interaction with

VEGF-A in Astrocytomas. J Neuropathol Exp Neurol. 2004; 63(9): 978-89.

21. Yu Q, Stamenkovic I. Angiopoietin-2 Is Implicated in the Regulation of

Tumor Angiogenesis. Am J Pathol. 2001; 158(2): 563-70.

22. Lakka SS, Gondi CS, Rao JS. Proteases and Glioma Angiogenesis. Brain Pathol. 2005;

15(4): 327-41.

23. Binder DK, Berger MS. Proteases and the biology of glioma invasion. J Neurooncol. 2002;

56(2): 149-58.

24. Kalluri R. Basement membranes: structure, assembly and role in tumour angiogenesis. Nat

Rev Cancer. 2003; 3(6): 422-33.

25. Sawaya RE, Yamamoto M, Gokaslan ZL, Wang SW, Mohanam S, Fuller GN, et al.

Expression and localization of 72 kDa type IV collagenase (MMP-2) in human malignant

gliomas in vivo. Clin Exp Metastasis. 1996; 14(1): 35-42.

26. Bergers G, Brekken R, McMahon G, Vu TH, Itoh T, Tamaki K, et al. Matrix

metalloproteinase-9 triggers the angiogenic switch during carcinogenesis. Nat Cell Biol.

2000; 2(10): 737-44.

27. Chandrasekar N, Jasti S, Alfred-Yung WK, Ali-Osman F, Dinh DH, Olivero WC, et al.

Modulation of endothelial cell morphogenesis in vitro by MMP-9 during glial-endothelial

cell interaction. Clin Exp Metastasis. 2000; 18(4): 337-42.

28. Gondi CS, Lakka SS, Dinh DH, Olivero WC, Gujrati M, Rao JS. Downregulation of uPA,

uPAR and MMP-9 using small, interfering, hairpin RNA (siRNA) inhibits glioma cell

invasion, angiogenesis and tumor growth. Neuron Glia Biol. 2004; 1(2): 165-76.

29. Brooks PC, Strömblad S, Sanders LC, von Schalscha TL, Aimes RT, Stetler-Stevenson WG,

et al. Localization of MMP-2 to the surface of invasive cells by interaction with integrin

alpha v beta 3. Cell. 1996; 85(5): 683-93.

30. Mohanam S, Gladson CL, Rao CN, Rao JS. Biological significance of the expression of

urokinase-type plasminogen activator receptors (uPARs) in brain tumors. Front Biosci.

1999; 4: D178-87.

31. Gladson CL, Pijuan-Thompson V, Olman MA, Gillespie GY, Yacoub IZ. Up-regulation of

urokinase and urokinase receptor genes in malignant astrocytoma. Am J Pathol. 1995;

146(5): 1150-60.

32. Ossowski L, Aguirre-Ghiso JA. Urokinase receptor and integrin partnership: coordination of

signaling for cell adhesion, migration, and growth. Curr Opin Cell Biol. 2000; 12(5):

613-20.

33. Yamamoto M, Sawaya R, Mohanam S, Rao VH, Bruner JM, Nicolson GL, et al. Expression

and localization of urokinase-type plasminogen activator receptor in human gliomas.

Cancer Res. 1994; 54(18): 5016-20.

34. Ding Q, Stewart J Jr, Olman MA, Klobe MR, Gladson CL. The pattern of enhancement of

Src kinase activity on platelet-derived growth factor stimulation of glioblastoma cells is

affected by the integrin engaged. J Biol Chem. 2003; 278(41): 39882-91.

35. Huttenlocher A, Horwitz AR. Integrins in cell migration. Cold Spring Harb Perspect Biol.

2011; 3(9): a005074.

36. Nobes CD, Hall A. Rho, rac and cdc42 GTPases regulate the assembly of mutilmolecular

focal complexes associated with actin stress fibers, lamellipodia, and filopodia. Cell.

1995; 81(1): 53-62.

37. Yamazaki D, Suetsugu S, Miki H, Kataoka Y, Nishikawa S, Fujiwara T, et al. WAVE2 is

required for directed cell migration and cardiovascular development. Nature. 2003;

424(6947): 452-6.

38. Kiosses WB, Shattil SJ, Pampori N, Schwartz MA. Rac recruits high-affinity integrin

alphavbeta3 to lamellipodia in endothelial cell migration. Nat Cell Biol. 2001; 3(3): 316-

20.

39. Klemke RL, Cai S, Giannini AL, Gallagher PJ, de Lanerolle P, Cheresh DA. Regulation of

cell motility by mitogen-activated protein kinase. J Cell Biol. 1997; 137(2): 481-92.

40. Chrzanowska-Wodnicka M, Burridge K. Rho-stimulated contractility drives the formation of

stress fibers and focal adhesions. J Cell Biol. 1996; 133(6): 1403-15.

41. Xu J, Rodriguez D, Petitclerc E, Kim JJ, Hangai M, Moon YS, et al. Proteolytic exposure of

a cryptic site within collagen type IV is required for angiogenesis and tumor growth in

vivo. J Cell Biol. 2001; 154(5): 1069-79.

42. Wang D, Anderson JC, Gladson CL. The role of the ECM in angiogenesis in malignant

glioma tumors. Brain Pathol. 2005; 15(4): 318-26.

43. Sudhakar A, Sugimoto H, Yang C, Lively J, Zeisberg M, Kalluri R. Human tumstatin and

human endostatin exhibit distinct antiangiogenic activities mediated by alpha v beta 3

and alpha 5 beta 1 integrins. Proc Natl Acad Sci U S A. 2003; 100(8): 4766-71.

44. Stratman AN, Saunders WB, Sacharidou A, Koh W, Fisher KE, Zawieja DC, et al. .

Endothelial cell lumen and vascular guidance tunnel formation requires MT1-MMP-

dependent proteolysis in 3-dimensional collagen matrices. Blood. 2009; 114(2): 237-47.

45. Gingras MC, Roussel E, Bruner JM, Branch CD, Moser RP. Comparison of cell adhesion

molecule expression between glioblastoma multiforme and autologous normal brain

tissue. J Neuroimmunol. 1995; 57(1-2): 143-53.

46. Haskell H, Natarajan M, Hecker TP, Ding Q, Stewart J Jr, Grammer JR, et al. Focal

Adhesion Kinase Is Expressed in the Angiogenic Blood Vessels of Malignant Astrocytic

Tumors in Vivo and Promotes Capillary Tube Formation of Brain Microvascular

Endothelial Cells. Clin Cancer Res. 2003; 9(6): 2157-65.

47. Liu Y, Senger DR. Matrix-specific activation of Src and Rho initiates capillary

morphogenesis of endothelial cells. FASEB J. 2004; 18(3): 457-68.

48. Kamei M, Saunders WB, Bayless KJ, Dye L, Davis GE, Weinstein BM. Endothelial tubes

assemble from intracellular vacuoles in vivo. Nature. 2006 Jul 27;442(7101):453-6.

49. Koh W, Mahan RD, Davis GE. Cdc42- and Rac1-mediated endothelial lumen formation

requires Pak2, Pak4 and Par3, and PKC-dependent signaling. J Cell Sci. 2008; 121(Pt 7):

989-1001.

50. Lampugnani MG, Orsenigo F, Rudini N, Maddaluno L, Boulday G, Chapon F. CCM1

regulates vascular-lumen organization by inducing endothelial polarity. J Cell Sci. 2010;

123(Pt 7): 1073-80.

51. Baluk P, Morikawa S, Haskell A, Mancuso M, McDonald DM. Abnormalities of basement

membrane on blood vessels and endothelial sprouts in tumors. Am J Pathol. 2003;

163(5): 1801-15.

52. Lindblom P, Gerhardt H, Liebner S, Abramsson A, Enge M, Hellstrom M. Endothelial

PDGF-B retention is required for proper investment of pericytes in the microvessel wall.

Genes Dev. 2003; 17(15): 1835-40.

53. Svensson A, Özen I, Genové G, Paul G, Bengzon J. Endogenous brain pericytes are widely

activated and contribute to mouse glioma microvasculature. PLoS One. 2015; 10(4):

e0123553.

54. Cheng L, Huang Z, Zhou W, Wu Q, Donnola S, Liu JK, et al. Glioblastoma stem cells

generate vascular pericytes to support vessel function and tumor growth. Cell. 2013;

153(1): 139-52.

55. Yancopoulos GD, Davis S, Gale NW, Rudge JS, Wiegand SJ, Holash J. Vascular- specific

growth factors and blood vessel formation. Nature. 2000; 407(6801): 242-8.

56. Lewis CE, De Palma M, Naldini L. Tie2-expressing monocytes and tumor angiogenesis:

regulation by hypoxia and angiopoietin-2. Cancer Res. 2007; 67(18): 8429-32.

57. Du R, Lu KV, Petritsch C, Liu P, Ganss R, Passegué E, et al. HIF1alpha induces the

recruitment of bone marrow-derived vascular modulatory cells to regulate tumor

angiogenesis and invasion. Cancer Cell. 2008; 13(3): 206-20.

58. Shaifer CA, Huang J, Lin PC. Glioblastoma cells incorporate into tumor vasculature and

contribute to vascular radioresistance. Int J Cancer. 2010; 127(9): 2063-75.

59. Yao X, Ping Y, Liu Y, Chen K, Yoshimura T, Liu M, et al. Vascular endothelial growth

factor receptor 2 (VEGFR-2) plays a key role in vasculogenic mimicry formation,

neovascularization, and tumor initiation by glioma stem-like cells. PLoS One. 2013; 8(3):

e57188.

60. Francescone R, Scully S, Bentley B, Yan W, Taylor SL, Oh D et al. Glioblastoma- derived

tumor cells induce vasculogenic mimicry through Flk-1 protein activation. J Biol Chem.

2012; 287(29): 24821-31.

61. Ricci-Vitiani L, Pallini R, Biffoni M, Todaro M, Invernici G, Cenci T. Tumour

vascularization via endothelial differentiation of glioblastoma stem-like cells. Nature.

2010; 468(7325): 824-8.

62. Wang R, Chadalavada K, Wilshire J, Kowalik U, Hovinga KE, Geber A, et al. Glioblastoma

stem-like cells give rise to tumour endothelium. Nature. 2010; 468(7325): 829-33.

63. Senger DR, Davis GE. Angiogenesis. Cold Spring Harb Perspect Biol. 2011; 3(8): a005090.

64. Cheng L, Wu Q, Guryanova OA, Huang Z, Huang Q, Rich JN, et al. Elevated invasive

potential of glioblastoma stem cells. Biochem Biophys Res Commun. 2011; 406(4): 643-

65. Gladson CL. Expression of integrin alpha v beta 3 in small blood vessels of glioblastoma

tumors. J Neuropathol Exp Neurol. 1996; 55(11): 1143-9.

66. Tabatabai G, Weller M, Nabors B, Picard M, Reardon D, Mikkelsen T. Targeting integrins in

malignant glioma. Target Oncol. 2010; 5(3): 175-81.

67. Ivaska J, Heino J. Adhesion receptors and cell invasion: mechanisms of integrin- guided

degradation of extracelluar matrix. Cell Mol Life Sci. 2000; 57(1): 16-24.

68. Brooks PC, Strömblad S, Sanders LC, von Schalscha TL, Aimes RT, Stetler-Stevenson WG,

et al. Localization of matrix metalloproteinase MMP-2 to the surface of invasive cells by

interaction with integrin alpha v beta 3. Cell. 1996; 85(5): 683-93.

69. Shattil S, Kim C, and Ginsberg MH. The final steps of integrin activation: the end game. Nat

Rev Mol Cell Biol. 2010; 11(4): 288-300.

70. Haskell H, Natarajan M, Hecker TP, Ding Q, Stewart J Jr, Grammer JR, et al. Focal adhesion

kinase is expressed in the angiogenic blood vessels of malignant astrocytic tumors in vivo

and promotes capillary tube formation of brain microvascular endothelial cells. Clin

Cancer Res. 2003; 9(6): 2157-65.

71. Ilic D, Kovacic B, McDonagh S, Jin F, Baumbusch C, Gardner DG, et al. Focal adhesion

kinase is required for blood vessel morphogenesis. Circ Res. 2003; 92(3): 300-7.

72. Zhai J, Lin H, Nie Z, Wu J, Cañete-Soler R, Schlaepfer WW, et al. Direct interaction of

focal adhesion kinase with p190RhoGEF. J Biol Chem. 2003; 278(27): 24865-73.

73. Schlaepfer DD, Mitra SK. Multiple connections link FAK to cell motility and invasion. Curr

Opin Genet Dev. 2004; 14(1): 92-101.

74. Cox BD, Natarajan M, Stettner MR, Gladson CL. New Concepts regarding FAK promotion

of cell migration and proliferation. J Cell Biochem. 2006; 99(1): 35-52.

75. Mitra SK, Mikolon D, Molina JE, Hsia DA, Hanson DA, Chi A, et al. Intrinsic FAK activity

and Y925 phosphorylation facilitate an angiogenic switch in tumors. Oncogene. 2006;

25(44): 5969-84.

76. Chuang JY, Chen PC, Tsao CW, Chang AC, Lein MY, Lin CC, et al. WISP-1, a novel

angiogenic regulator of the CCN family, promotes oral squamous cell carcinoma

angiogenesis through VEGF-A expression. Oncotarget. 2015; 6(6): 4239-52.

77. Schlaepfer DD, Broome MA, Hunter T. Fibronectin-Stimulated Signaling from a Focal

Adhesion Kinase-c-Src Complex: Involvement of the Grb2, p130Cas, and Nck Adaptor

Proteins. Mol Cell Biol. 1997; 17(3): 1702-13.

78. Cary LA, Han DC, Polte TR, Hanks SK, Guan JL. Identification of p130CAS as a mediator

of focal adhesion kinase-promoted cell migration. J Cell Biol. 1998; 140(1): 211-21.

79. Guerrero MS, Parsons JT, Bouton AH. Cas and NEDD9 contribute to tumor progression

through dynamic regulation of the cytoskeleton. Genes Cancer. 2012; 3(5-6): 371-81.

80. Abassi YA, Rehn M, Ekman N, Alitalo K, Vuori K. p130CAS couples the tyrosine kinase

BMX/Etk with regulation of the actin cytoskeleton and cell migration. J Biol Chem.

2003; 278(37): 35636-43.

81. Chen R, Kim O, Li M, Xiong X, Guan JL, Kung HJ. Regulation of the PH-domain-

containing tyrosine kinase Etk by focal adhesion kinase through the FERM domain. Nat

Cell Biol. 2001; 3(5): 439-44.

82. Renshaw MW, Ren XD, Schwartz MA. Growth factor activation of MAP kinase requires cell

adhesion. EMBO J. 1997; 16(18): 5592-9.

83. Borges E, Jan Y, Ruoslahti E. Platelet-derived growth factor beta and vascular endothelial

growth factor receptor 2 bind to the beta 3 integrin through its extracellular domain. J

Biol Chem. 2000; 275(51): 39867-73.

84. Somanath PR, Ciocea A, Byzova TV. Integrin and growth factor receptor alliance in

angiogenesis. Cell Biochem Biophys. 2009; 53(2): 53-64.

85. Mahabeleshwar GH, Chen J, Feng W, Somanath PR, Razorenova OV, Byzova TV. Integrin

affinity modulation in angiogenesis. Cell Cycle. 2008; 7(3): 335-47.

86. Rahman S, Patel Y, Murray J, Patel KV, Sumathipala R, Sobel M, et al. Novel hepatocyte

growth factor (HGF) binding domains in fibronectin and vitronectin coordinate a distinct

and amplified Met-integrin induced signaling pathway in endothelial cells. BMC Cell

Biol. 2005; 6(1): 8.

87. Gavert N, Ben-Shmuel A, Raveh S, Ben-Ze'ev A. L1-CAM in cancerous tissues. Expert Opin

Biol Ther. 2008; 8(11): 1749-57.

88. Bao S, Wu Q, Li Z, Sathornsumetee S, Wang H, McLendon RE, et al. Targeting cancer stem

cells through L1CAM suppresses glioma growth. Cancer Res. 2008; 68(15): 6043-8.

89. Held-Feindt J, Schmelz S, Hattermann K, Mentlein R, Mehdorn HM, Sebens S. The neural

adhesion molecule L1CAM confers chemoresistance in human glioblastomas.

Neurochem Int. 2012; 61(7): 1183-91.

90. Mechtersheimer S, Gutwein P, Agmon-Levin N, Stoeck A, Oleszewski M, Riedle S, et al.

Ectodomain shedding of L1 adhesion molecule promotes cell migration by autocrine

binding to integrins. J Cell Biol. 2001; 155(4): 661-73.

91. Tsuzuki T, Izumoto S, Ohnishi T, Hiraga S, Arita N, Hayakawa T. Neural cell adhesion

molecule L1 in gliomas: correlation with TGF-beta and p53. J Clin Pathol. 1998; 51(1):

13-7.

92. Yang M, Li Y, Chilukuri K, Brady OA, Boulos MI, Kappes JC, et al. L1 stimulation of

human glioma cell motility correlates with FAK activation. J Neurooncol. 2011; 105(1):

27-44.

93. Friedli A, Fischer E, Novak-Hofer I, Cohrs S, Ballmer-Hofer K, Schubiger PA, et al. The

soluble form of the cancer-associated L1 cell adhesion molecule is a pro-angiogenic

factor. Int J Biochem Cell Biol. 2009; 41(7): 1572-80.

94. Gast D, Riedle S, Kiefel H, Müerköster SS, Schäfer H, Schäfer MK, et al. The RGD integrin

binding site in human L1-CAM is important for nuclear signaling. Exp Cell Res. 2008;

314(13): 2411-8.

95. Mohanan V, Temburni MK, Kappes JC, Galileo DS. L1CAM stimulates glioma cell motility

and proliferation through the fibroblast growth factor receptor. Clin Exp Metastasis.

2013; 30(4): 507-20.

96. Islam R, Kristiansen LV, Romani S, Garcia-Alonso L, Hortsch M. Activation of EGF

Receptor Kinase by L1-mediated Homophilic Cell Interactions. Mol Biol Cell. 2004;

15(4): 2003-12.

97. Charles NA, Holland EC, Gilbertson R, Glass R, Kettenmann H. The Brain Tumor

Microenvironment. Glia. 2012; 60(3): 502-14.

98. Leaver HA, Rizzo MT, Whittle IR. Glioma Cell Death: Cell–Cell Interactions and Signalling

Networks. Mol Neurobiol. 2010; 42(1): 89-96.

99. Thelen K, Kedar V, Panicker AK, Schmid RS, Midkiff BR, Maness PF. The Neural Cell

Adhesion Molecule L1 Potentiates Integrin-Dependent migration to Extracellular Matrix

Proteins. J Neurosci. 2002; 22(12): 4918-31.

100. Felding-Habermann B, Silletti S, Mei F, Siu CH, Yip PM, Brooks PC, et al. A Single

Immunoglobulin-like Domain of the Human Neural Cell Adhesion Molecule L1 Supports

Adhesion by Multiple Vascular and Platelet Integrins. J Cell Biol. 1997; 139(6): 1567-81.

101. Silletti S, Mei F, Sheppard D, Montgomery AM. Plasmin-sensitive dibasic sequences in the

third-Fibronectin-like domain of L1-Cell Adhesion Molecule (CAM) Facilitate

Homomultimerization and concomitant integrin recruitment. J Cell Biol. 2000; 149(7):

1485-502.

102. Chen MM, Lee CY, Leland HA, Lin GY, Montgomery AM, Silletti S. Inside-out regulation

of L1 Conformation, Integrin Binding, Proteolysis, and Concomitant Cell Migration. Mol

Biol Cell. 2010; 21(10): 1671-85.

103. Hall H, Djonov V, Ehrbar M, Hoechli M, Hubbell JA. Heterophilic interactions between

cell adhesion molecule L1 and alphavbeta3-integrin induce HUVEC process extension in

vitro and angiogenesis in vivo. Angiogenesis. 2004; 7(3): 213-23.

104. Gutwein P, Stoeck A, Riedle S, Gast D, Runz S, Condon TP. Cleavage of L1 in exosomes

and apoptotic membrane vesicles released from ovarian carcinoma cells. Clin Cancer

Res. 2005; 11(7): 2492-501.

105. Ratajczak J, Wysoczynski M, Hayek F, Janowska-Wieczorek A, Ratajczak MZ. Membrane-

derived microvesicles: important and underappreciated mediators of cell-to-cell

communication. Leukemia. 2006; 20(9): 1487-95.

106. Gutwein P, Mechtersheimer S, Riedle S, Stoeck A, Gast D, Joumaa S, et al. ADAM10-

mediated cleavage of L1 adhesion molecule at the cell surface and in released membrane

vesicles. FASEB J. 2003; 17(2): 292-4.

107. Nakano I, Garnier D, Minata M, Rak J. Extracellular vesicles in the biology of brain

tumour stem cells--Implications for inter-cellular communication, therapy and biomarker

development. Semin Cell Dev Biol. 2015; 40: 17-26.

108. Hale JS, Li M, Lathia JD. The malignant social network: Cell-cell adhesion and

communication in cancer stem cells. Cell Adh Migr. 2012; 6(4): 346-55.

109. Gómez-Gaviro MV, Lovell-Badge R, Fernández-Avilés F, Lara-Pezzi E. The Vascular

Stem Cell Niche. J Cardiovasc Transl Res. 2012; 5(5): 618-30.

110. Gilbertson RJ, Rich JN. Making a tumour’s bed: glioblastoma stem cells and the vascular

niche. Nat Rev Cancer. 2007; 7(10): 733-6.

111. Shen Q, Wang Y, Kokovay E, Lin G, Chuang SM, Goderie SK. Adult SVZ stem cells lie in

a vascular niche: a quantitative analysis of niche cell-cell interactions. Cell Stem Cell.

2008; 3(3): 289-300.

112. Li Q, Ford MC, Lavik EB, Madri JA. Modeling the neurovascular niche: VEGF- and

BDNF-mediated cross-talk between neural stem cells and endothelial cells: an in vitro

study. J Neurosci Res. 2006; 84(8): 1656-68.

113. Calabrese C, Poppleton H, Kocak M, Hogg TL, Fuller C, Hamner B, et al. A perivascular

niche for brain tumor stem cells. Cancer Cell. 2007; 11(1): 69-82.

114. Lathia JD, Gallagher J, Heddleston JM, Wang J, Eyler CE, Macswords J, et al. Integrin

alpha 6 regulates glioblastoma stem cells. Cell Stem Cell. 2010; 6(5): 421-32.

115. Charles N, Ozawa T, Squatrito M, Bleau AM, Brennan CW, Hambardzumyan D, et al.

Perivascular nitric oxide activates notch signaling and promotes stem-like character in

PDGF-induced glioma cells. Cell Stem Cell. 2010; 6(2): 141-52.

116. Lathia JD, Li M, Hall PE, Gallagher J, Hale JS, Wu Q, et al. Laminin alpha-2 enables

glioblastoma stem cell growth. Ann Neurol. 2012; 72(5): 766-78.

117. Hovinga KE, Shimizu F, Wang R, Panagiotakos G, Van Der Heijden M, Moayedpardazi H,

et al. Inhibition of notch signaling in glioblastoma targets cancer stem cells via an

endothelial cell intermediate. Stem Cells. 2010; 28(6): 1019-29.

118. Hjelmeland AB, Lathia JD, Sathornsumetee S, Rich JN. Twisted tango: brain tumor

neurovascular interactions. Nat Neurosci. 2011; 14(11): 1375-81.

119. Ping YF, Yao XH, Jiang JY, Zhao LT, Yu SC, Jiang T, et al. The chemokine CXCL12 and

its receptor CXCR4 promote glioma stem cell-mediated VEGF production and tumour

angiogenesis via PI3K/AKT signaling. J Pathol. 2011; 224(3): 344-54.

120. Lugassy C, Wadehra M, Li X, Corselli M, Akhavan D, Binder SW, et al. Pilot study on

"pericytic mimicry" and potential embryonic/stem cell properties of angiotropic

melanoma cells interacting with the abluminal vascular surface. Cancer Microenviron.

2013; 6(1): 19-29.

121. Kesari S. Understanding glioblastoma tumor biology: the potential to improve current

diagnosis and treatments. Semin Oncol. 2011; 38 Suppl 4: S2-10.

122. Shojaei F. Anti-angiogenesis therapy in cancer: Current challenges and future perspectives.

Cancer Lett. 2012; 320(2): 130-7.

123. Pàez-Ribes M, Allen E, Hudock J, Takeda T, Okuyama H, Viñals F, et al. Antiangiogenic

therapy elicits malignant progression of tumors to increased local invasion and distant

metastasis. Cancer Cell. 2009; 15(3): 220-31.

124. Huang P, Rani MR, Ahluwalia MS, Bae E, Prayson RA, Weil RJ, et al. Endothelial

expression of TNF receptor-1 generates a proapoptotic signal inhibited by integrin

alpha6beta1 in glioblastoma. Cancer Res. 2012; 72(6): 1428-37.

125. Lathia JD, Gallagher J, Myers JT, Li M, Vasanji A, McLendon RE, et al. Direct in vivo

evidence for tumor propagation by glioblastoma cancer stem cells. PLoS One. 2011; 6(9):

e24807.

126. Wang D, Olman MA, Stewart J Jr, Tipps R, Huang P, Sanders PW, et al. Downregulation of

FIP200 induces apoptosis of glioblastoma cells and microvascular endothelial cells by

enhancing Pyk2 activity. PLoS One. 2011;6(5):e19629.

127. Meijering E, Dzyubachyk O, Smal I. Methods for cell and particle tracking. Methods

Enzymol. 2012; 504: 183-200.

128. Ding Q, Stewart J Jr, Prince CW, Chang PL, Trikha M, Han X, et al. Promotion of

Malignant Astrocytoma Cell Migration by Osteopontin Expressed in the Normal Brain:

Differences in Integrin Signaling During Cell Adhesion to Osteopontin Versus

Vitronectin. Cancer Res. 2002; 62(18): 5336-43.

129. Roth P, Silginer M, Goodman SL, Hasenbach K, Thies S, Maurer G, et al. Integrin control

of the transforming growth factor-beta pathway in glioblastoma. Brain. 2013; 136(Pt 2):

564-76.

130. Fael Al-Mayhani TM, Ball SL, Zhao JW, Fawcett J, Ichimura K, Collins PV, et al. An

efficient method for derivation and propagation of glioblastoma cell lines that conserves

the molecular profile of their original tumours. J Neurosci Methods. 2009; 176(2): 192-9.

131. Gangemi RM, Griffero F, Marubbi D, Perera M, Capra MC, Malatesta P, et al. SOX2

silencing in glioblastoma tumor-initiating cells causes stop of proliferation and loss of

tumorigenicity. Stem Cells. 2009; 27(1): 40-8.

132. Cheresh DA, Stupack DG. Regulation of angiogenesis: apoptotic cues from the ECM.

Oncogene. 2008; 27(48): 6285-98.

133. Huang C, Jacobson K, Schaller MD. MAP kinases and cell migration. J Cell Sci. 2004;

117(Pt 20): 4619-28.

134. Beadle C, Assanah MC, Monzo P, Vallee R, Rosenfeld SS, Canoll P. The role of myosin II

in glioma invasion of the brain. Mol Biol Cell. 2008; 19(8): 3357-68.

135. Chatterjee S, Matsumura A, Schradermeier J, Gillespie GY. Human malignant glioma

therapy using anti-alpha(v)beta3 integrin agents. J Neurooncol. 2000; 46(2): 135-44.

136. Mikkelsen T, Brodie C, Finniss S, Berens ME, Rennert JL, Nelson K, et al. Radiation

sensitization of glioblastoma by cilengitide has unanticipated schedule-dependency. Int J

Cancer. 2009; 124(11): 2719-27.

137. He Y, Luo Y, Tang S, Rajantie I, Salven P, Heil M, et al. Critical function of Bmx/Etk in

ischemia-mediated arteriogenesis and angiogenesis. J Clin Invest. 2006; 116(9): 2344-55.

138. Holopainen T, López-Alpuche V, Zheng W, Heljasvaara R, Jones D, He Y, et al. Deletion

of the endothelial Bmx tyrosine kinase decreases tumor angiogenesis and growth. Cancer

Res. 2012; 72(14): 3512-21.

139. Simon KO, Nutt EM, Abraham DG, Rodan GA, Duong LT. The alphavbeta3 integrin

regulates alpha5beta1-mediated cell migration toward fibronectin. J Biol Chem. 1997;

272(46): 29380-9.

140. Pijuan-Thompson V, Gladson CL. Ligation of integrin alpha5beta1 is required for

internalization of vitronectin by integrin alphavbeta3. J Biol Chem. 1997; 272(5): 2736-

43.

141. Friedlander M, Brooks PC, Shaffer RW, Kincaid CM, Varner JA, Cheresh DA. Definition

of two angiogenic pathways by distinct αv integrins. Science. 1995; 270(5241): 1500-2.

142. Hood JD, Frausto R, Kiosses WB, Schwartz MA, Cheresh DA. Differential αv integrin

mediated Ras-ERK signaling during two pathways of angiogenesis. J Cell Biol. 2003;

162(5): 933-43.

143. Peddibhotla SS, Brinkmann BF, Kummer D, Tuncay H, Nakayama M, Adams RH, et al.

CD9 links junctional adhesion molecule-A to αvβ3 integrin to mediate basic fibroblast

growth factor-specific angiogenic signaling. Mol Biol Cell. 2013; 24(7): 933-44.

144. Schmidt C, Künemund V, Wintergerst ES, Schmitz B, Schachner M. CD9 of mouse brain is

implicated in neurite outgrowth and cell migration in vitro and is associated with the

alpha 6/beta 1 integrin and the neural adhesion molecule L1. J Neurosci Res. 1996; 43(1):

12-31.

145. Williams EJ, Doherty P, Turner G, Reid RA, Hemperly JJ, Walsh FS. Calcium influx into

neurons can solely account for cell contact-dependent neurite outgrowth stimulated by

transfected L1. J Cell Biol. 1992; 119(4): 883-92.

146. Hough MR, Rosten PM, Sexton TL, Kay R, Humphries RK. Mapping of CD24 and

homologous sequences to multiple chromosomal loci. Genomics. 1994; 22(1): 154-61.

147. Pruszak J, Ludwig W, Blak A, Alavian K, Isacson O. CD15, CD24, and CD29 define a

surface biomarker code for neural lineage differentiation of stem cells. Stem Cells. 2009;

27(12): 2928-40.

148. Behera R, Kumar V, Lohite K, Karnik S, Kundu GC. Activation of JAK2/STAT3 signaling

by osteopontin promotes tumor growth in human breast cancer cells. Carcinogenesis.

2010; 31(2): 192-200.

149. Rice J, Courter DL, Giachelli CM, Scatena M. Molecular mediators of alphavbeta3-induced

endothelial cell survival. J Vasc Res. 2006; 43(5): 422-36.

150. Voura EB, Ramjeesingh RA, Montgomery AM, Siu CH. Involvement of integrin

alpha(v)beta(3) and cell adhesion molecule L1 in transendothelial migration of melanoma

cells. Mol Biol Cell. 2001; 12(9): 2699-710.

151. Magrini E, Villa A, Angiolini F, Doni A, Mazzarol G, Rudini N, et al. Endothelial

deficiency of L1 reduces tumor angiogenesis and promotes vessel normalization. J Clin

Invest. 2014; 124(10): 4335-50.

152. Issa Y, Nummer D, Seibel T, Müerköster SS, Koch M, Schmitz-Winnenthal FH, et al.

Enhanced L1CAM expression on pancreatic tumor endothelium mediates selective tumor

cell transmigration. J Mol Med (Berl). 2009 ; 87(1): 99-112.

153. Nabors LB, Fink KL, Mikkelsen T, Grujicic D, Tarnawski R, Nam do H, Mazurkiewicz M,

et al. Two cilengitide regimens in combination with standard treatment for patients with

newly diagnosed glioblastoma and unmethylated MGMT gene promoter: results of the

open-label, controlled, randomized phase II CORE study. Neuro Oncol. 2015; 17(5):

708-17.

154. Scheurer SB, Rybak JN, Rösli C, Neri D, Elia G. Modulation of gene expression by hypoxia

in human umbilical cord vein endothelial cells: A transcriptomic and proteomic study.

Proteomics. 2004; 4(6): 1737-60.

155. Gonzalez AM, Gonzales M, Herron GS, Nagavarapu U, Hopkinson SB, Tsuruta D, et al.

Complex interactions between the laminin alpha4 subunit and integrins regulate

endothelial cell behavior in vitro and angiogenesis in vivo. Proc Natl Acad Sci U S A.

2002; 99(25): 16075-80.

156. Iivanainen A, Kortesmaa J, Sahlberg C, Morita T, Bergmann U, Thesleff I, et al. Primary

structure, developmental expression, and immunolocalization of the murine laminin

alpha4 chain. J Biol Chem. 1997; 272(44): 27862-8.

157. Ljubimova JY, Fugita M, Khazenzon NM, Das A, Pikul BB, Newman D, et al. Association

between Laminin-8 and Glial Tumor Grade, Recurrence, and Patient Survival. Cancer.

2004; 101(3): 604-12.

158. Gutwein P, Mechtersheimer S, Riedle S, Stoeck A, Gast D, Joumaa S, et al. ADAM10-

mediated cleavage of L1 adhesion molecule at the cell surface and in released membrane

vesicles. FASEB J. 2003; 17(2): 292-4.

159. Mannino M, Chalmers AJ. Radioresistance of glioma stem cells: Intrinsic characteristic or

property of the ‘microenvironment-stem cell unit’? Mol Oncol. 2011; 5(4): 374-86.

160. Brown CK, Khodarev NN, Yu J, Moo-Young T, Labay E, Darga TE, et al. Glioblastoma

cells block radiation-induced programmed cell death of endothelial cells. FEBS Lett.

2004; 565(1-3): 167-70.

161. Tavora B, Reynolds LE, Batista S, Demircioglu F, Fernandez I, Lechertier T, et al.

Endothelial-cell FAK targeting sensitizes tumours to DNA-damaging therapy. Nature.

2014; 514(7520): 112-6.

162. Rivera LB, Bergers G. Tumor angiogenesis, from foe to friend. Science. 2015; 349(6249):

694-5.

163. Cheng L, Wu Q, Huang Z, Guryanova OA, Huang Q, Shou W, et al. L1CAM regulates

DNA damage checkpoint response of glioblastoma stem cells through NBS1. EMBO J.

2011; 30(5): 800-13.

164. Wong PP, Demircioglu F1, Ghazaly E2, Alrawashdeh W3, Stratford MR4, Scudamore CL,

et al. Dual-action combination therapy enhances angiogenesis while reducing tumor

growth and spread. Cancer Cell. 2015; 27(1): 123-37.

165. Pugacheva EN, Golemis EA. The focal adhesion scaffolding protein HEF1 regulates

activation of the Aurora-A and Nek2 kinases at the centrosome. Nat Cell Biol. 2005;

7(10): 937-46.

166. Kozyreva VK, McLaughlin SL, Livengood RH, Calkins RA, Kelley LC, Rajulapati A, et al.

NEDD9 Regulates Actin Dynamics through Cortactin Deacetylation in an

AURKA/HDAC6-dependent Manner. Mol Cancer Res. 2014; 12(5): 681-93.

APPENDICES

SUPPLEMENTAL DATA

Figure 27. Specificity of rabbit affinity purified anti-phospho-p130CAS (Y234) antibody. Rabbit phospho-specific antibody towards p130CAS (Y234) was generated using the following peptide: -AQPEQDE[pY]DIPRHL, corresponding to amino acids 227-240 in human p130CAS protein, and affinity purified as described (1, 2). The antibody was tested for specificity using MCF7 cells that were serum starved for 24h, treated with EGF for 30 min, lysed and subjected to disulfide-reduced SDS-PAGE (1, 2). Lysate was also treated with a cocktail of recombinant

phosphatases (PP1, PP2A, AP, SHIP2) for 5 min to dephosphorylate proteins before addition of the gel loading buffer (1, 2). The anti-phospho-p130CAS (Y234) antibody (dilution 1:1000) detects a band migrating at 130-kDa and does not cross react with phosphorylated NEDD9 (HEF1) protein that migrates as a doublet between 80 and 100-kDa. The mouse mAb used for detection of total p130CAS was purchased (BD Biosciences, 1:5000 dilution) and directed towards the C-terminus. Rabbit affinity-purified anti-phospho-NEDD9 (Y166) antibody (dilution 1:500) and mouse mAb 2G9 anti-NEDD9 (1:5000 dilution) used to detect phosphorylated and total NEDD9 protein, respectively, have been described previously (165-166).

100

Figure 28. Controls for the cell-cell adhesion assay and 2P-LSM; and an association of ECs and CSCs after co-seeding on Matrigel®. A, Representative standard curve of the fluorescence emitted from increasing numbers of adherent GFP-CSCs in the cell-cell adhesion assay. B, Red-fluorescent-ECs treated with control siRNA or β3 siRNA were mixed and co- injected with GFP-CSCs into mouse brain slices in organotypic culture, followed by 2P-LSM at 24 h. The total number of ECs were quantified and the total number of ECs with β3 siRNA normalized to ECs treated with control siRNA. Data are graphed as the mean+SEM. C&D, GFP- CSCs (08387 and 4302) were treated with control siRNA or L1CAM siRNA, mixed and co- injected with red-fluorescent ECs into mouse brain slices in organotypic culture, followed by 2P- LSM at 24 h. The total number of GFP-CSCs were quantified and the total number of GFP-CSCs with L1CAM siRNA normalized to GFP-CSCs treated with control siRNA. Data are graphed as the mean+SEM. Statistics for B-D, exact two-sided Wilcoxon rank-sum tests. E, Red-fluorescent ECs were seeded on top of Matrigel® in complete NBM (ECs alone), in CSC conditioned-media (CM), or when mixed with CSCs. An association between ECs and CSCs is visualized at 2, 4 and 24 h. (5X magnification).

101

102

Figure 29. Co-seeding of ECs and CSCs significantly increases the activation of ERK and JNK in ECs; and CSCs increase the directional 2D motility of ECs through secreted factors and through direct contact. A, ECs in CM/EC+CSC or mixed directly with CSCs (3832 or 3691) were seeded onto laminin in NBM with bFGF (10 ng/ml) and no EGF (3 h), double- labeled for pERK or pJNK (Alexa-594, red) and for vWf (an EC marker, Alexa-488, green) by double-label immunofluorescence, followed by the calculation of signal intensity using FIJI (ImageJ) based on the sum of the mean gray level (intensity) output of all pixels of the area within the enclosed cell traced along its outer edge. Intensity was normalized by dividing the sum of the mean gray level by the area of the cell. Data are graphed as the mean+SEM. Statistics: two-sided Wilcoxon rank-sum tests, n=50. B, Red-fluorescent ECs and GFP-CSCs were seeded on either side of an 500 µm insert onto laminin, at 18 h the insert removed, and live- imaging of migration into the 500 µm gap performed q 15 min (24 h). The effect on EC and CSC migration into the gap when plated opposite of each other is graphed as the mean+SEM number of cells/mm2 in the gap. Statistics: exact two-sided Wilcoxon rank-sum tests. C, Tracks (trajectories) of ECs (n=30) with and without the presence of CSCs in the first 7 h of migration before possible cell-cell contact. D, The trajectories and displacement of ECs plated opposite of CSCs between 7-24 h after contact with CSCs (left panel) or in the absence of CSC contact (right panel) during the same time frame. E, Trajectories of the random migration pattern of ECs throughout the entire 24 h in the absence of CSCs.

103

LIST OF ABBREVIATIONS

2PLSM – 2-Photon Laser Scanning Microscopy

ADAM10 – A Disintegrin And Metalloproteinase Domain 10

AKT – Protein Kinase B

Ang – 1 – Angiopoietin 1

Ang – 2 –Angiopoietin 2

AP – Alkaline Phosphatase

BCAR1 – Breast Cancer Anti-Estrogen Resistance 1

BDNF – Brain-derived Neurotrophic Factor bFGF – Basic Fibroblast Growth Factor

BM – Basement Membrane

BMX – Bone Marrow Tyrosine Kinase Gene In Chromosome X Protein. See Etk.

BSA – Bovine Serum Albumin

CD9 – CD9 Antigen / Tetraspanin-29

CD24 – CD24 Antigen / Cluster of Differentiation 24

CD31 – CD31 Antigen / Cluster of Differentiation 31

CD105 – CD105 Antigen / Endoglin

104

CD133 – CD133 Antigen / Prominin 1

Cdc42 – Cell Division Control Protein 42 Homolog

CHO – Chinese Hamster Ovary

CM – Conditioned Medium

CM/EC+CSC – Conditioned Medium from EC+CSC Co-culture

Crk – Proto-Oncogene C-Crk

CSC – Glioblastoma Tumor Cells with Cancer Stem-like Properties

CXCR4 – Chemokine (C-X-C Motif) Receptor 4

DG – Dentate Gyrus

DMEM – Dulbecco’s modified eagle’s medium

DNA – Deoxyribonucleic Acid

DOCK180 – Dedicator Of Cytokinesis 1

EC – Endothelial Cell

ECM – Extracellular Matrix

EGFR – Epidermal Growth Factor Receptor

EGM – Endothelial Cell Growth Medium

EPC – Endothelial Progenitor Cells

Ephrin-B2 – Ligand Of Eph-Related Kinase 5

ERK – Extracellular Signal-Regulated Kinase / Mitogen-activated Protein Kinase

Etk – Epithelial And Endothelial Tyrosine Kinase. See also BMX (alias)

FAK – Focal Adhesion Kinase

FAT – Focal Adhesion Targeting Domain

FBS – Fetal Bovine Serum

105

FERM – 4.1 protein, ezrin,, radixin , moesin domain

FGFR – Fibroblast Growth Factor Receptor

FN – Fibronectin

GBM – Glioblastoma

GFP – Green Fluorescent Protein

Grb2 – Growth Factor Receptor-bound Protein 2

GTPase – Guanosine Triphosphate Hydrolase Enzyme

HEF1 – Human Enhancer of Filamentation 1. See also NEDD9 (alias).

HEK293 – Human Embryonic Kidney 293 cells

HGF— Hepatocyte Growth Factor

HOXA9 – Homeobox A9

HuVEC – Human Umbilical Vein Endothelial Cell

ICAM-1 – Intercellular Adhesion Molecule 1

IDH – Isocitrate Dehydrogenase 1

IFN-γ – Interferon Gamma

IgCAM – Ig Cell-Adhesion Molecule

IL1-β – Interleukin-1β

IL-3 – Interleukin-3

JAK – Janus Kinase

JAMA – Junctional Adhesion Molecule-A

JNK – c-Jun N-terminal kinase

JSAP1 – JNK/Stress-Activated Protein Kinase-Associated Protein 1

L1CAM – L1 cell adhesion molecule

106

LN308 – Human Glioblastoma Cell Line 308

MAPK – mitogen-activated protein kinase

MCD – Methyl-β-Cyclodextrin

MEK – ERK Activator Kinase / Mitogen-Activated Protein Kinase Kinase

Met – Tyrosine Protein Kinase Met / Hepatocyte Growth Factor Receptor

MGMT – O6-methylguanine methyltransferase

MMP – Matrix Metalloproteases mRNA – Messenger Ribonucleic Acid

MV – Microvesicle

NBM – Neurobasal Medium

NCAM – Neural Cell Adhesion Molecule

NEDD9 – Neural precursor cell expressed developmentally down-regulated protein 9. See also

HEF1 (alias).

NEFL – Neurofilament light polypeptide

NF1 – Neurofibromin

NF-κB – Nuclear Factor Kappa B

NSC – Neural Stem Cell p130CAS – Cas Scaffolding Protein Family Member 1. See also BCAR1 (alias). p190RhoGEF – 190 KDa Guanine Nucleotide Exchange Factor

Par3 – Par-3 Family Cell Polarity Regulator

Par6b – Par-6 Family Cell Polarity Regulator Beta

PBS – Phosphate-buffered Saline

PDGF – Platelet Derived Growth Factor

107

PDGFRA – Platelet-Derived Growth Factor Receptor, Alpha Polypeptide

PH – Pleckstrin Homology Domain

PI3K – Phosphoinositide 3’-Kinase

PKC-β – Protein Kinase C Beta

PKC-ζ – Protein kinase C Zeta

PLC-γ – Protein Kinase C Gamma

PMA – 4-phorbol-12-myristate 13-acetate

PP1 – Protein Phosphatase 1

PP2 – 4-amino-5-(4- chlorophenyl)-7-(dimethylethyl)pyrazolo[3,4-α]pyrimidine

PP2A – Protein Phosphatase 2

PTEN – Phosphatase and Tensin Homolog qRT-PCR – quantitative real-time polymerase chain reaction

Rac1 – Ras-related C3 botulinum toxin substrate 1

RAD – cyclo(Arg-Ala-Asp-D-Phe-Lys)

Raf-1 – Raf-1 Proto-Oncogene, Serine/Threonine Kinase

RanBPM – RAN Binding Protein 9

Ras – Ras small GTPase subfamily

RhoA – Ras homolog gene family, member A

RGD – cyclo (Arg-Gly-Asp-D-Phe-Lys)

SDF-1α – Stromal Cell-derived Factor-1 Alpha

SDS-PAGE – Sodium Dodecyl Sulfate - Polyacrylamide Gel Electrophoresis

SH2 – Src Homology 2 domain

SH3 – Src Homology 3 Domain

108

Shc – SHC-transforming Protein

SHIP2 – Phosphatidylinositol-3,4,5-trisphosphate 5-phosphatase siRNA – Small Interfering Ribonucleic Acid

SNB19 – Human Glioblastoma Cell Line

Sos – Son Of Sevenless Homolog

SOX2 – SRY (Sex Determining Region Y)-Box 2

Src – Proto-oncogene Tyrosine-protein Kinase Src

STAT – Signal Transducer and Activator of Transcription

SVZ – Subventricular Zone

TGFβ – Transforming Growth Factor Beta

Tie2 – Angiopoietin-1 Receptor

TIMP – Tissue Inhibitor of Metalloproteinase

TNF-α – Tumor Necrosis Factor Alpha

TP53 – Tumor Suppressor p53

U251MG – Human Glioblastoma Astrocytoma Cell Line uPA – Urokinase uPAR – Urokinase Receptor vCAM-1 – Vascular Cell Adhesion Molecule 1

VEGF-A – vascular endothelial growth factor A

VEGFR1 – Vascular Endothelial Growth Factor Receptor 1

VEGFR2 – Vascular Endothelial Growth Factor Receptor 2 vWf – Von Willebrand Factor

WASP – Wiskott–Aldrich Syndrome Protein

109

WAVE – WASP-family Verprolin-homologous Protein

WHO – World Health Organization

WISP-1 – WNT1-inducible-signaling Pathway Protein 1

110