HEF1 is a necessary and specific downstream effector of FAK that promotes the migration of glioblastoma cells.

The highly invasive behavior of glioblastoma cells contributes to the morbidity and mortality associated with these tumors. The integrin-mediated adhesion and migration of glioblastoma cells on brain matrix proteins is enhanced by stimulation with growth factors, including platelet-derived growth factor (PDGF). As focal adhesion kinase (FAK), a nonreceptor cytoplasmic tyrosine kinase, has been shown to promote cell migration in various other cell types, we analysed its role in glioblastoma cell migration. Forced overexpression of FAK in serum-starved glioblastoma cells plated on recombinant (rec)-osteopontin resulted in a twofold enhancement of basal migration and a ninefold enhancement of PDGF-BB-stimulated migration. Both expression of mutant FAK(397F) and the downregulation of FAK with small interfering (si) RNA inhibited basal and PDGF-stimulated migration. FAK overexpression and PDGF stimulation was found to increase the phosphorylation of the Crk-associated substrate (CAS) family member human enhancer of filamentation 1 (HEF1), but not p130CAS or Src-interacting protein (Sin)/Efs, although the levels of expression of these proteins was similar. Moreover downregulation of HEF1 with siRNA, but not p130CAS, inhibited basal and PDGF-stimulated migration. The phosphorylated HEF1 colocalized with vinculin and was associated almost exclusively with 0.1% Triton X-100 insoluble material, consistent with its signaling at focal adhesions. FAK overexpression promoted invasion through normal brain homogenate and siHEF1 inhibited this invasion. Results presented here suggest that HEF1 acts as a necessary and specific downstream effector of FAK in the invasive behavior of glioblastoma cells and may be an effective target for treatment of these tumors.

TGF-beta1 up-regulates paxillin protein expression in malignant astrocytoma cells: requirement for a fibronectin substrate.

Cytokines can influence the interactions between members of the integrin cell adhesion receptor family and the extracellular matrix thereby potentially affecting cell function and promoting cell adhesion, growth and migration of malignant astrocytoma tumor cells. As malignant astrocytoma cells synthesize TGF-beta1 in vivo, we analysed the effects of TGF-beta1 on signaling events associated with integrin receptor ligation, focusing on the effects on paxillin, a phosphorylated adaptor protein, that acts as a scaffold for signaling molecules recruited to focal adhesions. TGF-beta1-stimulation of primary astrocytes and serum-starved U-251MG malignant astrocytoma cells attached to fibronectin induced a substantial increase in the levels of paxillin protein (fivefold increase at 2.0 ng/ml) in a dose- and time-dependent manner compared to the levels observed on plating onto fibronectin in the absence of stimulation. In the astrocytoma cells, this resulted in an increase in the pool of tyrosine-phosphorylated paxillin, although it did not appear to alter the extent of phosphorylation of the paxillin molecules. In contrast, in primary astrocytes the protein levels were upregulated in the absence of a parallel increase in phosphorylation. The TGF-beta1-stimulated increase in paxillin levels required ligation of the fibronectin receptor, as it was not induced when the cells were plated onto vitronectin, collagen or laminin. The increase in the pool of paxillin on TGF-beta1 stimulation of the fibronectin-plated astrocytoma cells was associated with an increase in translation, but was not associated with an increase in the steady-state levels of paxillin mRNA. Stimulation with TGF-beta1 on a fibronectin substrate increased subsequent attachment and spreading of U-251MG cells onto fibronectin and, to a lesser extent, vitronectin, but not collagen. Our results indicate that physiologic levels of TGF-beta1 stimulate the expression of paxillin protein at the level of translation through a process that requires engagement of the fibronectin receptor, and promotes attachment and spreading of malignant astrocytoma cells on fibronectin.

Transcriptional regulation of the human osteopontin promoter: functional analysis and DNA-protein interactions.

Synthesis of cell attachment proteins and cytokines, such as osteopontin (OPN), can promote tumor cell remodeling of the extracellular matrix into an environment that promotes tumor cell attachment and migration. We investigated the transcriptional regulation of OPN in the U-251MG and U-87MG human malignant astrocytoma cell lines. Deletion and mutagenesis analyses of the OPN promoter region identified a proximal promoter element (-24 to -94 relative to the transcription initiation site) that is essential for maintaining high levels of OPN expression in the tumor cells. This element, designated RE-1, consists of two cis-acting elements, RE-1a (-55 to -86) and RE-1b (-22 to -45), which act synergistically to regulate the activity of the OPN promoter. Gel shift assays using nuclear extracts of U-251MG cells demonstrated that RE-1a contains binding sites for transcription factors Sp1, the glucocorticoid receptor, and the E-box-binding factors, whereas RE-1b contains a binding site for the octamer motif-binding protein (OCT-1/OCT-2). Inclusion of antibodies directed toward Myc and OCT-1 in the gel shift assays indicated that Myc and OCT-1 participate in forming DNA-protein complexes on the RE-1a and RE-1b elements, respectively. Our results identify two previously unrecognized elements in the OPN promoter that act synergistically to promote upregulation of OPN synthesis by tumor cells but are regulated by different transcription factors.

p125 focal adhesion kinase promotes malignant astrocytoma cell proliferation in vivo.

p125 focal adhesion kinase (p125FAK) is a cytoplasmic tyrosine kinase that is activated upon engagement of integrin cell adhesion receptors, and initiates several signaling events that modulate cell function in vitro. To determine the biologic role of p125FAK in malignant astrocytic tumor cells, U-251MG human malignant astrocytoma cells were stably transfected with p125FAK cDNA using the TET-ON system, and stable clones isolated that exhibited an estimated 5- or 20-fold increase in p125FAK expression on administration of 0.1 or 2.0 microg/ml doxycycline, respectively. In vitro studies demonstrated that induction of p125FAK resulted in a 2- to 3-fold increase in cell migration, increased p130CAS phosphorylation, localization of exogenous p125FAK to focal adhesions, and a 2-fold increase in soft agar growth. To determine the role of p125FAK in vivo, clones were injected stereotactically into the brains of scid mice. A 4.5-fold estimated increase in p125FAK expression was induced by administration of doxycycline in the drinking water. Analysis of xenograft brains demonstrated that, upon induction of p125FAK, there was a 1.6- to 2.8-fold increase in tumor cell number, and an increase in mAb PCNA-labeling of tumor cells in the absence of a change in the apoptotic index. Compared to normal brain, the expression of p125FAK was elevated in malignant astrocytic tumor biopsies from patient samples. These data demonstrate for the first time that p125FAK promotes tumor cell proliferation in vivo, and that the underlying mechanism is not associated with a reduction in apoptosis.

Extracellular FGF-1 inhibits cytoskeletal organization and promotes fibroblast motility.

Previous efforts from this laboratory have established that acidic fibroblast growth factor (FGF-1), either added exogenously or secreted as a biologically active protein, induces a transformed phenotype in primary murine fibroblasts. Experimental studies described here demonstrate that constitutive exposure to extracellular FGF-I results in reduced cell attachment to multiple ligands, inhibition of cytoskeletal organization, and reduced collagen contraction, despite no detectable change in integrin cell surface expression. In addition, FGF-1-transduced fibroblasts demonstrated a > 10-fold increase in migration, an observation correlated with increased tyrosine phosphorylation of p125FAK and p130CAS. Collectively, these results suggest that FGF-1-induced fibroblast transformation includes the involvement of specific FGF receptor-mediated signal transduction cascades targeted to cytoskeletal and focal adhesion structures.

The extracellular matrix of gliomas: modulation of cell function.

The ECM of astrocytic tumors promotes and modulates a variety of cell functions, such as cell attachment, migration, proliferation, survival, and signaling. Recent studies indicate that there are extensive and complex interactions among ECM molecules and that these can modify the function of the participating molecules, interactions between the proteoglycan, phosphacan, and the ECM protein, tenascin, being an example (63). In addition, on nonastrocytic cell types it has been shown that an integrin receptor and the cell surface proteoglycan CD44 recognize the same ECM ligand osteopontin, and thus modulate each others function (77, 86). Thus, interacting components in the ECM and cell surface receptors likely cooperate in regulating cell function and tumor invasion (59, 77, 80, 85-87, 95). As tumor cells are capable of remodeling their ECM through synthesis of ECM proteins and proteoglycans, as well as upregulating integrin receptors and proteoglycans on their cell surface, tumor cells are capable of controlling their own destiny. ECM molecules which are concentrated in blood vessels of malignant astrocytomas, such as tenascin-C and the 250-kDa CSPG (NG2), are potentially therapeutic targets.

Adenovirus-mediated delivery of antisense gene to urokinase-type plasminogen activator receptor suppresses glioma invasion and tumor growth.

The urokinase-type plasminogen activator (uPA) and uPA receptor (UPAR) play important roles in the proteolytic cascade involved in the invasiveness of gliomas and other invasive tumors. High-level expression of uPAR has been correlated with high-grade glioma cell lines and tumors We report here that down-regulating uPAR levels by antisense strategy using an adenovirus construct (Ad-uPAR) inhibited glioma invasion in Matrigel and spheroid in vitro models. sc. (U87-MG) and intracranial (SNB19) injections of Ad-uPAR-infected glioma cells did not produce tumors in nude mice. However, injection of the Ad-uPAR construct into previously established so U87-MG tumors in nude mice caused regression of those tumors. Our results support the therapeutic potential of targeting the uPA-uPAR system for the treatment of gliomas and other cancers.

Vitronectin, a glioma-derived extracellular matrix protein, protects tumor cells from apoptotic death.

Vitronectin (VN) is an extracellular matrix (ECM) protein, the synthesis of which in vivo by glioma cells correlates with tumor grade. Although the role of VN as a permissive substrate for glioma migration has been well characterized, its role in conferring a survival advantage for tumor cells has not been addressed previously. By using an in vitro assay of DNA fragmentation as a quantitative measure of apoptotic cell death, we sought to determine whether the sensitivity of two human glioma cell lines (D54 and U251) to drug-induced apoptosis could be inhibited by VN. As well, the extent to which apoptosis could be inhibited was correlated with the levels of the Bcl-2 family of proteins that are known to modulate apoptosis and chemoresistance. Results of the study were: (a) VN coatings, in a dose-dependent manner, inhibited topoisomerase (Topo)-induced apoptosis by up to 50% (optimal coating density, 500 ng/cm2); in contrast, fibronectin (FN), an ECM protein present in abundance in the brain, demonstrated no protection; (b) in a dose-response study, VN clearly conferred a survival advantage (LD50 of Topo: on VN, 120 ng/ml; on FN, 35 ng/ml); (c) the protective effect of VN was not due to enhanced cell adhesion or alterations in the cell cycle distribution; (d) both of the classic integrin receptors that bind VN (alpha(v)beta3, alpha(v)beta5) were capable of mediating this protective effect, because ligation of either of the two classic integrins conferred chemoresistance to Topo; and (e) chemoresistance observed with VN was associated with an increase in expression of two antiapoptotic proteins, Bcl-2 and Bcl-X(L), with a consequent increase in the ratios for Bcl-2:Bax and Bcl-X(L):Bax. VN, an ECM protein preferentially expressed at the tumor-brain interface in vivo, may confer a survival advantage to glioma cells at the advancing tumor margin and may thus, in part, underlie the high level of tumor recurrence at this interface.

Glioma migration can be blocked by nontoxic inhibitors of myosin II.

Anaplastic gliomas are infiltrative tumors, and their ability to migrate through normal brain contributes to their highly malignant behavior. Invasion of brain requires cell motility, which in turn depends on the activity of the cytoskeleton. A cytoskeletal component central to this process is myosin II, the cytoplasmic analogue of smooth and skeletal muscle myosin. Myosin II activity is regulated by the enzyme myosin light chain kinase, which activates myosin II by phosphorylating it on its regulatory light chain. We have investigated the role of myosin II in glioma motility and invasiveness by examining the effects of two inhibitors of myosin light chain kinase, ML7 and KT5926. Both drugs are potent inhibitors of both glioma motility, as measured by a scrape motility assay, and an in vitro haptotaxis assay. The inhibition of in vitro haptotaxis follows the dose-response relationship expected for competitive inhibition of myosin light chain kinase by these drugs and is seen at drug concentrations that are nontoxic. These results highlight the important role that myosin II contributes to glioma invasiveness and suggest that it may serve as a target in future strategies at blocking invasion by these tumors.

Retinoic acid alters the mechanism of attachment of malignant astrocytoma and neuroblastoma cells to thrombospondin-1.

Based on the hypothesis that the attachment of neuroectodermal cells to thrombospondin-1 (TSP-1) may affect tumor spread and play a role in the anti-tumor effects of retinoic acid, we investigated the expression of TSP-1 in these cells in situ and the effect of retinoic acid on the morphology of TSP-1-adherent neuroblastoma (SK-N-SH) and malignant astrocytoma (U-251MG) cells in vitro. TSP-1-adherent SK-N-SH cells demonstrated process outgrowth, with further neuronal differentiation after retinoic acid treatment, consistent with the in situ studies showing that TSP-1 expression occurs in a differentiation-specific manner in neuroblastic tumors. TSP-1-adherent U-251MG cells failed to spread; however, after retinoic acid treatment the cells demonstrated broad lamellipodia containing radial actin fibers and organization of integrins alpha3beta1 and alpha5beta1 in clusters in lamellipodia and filopodia. The attachment of both SK-N-SH and U-251MG cells to TSP-1 was found to be mediated by heparan sulfate proteoglycans, integrins, and the CLESH-1 adhesion domain first identified in CD36. Heparin and heparitinase treatment inhibited TSP-1 attachment. Integrins alpha3beta1 and alpha5beta1 mediated TSP-1 attachment of SK-N-SH cells, and integrins alpha3beta1, alpha5beta1, and alphavbeta3 mediated TSP-1 attachment of U-251MG cells. Attachment was dependent on the RGD sequence which is located in the carboxy-terminus of TSP-1. Treatment with a pharmacologic dosage of retinoic acid altered the TSP-1 cell adhesion mechanism in both cell lines in that neither heparin nor micromolar concentrations of the RGD peptide inhibited attachment; after treatment, attachment was inhibited by the CSVTCG peptide located in the type I repeat domain of TSP-1 and a recombinant adhesion domain (CLESH-1) from CD36. Expression of CD36 was found in the retinoic acid-treated U-251MG cells. These data indicate that neuroectodermally derived cells utilize several mechanisms to attach to TSP-1, and these are differentially modulated by treatment with retinoic acid. These data also suggest that the CSVTCG sequence of TSP-1 modulates or directs cytoskeletal organization in neuroblastoma and astrocytoma cells.

Biological significance of the expression of urokinase-type plasminogen activator receptors (uPARs) in brain tumors.

The urokinase-type plasminogen activator receptor (uPAR) plays a critical role in the regulation of cell-surface plasminogen activation in several physiological and pathological conditions. Recent evidence suggests that the uPAR is also involved in processes that are not related to plasminogen activation, including cell adhesion and transmission of extracellular signals across the plasma membrane. The uPAR influences cell migration and spreading both in vivo and in vitro through the cell-surface activation of plasminogen. The uPAR can bind to vitronectin, an adhesive extracellular matrix protein that contains the Arg-gly-Asp (RGD) cell adhesion domain and that serves as a ligand for several integrin receptors. uPAR also forms complexes with (1, (2, and (3 integrins, thereby allowing mutual interactions and regulation between cell adhesion and proteolysis. Recently, uPAR has been shown to have strong prognostic value for predicting disease recurrence and overall survival in certain types of cancer. We discuss here the biological significance of uPAR in the glioblastoma invasion process. Strong correlations found between elevated uPAR levels in glioblastoma cells and tumor invasiveness have led to uPAR being selected as a target for therapy in experimental animal models. Using antisense vectors to down regulate uPAR expression at the level of the mRNA and protein in glioblastoma cells, has been shown to inhibit tumor formation in nude mice. These results provide a potential basis from which to develop novel therapeutic strategies to direct the expression of antisense uPAR and to evaluate the efficiency of this technique for cancer gene therapy in patients with brain tumor.

The role of integrins in the malignant phenotype of gliomas.

Integrins are cell surface receptors that mediate the physical and functional interactions between a cell and its surrounding extracellular matrix (ECM). Expressed as heterodimers, the specific alpha or beta chains that constitute the integrin receptor determine the repertoire of ECM proteins to which a specific integrin may bind (table 1). While classically, the role ascribed to integrins has been that of anchoring cells to the ECM, the more contemporary spectrum of integrin function greatly exceeds that of mere cell adhesion. Recent reports have demonstrated that the interaction between the ECM and cell surface integrins leads to intracellular signaling events that affect cell migration, proliferation, and survival, which in the context of neoplastic cells, can translate directly into the malignant phenotype (1). Indeed, the role of specific integrins in tumorigenesis has been demonstrated in numerous cancer types (table 2). In primary tumors of the nervous system, the contribution of integrins to the malignant phenotype of gliomas has been an area of significant attention and research in numerous laboratories, including that of ours. As illustrated in table 3, several integrins have been identified as being of key importance in glioma biology. In this article, we review the current knowledge of how these integrins influence the malignant characteristics of gliomas and, as such, how these cell surface receptors may thus represent potential targets in the design of future therapeutics for patients afflicted with gliomas.

Malignant astrocytoma cell attachment and migration to various matrix proteins is differentially sensitive to phosphoinositide 3-OH kinase inhibitors.

Phosphoinositide 3-OH kinase (PI3-K) has been shown to play an important role in the signaling pathway necessary for cytoskeletal reorganization in non-astrocytic cells. We investigated the role of PI3-K in U-251 MG human malignant astrocytoma cell adhesion and migration. Attachment of U-251 MG cells to vitronectin, fibronectin, laminin, and collagen was inhibited in a concentration-dependent manner by two specific inhibitors of PI3-K (Wortmannin and LY294002). Attachment to vitronectin, fibronectin, and laminin was more sensitive to inhibition of PI3-K (45% inhibition at 10 nM Wortmannin) than attachment to collagen (25% inhibition at 100 nM Wortmannin). Similarly, migration toward these substrates showed differential sensitivity to inhibition. Attachment of the cells to these matrix proteins resulted in an increase in PI3-K activity, as compared to that of cells in suspension, with attachment to vitronectin resulting in the greatest increase in PI3-K activity. p125 focal adhesion kinase (p125FAK) was found to co-immunoprecipitate with PI3-K from the NP40-soluble cell fraction of a 1% NP40 detergent lysate of cells in the early stages of adhesion to vitronectin and fibronectin, but not during adhesion to collagen. The expression of p125FAK protein and level of phosphorylation were similar on adherence to all three substrates. These data indicate that the sensitivity of U-251MG cell attachment and migration to PI3-K inhibitors is substrate-dependent, and that complex formation of PI3-K and p125FAK correlates with this sensitivity to PI3-K inhibitors. Our data suggest a role for PI3-K and p125FAK complex formation in PI3-K activation.

Vitronectin expression in differentiating neuroblastic tumors: integrin alpha v beta 5 mediates vitronectin-dependent adhesion of retinoic-acid-differentiated neuroblastoma cells.

The metastatic potential of undifferentiated neuroblastomas is typically lost when differentiation into ganglioneuroblastomas occurs spontaneously or is induced. Cell adhesion may play a role in metastasis, and we have shown recently that expression of integrin alpha v beta 5 protein and mRNA is up-regulated in ganglioneuroblastomas in vivo. To investigate whether interactions of alpha v beta 5 with matrix components play a role in the loss of metastatic potential, we used immunohistochemical and in situ hybridization to analyze neuroblastic tumors at various stages of differentiation for expression of the alpha v beta 5 ligands, vitronectin and osteopontin, and determined the ability of vitronectin to promote attachment and neurite outgrowth in vitro in a retinoic-acid-differentiated neuroblastoma cell model. We found that vitronectin, but not osteopontin, was expressed in 5 of 5 ganglioneuroblastomas but was absent or weakly expressed in 6 of 6 undifferentiated neuroblastomas. Neuronal cell vitronectin was detected in 7 of 9 ganglioneuromas, 5 of 8 peripheral ganglia, and 14 of 21 adrenal gland medullae, confirming expression of vitronectin in mature peripheral neurons. In vitro, vitronectin promoted attachment of both undifferentiated and retinoic-acid-differentiated neuroblastoma cells, which was inhibited 20 and 60%, respectively, by monoclonal antibody anti-integrin alpha v beta 5. Vitronectin-promoted neurite outgrowth of retinoic-acid-differentiated neuroblastoma cells was not inhibited by monoclonal antibody anti-alpha v beta 5. These data suggest that the synthesis of vitronectin and the ability of integrin alpha v beta 5 to mediate vitronectin adhesion on retinoic-acid-differentiated neuroblastoma cells may promote differentiation of neuroblastoma cells in vivo.

Ligation of integrin alpha5beta1 is required for internalization of vitronectin by integrin alphavbeta3.

Remodeling of the matrix by tumor cells is necessary for tumor invasion. We have shown previously that malignant astrocytomas, in contrast to normal astrocytes, synthesize vitronectin and express integrins alphavbeta3 and alphavbeta5. The activity states of these two integrins are differentially controlled. Thus, we investigated the regulation of the activity of integrins alphavbeta3 and alphavbeta5 with regard to their role in vitronectin internalization in U-251MG astrocytoma cell monolayers adherent to fibronectin, collagen, or laminin in serum-free conditions. Binding of [125I]vitronectin occurred in a specific, saturable manner that was partially inhibitable by monoclonal antibodies (mAbs) specific for integrins alphavbeta3 or alphavbeta5. Specific, lysosomally-mediated degradation of [125I]vitronectin was detectable at 1 h and increased over the 24-h assay period. The cell substrate affected the rate of turnover of [125I]vitronectin, which was 3.0 ng/min for cells plated on fibronectin but 0.35 ng/min for cells plated on collagen. Furthermore, although mAbs specific for either integrin alphavbeta3 or alphavbeta5 inhibited degradation (30%; combined effect 70%) of [125I]vitronectin by cells plated on fibronectin, only mAb anti-alphavbeta5 inhibited degradation (70-90%) by cells plated on collagen or laminin. To determine the requirement for integrin alpha5beta1 ligation in order for integrin alphavbeta3 to internalize its ligand, cells were plated on mAbs anti-integrin alpha5 or anti-integrin alpha3. When plated on mAb anti-alpha5, mAbs anti-alphavbeta3 and anti-alphavbeta5 both inhibited degradation. However, when plated on mAb anti-alpha3, mAb anti-alphavbeta3 had no effect whereas mAb anti-alphavbeta5 inhibited degradation. These data indicate that a signal from integrin alpha5beta1 is necessary for integrin alphavbeta3 to internalize vitronectin, whereas integrin alphavbeta5 constitutively internalizes vitronectin.

Expression of integrin alpha v beta 3 in small blood vessels of glioblastoma tumors.

Angiogenesis, which promotes tumor proliferation and invasiveness, is mediated by integrin-dependent cell adhesion mechanisms and requires expression of integrin alpha v beta 3. To determine whether integrin alpha v beta 3 is expressed in the small blood vessel hyperplasia characteristic of glioblastoma tumors, we explored the mRNA and protein expression of integrin alpha v, beta 3, beta 5, and beta 1 subunits in small blood vessels in gliomas of various grades and nontumorous brain biopsies. Antisense beta 3 and beta 5 riboprobes hybridized to small blood vessels (endothelial and adjacent mesenchymal cells) in 11 of 11 glioblastomas and 2 of 4 anaplastic astrocytomas, but failed to hybridize to small blood vessels in 12 non-neoplastic specimens and 4 low-grade astrocytomas. In contrast, antisense alpha v and beta 1 riboprobes hybridized to small blood vessels in all of the biopsies. The expression of integrin alpha v beta 3 protein in these vessels was determined by immunohistochemical analysis using anti-integrin subunit-specific antibodies. The findings were similar to those obtained using in situ hybridization: beta 3 subunit protein was detected on small blood vessels in 9 of 12 glioblastoma and 2 of 4 anaplastic astrocytomas, whereas it was not detected in small blood vessels of 10 non-neoplastic brains and 4 low-grade astrocytomas. In contrast, the alpha v and beta 1 subunit proteins were detected in small blood vessels in all of the brain biopsies, whereas the beta 5 subunit protein could not be detected. These data suggest that integrin alpha v beta 3 is expressed in small blood vessels of glioblastoma tumors.

Stage-specific expression of integrin alphaVbeta3 in neuroblastic tumors.

The ligand specificity of the integrin cell adhesion receptors probably determines the ability of specific integrins to promote tumor cell proliferation and metastasis. Therefore, we compared the expression of integrin alphaVbeta3, a promiscuous receptor that binds with high affinity to numerous cell matrix proteins, with the expression of integrin alphaVbeta5 and the integrin beta 1 subunit (which pairs with multiple alpha subunits) in neuroblastic tumors at various stages of differentiation. Undifferentiated neuroblastoma tumors rapidly invade and metastasize, whereas ganglioneuroblastomas rarely metastasize. Differentiating neuroblastomas are associated with an intermediate prognosis. Paraffin sections of neuroblastic tumors at various stages of differentiation obtained at biopsy from 17 patients were hybridized with antisense integrin subunit-specific alphaV, beta3, beta1, and beta5 riboprobes. All neuroblastic tumors and seven adrenal glands obtained at autopsy were analyzed immunohistochemically with antibodies directed toward the alphaV, beta3, beta1, and beta5 subunits. The alphaV subunit was expressed in neuroblastic tumors independent of the stage of differentiation, although mRNA and protein expression were generally weak in ganglioneuroblastomas, and was also detected in adrenal gland medullae. The beta1 subunit was detected in most neuroblastic tumors independent of the stage of differentiation as well as in adrenal gland medullae. In contrast, the beta3 subunit, which was not expressed in adrenal gland medullae, was expressed at the protein and mRNA levels in undifferentiated neuroblastomas (six of seven and seven of seven, respectively) but was not expressed in neuroblasts or ganglion cells in ganglioneuroblastomas (one case weakly positive out of five). The beta 5 subunit was expressed at the protein (five of five) and mRNA (four of five) levels in the ganglion cells of ganglioneuroblastomas and, although mRNA for this subunit was detectable in undifferentiated tumors, the protein was not detectable. The expression of integrin alphaVbeta3 in undifferentiated neuroblastomas may contribute to the rapid growth of these tumors and their tendency to metastasize.

Up-regulation of urokinase and urokinase receptor genes in malignant astrocytoma.

To understand the role of urokinase (u-PA) and the urokinase receptor (u-PAR) in malignant astrocytoma cell invasion of normal brain, astrocytic expression of u-PAR and u-PA mRNAs were analyzed by riboprobe in situ hybridization in astrocytoma and non-neoplastic brain biopsies. In eight of eight malignant astrocytomas (glioblastomas), u-PAR and u-PA mRNA expression was demonstrated, whereas in seven non-neoplastic brain biopsies, u-PAR and u-PA mRNAs were not expressed. In one of four low grade and all anaplastic astrocytomas u-PAR mRNA was expressed, although u-PA mRNA was undetectable. Consistent with the mRNA detection, u-PAR and u-PA proteins were expressed by malignant astrocytes in five of five glioblastoma biopsies. To study the tumor margin, U-251MG glioblastoma cells were propagated intracerebrally in a severe combined immunodeficient mouse xenograft (28 days), and u-PA mRNA was found to localize predominantly to the leading tumor edge, whereas u-PAR mRNA was expressed throughout the tumor. Furthermore, adherent human U-251MG glioblastoma cells in vitro expressed u-PAR and u-PA proteins, which localized to sites of integrin alpha nu beta 3 cell-matrix contacts. These data indicate that co-expression of u-PAR and u-PA mRNAs and proteins marks the malignant astrocyte phenotype and that u-PA bound to u-PAR may play a role in glioblastoma cell invasion of normal brain by virtue of its expression at the leading tumor edge.

Cerebral microenvironment influences expression of the vitronectin gene in astrocytic tumors.

Expression of the vitronectin gene was detected in advanced human astrocytoma by in situ hybridization, whereas vitronectin mRNA was undetectable in low grade tumors or in normal adult brain, indicating that vitronectin is a marker of malignant astrocytoma. We established a model of human astrocytoma by transplanting U-251MG human astrocytoma cells intracerebrally into acid mice (C.B.17 severe combined immunodeficient mice). In this model, tumors progressed rapidly and vitronectin mRNA was preferentially detected at the invading tumor margins, i.e. where tumor cells were adjacent to the normal brain tissue. Surprisingly, when U-251MG cells were injected subcutaneously into scid mice, vitronectin mRNA was undetectable throughout the tumor. Moreover, vitronectin mRNA or protein could not be detected among these cells in culture under a wide variety of growth conditions. These findings demonstrate that the cerebral microenvironment influences the expression of the vitronectin gene in malignant astrocytoma. Importantly, the vitronectin binding integrins alpha v beta 3 and alpha v beta 5 localized to distinct sites within these tumors, with beta 3 mRNA synthesized among invading cells, and alpha v and beta 5 mRNAs detected throughout the tumor. In vitro, both of these receptors were capable of promoting adhesion and invasion of astrocytoma cells on a vitronectin substratum. These findings implicate the expression of the vitronectin gene as a contributing factor to the biological behavior of astrocytomas within the cerebral microenvironment.