Autocrine = Factors That Sustain Glioma Invasion and Paracrine Biology

http://www.medscape.com/ =

To Print: Click = your browser's=20 PRINT button.

NOTE: To view = the article=20 with Web enhancements, go to:
http://www.medscape.c= om/viewarticle/566941

Autocrine Factors That Sustain Glioma Invasion = and=20 Paracrine Biology in the Brain Microenvironment

Dominique B. Hoelzinger; Tim Demuth; = Michael E.=20 Berens

J Natl Cancer Inst. 2007;99(21):1583-1593.=20 =A92007 Oxford University Press

Posted 12/17/2007

Abstract and Introduction

Abstract

Invasion is a defining hallmark of glioblastoma multiforme, = just as=20 metastasis characterizes other high-grade tumors. Glial tumors = invariably=20 recur due to the regrowth of invasive cells, which are unaffected = by=20 standard treatment modalities. Drivers of glioma invasion include=20 autocrine signals propagated by secreted factors that signal = through=20 receptors on the tumor. These secreted factors are able to diffuse = through=20 the peritumoral stroma, thereby influencing parenchymal cells that = surround the tumor mass. Here we describe various autocrine = motility=20 factors that are expressed by invasive glioma cells and explore = the=20 effects that they may have on normal cells present in the path of=20 invasion. Conversely, normal brain parenchymal cells secrete = ligands that=20 can stimulate receptors on invasive glioma cells and potentially=20 facilitate glioma invasion or create a permissive microenvironment = for=20 malignant progression. Parallel observations have been made for = solid=20 tumors of epithelial origin, in which parenchymal and stromal = cells either=20 support or suppress tumor invasion. Most autocrine and paracrine=20 interactions involved in glioma invasion constitute known = signaling=20 systems in stages of central nervous system development that = involve the=20 migration of precursor cells that populate the developing brain. = Key=20 paracrine interactions between glioma cells and the brain = microenvironment=20 can influence glioma pathobiology and therefore contribute to its = poor=20 prognosis. Current therapies for glioma that could have an impact = on=20 paracrine communication between tumors and normal cells are = discussed. We=20 suggest that cells in the normal brain parenchyma be considered as = potential targets for adjuvant therapies to control glioma growth = because=20 such cells are less likely to develop resistance than glioma = cells.

Introduction

Tumors of the central nervous system can arise from several = different=20 cellular lineages that include glia, such as astrocytes and=20 oligodendrocytes. Astrocytic tumors (astrocytomas) fall into two = distinct=20 categories based on how they interact with their immediate=20 microenvironment: diffuse and localized astrocytomas. Localized=20 astrocytomas have a circumscribed pattern of growth and limited = invasive=20 potential, whereas diffuse astrocytomas are characterized by their = cellular infiltration of the peritumoral margin and disperse to = distant=20 sites, regardless of tumor grade. There are three grades of = diffuse=20 astrocytic tumors:^[1] astrocytoma (World Health = Organization=20 [WHO] grade II), anaplastic astrocytoma (WHO grade III), and = glioblastoma=20 multiforme (WHO grade IV). Although astrocytic tumors of all three = grades=20 are invasive, the higher-grade tumors (i.e., grades III and IV) = are=20 progressively more proliferative and result in a shorter time to = tumor=20 recurrence than lower-grade tumors. All glial tumors (except for = pilocytic=20 astrocytomas [WHO grade I]) recur at high frequency after = treatment with=20 surgery, radiation, and chemotherapy and are essentially = incurable.=20 Recurrence is due mainly to the highly invasive behavior of even = low-grade=20 (i.e., WHO grade II) astrocytomas, which allows the active egress = of=20 glioma cells from the main tumor mass into the surrounding normal = brain=20 tissue. Therefore, invasive glioma cells are essentially beyond = the reach=20 of current therapies, which are localized to the tumor and = immediate=20 peri-tumoral environment. However, the survival of patients with = tumors=20 with similar degrees of invasiveness (such as low-grade and = high-grade=20 astrocytoma) differs dramatically (median survival ranges from = 10-15 years=20 in patients with low-grade astrocytomas to 15 months in patients = with=20 glioblastoma multiforme) for reasons that are still unknown.

Glioblastoma multiforme is the most common and most malignant = of all=20 glial tumors: it accounts for 60% of all glial tumors and occurs = at a=20 frequency of 5 cases per 100000 people.^[2] Although = molecular=20 markers for glioblastoma multiforme have helped to identify = patients=20 responsive to current therapies,^[3] the overall = survival of=20 responsive patients has not changed substantially in the last 20=20 years.^[4] Glioblastoma multiforme cell invasion is not = random;=20 transformed glial cells preferentially invade along anatomic = features such=20 as myelinated axons, vascular basement membranes, and the=20 subependyma,^[5] which suggests that they interact with = specific=20 features in their immediate microenvironment. These interactions = may also=20 play a role in patient survival: patients with glioblastoma = multiforme=20 located in the deep gray matter survive statistically = significantly longer=20 than patients with lobar gliomas (regardless of tumor size at=20 diagnosis).^[6] It has been theorized^[6] that = these=20 differences in survival reflect the differential invasion capacity = of=20 glioma cells that are located in different microenvironments. = Indeed,=20 mounting evidence indicates that normal peritumoral cells play = active=20 roles in tumor progression in other = carcinomas.^[7,8]

Glioblastoma multiforme invasion is likely triggered by signals = that=20 prompt tumor cells to egress from the tumor mass, including = signals that=20 are activated by an acidic and hypoxic tumor environment. The best = documented of these signals is the transcription factor hypoxia = inducible=20 factor-1, which increases the expression of invasion-related genes = such as=20 proteases and transforming growth factor-α = (TGF-α).^[9] As cells=20 exit the tumor mass, they encounter a different set of = microenvironmental=20 cues, including a disrupted extracellular matrix and normoxic = physiologic=20 conditions. Invasion maintenance may depend on autocrine motility = loops=20 and their paracrine effects. As invading glioblastoma cells = transit=20 through the brain parenchyma, they interact with a changing = landscape of=20 extracellular stimuli that perpetuate the trajectory of invasion = much like=20 the signaling systems secreted during neurodevelopmental migration = events.=20 The directional motility of neural precursors, glial precursors, = and=20 nascent axons through the embryonic central nervous system is = achieved and=20 maintained by autocrine signals that are secreted at the leading = edges of=20 these cells in response to paracrine cues. For example, embryonic=20 oligodendrocyte migration depends on platelet-derived growth = factor (PDGF)=20 secretion as well as on expression of the PDGF receptor=20 (PDGFR).^[10] The highly orchestrated migration of = neurons of=20 the developing brain is coordinated by autocrine secretion of = ligands such=20 as pleiotrophin (PTN) and amphoterin and their receptors syndecan, = receptor protein tyrosine phosphatase beta/zeta (RPTPζ), and = renal tumor=20 antigen.^[11,12] It is reasonable to anticipate that = invasive=20 glioblastoma multiforme cells re-engage developmental migration = mechanisms=20 to affect motility and that the microenvironment plays an active = role in=20 glioma invasion.

This review describes candidate autocrine-paracrine motility = signals in=20 the setting of glioma pathobiology and focuses on how these = specific=20 ligand-receptor systems (summarized in Table 1 ) may contribute to the invasive = phenotype of=20 glioblastoma multiforme. We do not address autocrine signaling = systems=20 that promote invasion but that do not have documented paracrine = effects on=20 cells in the brain parenchyma (such as the receptor tyrosine = kinases erb-2=20 and erb-3 and neuregulin, autocrine motility factor and its = receptor=20 angiopoietin-2, as well as the angiogenic factor cysteine-rich 61, = the=20 hormone ghrelin, and their respective signal transducing systems). =

Autocrine Signals in Glioma Invasion

Glioblastoma multiforme cells secrete several factors that = result in=20 autocrine motility signaling. These factors have been the focus of = research that seeks to understand the basis for the pervasive = invasive=20 capacity of diffuse astrocytomas as well as research to identify = points of=20 vulnerability in invasive cells that might yield specific = anti-invasive=20 therapies. In this section, we list representative autocrine = motility=20 signaling systems that play a role in enhancing glioma = invasion.

Epidermal growth factor receptor (EGFR) signaling is one of the = salient=20 drivers of glioma proliferation. However, autocrine proliferation=20 signaling can also stimulate glioma invasion. In 1997, Chicoine = and=20 Silbergeld^[13] speculated that the growth factors EGF = and TGF-α=20 might also be the motogens that drive glioblastoma multiforme = invasion.=20 Recently, both TGF-α and heparin-binding EGF (HB-EGF) have = been confirmed=20 to participate in autocrine loops in glioma.^[14] = TGF-α and=20 HB-EGF secretion is likely to have pleiotropic effects on the=20 microenvironment of the tumor, given that EGFR is expressed by = neurons,=20 astrocytes, microglia, and reactive astrocytes. The most common = hallmark=20 of glioblastoma multiforme biology is aberrant EGF/EGFR signaling, = followed by dysregulation in PDGF/PDGFR signaling. PDGF and PDGFR=20 expression increases are observed during progression of astrocytic = tumors=20 from lower grades to glioblastoma multiforme.^[10] = Although PDGF=20 and its receptor PDGFR are most frequently linked to the = proliferation of=20 glial tumors and to angiogenesis,^[10,15] several=20 reports^[16,17] suggest that signaling through this = autocrine=20 pair results in increased glioblastoma multiforme cell motility in = vitro.

Glioblastoma multiforme cells also secrete hepatocyte growth=20 factor/scatter factor (HGF/SF), which binds to its receptor, the = Met=20 tyrosine kinase; the expression of both players in this = ligand-receptor=20 system increases with increasing grade of = malignancy.^[18]=20 Although the proliferative and angiogenic potentials of this = autocrine=20 loop have been well documented,^[19] HGF/SF expression = is also=20 associated with invasion. Indeed, both Met and HGF/SF are = expressed in=20 invasive cells from primary human glioma samples.^[18] = In=20 addition, an in vitro study^[20] that analyzed the = effect of a=20 panel of chemokines on glioma motility found that HGF/SF was the = most=20 potent chemotactic factor for the cell lines tested, suggesting = that=20 Met-HGF/SF signaling could also represent an autocrine motility = loop. A=20 number of cell types in the brain express Met ( Table 1 ) and may therefore be responsive to=20 glioma-generated HGF/SF signaling.

In addition to kinases, phosphatases can also drive = glioblastoma=20 invasion. RPTPζ is expressed in the central nervous system = during=20 development^[21] and was initially associated with the=20 modulation of neuronal and glial cell adhesion.^[22] = Although=20 various cell adhesion molecules and components of the = extracellular matrix=20 are known to interact with RPTPζ,^[23,24] PTN is = the only=20 soluble ligand that has been shown to bind = RPTPζ.^[25] PTN=20 binding to RPTPζ induces inactivation of the phosphatase = activity of=20 RPTPζ,^[26] which results in a net increase in the=20 phosphorylation of beta-catenin, beta-adducin, and Fyn, proteins = that=20 increase cell motility by disrupting the stability of the=20 cytoskeleton.^[27] Gene expression profiling has = revealed=20 overexpression of the genes encoding RPTPζ and PTN in a = subset of=20 glioblastoma multiforme tumors compared with normal = brain.^[28]=20 Migration assays confirmed that PTN stimulated the motility of = these=20 glioblastoma multiforme cell lines in an RPTPζ-dependent=20 manner.^[29]

Chemokines also play a role in glioblastoma invasion. The = chemokine=20 (CXC motif) receptor 4 (CCR4) was initially found to be = overexpressed in=20 glioblastoma multiforme compared with normal brain tissue during a = cDNA=20 microarray screen;^[30] in this study it was thought to = be a=20 proliferation marker. Subsequent studies confirmed expression of = CXCR4=20 protein in glioblastoma multiforme, as well as persistent = co-expression of=20 its ligand chemokine (C-X-C motif) ligand 12 = (CXCL12).^[31-33]=20 CXCL12 stimulates the invasiveness of glioma cell lines in=20 vitro.^[34,35] The increased transcription of both CXCR4 = and=20 CXCL12 in laser-capture microdissected invasive cells in a rat C6 = glioma=20 model provided the first clue that these proteins mediate glioma = cell=20 invasiveness in vivo.^[34] This chemokine-receptor pair = also=20 plays a role in breast cancer progression^[36] and = metastasis=20 and has been shown to contribute to the homing of breast cancer = cells to=20 secondary organs such as the lung.^[37]

More complex autocrine motility signals are represented by a = three-part=20 signaling system that involves glial cell line-derived = neurotrophic factor=20 (GDNF), which signals through two extracellular proteins, the GDNF = receptor α1 (GDNFRα1) (the ligand-binding moiety) and = its signal=20 transducing partner, rearranged during transfection protooncogene = (c-Ret).=20 This autocrine trio is overexpressed in glioblastoma multiforme = compared=20 with normal brain as well as in the rat C6 glioma cell line = compared with=20 rat brain.^[38] Stimulation of rat C6 glioma cells that=20 overexpressed GDNFRα1 and GDNF resulted in increased cell = migration=20 compared with unstimulated cells, whereas a low-grade glioma cell = line=20 that expressed lower amounts of GDNFRα1 reacted to = stimulation with a=20 modest increase in GDNF-dependent cell motility.^[39] = GDNF,=20 which is nearly universally expressed throughout the brain, could = foster=20 the maintenance of glioma cell invasion through its receptors that = are=20 expressed on invasive glioma cells.

In addition to conventional autocrine loops, there are = autocrine=20 motility signals that engage in phospholipid signaling. One such = signaling=20 system involves autotaxin (ATX), lysophosphatidic acid (LPA), and = the LPA=20 receptors (LPA1, LPA2, LPA3, and LPA4). ATX is a secreted = motogenic=20 protein that was first identified in conditioned medium from a = human=20 melanoma cell line;^[40] ATX is also expressed by other = tumors=20 that have an invasive phenotype, such as breast = cancer,^[41]=20 non-small-cell lung cancer,^[42] and glioblastoma=20 multiforme.^[43,44] The motogenic properties of ATX are = linked=20 to its lysophospholipase D activity: it catalyzes the production = of LPA,=20 signaling through LPA receptors present on the surface of invasive = tumor=20 cells. ATX is expressed by all grades of diffuse astrocytomas and = is=20 transcribed at higher levels in glioblastoma multiforme cells that = have=20 invaded the brain parenchyma than in cells that remain in the = tumor mass.=20 The ATX protein is detectable by immunohistochemistry in all = astrocytic=20 tumors but not in normal astrocytes, neurons, or mature=20 oligodendrocytes.^[43] ATX is secreted by a variety of=20 glioblastoma multiforme cell lines and, when glioblastoma = multiforme cells=20 are exposed to the lysophospholipase D substrate = lysophosphatidylcholine,=20 ATX promotes cell motility in an LPA1-dependent = manner.^[44] ATX=20 also promotes invasion of glioblastoma multiforme cells into rat = brain=20 slices ex vivo.^[45] This finding echoes the earlier = finding of=20 an LPA dose-dependent increase in migration of glioblastoma = multiforme=20 cell lines.^[46] Given that LPA1 is expressed = ubiquitously in=20 the brain,^[47] it is likely that every cell type in the = central=20 nervous system would be able to respond to LPA generated by = glioblastoma=20 multiforme cells during invasion.

Another bioactive phospholipid that is involved in glioma = invasion,=20 sphingosine-1-phosphate (S1P), is secreted by glioblastoma = multiforme=20 cells that express sphingosine kinase 1 (SphK1) and sphingosine = kinase=20 2.^[48] Van Brocklyn et al.^[48] reported that = glioblastoma patients whose tumors expressed high levels of SphK1 = had=20 statistically significantly shorter survival than patients whose = tumors=20 expressed low levels of SphK1 (102 versus 357 days; P =3D = .002). S1P=20 induces cell motility in glioblastoma multiforme cell lines that = express=20 S1P receptor-1 and S1P receptor-3; S1P receptor-2 is described as = a=20 receptor that inhibits cell motility when = activated.^[49,50] The=20 brain has the highest levels of S1P of various tissues=20 tested;^[51] however, the only cells in the brain = parenchyma=20 that are known to secrete S1P are cerebellar=20 astrocytes.^[52]

Paracrine Interactions Between Invasive Tumor Components and=20 Nonneoplasic Cells of the Central Nervous System

Microglia and Tumor-Associated Macrophages

Tumors secrete a number of factors that recruit inflammatory = cells to=20 the tumor periphery, including cytokines, chemokines, = colony-stimulating=20 factors, and growth factors. The effects of these immune cells on = the=20 neoplastic process range from the initiation of malignancies to = the=20 driving of tumor progression through promotion of angiogenesis,=20 lymphangiogenesis, proliferation, invasion, and = metastasis.^[8]=20 Whereas in extracranial solid tumors it is the tumor-associated = macrophage=20 that is closely associated with promotion of=20 malignancies,^[8,53] in intracranial tumors it is = microglia that=20 are the main responders to tumor burden. Both immune cell types = respond to=20 nearly identical set of tumor-secreted factors (e.g., EGF, HGF/SF, = tumor=20 necrosis factor-α, and cytokines) by expressing distinct = functional=20 programs that recapitulate normal developmental roles to promote = tumor=20 progression.^[8] However, in histologic sections of the=20 brain/tumor margin, microglia are present at the invasion = interface at=20 threefold higher numbers than tumor-associated = macrophages.^[54]=20 The precise function of peritumoral microglia is not fully = understood;=20 they may merely represent a reactive immune response to the tumor = or they=20 may actively influence tumor growth or invasion through cytokine=20 secretion^[55] (Fig. 1). Microglia facilitate = glioblastoma=20 multiforme invasion in an ex vivo rat brain slice model by = secreting=20 inactive matrix metalloprotease-2 (pro-MMP2).^[56] = Glioblastoma=20 multiforme cells, on the other hand, produce soluble factors that = activate=20 pro-MMP2, thereby increasing the amount of active MMP2, which = promotes=20 glioma invasion.^[56] This intercellular cooperation is=20 reflective of the finding that cocultures of glioblastoma = multiforme cells=20 and astrocytes display increased MMP2 activity,^[57] = which=20 suggests that astrocytes are capable of playing a part in = glioblastoma=20 multiforme invasion. In addition, microglia secrete multiple = cytokines,=20 including tumor necrosis factor-α and interleukin 1β = (IL-1β). The=20 microglial contribution of interleukin 1β to the glioma = microenvironment=20 can elicit glioma transcription of TGF-β, which, in turn, = results in=20 immunosuppression,^[58] increased expression and = secretion of=20 vascular endothelial growth factor (VEGF; which is involved in=20 angiogenesis), EGFR (which is involved in cell proliferation), and = matrix=20 metalloproteinase 9 (MMP9; which is involved in = invasion),^[59]=20 facilitating all three manifestations of the malignant phenotype = of glioma=20 cells.

Figure 1.
Glioblastoma multiforme tumor core and invasive cells in the = context=20 of cells found in the brain parenchyma. Invasive glioblastoma = multiforme=20 cells (dark nuclei) secrete various autocrine motility factors = that=20 maintain the process of invasion. Cells in the central nervous = system=20 express cognate receptors for ligands secreted by glioblastoma=20 multiforme cells, which enable communication between = glioblastoma=20 multiforme cells and the normal cells in the path of invasion.=20 Conversely, normal parenchymal cells secrete growth factors,=20 neurotrophins, chemokines, axonal guidance factors, and = bioactive lipids=20 that have the capacity to play a paracrine role in the = enhancement of=20 tumor invasion. Nearly all of the motogenic ligands that augment = glioblastoma multiforme invasion have distinct functions in=20 neurodevelopmental phases of migration, directing the movement = of neural=20 and glial progenitor cells from germinal centers to the final = functional=20 destinations.

There are several opportunities for cross-communication between = microglia and the invasive tumor cell. Microglia respond to EGF = secreted=20 by tumor cells with increased motility, but not by increasing = their=20 proliferation,^[60] which suggests that microglia may be = directed to lesions in the central nervous system via paracrine = signaling.=20 Activated microglia also secrete EGF, making it likely that = glioblastoma=20 multiforme cells that express EGFR can, in turn, respond to the = presence=20 of microglia with heightened invasion. This scenario is likely = also true=20 for Met-HGF/SF signaling between glioblastoma multiforme and = microglia,=20 given that both cell types express the receptor and the=20 ligand^[61] and respond to HGF/SF by increasing their=20 migration.^[55] Because microglia are frequently found=20 surrounding the tumor mass, it is possible that HGF/SF causes = microglial=20 migration toward the tumor. The same could also hold true for GDNF = and=20 GDNFR, which are thought to attract microglia to sites of brain = injury to=20 stimulate axonal sprouting.^[62,63]

Most tumors secrete chemokines, which frequently results in the = recruitment of a host of inflammatory cells. In glioblastoma = multiforme,=20 chemokine signaling through CXCR4 and CXCL12 not only induces = microglial=20 migration^[64] and potentially recruits microglia to the = peritumoral area but also influences T-regulatory cell motility. = Like many=20 other tumor types, gliomas create an immunosuppressed zone in = their=20 immediate vicinity.^[65] Among the factors that = contribute to=20 this phenomenon, T-regulatory cell recruitment to and accumulation = at the=20 tumor and its periphery appears to be the most=20 important.^[58,66] CXCR4-CXCL12 signals are also known = effectors=20 of T-regulatory cell homing to bone marrow.^[33] = Therefore, it=20 is possible that glioma-secreted CXCL12 may attract natural = T-regulatory=20 cells as well as CD4-positive T-lymphocytes, which are then = converted to=20 T-regulatory cells by tumor-secreted TGF-β,^[58] to = gliomas,=20 where they could actively suppress immune responses to the = tumor.

Microglia and most astrocytomas secrete ATX, which suggests = that they=20 are capable of generating the phospholipid motogen LPA. Both tumor = and=20 immune cells also express receptors for LPA such as LPA1 and=20 LPA2.^[45] This finding suggests that phospholipid = signaling=20 through ATX could contribute to microglial cell motility in both = an=20 autocrine (microglia driving its own motility) and paracrine = (microglia=20 motility being driven by glioblastoma-secreted ATX) manner. This=20 possibility is supported by the observation that LPA enhances = chemokinetic=20 migration of murine microglial cells in vitro.^[67]

Glial and Neuronal Progenitors

Among the growth factors that are known to drive glioma = tumorigenesis,=20 PDGF has the strongest effect on normal cells in the central = nervous=20 system, especially on all stages of neural and glial precursor = cells (Fig.=20 1). Both isoforms of PDGF (PDGFA and PDGFB) are expressed in = neurons=20 during embryogenesis,^[68,69] while their receptor, = PDGFRα, is=20 expressed in glial cells of the oligodendrocyte = lineage.^[70]=20 This coordinated expression suggests the presence of a paracrine = signaling=20 system in which neurons produce the ligand and glial cells of the=20 oligodendrocyte lineage express the receptor. Glial progenitor = cells are=20 exquisitely sensitive to PDGF; glial progenitor cells from = neonatal mice=20 that were transduced with PDGF-expressing retroviruses and = injected into=20 mice proliferated rapidly and formed tumors that resembled=20 oligoastrocytomas.^[71] Recently, in a glioma mouse = model, it=20 was discovered that PDGF-expressing retroviruses could also infect = cycling=20 glial progenitor cells from adult mice, which comprise 4% of the = cells in=20 the adult white matter.^[72] PDGF that was secreted by=20 retroviral-infected adult glial progenitor cells had a potent = oncogenic=20 effect, and the tumors that arose had proliferative and invasive=20 phenotypes that resembled those of human glioblastoma multiforme=20 cells.^[72] Surprisingly, however, the newly formed = tumors=20 consisted of a mixture of infected and noninfected cells; this = finding=20 suggests that the PDGF-expressing cells can recruit and transform = other=20 glial progenitor cells in the vicinity via paracrine signaling, = thereby=20 potentially increasing the tumor bulk. Such recruitment of = adjacent glial=20 cells may explain the observed extensive genetic heterogeneity of=20 glioblastoma multiforme.

Neural stem cells also respond to HGF/SF and CXCL12 with = increased=20 motility,^[55,73,74] which reflects the effect of these=20 autocrine factors on bone marrow and mesenchymal stem cell=20 migration.^[75] We speculate that glioma-secreted = motility=20 factors are responsible, in part, for the phenomenon of neural = stem cell=20 homing to tumors.^[76] The observation that CXCR4 and = CXCL12 are=20 involved in neuronal guidance in the developing = brain,^[77] in=20 microglia migration,^[64] and in neural stem cell = migration=20 toward gliomas^[78,79] has sparked interest in the role = of this=20 autocrine signaling system in glioma invasion. CXCL12 may be an=20 ontogenetically established guidance cue that is normally active = during=20 development but that is re-expressed by glioblastoma multiforme = cells to=20 maintain the invasion process.

Invasive glioblastoma multiforme cells secrete PTN and S1P, two = additional glioma motility factors with potential paracrine = effects on=20 neural precursor cells. Neural stem cells from the embryonic=20 subventricular zone and radial glia express RPTPζ during the = migratory=20 phases of central nervous system development.^[80] = Neural=20 progenitor cells from the embryonic rat brain also express most = S1P=20 receptors and thus display increased proliferation and cell-cell=20 aggregation when exposed to S1P.^[81] Taken together, = these two=20 less well known factors in glioma motility represent signaling = systems=20 that are similar to the CXCR4-CXCL12 signaling system in that they = involve=20 developmental migration cues.

Astrocytes

Glial cells function as a trophic support system for neurons = (i.e.,=20 they supply growth factors and cytokines). Some of the = neurotrophic=20 factors secreted by astrocytes, such as TGF-α, CXCL12, S1P, = and GDNF, have=20 the potential to increase the invasive capacity of glioblastoma = multiforme=20 cells (Fig. 1). However, despite evidence that astrocytes = constitutively=20 express EGFR, Met, RPTPζ, and LPA1,^[39,82-85] = little is known=20 about the effects of signaling through these receptors on normal=20 astrocytes in vivo. Ligands of the above-mentioned receptors can = have a=20 variety of effects on astrocytes in culture. For example, exposure = of=20 astrocytes to LPA in vitro results in cell contraction through = stress=20 fiber formation.^[86] It is interesting to note that = invasive=20 glioma cells secrete ATX, an extracellular enzyme that catalyzes = the=20 production of LPA. In addition, Sharif et al.^[87] found = that=20 normal mature astrocytes that were exposed to TGF-α reverted = to a radial=20 glial cell phenotype that was similar to that of glial progenitor = cells=20 found in earlier stages of neural development. Prolonged exposure = to TGF-α=20 resulted in the conversion of mature astrocytes into cells with an = even=20 earlier ontogenetic phenotype, neural stem cells. Other studies = [reviewed=20 in^[88]] indicate that neural stem cells contribute, in = part, to=20 the glioblastoma multiforme tumor mass. The capacity of TGF-α = to=20 dedifferentiate mature astrocytes to radial glia and stem cell = phenotypes=20 is similar to the dedifferentiating effect of PDGF on = oligodendrocytes=20 (see "Glial and Neuronal Progenitors"). It is interesting that the = two=20 best-described growth factor signaling systems that, together, are = hallmarks of glioblastoma multiforme biology are able to revert = mature=20 cells in the central nervous system to more plastic, pluripotent = cells,=20 which may contribute to the aberrant growth and dispersion of = glioblastoma=20 multiforme.

Oligodendrocytes

Little is known about how invasive astrocytoma cells might = affect=20 oligodendrocytes, which function to support neuronal health and = function=20 in the normal brain and to protect neurons from neurotoxic = insults. In=20 these capacities, oligodendrocytes secrete HB-EGF^[89] = and GDNF,=20 which are also expressed by astrocytes^[89] and bind to=20 receptors that stimulate glioma motility.^[14,39] = Differentiated=20 oligodendrocytes express S1P receptors^[90] and are = therefore=20 capable of receiving signals from invasive glioma cells that = secrete this=20 phospholipid, but little is known about the effect of S1P = signaling on=20 mature oligodendrocytes.

During the myelinating stages of central nervous system = development in=20 rat neonates, oligodendrocytes also secrete ATX,^[91] = which has=20 the effect of reducing oligodendrocyte adhesion, presumably to = support=20 changes in oligodendrocyte function and morphology that occur = during the=20 initial stages of myelination. Glioblastoma multiforme-derived ATX = may=20 provoke the de-adhesion of oligodendrocytes found in the path of = invading=20 glioblastoma multiforme cells from the extracellular matrix (Fig. = 1),=20 creating an environment that is permissive for invasion. In this = context,=20 it is interesting that oligodendrocyte-rich white matter tracts = are one of=20 the preferred conduits of invasion by glioblastoma multiforme.

Neurons

RPTPζ is not only expressed by astrocytes throughout the = central=20 nervous system but also by certain subsets of = neurons.^[21] The=20 RPTPζ ligand, PTN, is expressed along pathways of developing=20 axons,^[92] promotes neurite outgrowth in vitro, and is = also=20 expressed in reactive astrocytes following neuronal = injury.^[93]=20 Following experimental injury to the central nervous system, = transcription=20 of the RPTPζ gene increases substantially in areas of axonal = sprouting as=20 well as in areas of glial scaring, suggesting that RPTPζ is = part of=20 central nervous system repair mechanisms. These observations imply = that=20 cells in the central nervous system may react to injury caused by = the=20 physical expansion of the tumor in a manner similar to its = reaction to=20 physical injury. This injury response likely includes the = secretion of=20 factors such as PTN which could ultimately stimulate glioblastoma=20 multiforme dispersion away from the tumor mass. Neurons may also = be=20 sensitive to glioma-derived phospholipid signaling: they undergo = growth=20 cone collapse in response to LPA^[94] and express S1P=20 receptors,^[95] indicating that they are capable of = receiving=20 signals from invasive glioma cells that secrete these = phospholipids=20 (Figure 1).

Vasculature

The role of the microenvironment in glioblastoma multiforme = invasion=20 was recently and elegantly captured by time-lapse microscopy of = rat=20 glioblastoma multiforme invading into a vital rat brain=20 slice.^[96] In this model system, glioblastoma = multiforme cells=20 were observed to travel along blood vessels and to pause at select = vascular branch points, where they proliferated, supporting the = hypothesis=20 that glioblastoma multiforme cells may respond to = endothelial-derived cues=20 (Figure 1). This study also revealed that the invading = glioblastoma=20 multiforme cells disrupt contacts between astrocytic processes = called=20 podocytes and endothelial cells. Although the authors of the = time-lapse=20 study did not assign a mechanistic basis for the disruption in = cell-cell=20 contact (i.e., proteolytic digestion of contacts, = paracrine-activated=20 deattachment, or both), their observations demonstrate that, = during the=20 process of invasion, glioblastoma multiforme cells cause the = disengagement=20 of normal glial cells from their functional context.

Axonal Guidance Molecules May Be Linked To Glioma = Invasion

In addition to PTN, other axonal guidance molecules may = stimulate=20 glioma invasion. The semaphorins and their receptors, the = neuropilins and=20 the plexins, are also expressed by glioblastoma multiforme=20 cells^[97] and have been linked to cancer invasion = through the=20 observation that semaphorin-3A modulates the migration of neural=20 progenitor cells.^[98,99] Neuropilin 2 transcription was = increased in G112 glioma cells that were stimulated to migrate in = vitro=20 compared with stationary cells.^[100] In addition, we = recently=20 found that semaphorin-3B transcription was higher in laser capture = microdissected invasive glioblastoma multiforme cells than in = noninvasive=20 cells at the tumor core (Demuth T, Reavie L, Hoelzinger D, Rennert = J,=20 Nakada M, Berens ME: unpublished observations).

The netrins and their receptors, deleted in colorectal = carcinoma (DCC)=20 and the Unc5 homolog, could represent an additional group of = axonal=20 guidance factors that are involved in glioblastoma multiforme = invasion.=20 For example, netrin 4 transcription was higher in invasive = glioblastoma=20 multiforme cells than in stationary cells in the tumor=20 core,^[43] suggesting that this axonal and glial = progenitor=20 guidance system also drives glioblastoma multiforme invasion. = During=20 development, netrin 1 (NTN1) expressed in the spinal chord = attracts axonal=20 protrusions from neurons that express DCC but repels neurons that = express=20 Unc5.^[101] NTN1 also elicits directional motility in=20 neuroblastoma cells that express DCC.^[102] Little is = known=20 about netrin signaling in glia, but NTN1 is involved in migration = of glial=20 cells of the oligodendrocyte lineage during = development.^[102]=20 In addition, both glial cells of the oligodendrocyte lineage and = presumed=20 astrocyte precursors respond to NTN1 and semaphorin-3A = concentration=20 gradients in the central nervous system with directional migration = along=20 the optic nerve.^[103]

Therapeutic Targeting of Autocrine and Paracrine Invasion=20 Signaling

A growing understanding of the interplay between a tumor and = its=20 microenvironment has generated efforts to therapeutically target = the=20 microenvironment. One of the challenges of such an approach is to = target=20 the tumor-supporting functions of the microenvironment while = leaving=20 intact any tumor-inhibiting functions that it may have. Two = reviews have=20 comprehensively addressed this issue in extracranial=20 malignancies.^[104,105] Glial tumors, however, present a = unique=20 type of malignancy: their microenvironment (the brain) is = particularly=20 delicate, and the blood-brain barrier poses a substantial hurdle = for drug=20 delivery.

The most extensively studied class of targeted agents in glioma = are the=20 receptor tyrosine kinase inhibitors, which were developed to = target=20 receptor tyrosine kinases that are expressed on tumor cells but = may also=20 have additional therapeutic benefit because of their effects on = the tumor=20 environment. The most prominent of these agents is the rationally = designed=20 drug imatinib, which targets the Bcr-Abl fusion protein and the = PDGFR=20 tyrosine kinase in tumor cells.^[106,107] Imatinib has = also been=20 shown to interfere with angiogenesis by decreasing the degree of = pericyte=20 coverage of tumor-associated endothelial = neovasculature.^[108]=20 Despite the impressive antitumor activity of imatinib in chronic=20 myelogenous leukemia and against gastrointestinal stromal=20 tumors,^[109,110] results in glioblastoma multiforme = have been=20 disappointing^[111] ( Table 2 ). Among the factors thought to = contribute to=20 the poor results in glioblastoma multiforme is the limited = penetration of=20 the central nervous system by imatinib, combined with active=20 P-glycoprotein-mediated efflux of drug at the blood-brain=20 barrier.^[112] The blood-brain barrier is likely also = the=20 limiting step in the poor efficacy of imatinib in cases of central = nervous=20 system relapse of chronic myelogenous and acute lymphoblastic=20 leukemia,^[113] two normally imatinib-sensitive tumors = types.=20

The autocrine loop of EGFR and its ligands that is present in = brain=20 tumors lends itself well to targeting by small-molecule inhibitors = such as=20 gefinib and erlotinib. However, despite the high frequency of EGFR = gene=20 overexpression and gene amplification in glioblastoma multiforme, = as well=20 as the presence of EGFR in the brain microenvironment, gefitinib = has not=20 been found to provide therapeutic benefit in patients with = glioblastoma=20 multiforme.^[111] It is surprising that no association = between=20 EGFR gene amplification and response to EGFR inhibitors has been = found.=20 Nevertheless, coexpression of the vIII mutation in EGFR and PTEN = was=20 statistically significantly associated with response to EGFR = kinase=20 inhibitors.^[114] However, in a study with erlotinib, no = association between vIII mutation status and response to EGFR = inhibition=20 was found, although high tumor expression of EGFR coincident with = low=20 levels of protein kinase B/Akt phosphorylation were associated = with=20 response to erlotinib.^[115]

The promises of targeted agents have not been fulfilled in = glioblastoma=20 multiforme. Given the disappointing results from these clinical = trials, it=20 seems clear that novel biomarker-guided clinical trial designs are = needed.=20 One such trial design uses a pharmacodynamic marker to monitor the = efficacy of a given therapeutic agent. Thus, after a stereotactic = biopsy,=20 treatment-naive patients are administered the novel agent during a = 1-2=20 week period prior to surgical tumor removal. Drug levels and = status of=20 biomarkers in posttreatment samples relative to pretreatment = samples=20 should reveal the degree of penetration of the central nervous = system by=20 the drug and help to define the biologic effective dose based on = biomarker=20 engagement.

Because glioblastoma multiforme is characterized by extensive=20 neovascularization, it may respond to treatments that disrupt=20 angiogenesis. The monoclonal antibody bevacizumab targets VEGF, = the=20 paracrine stimulator of angiogenesis. The combination of = antiangiogenic=20 and cytotoxic treatment (i.e., bevacizumab and irinotecan) has = been very=20 promising in clinical trials^[116] ( Table 2 ) and may provide the rationale for = future=20 testing of VEGF-Trap, an angiogenesis inhibitor, and = small-molecule=20 inhibitors of VEGFR.

Interaction of glioblastoma multiforme cells with cells present = in the=20 normal brain stroma can enhance the invasive process. Chemokines = (such as=20 CXCL12) and chemokine receptors (such as CXCR4) are well known = agents of=20 pro-invasive inter-cellular communication, they have therefore = become the=20 focus of new anti-invasive therapies. The histone deacetylase = inhibitors=20 suberoylanilide hydroxamic acid (SAHA) and butyrate have been = found to=20 reduce CXCR4 mRNA and protein levels^[117] and are = therefore=20 further along in clinical development than other agents that = interfere=20 with CXCL12-CXCR4 signaling. In preclinical studies, histone = deacetylase=20 inhibitors were found to inhibit the migration of leukemia cells = and=20 lymphoblasts that were stimulated with CXCL12^[117] and = to=20 inhibit in vitro invasion and increase tumor necrosis = factor-induced=20 apoptosis by suppressing activation of nuclear factor-kappa=20 B.^[118]

Successful treatment of invasive brain tumors will depend on = blending=20 cocktails of targeted agents that are tailored for an individual = patient.=20 A thorough understanding of the interaction between the tumor and = its=20 microenvironment will help to define the context of vulnerability, = such as=20 the dependence on specific signaling pathways, which may lead to = novel=20 antitumor approaches. Clinical testing of such targeted = interventions will=20 require careful monitoring of the effects of these agents on both = cells in=20 the tumor microenvironment and in the tumor. Advances in imaging=20 modalities will allow for better assessment of the tumor=20 microenvironment's contribution to tumor biology.

Conclusions

Multiple autocrine motility-enhancing signaling systems are = likely to=20 be involved in the maintenance of glioblastoma multiforme = invasion. The=20 observation that glioblastoma multiforme invasion occurs along = specific=20 anatomic features suggests the presence of scaffolding that = enables the=20 traction necessary for cell motility. Conversely, invasive = glioblastoma=20 multiforme cells receive distinct signals that are generated by = the=20 cellular context along which they transit; such proximity of tumor = and=20 normal cells enables communication via paracrine signaling systems = (Fig.=20 1), which may support tumor invasion. Most of the autocrine=20 receptor/ligand groups discussed herein have documented roles in = central=20 nervous system development, specifically in association with glial = and=20 neuronal precursor cell waves of colonization. This similarity = suggests=20 that normal cells in the brain express, or have the capacity to = express,=20 receptors for tumor-secreted ligands or that brain cells produce = ligands=20 that activate receptors on glioma cells ( Table 1 ). The hypothesis that neoplastic = cells use=20 signals that influence cell types involved in normal developmental = tissue=20 morphogenesis is not new; a hallmark of this hypothesis is the = recruitment=20 of macrophages, fibroblasts, and the vasculature to tumor sites. = Such=20 phenomena have been observed in several cancers, including breast = cancer=20 and melanoma.^[7] However, until recently it has been = difficult=20 to demonstrate communication between invasive tumor cells and the=20 surrounding stroma. The use of gene expression profiling of = specific=20 subpopulations of tumor cells has begun to shed light on the = differences=20 between invasive tumor cells and the original tumor mass from = which they=20 emanated. Gene expression signatures of tumor invasion that have = been=20 identified for glioblastoma multiforme invasion^[43] and = for=20 breast cancer invasion^[119] revealed possible = interactions=20 between the invasive cells and immune cells. Confirmation of this=20 interaction revealed that macrophages are indeed required elements = in=20 tumor invasion.^[120]

Studies of the influence of the tumor microenvironment on = tumorbiology=20 are further advanced in the field of breast cancer. Normal = myoepithelium=20 has a recognized tumor suppressor role in breast cancer=20 pathogenesis.^[121] This normal tissue surrounding the = neoplasm=20 expresses antiangiogenic factors and protease inhibitors that slow = tumor=20 progression.^[122] On the other hand, breast tumors have = an=20 impact on normal breast tissue, even at the molecular level: there = are=20 statistically significant differences in gene expression patterns = of=20 myoepithelial cells derived from normal breast tissue and of those = derived=20 from invasive breast carcinoma.^[36] These observations = suggest=20 that invasive stages of breast carcinoma induce changes in the = host tissue=20 that could synergize with the tumor to promote invasion. It is = interesting=20 that the majority of differentially expressed genes encode = secreted=20 proteins and/or their receptors,^[36] including CXCR4 = and=20 CXCL12, indicating the involvement of paracrine interactions. It = is highly=20 likely that glioblastoma multiforme tumor burden may have similar = impact=20 on normal cells in the brain parenchyma.

In this review, we have examined select autocrine = receptor/ligand=20 groups that are expressed by glioblastoma multiforme and linked to = glioblastoma multiforme invasion, and we hypothesize that = glioblastoma=20 multiforme motility signals may have paracrine effects on proximal = normal=20 cells in the brain. Glioblastoma multiforme re-engages = pre-established=20 ontogenetic motility signals that drive central nervous system = migration=20 events that involve every cell type present in the brain = parenchyma. The=20 expression and function of paracrine receptor-ligand pairs in the = context=20 of glioblastoma multiforme and its microenvironment suggest that=20 glioblastoma multiforme autocrine motility signals may elicit = responses=20 from the brain parenchyma that support or even facilitate = invasion.

The interaction of neoplastic cells with the peritumoral stroma = can=20 also result in the disruption of tissue homeostasis. For example, = tumor=20 burden frequently results in the recruitment of leukocytes to the = tumor=20 site (through cytokines and chemokines), which leads to an = inflammatory=20 reaction.^[123] Even though this process reflects the = natural=20 course of wound healing, these inflammatory responses are not=20 self-limiting, and thus it has been proposed that tumors resemble = wounds=20 that do not heal.^[124] Reinstating tissue homeostasis = by=20 inhibiting tumor-stroma communication (such as by inhibiting=20 proinflammatory or proinvasive signals) may be one way to slow = tumor=20 progression. Inhibiting the activity of chemokine = colony-stimulating=20 factor 1, which attracts inflammatory cells to the tumor site, has = been=20 shown to reduce invasion and metastasis in breast = cancer.^[125]=20 Similar inhibition of the autocrine motility loops of glioblastoma = multiforme could also interrupt their collateral paracrine = pro-invasion=20 signaling effects. There are two potential benefits to targeting = autocrine=20 motility stimulation: one is to arrest tumor-driven motility, the = other is=20 to impede tumor motility that is supported and even sustained by = the=20 peritumoral stroma. A better understanding of the influence of = tumors on=20 their microenvironment could lead to therapies that target the = more=20 genetically stable normal cells that surround the glioma rather = than the=20 genetically unstable tumor cells themselves, which invariably = develop=20 resistance to therapy. Making the host tissue more resistant to = the=20 aberrant signaling of invasive tumor cells may be an avenue of = tumor=20 management that could produce palliative adjuvant therapy slowing = tumor=20 recurrence.

Table 1. Gioblastoma Multiforme Autocrine Motility = Factors and=20 Their Receptors That are Also Expressed by Cells in the Central = Nervous=20 System

Cell type Receptors expressed Ligands/motogens secreted

Astrocytes Receptor protein tyrosine phosphatase beta/zeta,=20 lysophosphatidic acid receptor 1, Met protooncogene, = epidermal=20 growth factor receptor Glial cell line-derived neurotrophic factor,=20 sphingosine-1-phosphate, chemokine CXC motif ligand 12,=20 transforming growth factor-α

Oligodendrocytes Lysophosphatidic acid receptor 1, lysophosphatidic acid=20 receptor 3, sphingosine-1-phosphate receptor,Met = protooncogene Autotaxin, glial cell line-derived neurotrophic factor,=20 heparin-binding epidermal growth factor

Microglia Lysophosphatidic acid receptor 1, lysophosphatidic acid=20 receptor 2, Met protooncogene, chemokine CXC motif = receptor 4,=20 sphingosine-1-phosphate receptor, glial cell line-derived=20 neurotrophic factor receptor, neuropilin 1, plexinA, = epidermal=20 growth factor receptor Hepatocyte growth factor and scatter factor chemokine = CXC=20 motif ligand 12, autotaxin, glial cell line-derived = neurotrophic=20 factor, epidermal growth factor

Reactive astrocytes Met protooncogene Pleiotrophin, autotaxin, hepatocyte growth factor and = scatter=20 factor

Glial progenitors Platelet-derived growth factor receptor α, = lysophosphatidic=20 acid receptor 1, lysophosphatidic acid receptor 2, Unc5, = deleted=20 in colorectal carcinoma, neuropilin 1, neuropilin 2, = rearranged=20 during transfection protooncogene, glial cell line-derived = neurotrophic factor receptor Glial cell line-derived neurotrophic factor

Neurons Receptor protein tyrosine phosphatase beta/zeta,=20 lysophosphatidic acid receptor 1, plexin B, Met = protooncogene,=20 epidermal growth factor receptor Platelet-derived growth factor A, platelet-derived = growth=20 factor B, chemokine CXC motif ligand 12, pleiotrophin,=20 semaphorin-3A, hepatocyte growth factor and scatter = factor,=20 transforming growth factor-α

Neural progenitors Sphingosine-1-phosphate receptor 1, = sphingosine-1-phosphate=20 receptor 2, sphingosine-1-phosphate receptor 3, = lysophosphatidic=20 acid receptor 1, neuropilin 1, plexin, deleted in = colorectal=20 carcinoma, Met protooncogene, platelet-derived growth = factor=20 receptor Glial cell line-derived neurotrophic factor, = platelet-derived=20 growth factor

Neural stem cells Met protooncogene, chemokine CXC motif receptor 4 Chemokine CXC motif ligand 12, pleiotrophin

Vascular endothelium Met protooncogene, sphingosine-1-phosphate receptor 1,=20 sphingosine-1-phosphate receptor 3 None identified

Table 2. Agents That Target Autocrine and Paracrine = Signaling=20 Partners*

Target Agent Clinical trial (reference)

VEGF Bevacizumab Phase II in recurrent GBM and anaplastic astrocytoma=20 (116)

PDGFR Imatinib Phase I/II in recurrent glioma showed minimal activity; = 9%=20 intracranial hemorrhage (126)

Phase II of imatinib + hydroxzurea: progression-free = survival=20 14.4 weeks (127)

Phase II of imatinib + temozolamide (ongoing)

EGFR Gefitinib Phase II trial did not show therapeutic benefit = (111)

HGF/SF AMG-102 Phase II trial in advanced malignant glioma = (ongoing)

TGF-β AP 12009 Phase II trial in glioma = (ongoing)

*VEGF =3D vascular endothelial growth factor; GBM =3D = glioblastoma=20 multiforme; PDGFR =3D platelet-derived growth factor receptor; = EGFR =3D=20 epidermal growth factor receptor; HGF/SF =3D hepatocyte growth = factor and=20 scatter factor; TGF-β =3D transforming growth = factor-β.

References

Kleihues P, Louis DN, Scheithauer BW, Rorke LB, = Reifenberger=20 G, Burger PC, et al. The WHO classification of tumors of = the=20 nervous system. J Neuropathol Exp Neurol (2002) 61:215-25. = discussion 226-9.
McCarthy BJ, Kruchko C. Consensus conference on cancer = registration of brain and central nervous system tumors. = Neuro=20 Oncol (2005) 7:196-201.
Hegi ME, Diserens AC, Gorlia T, Hamou MF, de Tribolet = N,=20 Weller M, et al. MGMT gene silencing and benefit from = temozolomide=20 in glioblastoma. N Engl J Med (2005) 352:997-1003.
Demuth T, Berens ME. Molecular mechanisms of glioma = cell=20 migration and invasion. J Neurooncol (2004) 70:217-28.
Scherer Hea. Cerebral astrocytomas and their = derivatives. Am J=20 Cancer (1940) 1:159-98.
Ramnarayan R, Dodd S, Das K, Heidecke V, Rainov NG. = Overall=20 survival in patients with malignant glioma may be = significantly=20 longer with tumors located in deep grey matter. J Neurol = Sci=20 (2007) 260:49-56.
Mueller MM, Fusenig NE. Friends or foes=97bipolar = effects of the=20 tumour stroma in cancer. Nat Rev Cancer (2004) 4:839-49.
Pollard JW. Tumour-educated macrophages promote tumour = progression and metastasis. Nat Rev Cancer (2004) 4:71-8.
Kaur B, Khwaja FW, Severson EA, Matheny SL, Brat DJ, = Van Meir=20 EG. Hypoxia and the hypoxia-inducible-factor pathway in = glioma=20 growth and angiogenesis. Neuro Oncol (2005) 7:134-53.
Shih AH, Holland EC. Platelet-derived growth factor = (PDGF) and=20 glial tumorigenesis. Cancer Lett (2006) 232:139-47.
Raulo E, Chernousov MA, Carey DJ, Nolo R, Rauvala H. = Isolation=20 of a neuronal cell surface receptor of heparin binding=20 growth-associated molecule (HB-GAM). Identification as = N-syndecan=20 (syndecan-3). J Biol Chem (1994) 269:12999-3004.
Rauvala H, Huttunen HJ, Fages C, Kaksonen M, Kinnunen = T, Imai=20 S, et al. Heparin-binding proteins HB-GAM (pleiotrophin) = and=20 amphoterin in the regulation of cell motility. Matrix Biol = (2000)=20 19:377-87.
Chicoine MR, Silbergeld DL. Mitogens as motogens. J = Neurooncol=20 (1997) 35:249-57.
Ramnarain DB, Park S, Lee DY, Hatanpaa KJ, Scoggin SO, = Otu H,=20 et al. Differential gene expression analysis reveals = generation of=20 an autocrine loop by a mutant epidermal growth factor = receptor in=20 glioma cells. Cancer Res (2006) 66:867-74.
Lamszus K, Heese O, Westphal M. Angiogenesis-related = growth=20 factors in brain tumors. Cancer Treat Res (2004) = 117:169-90.
Cattaneo MG, Gentilini D, Vicentini LM. Deregulated = human=20 glioma cell motility: inhibitory effect of somatostatin. = Mol Cell=20 Endocrinol (2006) 256:34-9.
Natarajan M, Stewart JE, Golemis EA, Pugacheva EN,=20 Alexandropoulos K, Cox BD, et al. HEF1 is a necessary and = specific=20 downstream effector of FAK that promotes the migration of=20 glioblastoma cells. Oncogene (2006) 25:1721-32.
Koochekpour S, Jeffers M, Rulong S, Taylor G, = Klineberg E,=20 Hudson EA, et al. Met and hepatocyte growth factor/scatter = factor=20 expression in human gliomas. Cancer Res (1997) 57:5391-8.
Abounader R, Laterra J. Scatter factor/hepatocyte = growth=20 factor in brain tumor growth and angiogenesis. Neuro Oncol = (2005)=20 7:436-51.
Brockmann MA, Ulbricht U, Gruner K, Fillbrandt R, = Westphal M,=20 Lamszus K. Glioblastoma and cerebral microvascular = endothelial=20 cell migration in response to tumor-associated growth = factors.=20 Neurosurgery (2003) 52:1391-9. discussion 1399.
Snyder SE, Li J, Schauwecker PE, McNeill TH, Salton = SR.=20 Comparison of RPTP zeta/beta, phosphacan, and trkB mRNA = expression=20 in the developing and adult rat nervous system and = induction of=20 RPTP zeta/beta and phosphacan mRNA following brain injury. = Brain=20 Res Mol Brain Res (1996) 40:79-96.
Milev P, Friedlander DR, Sakurai T, Karthikeyan L, = Flad M,=20 Margolis RK, et al. Interactions of the chondroitin = sulfate=20 proteoglycan phosphacan, the extracellular domain of a=20 receptor-type protein tyrosine phosphatase, with neurons, = glia,=20 and neural cell adhesion molecules. J Cell Biol (1994) = 127((Pt=20 1)):1703-15.
Grumet M, Milev P, Sakurai T, Karthikeyan L, Bourdon = M,=20 Margolis RK, et al. Interactions with tenascin and = differential=20 effects on cell adhesion of neurocan and phosphacan, two = major=20 chondroitin sulfate proteoglycans of nervous tissue. J = Biol Chem=20 (1994) 269:12142-6.
Milev P, Maurel P, Haring M, Margolis RK, Margolis RU. = TAG-1/axonin-1 is a high-affinity ligand of neurocan,=20 phosphacan/protein-tyrosine phosphatase-zeta/beta, and = N-CAM. J=20 Biol Chem (1996) 271:15716-23.
Meng K, Rodriguez-Pena A, Dimitrov T, Chen W, Yamin M, = Noda M,=20 et al. Pleiotrophin signals increased tyrosine = phosphorylation of=20 beta beta-catenin through inactivation of the intrinsic = catalytic=20 activity of the receptor-type protein tyrosine phosphatase = beta/zeta. Proc Natl Acad Sci USA (2000) 97:2603-8.
Fukada M, Fujikawa A, Chow JP, Ikematsu S, Sakuma S, = Noda M.=20 Protein tyrosine phosphatase receptor type Z is = inactivated by=20 ligand-induced oligomerization. FEBS Lett (2006) = 580:4051-6.
Pariser H, Ezquerra L, Herradon G, Perez-Pinera P, = Deuel TF.=20 Fyn is a downstream target of the pleiotrophin/receptor = protein=20 tyrosine phosphatase beta/zeta-signaling pathway: = regulation of=20 tyrosine phosphorylation of Fyn by pleiotrophin. Biochem = Biophys=20 Res Commun (2005) 332:664-9.
Muller S, Kunkel P, Lamszus K, Ulbricht U, Lorente GA, = Nelson=20 AM, et al. A role for receptor tyrosine phosphatase zeta = in glioma=20 cell migration. Oncogene (2003) 22:6661-8.
Ulbricht U, Brockmann MA, Aigner A, Eckerich C, Muller = S,=20 Fillbrandt R, et al. Expression and function of the = receptor=20 protein tyrosine phosphatase zeta and its ligand = pleiotrophin in=20 human astrocytomas. J Neuropathol Exp Neurol (2003) = 62:1265-75.
Sehgal A, Keener C, Boynton AL, Warrick J, Murphy GP. = CXCR-4,=20 a chemokine receptor, is overexpressed in and required for = proliferation of glioblastoma tumor cells. J Surg Oncol = (1998)=20 69:99-104.
Bajetto A, Barbieri F, Dorcaratto A, Barbero S, Daga = A,=20 Porcile C, et al. Expression of CXC chemokine receptors = 1-5 and=20 their ligands in human glioma tissues: role of CXCR4 and = SDF1 in=20 glioma cell proliferation and migration. Neurochem Int = (2006)=20 49:423-32.
Barbero S, Bajetto A, Bonavia R, Porcile C, Piccioli = P, Pirani=20 P, et al. Expression of the chemokine receptor CXCR4 and = its=20 ligand stromal cell-derived factor 1 in human brain tumors = and=20 their involvement in glial proliferation in vitro. Ann N Y = Acad=20 Sci (2002) 973:60-9.
Zhou Y, Larsen PH, Hao C, Yong VW. CXCR4 is a major = chemokine=20 receptor on glioma cells and mediates their survival. J = Biol Chem=20 (2002) 277:49481-7.
Ehtesham M, Winston JA, Kabos P, Thompson RC. CXCR4 = expression=20 mediates glioma cell invasiveness. Oncogene (2006) = 25:2801-6.
Hong X, Jiang F, Kalkanis SN, Zhang ZG, Zhang XP, = DeCarvalho=20 AC, et al. SDF-1 and CXCR4 are up-regulated by VEGF and = contribute=20 to glioma cell invasion. Cancer Lett (2006) 236:39-45.
Allinen M, Beroukhim R, Cai L, Brennan C, = Lahti-Domenici J,=20 Huang H, et al. Molecular characterization of the tumor=20 microenvironment in breast cancer. Cancer Cell (2004) = 6:17-32.
Muller A, Homey B, Soto H, Ge N, Catron D, Buchanan = ME, et al.=20 Involvement of chemokine receptors in breast cancer = metastasis.=20 Nature (2001) 410:50-6.
Wiesenhofer B, Stockhammer G, Kostron H, Maier H, = Hinterhuber=20 H, Humpel C. Glial cell line-derived neurotrophic factor = (GDNF)=20 and its receptor (GFR-alpha 1) are strongly expressed in = human=20 gliomas. Acta Neuropathol (Berl) (2000) 99:131-7.
Song H, Moon A. Glial cell-derived neurotrophic factor = (GDNF)=20 promotes low-grade Hs683 glioma cell migration through = JNK,=20 ERK-1/2 and p38 MAPK signaling pathways. Neurosci Res = (2006)=20 56:29-38.
Stracke ML, Krutzsch HC, Unsworth EJ, Arestad A, Cioce = V,=20 Schiffmann E, et al. Identification, purification, and = partial=20 sequence analysis of autotaxin, a novel = motility-stimulating=20 protein. J Biol Chem (1992) 267:2524-9.
Yang SY, Lee J, Park CG, Kim S, Hong S, Chung HC, et = al.=20 Expression of autotaxin (NPP-2) is closely linked to = invasiveness=20 of breast cancer cells. Clin Exp Metastasis (2002) = 19:603-8.
Yang Y, Mou L, Liu N, Tsao MS. Autotaxin expression in = non-small-cell lung cancer. Am J Respir Cell Mol Biol = (1999)=20 21:216-22.
Hoelzinger D, Mariani L, Weis J, Woyke T, Berens T, = McDonough=20 W, et al. Gene expression profile of glioblastoma = multiformes=20 invasive phenotype points to new therapeutic targets. = Neoplasia=20 (2005) 7:7-16.
Kishi Y, Okudaira S, Tanaka M, Hama K, Shida D, = Kitayama J, et=20 al. Autotaxin is overexpressed in glioblastoma multiforme = and=20 contributes to cell motility of glioblastoma by converting = lysophosphatidylcholine to lysophosphatidic acid. J Biol = Chem=20 (2006) 281:17492-500.
Hoelzinger D, Nakada M, Demuth T, Rosensteel T, Reavie = L,=20 Berens M. Autotaxin: a secreted autocrine/paracrine factor = that=20 promotes glioma invasion. J Neurooncol (2007) In press.
Manning TJ Jr, Parker JC, Sontheimer H. Role of=20 lysophosphatidic acid and rho in glioma cell motility. = Cell Motil=20 Cytoskeleton (2000) 45:185-99.
Chun J. Lysophospholipids in the nervous system.=20 Prostaglandins Other Lipid Mediat (2005) 77:46-51.
Van Brocklyn JR, Jackson CA, Pearl DK, Kotur MS, = Snyder PJ,=20 Prior TW. Sphingosine kinase-1 expression correlates with = poor=20 survival of patients with glioblastoma multiforme: roles = of=20 sphingosine kinase isoforms in growth of glioblastoma cell = lines.=20 J Neuropathol Exp Neurol (2005) 64:695-705.
Malchinkhuu E, Sato K, Horiuchi Y, Mogi C, Ohwada S, = Ishiuchi=20 S, et al. Role of p38 mitogen-activated kinase and c-Jun = terminal=20 kinase in migration response to lysophosphatidic acid and=20 sphingosine-1-phosphate in glioma cells. Oncogene (2005)=20 24:6676-88.
Van Brocklyn JR, Young N, Roof R. = Sphingosine-1-phosphate=20 stimulates motility and invasiveness of human glioblastoma = multiforme cells. Cancer Lett (2003) 199:53-60.
Edsall LC, Spiegel S. Enzymatic measurement of = sphingosine=20 1-phosphate. Anal Biochem (1999) 272:80-6.
Bassi R, Anelli V, Giussani P, Tettamanti G, Viani P, = Riboni=20 L. Sphingosine-1-phosphate is released by cerebellar = astrocytes in=20 response to bFGF and induces astrocyte proliferation = through=20 Gi-protein-coupled receptors. Glia (2006) 53:621-30.
Mantovani A, Schioppa T, Porta C, Allavena P, Sica A. = Role of=20 tumor-associated macrophages in tumor progression and = invasion.=20 Cancer Metastasis Rev (2006) 25:315-22.
Badie B, Schartner JM. Flow cytometric = characterization of=20 tumor-associated macrophages in experimental gliomas. = Neurosurgery=20 (2000) 46:957-61. discussion 961-2.
Badie B, Schartner J, Klaver J, Vorpahl J. In vitro = modulation=20 of microglia motility by glioma cells is mediated by = hepatocyte=20 growth factor/scatter factor. Neurosurgery (1999) = 44:1077-82.=20 discussion 1082-3.
Markovic DS, Glass R, Synowitz M, Rooijen N, = Kettenmann H.=20 Microglia stimulate the invasiveness of glioma cells by = increasing=20 the activity of metalloprotease-2. J Neuropathol Exp = Neurol (2005)=20 64:754-62.
Le DM, Besson A, Fogg DK, Choi KS, Waisman DM, Goodyer = CG, et=20 al. Exploitation of astrocytes by glioma cells to = facilitate=20 invasiveness: a mechanism involving matrix = metalloproteinase-2 and=20 the urokinase-type plasminogen activator-plasmin cascade. = J=20 Neurosci (2003) 23:4034-43.
Grauer OM, Nierkens S, Bennink E, Toonen LW, Boon L, = Wesseling=20 P, et al. CD4+FoxP3+ regulatory T cells gradually = accumulate in=20 gliomas during tumor growth and efficiently suppress = antiglioma=20 immune responses in vivo. Int J Cancer (2007) 121:95-105.
Watters JJ, Schartner JM, Badie B. Microglia function = in brain=20 tumors. J Neurosci Res (2005) 81:447-55.
Nolte C, Kirchhoff F, Kettenmann H. Epidermal growth = factor is=20 a motility factor for microglial cells in vitro: evidence = for EGF=20 receptor expression. Eur J Neurosci (1997) 9:1690-8.
Kunkel P, Muller S, Schirmacher P, Stavrou D, = Fillbrandt R,=20 Westphal M, et al. Expression and localization of scatter=20 factor/hepatocyte growth factor in human astrocytomas. = Neurooncol=20 (2001) 3:82-8.
Batchelor PE, Liberatore GT, Wong JY, Porritt MJ, = Frerichs F,=20 Donnan GA, et al. Activated macrophages and microglia = induce=20 dopaminergic sprouting in the injured striatum and express = brain-derived neurotrophic factor and glial cell = line-derived=20 neurotrophic factor. J Neurosci (1999) 19:1708-16.
Batchelor PE, Porritt MJ, Martinello P, Parish CL, = Liberatore=20 GT, Donnan GA, et al. Macrophages and microglia produce = local=20 trophic gradients that stimulate axonal sprouting toward = but not=20 beyond the wound edge. Mol Cell Neurosci (2002) 21:436-53.
Odemis V, Moepps B, Gierschik P, Engele J. = Interleukin-6 and=20 cAMP induce stromal cell-derived factor-1 chemotaxis in = astroglia=20 by up-regulating CXCR4 cell surface expression. = Implications for=20 brain inflammation. J Biol Chem (2002) 277:39801-8.
Dix AR, Brooks WH, Roszman TL, Morford LA. Immune = defects=20 observed in patients with primary malignant brain tumors. = J=20 Neuroimmunol (1999) 100:216-32.
El Andaloussi A, Han Y, Lesniak MS. Prolongation of = survival=20 following depletion of CD4+CD25+ regulatory T cells in = mice with=20 experimental brain tumors. J Neurosurg (2006) 105:430-7.
Schilling T, Stock C, Schwab A, Eder C. Functional = importance=20 of Ca2+-activated K+ channels for lysophosphatidic = acid-induced=20 microglial migration. Eur J Neurosci (2004) 19:1469-74.
Sasahara M, Fries JW, Raines EW, Gown AM, Westrum LE, = Frosch=20 MP, et al. PDGF B-chain in neurons of the central nervous = system,=20 posterior pituitary, and in a transgenic model. Cell = (1991)=20 64:217-27.
Yeh HJ, Ruit KG, Wang YX, Parks WC, Snider WD, Deuel = TF. PDGF=20 A-chain gene is expressed by mammalian neurons during = development=20 and in maturity. Cell (1991) 64:209-16.
Pringle NP, Mudhar HS, Collarini EJ, Richardson WD. = PDGF=20 receptors in the rat CNS: during late neurogenesis, PDGF=20 alpha-receptor expression appears to be restricted to = glial cells=20 of the oligodendrocyte lineage. Development (1992) = 115:535-51.
Shih AH, Dai C, Hu X, Rosenblum MK, Koutcher JA, = Holland EC.=20 Dose-dependent effects of platelet-derived growth factor-B = on=20 glial tumorigenesis. Cancer Res (2004) 64:4783-9.
Assanah M, Lochhead R, Ogden A, Bruce J, Goldman J, = Canoll P.=20 Glial progenitors in adult white matter are driven to form = malignant gliomas by platelet-derived growth = factor-expressing=20 retroviruses. J Neurosci (2006) 26:6781-90.
Heese O, Disko A, Zirkel D, Westphal M, Lamszus K. = Neural stem=20 cell migration toward gliomas in vitro. Neurooncol (2005)=20 7:476-84.
Schmidt NO, Westphal M, Hagel C, Ergun S, Stavrou D, = Rosen EM,=20 et al. Levels of vascular endothelial growth factor, = hepatocyte=20 growth factor/scatter factor and basic fibroblast growth = factor in=20 human gliomas and their relation to angiogenesis. Int J = Cancer=20 (1999) 84:10-8.
Son BR, Marquez-Curtis LA, Kucia M, Wysoczynski M, = Turner AR,=20 Ratajczak J, et al. Migration of bone marrow and cord = blood=20 mesenchymal stem cells in vitro is regulated by = stromal-derived=20 factor-1-CXCR4 and hepatocyte growth factor-c-met axes and = involves matrix metalloproteinases. Stem Cells (2006) = 24:1254-64.
Aboody KS, Brown A, Rainov NG, Bower KA, Liu S, Yang = W, et al.=20 Neural stem cells display extensive tropism for pathology = in adult=20 brain: evidence from intracranial gliomas. Proc Natl Acad = Sci USA=20 (2000) 97:12846-51.
Banisadr G, Skrzydelski D, Kitabgi P, Rostene W, = Parsadaniantz=20 SM. Highly regionalized distribution of stromal = cell-derived=20 factor-1/CXCL12 in adult rat brain: constitutive = expression in=20 cholinergic, dopaminergic and vasopressinergic neurons. = Eur J=20 Neurosci (2003) 18:1593-606.
Ehtesham M, Yuan X, Kabos P, Chung NH, Liu G, Akasaki = Y, et=20 al. Glioma tropic neural stem cells consist of astrocytic=20 precursors and their migratory capacity is mediated by = CXCR4.=20 Neoplasia (2004) 6:287-93.
Imitola J, Raddassi K, Park KI, Mueller FJ, Nieto M, = Teng YD,=20 et al. Directed migration of neural stem cells to sites of = CNS=20 injury by the stromal cell-derived factor 1alpha/CXC = chemokine=20 receptor 4 pathway. Proc Natl Acad Sci USA (2004) = 101:18117-22.
Lamprianou S, Harroch S. Receptor protein tyrosine = phosphatase=20 from stem cells to mature glial cells of the central = nervous=20 system. J Mol Neurosci (2006) 29:241-55.
Harada J, Foley M, Moskowitz MA, Waeber C.=20 Sphingosine-1-phosphate induces proliferation and = morphological=20 changes of neural progenitor cells. J Neurochem (2004) = 88:1026-39.
Liu B, Neufeld AH. Activation of epidermal growth = factor=20 receptors in astrocytes: from development to neural = injury. J=20 Neurosci Res (2007) In press.
Shintani T, Watanabe E, Maeda N, Noda M. Neurons as = well as=20 astrocytes express proteoglycan-type protein tyrosine = phosphatase=20 zeta/RPTPbeta: analysis of mice in which the = PTPzeta/RPTPbeta gene=20 was replaced with the LacZ gene. Neurosci Lett (1998) = 247:135-8.
Sorensen SD, Nicole O, Peavy RD, Montoya LM, Lee CJ, = Murphy=20 TJ, et al. Common signaling pathways link activation of = murine=20 PAR-1, LPA, and S1P receptors to proliferation of = astrocytes. Mol=20 Pharmacol (2003) 64:1199-209.
Welch WC, Kornblith PL, Michalopoulos GK, Petersen BE, = Beedle=20 A, Gollin SM, et al. Hepatocyte growth factor (HGF) and = receptor=20 (c-met) in normal and malignant astrocytic cells. = Anticancer Res=20 (1999) 19:1635-40.
Manning TJ Jr, Rosenfeld SS, Sontheimer H. = Lysophosphatidic=20 acid stimulates actomyosin contraction in astrocytes. J = Neurosci=20 Res (1998) 53:343-52.
Sharif A, Legendre P, Prevot V, Allet C, Romao L, = Studler JM,=20 et al. Transforming growth factor alpha promotes = sequential=20 conversion of mature astrocytes into neural progenitors = and stem=20 cells. Oncogene (2007) 26:2695-706.
Hulleman E, Helin K. Molecular mechanisms in = gliomagenesis.=20 Adv Cancer Res (2005) 94:1-27.
Du Y, Dreyfus CF. Oligodendrocytes as providers of = growth=20 factors. J Neurosci Res (2002) 68:647-54.
Yu N, Lariosa-Willingham KD, Lin FF, Webb M, Rao TS.=20 Characterization of lysophosphatidic acid and=20 sphingosine-1-phosphate-mediated signal transduction in = rat=20 cortical oligodendrocytes. Glia (2004) 45:17-27.
Fox MA, Colello RJ, Macklin WB, Fuss B.=20 Phosphodiesterase-Ialpha/autotaxin: a counteradhesive = protein=20 expressed by oligodendrocytes during onset of myelination. = Mol=20 Cell Neurosci (2003) 23:507-19.
Wewetzer K, Rauvala H, Unsicker K. Immunocytochemical=20 localization of the heparin-binding growth-associated = molecule=20 (HB-GAM) in the developing and adult rat cerebellar = cortex. Brain=20 Res (1995) 693:31-8.
Takeda A, Onodera H, Sugimoto A, Itoyama Y, Kogure K, = Rauvala=20 H, et al. Induction of heparin-binding growth-associated = molecule=20 expression in reactive astrocytes following hippocampal = neuronal=20 injury. Neuroscience (1995) 68:57-64.
Fukushima N, Weiner JA, Chun J. Lysophosphatidic acid = (LPA) is=20 a novel extracellular regulator of cortical neuroblast = morphology.=20 Dev Biol (2000) 228:6-18.
Buccoliero R, Futerman AH. The roles of ceramide and = complex=20 sphingolipids in neuronal cell function. Pharmacol Res = (2003)=20 47:409-19.
Farin A, Suzuki SO, Weiker M, Goldman JE, Bruce JN, = Canoll P.=20 Transplanted glioma cells migrate and proliferate on host = brain=20 vasculature: a dynamic analysis. Glia (2006) 53:799-808.
Rieger J, Wick W, Weller M. Human malignant glioma = cells=20 express semaphorins and their receptors, neuropilins and = plexins.=20 Glia (2003) 42:379-89.
Eickholt BJ, Mackenzie SL, Graham A, Walsh FS, Doherty = P.=20 Evidence for collapsin-1 functioning in the control of = neural=20 crest migration in both trunk and hindbrain regions. = Development=20 (1999) 126:2181-9.
Marin O, Yaron A, Bagri A, Tessier-Lavigne M, = Rubenstein JL.=20 Sorting of striatal and cortical interneurons regulated by = semaphorin-neuropilin interactions. Science (2001) = 293:872-5.
Mariani L, Beaudry C, McDonough WS, Hoelzinger DB, = Demuth T,=20 Ross KR, et al. Glioma cell motility is associated with = reduced=20 transcription of proapoptotic and proliferation genes: a = cDNA=20 microarray analysis. J Neurooncol (2001) 53:161-76.
Barallobre MJ, Pascual M, Del Rio JA, Soriano E. The = Netrin=20 family of guidance factors: emphasis on Netrin-1 = signalling. Brain=20 Res Brain Res Rev (2005) 49:22-47.
Shekarabi M, Kennedy TE. The netrin-1 receptor DCC = promotes=20 filopodia formation and cell spreading by activating Cdc42 = and=20 Rac1. Mol Cell Neurosci (2002) 19:1-17.
Sugimoto Y, Taniguchi M, Yagi T, Akagi Y, Nojyo Y, = Tamamaki N.=20 Guidance of glial precursor cell migration by secreted = cues in the=20 developing optic nerve. Development (2001) 128:3321-30.
Albini A, Sporn MB. The tumour microenvironment as a = target=20 for chemoprevention. Nat Rev Cancer (2007) 7:139-47.
Joyce JA. Therapeutic targeting of the tumor = microenvironment.=20 Cancer Cell (2005) 7:513-20.
Buchdunger E, Cioffi CL, Law N, Stover D, Ohno-Jones = S, Druker=20 BJ, et al. Abl protein-tyrosine kinase inhibitor STI571 = inhibits=20 in vitro signal transduction mediated by c-kit and=20 platelet-derived growth factor receptors. J Pharmacol Exp = Ther=20 (2000) 295:139-45.
Capdeville R, Buchdunger E, Zimmermann J, Matter A. = Glivec=20 (STI571, imatinib), a rationally developed, targeted = anticancer=20 drug. Nat Rev Drug Discov (2002) 1:493-502.
Yokoi K, Sasaki T, Bucana CD, Fan D, Baker CH, Kitadai = Y, et=20 al. Simultaneous inhibition of EGFR, VEGFR, and = platelet-derived=20 growth factor receptor signaling combined with gemcitabine = produces therapy of human pancreatic carcinoma and = prolongs=20 survival in an orthotopic nude mouse model. Cancer Res = (2005)=20 65:10371-80.
Demetri GD, von Mehren M, Blanke CD, Van den Abbeele = AD,=20 Eisenberg B, Roberts PJ, et al. Efficacy and safety of = imatinib=20 mesylate in advanced gastrointestinal stromal tumors. N = Engl J Med=20 (2002) 347:472-80.
Druker BJ, Talpaz M, Resta DJ, Peng B, Buchdunger E, = Ford JM,=20 et al. Efficacy and safety of a specific inhibitor of the = BCR-ABL=20 tyrosine kinase in chronic myeloid leukemia. N Engl J Med = (2001)=20 344:1031-7.
Rich JN, Reardon DA, Peery T, Dowell JM, Quinn JA, = Penne KL,=20 et al. Phase II trial of gefitinib in recurrent = glioblastoma. J=20 Clin Oncol (2004) 22:133-42.
Dai H, Marbach P, Lemaire M, Hayes M, Elmquist WF.=20 Distribution of STI-571 to the brain is limited by=20 P-glycoprotein-mediated efflux. J Pharmacol Exp Ther = (2003)=20 304:1085-92.
Leis JF, Stepan DE, Curtin PT, Ford JM, Peng B, = Schubach S, et=20 al. Central nervous system failure in patients with = chronic=20 myelogenous leukemia lymphoid blast crisis and = Philadelphia=20 chromosome positive acute lymphoblastic leukemia treated = with=20 imatinib (STI-571). Leuk Lymphoma (2004) 45:695-8.
Mellinghoff IK, Wang MY, Vivanco I, Haas-Kogan DA, Zhu = S, Dia=20 EQ, et al. Molecular determinants of the response of = glioblastomas=20 to EGFR kinase inhibitors. N Engl J Med (2005) = 353:2012-24.
Haas-Kogan DA, Prados MD, Tihan T, Eberhard DA, = Jelluma N,=20 Arvold ND, et al. Epidermal growth factor receptor, = protein kinase=20 B/Akt, and glioma response to erlotinib. J Natl Cancer = Inst (2005)=20 97:880-7.
Vredenburgh JJ, Desjardins A, Herndon JE 2nd, Dowell = JM,=20 Reardon DA, Quinn JA, et al. Phase II trial of bevacizumab = and=20 irinotecan in recurrent malignant glioma. Clin Cancer Res = (2007)=20 13:1253-9.
Crazzolara R, Johrer K, Johnstone RW, Greil R, Kofler = R,=20 Meister B, et al. Histone deacetylase inhibitors potently = repress=20 CXCR4 chemokine receptor expression and function in acute=20 lymphoblastic leukaemia. Br J Haematol (2002) 119:965-9.
Takada Y, Gillenwater A, Ichikawa H, Aggarwal BB.=20 Suberoylanilide hydroxamic acid potentiates apoptosis, = inhibits=20 invasion, and abolishes osteoclastogenesis by suppressing = nuclear=20 factor-kappaB activation. J Biol Chem (2006) 281:5612-22.
Wang W, Goswami S, Lapidus K, Wells AL, Wyckoff JB, = Sahai E,=20 et al. Identification and testing of a gene expression = signature=20 of invasive carcinoma cells within primary mammary tumors. = Cancer=20 Res (2004) 64:8585-94.
Goswami S, Sahai E, Wyckoff JB, Cammer M, Cox D, = Pixley FJ, et=20 al. Macrophages promote the invasion of breast carcinoma = cells via=20 a colony-stimulating factor-1/epidermal growth factor = paracrine=20 loop. Cancer Res (2005) 65:5278-83.
Barsky SH, Karlin NJ. Myoepithelial cells: autocrine = and=20 paracrine suppressors of breast cancer progression. J = Mammary=20 Gland Biol Neoplasia (2005) 10:249-60.
Barsky SH. Myoepithelial mRNA expression profiling = reveals a=20 common tumor-suppressor phenotype. Exp Mol Pathol (2003)=20 74:113-22.
Coussens LM, Werb Z. Inflammation and cancer. Nature = (2002)=20 420:860-7.
Dvorak HF. Angiogenesis: update 2005. J Thromb Haemost = (2005)=20 3:1835-42.
Lin EY, Nguyen AV, Russell RG, Pollard JW. = Colony-stimulating=20 factor 1 promotes progression of mammary tumors to = malignancy. J=20 Exp Med (2001) 193:727-40.
Wen PY, Yung WK, Lamborn KR, Dahia PL, Wang Y, Peng B, = et al.=20 Phase I/II study of imatinib mesylate for recurrent = malignant=20 gliomas: North American Brain Tumor Consortium Study = 99-08. Clin=20 Cancer Res (2006) 12:4899-907.
Reardon DA, Egorin MJ, Quinn JA, Rich JN, Gururangan = S,=20 Vredenburgh JJ, et al. Phase II study of imatinib mesylate = plus=20 hydroxyurea in adults with recurrent glioblastoma = multiforme. J=20 Clin Oncol (2005) = 23:9359-68.

Acknowledgements

We would like to thank David Hoelzinger, MD, for creating the=20 illustration for the fi gure and Mitsutoshi Nakada, MD, PhD, for = critical=20 evaluation of the manuscript.

Funding Information

National Institutes of = Health=20 (R01NS42262-4 to M. E. B.).

Reprint Address

Michael E. Berens, PhD, Cancer and Cell Biology Division, = Translational=20 Genomics Research Institute, 445 North Fifth Street, Phoenix, AZ = 85004=20 (e-mail: mberens@tgen.org = ).=20

Dominique B. Hoelzinger, Tim = Demuth,=20 Michael E. Berens, Cancer and Cell Biology Division, = Translational=20 Genomics Research Institute, Phoenix, AZ