SPECIAL ISSUE REVIEWS–A PEER REVIEWED FORUM

Deciphering the Signaling Events thatPromote Melanoma Tumor Cell VasculogenicMimicry and Their Link to EmbryonicVasculogenesis: Role of the Eph ReceptorsAngela R. Hess,*1,2 Naira V. Margaryan,1 Elisabeth A. Seftor,1 and Mary J.C. Hendrix1

During embryogenesis, the primordial microcirculation is formed through a process known asvasculogenesis. The term “vasculogenic mimicry” has been used to describe the manner in which highlyaggressive, but not poorly aggressive melanoma tumor cells express endothelial and epithelial markers andform vasculogenic-like networks similar to embryonic vasculogenesis. Vasculogenic mimicry is oneexample of the remarkable plasticity demonstrated by aggressive melanoma cells and suggests that thesecells have acquired an embryonic-like phenotype. Since the initial discovery of tumor cell vasculogenicmimicry by our laboratory, we have been focusing on understanding the molecular mechanisms thatregulate this process. This review will highlight recent findings identifying key signal transduction eventsthat regulate melanoma vasculogenic mimicry and their similarity to the signal transduction eventsresponsible for promoting embryonic vasculogenesis and angiogenesis. Specifically, this review will focuson the role of the Eph receptors and ligands in embryonic vasculogenesis, angiogenesis, and vasculogenicmimicry. Developmental Dynamics 236:3283–3296, 2007. © 2007 Wiley-Liss, Inc.

Key words: EphA2; vasculogenic mimicry; melanoma; VE-cadherin; metastasis

Accepted 16 April 2007

INTRODUCTION

During embryonic development theprimordial microcirculation is formedthrough a process known as vasculo-genesis (reviewed in Conway et al.,2001). Embryonic vasculogenesis in-volves the differentiation of endothe-lial precursors, known as angioblasts,which then coalesce to form a primi-tive microcirculation, often resem-bling a “honeycomb-like network” oftubular structures that are homoge-

nous in size and length. This primitivemicrocirculation is then remodeledthrough the process of angiogenesis,which involves the sprouting of newendothelial-lined vessels from pre-ex-isting vessels. This remodeling pro-cess results in the formation of endo-thelial-lined vessels of varying sizesthat eventually give rise to the adultvasculature. The factors that regulatethe process of vasculogenesis and an-giogenesis in the developing embryo

are the result of coordinated signaltransduction events in addition tophysiological factors such as shearstress and hemodynamics occurringwithin the developing embryo (re-viewed in Jones et al., 2006). Angio-genesis continues to occur throughoutadulthood during periods of woundhealing or tissue repair. Additionally,pathological forms of angiogenesis canoccur such as that which takes placeduring tumor formation.

1Children’s Memorial Research Center, Program in Cancer Biology and Epigenomics, Northwestern University Feinberg School of Medicine,Robert H. Lurie Comprehensive Cancer, Chicago, Illinois2Department of Biological and Allied Health Sciences, Bloomsburg University, Bloomsburg, PennsylvaniaGrant sponsor: Melanoma Research Foundation; Grant sponsor: US National Institutes of Health (NIH); Grant numbers: CA59702 andCA80318.*Correspondence to: Angela R. Hess, Ph.D, Children’s Memorial Research Center, Program in Cancer Biology and Epigenom-ics, Northwestern University Feinberg School of Medicine, 2300 Children’s Plaza, Box 222, Chicago, IL 60614-3394. E-mail:ahess@childrensmemorial.org

DOI 10.1002/dvdy.21190Published online 6 June 2007 in Wiley InterScience (www.interscience.wiley.com).

DEVELOPMENTAL DYNAMICS 236:3283–3296, 2007

© 2007 Wiley-Liss, Inc.

Several years ago, we described aunique characteristic of highly aggres-sive melanoma tumor cells and coinedthe term vasculogenic mimicry (VM)(reviewed in Hendrix et al., 2003;Zhang et al., 2007). VM describes theability of highly aggressive melanomatumor cells, but not poorly aggressivemelanoma tumor cells, to expressmultiple cellular phenotypes, includ-ing endothelial and epithelial associ-ated markers, and to form a vasculo-genic-like network of matrix-richpatterns when cultured on a three-dimensional matrix in vitro in a man-ner similar to the process of embry-onic vasculogenesis (Maniotis et al.,1999). Remarkably, the formation ofthese structures in vitro recapitulatesmatrix-rich patterned networks foundin patients’ melanoma tissues corre-lating with an increased risk of mela-noma metastasis resulting in a poorclinical outcome (Folberg et al., 1993).Further studies revealed that the net-works formed by aggressive mela-noma tumor cells were capable of con-ducting fluorescent dyes in vitro(Maniotis et al., 1999). Additionally, itwas established that these extravas-cular networks physiologically con-nected with the mouse vasculature incutaneous melanoma xenografts (Rufet al., 2003). Together, these data sug-gested that this primitive microcircu-lation may act both as a complimen-tary means of tumor perfusion as wellas an additional conduit for metasta-sis. In recent years, VM has been re-ported in a number of different tumortypes including breast, prostate, ovar-ian, Ewing sarcoma, and lung carci-noma (Kobayashi et al., 2002; Liu etal., 2002; Passalidou et al., 2002;Sharma et al., 2002; Shirakawa et al.,2002; Sood et al., 2002; van der Schaftet al., 2005). VM is one example of theplasticity of aggressive tumor cellsand suggests a reversion to a moreprimitive embryonic phenotype.

To better understand the process ofVM, our laboratory has been investi-gating the molecular mechanisms un-derlying this newly characterizedform of tumor neovascularization asan attempt to identify new therapeu-tic targets, particularly for aggressivemelanoma. We have focused our at-tention on the signal transductionmechanisms that promote VM and todate have identified various roles for

EphA2, vascular endothelial cadherin(VE-cadherin), focal adhesion kinase(FAK), phosphoinositide 3-kinase(PI3K), and extracellular regulatedkinase 1 and 2 (Erk1/2) in regulatingthe process of melanoma VM as wellas melanoma aggressiveness in gen-eral (Hendrix et al., 2001; Hess et al.,2001, 2003, 2005, 2006). Interestingly,all of these signaling molecules play arole in embryonic vasculogenesis aswell as adult angiogenesis, particu-larly with respect to the Eph recep-tors. Therefore, the goal of this reviewis to understand the link between thesignal transduction mechanisms thatregulate embryonic vasculogenesisand angiogenesis with that of mela-noma vasculogenic mimicry, with aparticular focus on the role of the Ephreceptors and their downstream effec-tors.

EPH RECEPTORSIGNALING DURINGEMBRYONICVASCULOGENESIS

The Eph family of receptor tyrosinekinases (RTKs) is the largest family ofRTKs, consisting of 14 members. Thereceptors are divided into two classes,EphA and EphB, based on the homol-ogy of their extracellular domain (EphNomenclature Committee, 1997). Theligands for this family are also mem-brane bound and/or secreted (termedthe ephrins) and are divided into twoclasses based on their membrane link-age: class A contains ligands that arebound by glycosylphosphatidylinositol(GPI) linkage, while the B class of li-gands contains members that possessboth transmembrane and cytoplasmicregions (Eph Nomenclature Commit-tee, 1997). Receptor/ligand engage-ment results in phosphorylation of thereceptor on tyrosine residues, thus re-sulting in an induction of the recep-tor’s kinase activity. Additionally, ithas been noted that several receptor/ligand pairs, particularly within theEphB/ephrin-B subclass, are capableof bi-directional signaling. The Ephfamily of receptors and ligands hasbeen found to play key roles duringdevelopment where they regulate var-ious processes such as directing ax-onal (Henkemeyer et al., 1996; Orioliet al., 1996; Park et al., 1997; Wangand Anderson, 1997) and neural crest

cell migrations (Krull et al., 1997;Smith et al., 1997; Wang and Ander-son 1997), regulating axonal bundling(Orioli et al., 1996; Winslow et al.,1995), preventing mixing of differentcell populations during embryogene-sis (Mellitzer et al., 1999), and me-diating angiogenesis (Adams et al.,1999; Brantley-Sieders and Chen,2004; Brantley-Sieders et al., 2004,2005; Chen et al., 2006; Cheng et al.,2002; Gale et al., 2001; Gerety et al.,1999; Shin et al., 2001; Wang et al.,1998; Zhong et al., 2001).

Within the last decade, several re-ports of Eph RTKs and ephrins haveindicated that these molecules playkey roles during embryonic vasculo-genesis, specifically, the EphB2,EphB3, and EphB4 receptors alongwith their respective ligands, eph-rin-B1 and ephrin-B2. Studies haveshown that ephrin-B2 marks endothe-lial cells destined to form the lining ofarteries, whereas its principle recep-tor EphB4 marks endothelial cellsdestined to form the lining of veins(Adams et al., 1999; Gerety et al.,1999; Shin et al., 2001; Wang et al.,1998). The importance of these pro-teins in embryonic vasculogenesis ishighlighted in homozygous knockoutmice of either ephrin-B2 or EphB4,both of which result in embryonic le-thality due to a failure in remodelingof the capillary plexus into functionalveins and arteries (Gerety et al., 1999;Wang et al., 1998). The remarkablefinding that homozygous knockouts ofeither ephrin-B2 or EphB4 resulted inessentially the same phenotypical out-come is indicative of bi-directional sig-naling between this ligand and recep-tor pair. Moreover, it has beendemonstrated that the expression ofboth ephrin-B2 and EphB4 can persistthroughout adulthood. Shin and col-leagues utlized ephrin-B2taulacZ/� in-dicator mice to provide evidence sug-gesting that ephrin-B2 can mediatevarious processes of adult angiogene-sis including tumor neovasculariza-tion (Shin et al., 2001).

In addition to the expression ofEphB4 and ephrin-B2 during embry-onic vasculogenesis and angiogenesis,it has been demonstrated that EphB2,EphB3, and ephrin-B1 are also ex-pressed in the embryo and play a roleduring embryonic vasculogenesis andangiogenesis (Adams et al., 1999). In-

3284 HESS ET AL.

terestingly when the phenotype ofdouble ephB2/ephB3 knockout micewas compared with the phenotype ofephrin-B2 knockout mice, strikingsimilarities in vascular defects werefound (Adams et al., 1999). Given thatEphB2 was found to be expressedwithin the embryonic mesenchymewhereas EphB3 was found to be ex-pressed in endothelial cells (Adams etal., 1999), it was suggested that thecommunication between the embry-onic mesenchyme and endothelialcells is critical for embryonic vasculo-genesis and angiogenesis and thatthis communication is in part medi-ated by the Eph receptors and ligands.

Although the role of EphB2, EphB3,EphB4, ephrin-B1, and ephrin-B2 inmediating embryonic vasculogenesisis clear, the role of EphA2 and its prin-ciple ligand ephrin-A1 remains to bedetermined. Early studies reportedthe presence of ephrin-A1 in areas ofembryonic vasculogenesis and angio-genesis, suggesting a role for eph-rin-A1 in this process; however,EphA2 has not been found to be ex-pressed during embryonic vasculogen-esis or angiogenesis, and in factEphA2 knock-out mice are viable(Brantley-Sieders et al., 2004; Flenni-ken et al., 1996; McBride and Ruiz1998). Despite a lack of evidence con-necting EphA2 and ephrin-A1 withembryonic vasculogenesis, there is asignificant amount of evidence for thisEph receptor/ligand pair in adult an-giogenesis, which will be the focus ofthe next section.

THE ROLE OF EPHA2 ANDEPHRIN-A1 IN ADULTANGIOGENESIS

Angiogenesis, which involves the for-mation of a new endothelial-lined vas-culature from a preexisting one, occursthroughout adulthood under both nor-mal and pathological conditions. One ofthe first indications implementingEphA2 and ephrin-A1 as mediators ofangiogenesis was reported by Pandeyand colleagues who demonstrated thatTNF-� could upregulate the expressionof ephrin-A1 in human umbilical veinendothelial cells (HUVECs), resultingin EphA2 phosphorylation and stimula-tion of angiogenesis in vivo as well asendothelial cell chemotaxis in vitro(Pandey et al., 1995). Since those initial

observations, there have been manypublished reports aiming to address therole of both EphA2 and ephrin-A1 inmediating angiogenesis.

Using various models for angiogen-esis, it has been demonstrated thataddition of soluble EphA2 can signifi-cantly inhibit ephrin-A1 and VEGFinduced angiogensis in vivo (Chen etal., 2006; Cheng et al., 2002). Interest-ingly, addition of soluble EphA2 hadno effect on bFGF-stimulated angio-genesis, suggesting that EphA2 sig-naling is specific to ephrin-A1- andVEGF-induced angiogenesis (Chenget al., 2002). In vitro studies usingisolated endothelial cells uncoveredseveral potential mechanisms for howEphA2 promotes angiogenesis. Chenand colleagues demonstrated that sol-uble EphA2 inhibited both ephrin-A1and VEGF induced endothelial migra-tion and tube formation (Chen et al.,2006). Additionally, Cheng and col-leagues demonstrated that solubleEphA2 abrogated the positive effectsof VEGF on endothelial cell survivalunder serum-free conditions, sug-gesting that EphA2 serves as a sur-vival factor for endothelial cells inconjunction with VEGF (Cheng etal., 2002). Interestingly, these inves-tigators found that treatment of hu-man microvascular endothelial cells(HMECs) and HUVECs with VEGFresulted in the upregulation of eph-rin-A1 in these cells concomitantwith EphA2 phosphorylation, simi-lar to the observations reported byPandey and colleagues with respectto TNF-� treatment (Cheng et al.,2002). Based on these data, it is ap-parent that signaling through theEphA2 receptor is important for me-diating many of the properties char-acteristic of endothelial cells engag-ing in angiogenesis.

Several signal transduction path-ways have been demonstrated down-stream of EphA2/ephrin-A1 duringangiogenesis. Recently, it was shownthat ephrin-A1 stimulation of EphA2phosphorylation on endothelial cellscan induce PI3K activation concomi-nant with an increase in Rac1 activity,thus promoting endothelial cell migra-tion and vascular assembly (Brantley-Sieders et al., 2004). Moreover, thereis evidence to suggest that PI3K-me-diated activation of Rac1 GTPase maybe partially regulated through Vav2

and Vav3 guanine nucleotide ex-change factors (Hunter et al., 2006).Another study by Ojima and col-leagues using bovine retinal endothe-lial cells (BRECs) demonstrated thatVEGF-induced extracellular regu-lated kinase 1 and 2 (Erk1/2) and Aktphosphorylation was inhibited by pre-treatment with ephrin-A1 (Ojima etal., 2006). Furthermore, this inhibi-tion was found to reduce the effects ofVEGF-stimulated endothelial cell mi-gration, tube formation, and prolifer-ation. Together, these data offer in-sight into the complexity of thesignaling events that lie downstreamof EphA2 and ephrin-A1, which act topromote angiogeneis.

THE ROLE OF EPHA2 ANDEPHRIN-A1 IN TUMORNEOVASCULARIZATION

As with any other tissue within thebody, a growing tumor mass needs aconstant supply of nutrients as well asan avenue for the removal of wasteproducts; therefore, development of afunctional vascular system is neces-sary to sustain the growth of a tumor.The notion that a tumor’s blood supplycan be acquired through a process ofangiogenesis was first proposed over30 years ago (Folkman, 1995). Sincethen, it is now recognized that tumorscan acquire a blood supply throughalternative means including vessel co-option, intussuseception, recruitmentof endothelial progenitor cells, andvasculogenic mimicry (reviewed inDome et al., 2007).

EphA2 is overexpressed in a largenumber of tumor types including mel-anoma, prostate, breast, ovarian, pan-creatic, and lung and is often linkedwith a poor clinical outcome (Duxburyet al., 2004a; Kinch et al., 2003; Mu-dali et al., 2006; Thaker et al., 2004;Walker-Daniels et al., 1999; Zelinskiet al., 2001; Zeng et al., 2003). Giventhat EphA2 is often overexpressed invarious tumor types combined with in-creasing amounts of evidence support-ing a role for EphA2 in promoting an-giogenesis, it is natural to hypothesizethat overexpression of EphA2 by tu-mor cells may promote tumor neovas-cularization. Recently, there havebeen a number of reports that supportthe hypothesis that upregulation ofEphA2 promotes tumor neovascular-

EPH RECEPTOR SIGNALING AND MELANOMA VM 3285

ization. One of the earliest examplesusing immunohistochemistry (IHC)performed on xenografts of humanMDA-MB-435 breast cancer cells orKS1767 Kaposi’s sarcoma cells re-vealed that EphA2 and ephrin-A1were expressed within the tumor vas-culature as well as the tumor cellsthemselves (Ogawa et al., 2000). Toconfirm the clinical relevance of thisexpression pattern for EphA2 andephrin-A1 in xenografted tumors,Ogawa and colleagues performed IHCon a number of different clinical spec-imens covering a wide variety of tu-mor types with similar results. Thesedata set the stage for numerous re-ports supporting a role for EphA2 inmediating tumor neovascularization.

Much of the work investigating arole for EphA2 in mediating tumor an-giogenesis has been done using twodifferent tumor models: RIP-Tagtransgenic model of angiogenic depen-dent pancreatic islet cell carcinomaand the 4T1 model of metastatic mam-mary adenocarcinoma (Brantley-Sied-ers et al., 2005, 2006; Brantley et al.,2002; Cheng et al., 2003). Using thesemodels, it has been shown that block-ing EphA2 inhibits tumor angiogene-sis manifested as a decrease in tumorvolume and microvessel density. IHCanalysis of tumors taken from thesetwo models revealed that ephrin-A1was predominantly expressed by thetumor cells whereas EphA2 was pri-marily expressed within the tumorvasculature, suggesting that eph-rin-A1 expressed on the tumor cellswas responsible for attracting EphA2expressing endothelial cells. In sup-port of this hypothesis, Brantley-Sied-ers and colleagues demonstrated thattransplanting 4T1 mammary adeno-carcinomas cells expressing ephrin-A1into EphA2-deficient mice resulted ina significant decrease in tumor vol-ume concomitant with a decrease inmicrovessel density (Brantley-Siederset al., 2005). Additionally, EphA2-de-ficient endothelial cells displayed sig-nificantly reduced migratory capacityin response to 4T1 mammary adeno-carcinoma cells in vitro (Brantley-Sieders et al., 2005). Moreover, down-regulation of ephrin-A1 in metastaticmammary tumor cells using siRNAresulted in a reduction in tumor mi-crovessel density in vivo as well astumor cell induced endothelial migra-

tion in vitro (Brantley-Sieders et al.,2006). Lastly, as proof of principle,these authors overexpressed eph-rin-A1 in non-metastatic mammarytumor cells, which resulted in an in-crease in microvessel density in vivoand tumor-cell-induced migration invitro (Brantley-Sieders et al., 2006).Together, these results highlight animportant role for both eprhin-A1 andEphA2 in promoting tumor neovascu-larization. However, the mechanismunderlying this pathogenesis is justbeginning to emerge.

To understand how EphA2 is mediat-ing tumor neovascularization, Chengand colleagues investigated the role ofVEGF in promoting angiogenesis usingthe RIP-Tag transgenic model (Chenget al., 2003). They found that condi-tioned media from a � islet carcinomacell line induced endothelial cell migra-tion, and that this response was inhib-ited by the addition of blocking antibod-ies to VEGF or addition of solubleEphA2 both in vitro and in vivo, sug-gesting that EphA2 can mediate VEGFinduced angiogenesis in this model.Similar results have been reported us-ing an ovarian cancer model where itwas demonstrated that treatment withagonistic EphA2 antibodies, which actto decrease the levels of EphA2 ex-pressed by tumor cells, resulted in adecrease in the levels of VEGF but notbFGF, and that the decrease in VEGFwas concomitant with a decrease in Srcphosphorylation as well as a decrease inmicrovessel density in vivo (Landen etal., 2006). Furthermore, it was recentlyreported that downregulation of eph-rin-A1 in metastatic mammary cells re-sulted in a downregulation of VEGF ex-pression in these cells; likewise,overexpression of ephrin-A1 in non-metastatic mammary cells resulted in asignificant increase in VEGF produc-tion (Brantley-Sieders et al., 2006).These results suggested that ephrin-A1can regulate angiogenesis through boththe secretion of VEGF and by activationof EphA2 on the surface of tumor-asso-ciated endothelial cells.

THE ROLE OF EPHA2 INMEDIATING MELANOMATUMOR CELLVASCULOGENIC MIMICRY

As mentioned previously, a tumor’sblood supply can be acquired through

a number of different mechanisms.Vasculogenic mimicry describes theunique ability of highly aggressivemelanoma tumor cells to express mul-tiple cellular phenotypes, includingendothelial and epithelial associatedmarkers and matrix-rich vasculo-genic-like networks when cultured ona three-dimensional matrix in vitro,mimicking embryonic vasculogenesis,and resembling the matrix-rich pat-terned networks found in patient tu-mors correlating with an increasedrisk of metastatic disease (Folberg etal., 1993; Maniotis et al., 1999). Stud-ies using mouse xenograft modelshave suggested that VM can serve as afunctional means for tumor perfusionthat may complement classical angio-genesis (Ruf et al., 2003). Addition-ally, VM may serve as an additionalroute for tumor metastasis. Giventhat the most significant health threatto patients with melanoma is deathdue to metastatic disease (Balch et al.,2004), identifying the unique charac-teristics of aggressive melanoma cellsthat enable them to engage in VMmay offer new therapeutic avenues inwhich to target malignant melanoma.

To compare the molecular signatureof highly versus poorly aggressive mel-anoma tumor cells, microarray analysiswas performed (Bittner et al., 2000;Seftor et al., 2002). These data revealeda dysregulated genotype wherebyhighly aggressive melanoma tumorcells expressed various endothelial andepithelial genes important for both em-bryonic vasculogenesis and angiogene-sis, including many different Eph recep-tors and ligands (Table 1; Bittner et al.,2000; Seftor et al., 2002). Given the roleof the Eph receptors and their ligandsin mediating many aspects of embry-onic development combined with re-ports identifying the expression of mul-tiple Eph-family receptor tyrosinekinases in human and mouse embry-onic stem cells (Baharvand et al., 2006;Lickliter et al., 1996) as well as in he-matopoietic stem cells (Lazarova et al.,2006) suggested that the aggressivemelanoma tumor cells had undergonetransdifferentiation to acquire a moreembryonic-like phenotype. These data,in combination with data supporting arole for Eph receptors in embryonic vas-culogenesis as well as angiogenesis,suggest that expression of these Eph

3286 HESS ET AL.

receptors and ligands are playing a rolein mediating melanoma VM.

Our studies have utilized an in vitromodel system in which to study theprocess of melanoma VM, whereby ag-gressive melanoma tumor cells arecultured on a three-dimensional type1 collagen matrix for a period of 6days, thus allowing the investigationof the signaling mechanisms regulat-ing this process over time (Fig. 1). Us-ing this model, we demonstrated thatduring VM, there is an increase intyrosine phosphorylation within theaggressive melanoma tumor cells asthey form vasculogenic-like structuresin vitro, as illustrated in Figure 2,suggesting that aggressive melanomacells engaged in VM have an increasein signal transduction events medi-ated by tyrosine phosphorylation(Hess et al., 2001). To confirm thesefindings, we analyzed the difference intyrosine phosphorylated proteins be-tween aggressive melanoma tumorcells engaged in VM and poorly ag-gressive melanoma tumor cells unableto engage in VM. We reported thataggressive melanoma tumor cells hada different profile of tyrosine phos-phorylated proteins compared topoorly aggressive melanoma tumorcells (Hess et al., 2001), and identifiedEphA2 as a predominant receptor ty-rosine kinase found to be phosphory-lated in aggressive melanoma cells.These results, in combination withthose obtained from the microarrayanalysis identifying EphA2 as themost overexpressed of all Eph recep-

tors in the aggressive melanoma tu-mors cells, suggested that increasedEphA2 may potentiate downstreamsignaling events necessary for mela-noma VM. As proof of principle, weutilized antisense oligonucleotidestrategies to transiently downregu-late the expression of EphA2 in ag-gressive melanoma tumor cells,which resulted in the inability ofthese tumor cells to engage in VM(Hess et al., 2001). These data iden-tified EphA2 as an important medi-ator of melanoma VM and likewisebegan to establish a signal transduc-tion–mediated mechanism for thisphenomenon.

In an effort to better understand therole of EphA2 in mediating melanomaVM and metastasis as a whole, wehave since developed aggressive mel-anoma tumor cells that have stablydownregulated levels of EphA2 andhave been characterizing them basedon VM potential, invasion, prolifera-tion, and tumor formation potential invivo. As our previous report suggested(Hess et al., 2001), downregulation ofEphA2 inhibits the ability of aggres-sive melanoma cells to engage in VMin vitro (Fig. 3). Furthermore, down-regulation of EphA2 resulted in a de-crease in the invasive capacity of ag-gressive melanoma tumor cells invitro, as demonstrated in Figure 4. Tofurther understand the role of EphA2in promoting melanoma aggressive-ness, we utilized an orthotopic mousemodel for cutaneous melanoma andchallenged the aggressive melanoma

cells having stably downregulatedEphA2 to form tumors in vivo. Thedata presented in Table 2 demon-strate a significant inability of theseaggressive tumor cells lacking suffi-cient amounts of EphA2 to form tu-mors in vivo, thus underscoring theimportance of EphA2 in mediatingmelanoma tumor formation. Thequestion remains as to the underlyingmechanism resulting in tumor inhibi-tion. There are at least two possibili-ties: (1) a decrease in microvessel den-sity within the tumor, as a result ofeither inhibiting tumor angiogenesisand/or vasculogenic mimicry; and/or(2) tumor growth inhibited due to aninability of the tumor cells to prolifer-ate within a mouse microenviron-ment.

Recently, there have been reportsattempting to investigate a correla-tion between EphA2 and microvesseldensity in both colorectal and ovariancarcinomas (Kataoka et al., 2004; Linet al., 2007). In both cases, there was asignificant association between in-creased EphA2 expression in the tu-mor cells with increased microvesseldensity. Furthermore, Lin and col-leagues reported that increases inEphA2 expression and microvesseldensity corresponded to a decrease inpatient survival (Lin et al., 2007).These data suggest that the inabilityof aggressive melanoma cells lackingsufficient EphA2 to form tumors inmice may be due to impaired tumorneovascularization, especially givenprevious reports that firmly establisha role for EphA2 in mediating angio-genesis in conjunction with the inabil-ity of these cells to undergo VM invitro.

Early reports first identifying theoverexpression of EphA2 in melanomasuggested that eprhin-A1 could act asa growth factor for melanoma cells(Easty et al., 1995), likewise it hasbeen reported that EphA2 may act asa survival factor for endothelial cells(Cheng et al., 2003). Additionally,Straume and Akslen (2002) found thatexpression of EphA2 in patient sam-ples of melanoma was associated withincreased tumor cell proliferation asmeasured by Ki-67 positively. To as-sess the role of EphA2 in mediatingmelanoma tumor cell proliferation, weexamined the proliferation capacity ofthe aggressive melanoma tumor cells

TABLE 1. Expression of Eph Receptors and Ligands in AggressiveVersus Poorly Aggressive Human Melanoma Tumor Cellsa

Gene Unigene Fold change

EphA2 Hs. 171596 1 14.8–58EphA3 Hs. 123642 2 2.3–45EphB1 Hs. 116092 1 4.3EphB2 Hs. 523329 1 2.8–34.7EphB4 Hs. 437008 2 1.6–2.3Ephrin-A1 Hs. 516664 2 3.0–10.8Ephrin-A4 Hs. 449913 1 1.3Ephrin-B1 Hs. 144700 1 3.9Ephrin-B2 Hs. 149239 1 3.7–7.5Ephrin-B3 Hs. 26988 2 3.7–6.5

aAltered gene expression of Eph receptors and ligands was identified by cDNA and/oroligonucleotide microarray analysis. Data are reported as a fold change in geneexpression between highly aggressive melanoma tumor cells versus poorlyaggressive melanoma tumor cells.

EPH RECEPTOR SIGNALING AND MELANOMA VM 3287

with stably downregulated EphA2and found a significant decrease in theability of these cells to proliferate invitro, as demonstrated in Figure 5.These data suggest that EphA2 maymediate melanoma tumor formationin vivo due to its direct effects on theproliferation capacity of these cells.

VE-CADHERIN MEDIATESEPHA2 EXPRESSION ANDPHOSPHORYLATIONDURING MELANOMA VM

In addition to the upregulation ofEphA2 in aggressive melanoma cells,microarray analysis also revealed the

upregulation of VE-cadherin. VE-cad-herin is an endothelial-specific cad-herin that plays an important role inregulating vascular morphology andstability (Dejana et al., 1999). VE-cad-herin is also critically important forembryonic vasculogenesis as VE-cad-herin knock-out mice die midgestationdue to large vascular malformations(Carmeliet et al., 1999). It is interest-ing to note that a link between EphA2and VE-cadherin has never been es-tablished in endothelial cells, eventhough they both are expressed in en-dothelial cells and both play a role intumor neovascularization; however, alink between EphA2 and E-cadherin

has been established in breast cancercells as well as in embryonic stem cells(Orsulic and Kemler 2000; Zantek etal., 1999). Studies have shown thatE-cadherin is necessary for the stabi-lization of EphA2 on the surface of themembrane at cell–cell adhesions al-lowing for proper interaction betweenEphA2 and ephrin-A1. Restoring ap-propriate interactions between EphA2and ephrin-A1 can downregulatemany of the aggressive properties ofbreast cancer cells overexpressingEphA2. Based on these observationsand the fact that both EphA2 and VE-cadherin are necessary for melanomaVM (Hendrix et al., 2001; Hess et al.,

Fig. 1. Bright-field microscopy of aggressive human melanoma tumor cells (A) and poorly aggressive melanoma tumor cells (B) cultured onthree-dimensional type 1 collagen matrix for 6 days and stained with Periodic-Acid Schiff reagent (PAS) without hematoxylin counterstain. The insetin A shows a cross-section of the extracellular matrix-rich networks that contain lumen-like structures. Scale bar � 100 �m (A,B). Inset: Image wasviewed using a 63� oil immersion lens.

Fig. 2. Bright-field and immunofluorescence microscopy showing co-localization of phosphotyrosine and F-actin proteins within areas of VM inaggressive melanoma tumor cells cultured on three-dimensional type 1 collagen matrix for 4 days. A,B: Three-dimensional cultures demonstratingbright-field image (A) with corresponding fluorescent image (B), dual-labeled with phosphotyrosine proteins (PY-20; FITC) and F-actin (phalloiden;Texas Red). Arrows indicate matrix-rich VM networks. Scale bar � 100 �m for A and B.

3288 HESS ET AL.

2001), we sought to understand therelationship between EphA2 and VE-cadherin in aggressive melanomacells. Using dual-labeled immunofluo-rescence, we observed that both VE-cadherin and EphA2 localized to areasof cell–cell adhesion, both in vitro andin patients’ tumors classified as hav-ing a high metastatic potential (Hesset al., 2006). We validated these obser-vations using co-immunoprecipitationexperiments and found that EphA2and VE-cadherin formed an associa-tion. Furthermore, increased phos-phorylation of EphA2 or knock-downof EphA2 using anti-sense oligonucle-otides resulted in no change in thelocalization of VE-cadherin on the sur-face of aggressive melanoma cells.However, knock-down of VE-cadherinresulted in a dramatic redistributionof EphA2 on the cell surface as well asa downregulation of EphA2 phosphor-ylation. These results suggested thatVE-cadherin may act to stabilizeEphA2 on the surface of the aggres-sive melanoma cells, thus potentiat-ing signals important for melanomaVM. These results are contrary tothose found in breast cancer cells, andsuggest that tumors derived from non-epithelial cells use EphA2 signaltransduction–mediated mechanisms

to promote tumor cell aggressiveness,rather then suppress it. These dataalso suggest that perhaps EphA2 ex-pression in aggressive melanoma sig-nals in a manner similar to EphA2expressed on the surface of endothe-lial cells undergoing angiogenesis,thereby promoting the formation ofvasculogenic-like networks. The sig-nal transduction pathways that liedownstream of these two importantcell surface–associated proteins arecurrently under investigation. Thereare many possibilities including sig-naling through phosphoinositide 3-ki-nase and focal adhesion kinase, bothof which have been found to play rolesin tumor neovascularization and sig-nal downstream of EphA2 and VE-cadherin (Carmeliet et al., 1999;Carter et al., 2002; Duxbury et al.,2004b; Miao et al., 2000; Qi andClaesson-Welsh, 2001).

PHOSPHOINOSITIDE 3-KINASE SIGNALING AS AMEDIATOR OF MELANOMAVM

Phosphoinositide 3-kinase (PI3K) is acytoplasmic lipid kinase consisting of aregulatory subunit, p85, and a catalytic

subunit, p110. Once activated, PI3Kphosphorylates phosphoinositide-4,5-bisphosphate (PIP2) to form phospho-inositide-3,4,5-trisphosphate (PIP3),which regulates a variety of cellularfunctions (reviewed in Vanhaesebroeckand Waterfield 1999). PI3K activity hasbeen implicated in the recruitment of avariety of cell-signaling components tothe plasma membrane where they caninduce cell motility and survival. Therole of PI3K in promoting tumor pro-gression through increased invasion,migration, survival, and induction of tu-mor angiogenesis has been reported inmany different tumor types (reviewedin Brader and Eccles, 2004). PI3K andthe signal transduction pathways thatit activates have become importanttherapeutic targets for a number of dif-ferent tumor types.

PI3K has been shown by two inde-pendent laboratories to interact withEphA2, using a yeast two-hybridscreen, and has been reported toplay a role in mediating angiogene-sis (Brantley-Sieders et al., 2004;Pandey et al., 1994). Based on thesedata we began investigating the roleof PI3K in mediating melanoma VM.We found that addition of LY294002,a specific inhibitor of PI3K, inhibitedthe ability of aggressive melanomatumor cells to engage in VM (Hess etal., 2003). In order to elucidate amechanism for this apparent role ofPI3K in mediating melanoma VM,we investigated the ability of PI3Kto mediate the activities of mem-brane type 1 matrix metalloprotein-ase (MT1-MMP) and matrix metallo-proteinase-2 (MMP-2), and thecleavage of the laminin 5�2 chaininto promigratory fragments, eventspreviously shown to play a role inpromoting melanoma VM (Hess etal., 2003; Seftor et al., 2001). Wefound that treatment with LY294002reversibly blocked the activity ofMMP-2 as well as the expression andactivity of MT1-MMP, and the cleav-age of the laminin 5�2 chain. Theseresults suggested that signalingthrough PI3K could promote mela-noma VM by regulating the activi-ties of MT1-MMP and MMP-2 and,ultimately, the cleavage of the lami-nin 5�2 chain into promigratoryfragments.

Fig. 3. Bright-field microscopy of untransfected human cutaneous C8161 metastatic melanomacells (A; C8161), neo-transfected (B; C8161-Neo), or C8161 EphA2 knock-downs (C; EphA2 AS[AS; antisense] 0.2-1 and EphA2 AS 0.2-2) seeded on three-dimensional type 1 collagen matrix for4 days. Arrows indicate VM networks. Scale bar � 100 �m (A–D).

EPH RECEPTOR SIGNALING AND MELANOMA VM 3289

FOCAL ADHESION KINASESIGNALING AS AMEDIATOR OF MELANOMAVM

Focal adhesion kinase (FAK) is a cy-toplasmic kinase largely responsiblefor signaling events downstream of in-tegrins (extracellular matrix recep-tors). Integrin clustering as triggeredby binding to extracellular matrixcomponents results in FAK phosphor-ylation and subsequent activation ofthis kinase domain. Activation of FAKin turn activates a plethora of signal-ing events that act to promote manydifferent cellular processes, includingcell survival, migration, and invasion(reviewed in Cox et al., 2006; Schla-epfer et al., 1999). Unregulated in-creases in cell survival mechanisms,migration, and invasion can contrib-ute to a tumor’s ability to grow andmetastasize to distant sites within thebody. Therefore, FAK is often found tobe overexpressed and/or constitutivelyactive in numerous tumor types in-cluding melanoma, prostate, thyroid,colorectal, ovarian, and oral tumors(Han et al., 1997; Maung et al., 1999;Owens et al., 1996; Schneider et al.,2002; Sood et al., 2004; Tremblay etal., 1996). Additionally FAK mediatesmany aspects of angiogenesis by reg-ulating endothelial cell proliferation,survival, and migration. The impor-tance of FAK signaling in mediatingthese events is highlighted in a recentreport by Shen and colleagues whodemonstrated that endothelial cellspecific knockout of FAK resulted indefective angiogenesis during late em-bryogenesis resulting in embryonic le-thality (Shen et al., 2005).

Several reports have linked EphA2and ephrin-A1 to FAK signaling. Miaoand colleagues first reported that inPC-3 prostate cancer cells, stimula-tion of EphA2 with ephrin-A1 resultedin a decrease in FAK phosphorylationconcomitant with a decrease in inte-grin-mediated cell adhesion, spread-ing, and migration, suggesting thatsignaling through EphA2 acts tonegatively regulate FAK function(Miao et al., 2000). In contrast tothese observations, Carter and col-leagues demonstrated that eph-rin-A1 induced spreading in EphA2expressing NIH3T3 cells or mouseembryonic fibroblasts (MEF) derived

from FAK�/� or p130Cas�/� mice,but not in NIH3T3 cells or mouseembryonic fibroblasts derived fromFAK�/� or p130cas�/� mice (Carteret al., 2002). Moreover, expression ofconstitutively active EphA2 inNIH3T3 cells allowed them to spreadwhen plated on poly-L-lysine in amanner similar to that observedwhen plated on ephrin-A1 suggest-ing that eprhin-A1 and EphA2 sig-naling through FAK can promote celladhesion and spreading (Carter etal., 2002). In support of these obser-vations, Duxbury and colleaguesdemonstrated that in pancreatic ad-enocarcinoma cells, overexpressionof EphA2 could induce a FAK-depen-dent increase in MMP-2 expressionrelative to an increase in invasivepotential (Duxbury et al., 2004b).Furthermore, treatment of thesecells with ephrin-A1 resulted in thedownregulation of EphA2 and subse-quent dephosphorylation of FAKconcomitant with a decrease inMMP-2. These results demonstrated

that EphA2 signaling through FAKcan promote cellular adhesion andinvasion. Similar results were re-cently reported by Liu and col-leagues, who observed that forcedexpression of ephrin-A1 in gliomacell lines resulted in the degradationof EphA2 concomitant with a down-regulation of FAK resulting in aninhibition of cellular migration, pro-liferation, and anchorage indepen-dent growth (Liu et al., 2007). Col-lectively, these studies suggest thatsignaling through EphA2 and eph-rin-A1 can result in either promotionor inhibition of cellular adhesion,migration, and invasion dependingon the cell type.

Given that FAK plays a role in me-diating angiogenesis and that signal-ing between EphA2 and FAK can pro-mote tumor cell aggressiveness, weinvestigated a role for FAK in mediat-ing many of the characteristics of anaggressive melanoma tumor cell in-cluding VM (Hess et al., 2005; Hessand Hendrix, 2006). We showed that

Fig. 4. Percent invasion for untransfected C8161 cells (C8161), neo-transfected C8161 cells(C8161-Neo), or C8161 EphA2 knock-down cells (EphA2 AS 0.2-1 and EphA2 AS 0.2-2) wascalculated as a percentage of cells able to invade through a matrix (collagen IV, laminin, andgelatin)-coated polycarbonate membrane within a 24-hr period using an in vitro invasion chambercompared with the total number of cells seeded and normalized to control. The 50–60% decreasein invasion in both C8161 EphA2 AS clones compared to untransfected C8161 cells was found tobe statistically significant using a student’s t-test (*P 0.001).

3290 HESS ET AL.

by expressing FAK-related non-ki-nase (FRNK), which acts as a domi-nant negative FAK protein, in aggres-sive melanoma cells we could inhibitmelanoma VM, and significantly re-duce tumor cell invasion, migration,proliferation, and clonogenicity (Hesset al., 2005; Hess and Hendrix, 2006).

As an attempt to understand amechanism for how FAK signalingmay be regulating melanoma tumorcell invasion, migration, and VM, wefirst examined the proteolytic en-

zymes known to be important forthese biological functions in severaltumor cell types including mela-noma. In particular, there have beenreports linking both the urokinase-type plasminogen activator receptor(uPAR)/urokinase system and ma-trix metalloproteinases (MMPs) withFAK signaling (guirre Ghiso, 2002;Hauck et al., 2001, 2002; Nguyen etal., 2000). We found urokinase activ-ity to be greatly reduced by FRNKexpression in aggressive melanoma

cells; however, we did not see an ef-fect on MMP-2 or MT1-MMP (Hesset al., 2005). These data are in con-trast to other reports linking FAKsignaling with the regulation ofMMPs. However, this discrepancycould be due to the specificity of FAKsignaling in different cell types (epi-thelial vs. mesenchymal) or in re-sponse to stimulation from differentextracellular matrix components(i.e., fibronectin vs. collagen). To-gether these results support previ-ous observations linking EphA2 andFAK signaling in tumor cell aggres-siveness; however, the signal trans-duction events that lie downstreamof EphA2 and FAK remain to beidentified.

MITOGEN ACTIVATEDKINASE SIGNALING ANDMELANOMA VM

Extracellular regulated kinase 1 and2 (Erk1/2) is one of the mitogen-acti-vated protein kinase (MAPK) familymembers. It is a component of theRas-Raf-Mek1/2-Erk1/2 signal trans-duction pathway and is activated inresponse to growth factors, cytokines,and hormones. Activation of thispathway results in numerous cellu-lar responses including, survival,proliferation, adhesion, invasion,and migration (reviewed in Gray-Schopfer et al., 2005). Mutations ofRas and/or Raf resulting in constitu-tive activation of Erk1/2 have thecapacity to transform mammaliancells in vitro (Cowley et al., 1994;Mansour et al., 1994). Furthermore,an association has been reported be-tween activating mutations of bothRas and Raf and tumorigenesis innumerous tumor types includingmelanoma (Davies et al., 2002;Giehl, 2005).

Recently, there have been severalreports linking EphA2 signal trans-duction events with Erk1/2 phosphor-ylation. Studies by Pratt and col-leagues in addition to those initiatedby Macrae and colleagues utilizingvarious breast cancer cell lines indi-cate that ligand (ephrin-A1) binding ofEphA2 results in an increase inErk1/2 phosphorylation, and subse-quent upregulation of EphA2 mRNA,suggesting an autocrine feedback loopbetween these pathways (Giehl, 2005;

Fig. 5. To assess proliferation, 2.5 � 104 untransfected C8161 cells (C8161), neo-transfectedC8161 cells (C8161-Neo), or C8161 EphA2 knock-down cells (EphA2 AS 0.2-1 and EphA2 AS0.2-2) were plated on 24-well tissue culture plates in RPMI containing 5% serum. Cells wereharvested with trypsin/EDTA and counted every 24 hr for a period of 4 days. Statistical significancewas determined using the student’s t-test. *P 0.01; for each time point n � 12.

TABLE 2. Ability of EphA2 Knock-Downs to Form Tumors In Vivoa

Cell lineNumber ofanimals Tumor volume (mm3)b P*

C8161 5 3,034 385 1.0C8161-Neo 5 3,727 673 0.8EphA2 AS 0.2–1 5 114 34 �0.001EphA2 AS 0.2–2 5 190 75 �0.001

a2.5 � 105 untransfected human cutaneous metastatic melanoma cells (C8161),neo-transfected (C8161-Neo), or C8161 EphA2 knock-downs (EphA2 AS [AS,antisense) 0.2–1 and EphA2 AS 0.2–2), were injected into the subscapular region ofnu/nu female mice.

bAfter 5 weeks, mice were sacrificed and tumor volume was calculated by multiplyingthe height, length, and width of each tumor, as measured using a microcaliper, andvolume is presented as mm3

*Statistical significance was determined using a Student’s t-test analysis.

EPH RECEPTOR SIGNALING AND MELANOMA VM 3291

Macrae et al., 2005; Pratt and Kinch,2002, 2003). On the contrary, Miaoand colleagues have reported thatstimulation of EphA2 with its ligandephrin-A1 results in a decrease inErk1/2 phosphorylation in prostatecancer and endothelial cells (Miao etal., 2001). Additionally, Maio and col-leagues reported that the stimulationof the MAPK pathway using growthfactors such as placental derivedgrowth factor (PDGF) and epidermalgrowth factor (EGF) can be attenu-ated when simultaneously treatedwith ephrin-A1 (Miao et al., 2001).These latter results are in agreementwith the study initiated by Macraeand colleagues, who reported similarresults using EGF to stimulate Erk1/2phosphorylation in a panel of breastcancer cells, and found those effects tobe attenuated by ephrin-A1 (Macraeet al., 2005). Furthermore, they foundthat cells containing mutations in Rasresulted in Erk1/2 phosphorylationthat was independent of EGF stimu-lation and likewise unresponsive tothe attenuation effects of ephrin-A1stimulation (Macrae et al., 2005). Co-incidently these same cells expressedhigh levels of EphA2 expression. To-gether these results suggest that con-stitutively active Erk1/2, possibilitydue to mutations in Ras and/or Raf,are responsible for the high levels ofEphA2 expression found in certain tu-mor types.

Although we have yet to establisha link between EphA2 and Erk1/2phosphorylation in our aggressivemelanoma cells, we have establisheda role for Erk1/2 in promoting anaggressive melanoma phenotypewith respect to FAK signaling. Wefound that expressing FRNK, a dom-inant negative antagonist of FAKsignaling, in aggressive melanomacells reduced Erk1/2 phosphoryla-tion (Hess et al., 2005). Further-more, we found that treatment of theaggressive cutaneous melanomacells with specific inhibitors toMek1/2 (PD98059 and U0126), thusdecreasing the levels of Erk1/2 phos-phorylation, resulted in a decreasein the invasive capacity and VM po-tential of these cells. Interestingly,we failed to see a decrease in themigratory behavior of these cells,suggesting that Erk1/2 is necessaryfor some but not all aspects of mela-noma aggressiveness. Subsequently,we explored specific proteolytic en-zymes that are important for inva-sion and VM. Using the MEK1/2 in-hibitor, U0126, to down-regulateErk1/2 phosphorylation, we founddecreases in secreted urokinase,MMP-2 and MT1-MMP activity,which could result in the decreasesobserved in invasion and VM. Wehypothesize that increased levelsof secreted urokinase, MMP-2,and MT1-MMP, mediated through

Erk1/2 signaling, allow aggressivemelanoma cells to remodel, degrade,and invade the extracellular matrixenabling VM and metastasis.

Based on our investigations intothe molecular mechanisms that me-diate melanoma VM, we have devel-oped a hypothetical model that illus-trates our current understanding ofthe signaling events involved in thisprocess, as shown in Figure 6. Al-though we have made considerableprogress in deciphering the molecu-lar mechanisms underling variouscharacteristics of aggressive mela-noma cells with a particular focuson melanoma VM, there remainmany unanswered questions. Cur-rent studies are aimed at linkingthe signaling events mediated byPI3K, FAK, and Erk1/2 with that ofEphA2. Additionally, we are inter-ested in understanding the molecu-lar mechanisms involved in pro-moting the deregulated genotypecharacteristic of aggressive mela-noma cells, specifically the aber-rantly regulated signaling eventsthat are responsible for the in-creased expression of EphA2. It isinteresting to note that many of thedifferent tumor types that have beenfound to engage in VM, also haveincreased EphA2, suggesting thatsignaling events regulated by EphA2may mediate VM in several tumortypes.

Fig. 6. Hypothetical model illustrating signal transduction pathways that promote melanoma VM. In this model, EphA2 and VE-cadherin co-localizeon the surface of aggressive melanoma cells at sites of cell-cell adhesion. This association may allow for signal transduction events to be mediatedthrough either PI3K and/or FAK. Increased FAK phosphorylation results in an increase in Erk1/2 phosphorylation thus promoting an increase inurokinase production. Additionally, increased Erk1/2 phosphorylation can increase the activities of MT1-MMP and MMP-2 through, as yet unidentified,upstream effectors. Lastly, activation of PI3K can regulate the expression and activity of MT1-MMP, which in turn promotes the conversion of MMP-2into its active confirmation through an interaction with TIMP-2. Both enzymatically active MT1-MMP and MMP-2 may then promote the cleavage ofthe laminin 5�2 chain into pro-migratory fragments. The combination of these signaling events work in concert to promote properties of aggressivemelanoma tumor cells, including invasion, migration, proliferation, and VM, which are associated with an increased risk for metastasis.

3292 HESS ET AL.

FUTURE DIRECTIONS

EphA2 has been implicated in medi-ating two different avenues of tumorneovascularization: angiogenesisand vasculogenic mimicry. The over-expression of EphA2 in several dif-ferent tumor types and its correla-tion with poor clinical outcomemakes it a prime target for the de-velopment of therapeutic interven-tion strategies. Recently, reportshave investigated the feasibility ofdownregulating EphA2 in breast,ovarian, and pancreatic tumors us-ing agonistic antibodies and siRNAtechnologies (Carles-Kinch et al.,2002; Duxbury et al., 2004a; LandenJr. et al., 2005, 2006). Using ortho-topic mouse models and systemic de-livery of either agonistic antibodiesor EphA2 targeted siRNA, there hasbeen success in reducing tumor bur-den and increasing survival with notoxic side effects. Our current under-standing of the consequences ofEphA2 over-expression in melanomaclearly demonstrated that EphA2 isa potential therapeutic target formetastatic melanoma as well. Thus,we are currently investigating thefeasibility of targeting melanoma tu-mors using similar siRNA strategies.

Understanding the ability of ag-gressive melanoma cells to acquire adysregulated genotype and engagein VM is just one example of theplasticity potential of aggressive tu-mor cells, and suggests a reversionto a more embryonic phenotype. Asadditional reports emerge linkingoverexpression of EphA2 or otherEph receptors with tumor aggres-siveness, new information regardingthe convergence of embryonic andtumorigenic signaling pathways mayprovide new insights into the regu-lation of the plastic tumor cell phe-notype. Deciphering the molecularunderpinnings promoting tumor cellplasticity will offer promise for thedevelopment of new therapeutic in-tervention strategies to target awide variety of tumor types.

ACKNOWLEDGMENTSWe thank Dr. Kevin Knudston at theUniversity of Iowa DNA sequencingfacility in Iowa City, IA, for perform-ing the microarray analysis and Dr.Dawn A. Kirschmann for her artistic

contribution of the hypothetical sig-naling model. This work was sup-ported in part by the Melanoma Re-search Foundation (to A.R.H) and theUS National Institutes of Health(NIH) grants CA59702 and CA80318(to M.J.C.H)

REFERENCES

Adams RH, Wilkinson GA, Weiss C, DiellaF, Gale NW, Deutsch U, Risau W, KleinR. 1999. Roles of ephrinB ligands andEphB receptors in cardiovascular devel-opment: demarcation of arterial/venousdomains, vascular morphogenesis, andsprouting angiogenesis. Genes Dev 13:295–306.

Baharvand H, Ashtiani SK, Taee A, Mas-sumi M, Valojerdi MR, Yazdi PE, MoradiSZ, Farrokhi A. 2006. Generation of newhuman embryonic stem cell lines withdiploid and triploid karyotypes. DevGrowth Differ 48:117–128.

Balch CM, Soong SJ, Atkins MB, BuzaidAC, Cascinelli N, Coit DG, Fleming ID,Gershenwald JE, Houghton A, Jr., Kirk-wood JM, McMasters KM, Mihm MF,Morton DL, Reintgen DS, Ross MI, SoberA, Thompson JA, Thompson JF. 2004.An evidence-based staging system for cu-taneous melanoma. CA Cancer J Clin54:131–149.

Bittner M, Meltzer P, Chen Y, Jiang Y,Seftor E, Hendrix M, Radmacher M, Si-mon R, Yakhini Z, Ben Dor A, Sampas N,Dougherty E, Wang E, Marincola F,Gooden C, Lueders J, Glatfelter A, Pol-lock P, Carpten J, Gillanders E, Leja D,Dietrich K, Beaudry C, Berens M, Al-berts D, Sondak V. 2000. Molecular clas-sification of cutaneous malignant mela-noma by gene expression profiling.Nature 406:536–540.

Brader S, Eccles SA. 2004. Phosphoinosi-tide 3-kinase signalling pathways in tu-mor progression, invasion and angiogen-esis. Tumori 90:2–8.

Brantley DM, Cheng N, Thompson EJ, LinQ, Brekken RA, Thorpe PE, MuraokaRS, Cerretti DP, Pozzi A, Jackson D, LinC, Chen J. 2002. Soluble Eph A receptorsinhibit tumor angiogenesis and progres-sion in vivo. Oncogene 21:7011–7026.

Brantley-Sieders DM, Chen J. 2004. Ephreceptor tyrosine kinases in angiogene-sis: from development to disease. Angio-genesis 7:17–28.

Brantley-Sieders DM, Caughron J, HicksD, Pozzi A, Ruiz JC, Chen J. 2004.EphA2 receptor tyrosine kinase regu-lates endothelial cell migration and vas-cular assembly through phosphoinosi-tide 3-kinase-mediated Rac1 GTPaseactivation. J Cell Sci 117:2037–2049.

Brantley-Sieders DM, Fang WB, Hicks DJ,Zhuang G, Shyr Y, Chen J. 2005. Im-paired tumor microenvironment inEphA2-deficient mice inhibits tumor an-giogenesis and metastatic progression.FASEB J 19:1884–1886.

Brantley-Sieders DM, Fang WB, Hwang Y,Hicks D, Chen J. 2006. Ephrin-A1 fa-cilitates mammary tumor metastasisthrough an angiogenesis-dependentmechanism mediated by EphA receptorand vascular endothelial growth factorin mice. Cancer Res 66:10315–10324.

Carles-Kinch K, Kilpatrick KE, StewartJC, Kinch MS. 2002. Antibody targetingof the EphA2 tyrosine kinase inhibitsmalignant cell behavior. Cancer Res 62:2840–2847.

Carmeliet P, Lampugnani MG, Moons L,Breviario F, Compernolle V, Bono F, Bal-coni G, Spagnuolo R, Oostuyse B, Dew-erchin M, Zanetti A, Angellilo A, MattotV, Nuyens D, Lutgens E, Clotman F, deRuiter MC, Gittenberger-de GA, Poel-mann R, Lupu F, Herbert JM, Collen D,Dejana E. 1999. Targeted deficiency orcytosolic truncation of the VE-cadheringene in mice impairs VEGF-mediatedendothelial survival and angiogenesis.Cell 98:147–157.

Carter N, Nakamoto T, Hirai H, Hunter T.2002. EphrinA1-induced cytoskeletalre-organization requires FAK andp130(cas). Nat Cell Biol 4:565–573.

Chen J, Hicks D, Brantley-Sieders D,Cheng N, McCollum GW, Qi-Werdich X,Penn J. 2006. Inhibition of retinal neo-vascularization by soluble EphA2 recep-tor. Exp Eye Res 82:664–-673.

Cheng N, Brantley DM, Liu H, Lin Q, En-riquez M, Gale N, Yancopoulos G, Cer-retti DP, Daniel TO, Chen J. 2002.Blockade of EphA receptor tyrosine ki-nase activation inhibits vascular endo-thelial cell growth factor-induced angio-genesis. Mol Cancer Res 1:2–11.

Cheng N, Brantley D, Fang WB, Liu H,Fanslow W, Cerretti DP, Bussell KN, Re-ith A, Jackson D, Chen J. 2003. Inhibi-tion of VEGF-dependent multistage car-cinogenesis by soluble EphA receptors.Neoplasia 5:445–456.

Conway EM, Collen D, Carmeliet P. 2001.Molecular mechanisms of blood vesselgrowth. Cardiovasc Res 49:507–521.

Cowley S, Paterson H, Kemp P, MarshallCJ. 1994. Activation of MAP kinase ki-nase is necessary and sufficient for PC12differentiation and for transformation ofNIH 3T3 cells. Cell 77:841–852.

Cox BD, Natarajan M, Stettner MR, Glad-son CL. 2006. New concepts regardingfocal adhesion kinase promotion of cellmigration and proliferation. J Cell Bio-chem 99:35–52.

Davies H, Bignell GR, Cox C, Stephens P,Edkins S, Clegg S, Teague J, WoffendinH, Garnett MJ, Bottomley W, Davis N,Dicks E, Ewing R, Floyd Y, Gray K, HallS, Hawes R, Hughes J, Kosmidou V,Menzies A, Mould C, Parker A, StevensC, Watt S, Hooper S, Wilson R, Jayati-lake H, Gusterson BA, Cooper C, ShipleyJ, Hargrave D, Pritchard-Jones K, Mait-land N, Chenevix-Trench G, Riggins GJ,Bigner DD, Palmieri G, Cossu A, Flana-gan A, Nicholson A, Ho JW, Leung SY,Yuen ST, Weber BL, Seigler HF, DarrowTL, Paterson H, Marais R, Marshall CJ,Wooster R, Stratton MR, Futreal PA.

EPH RECEPTOR SIGNALING AND MELANOMA VM 3293

2002. Mutations of the BRAF gene inhuman cancer. Nature 417:949–954.

Dejana E, Bazzoni G, Lampugnani MG.1999. Vascular endothelial (VE)-cad-herin: only an intercellular glue? ExpCell Res 252:13–19.

Dome B, Hendrix MJ, Paku S, Tovari J,Timar J. 2007. Alternative vasculariza-tion mechanisms in cancer: Pathologyand therapeutic implications. Am JPathol 170:1–15.

Duxbury MS, Ito H, Zinner MJ, AshleySW, Whang EE. 2004a. EphA2: a deter-minant of malignant cellular behaviorand a potential therapeutic target inpancreatic adenocarcinoma. Oncogene23:1448–1456.

Duxbury MS, Ito H, Zinner MJ, AshleySW, Whang EE. 2004b. Ligation ofEphA2 by Ephrin A1-Fc inhibits pancre-atic adenocarcinoma cellular invasive-ness. Biochem Biophys Res Commun320:1096–1102.

Easty DJ, Guthrie BA, Maung K, Farr CJ,Lindberg RA, Toso RJ, Herlyn M, Ben-nett DC. 1995. Protein B61 as a newgrowth factor: expression of B61 and up-regulation of its receptor epithelial cellkinase during melanoma progression.Cancer Res 55:2528–2532.

Eph Nomenclature Committee. 1997. Uni-fied Nomenclature for the Eph family ofreceptors and their ligands, the ephrins.Cell 90:403–404.

Flenniken AM, Gale NW, Yancopoulos GD,Wilkinson DG. 1996. Distinct and over-lapping expression patterns of ligandsfor Eph-related receptor tyrosine kinasesduring mouse embryogenesis. Dev Biol179:382–401.

Folberg R, Rummelt V, Parys-van GR,Hwang T, Woolson RF, Pe’er J, GrumanLM. 1993. The prognostic value of tumorblood vessel morphology in primaryuveal melanoma. Ophthalmology 100:1389–1398.

Folkman J. 1995. Seminars in Medicine ofthe Beth Israel Hospital, Boston. Clini-cal applications of research on angiogen-esis. N Engl J Med 333:1757–1763.

Gale NW, Baluk P, Pan L, Kwan M, HolashJ, DeChiara TM, McDonald DM, Yanco-poulos GD. 2001. Ephrin-B2 selectivelymarks arterial vessels and neovascular-ization sites in the adult, with expres-sion in both endothelial and smooth-muscle cells. Dev Biol 230:151–160.

Gerety SS, Wang HU, Chen ZF, AndersonDJ. 1999. Symmetrical mutant pheno-types of the receptor EphB4 and its spe-cific transmembrane ligand ephrin-B2 incardiovascular development. Mol Cell 4:403–414.

Giehl K. 2005. Oncogenic Ras in tumourprogression and metastasis. Biol Chem386:193–205.

Gray-Schopfer VC, da Rocha DS, Marais R.2005. The role of B-RAF in melanoma.Cancer Metastasis Rev 24:165–183.

guirre Ghiso JA. 2002. Inhibition of FAKsignaling activated by urokinase recep-tor induces dormancy in human carci-noma cells in vivo. Oncogene 21:2513–2524.

Han NM, Fleming RY, Curley SA, GallickGE. 1997. Overexpression of focal adhe-sion kinase (p125FAK) in human colo-rectal carcinoma liver metastases: inde-pendence from c-src or c-yes activation.Ann Surg Oncol 4:264–268.

Hauck CR, Sieg DJ, Hsia DA, Loftus JC,Gaarde WA, Monia BP, Schlaepfer DD.2001. Inhibition of focal adhesion kinaseexpression or activity disrupts epidermalgrowth factor-stimulated signaling pro-moting the migration of invasive humancarcinoma cells. Cancer Res 61:7079–7090.

Hauck CR, Hsia DA, Puente XS, ChereshDA, Schlaepfer DD. 2002. FRNK blocksv-Src-stimulated invasion and experi-mental metastases without effects oncell motility or growth. EMBO J 21:6289–6302.

Hendrix MJ, Seftor EA, Hess AR, SeftorRE. 2003. Vasculogenic mimicry and tu-mour-cell plasticity: lessons from mela-noma. Nat Rev Cancer 3:411–421.

Hendrix MJ, Seftor EA, Meltzer PS, Gard-ner LM, Hess AR, Kirschmann DA,Schatteman GC, Seftor RE. 2001. Ex-pression and functional significance ofVE-cadherin in aggressive human mela-noma cells: role in vasculogenic mimicry.Proc Natl Acad Sci USA 98:8018–8023.

Henkemeyer M, Orioli D, Henderson JT,Saxton TM, Roder J, Pawson T, Klein R.1996. Nuk controls pathfinding of com-missural axons in the mammalian cen-tral nervous system. Cell 86:35–46.

Hess AR, Hendrix MJ. 2006. Focal adhe-sion kinase signaling and the aggressivemelanoma phenotype. Cell Cycle 5:478–480.

Hess AR, Seftor EA, Gardner LM, Carles-Kinch K, Schneider GB, Seftor RE,Kinch MS, Hendrix MJ. 2001. Molecularregulation of tumor cell vasculogenicmimicry by tyrosine phosphorylation:role of epithelial cell kinase (Eck/EphA2). Cancer Res 61:3250–3255.

Hess AR, Seftor EA, Seftor RE, HendrixMJ. 2003. Phosphoinositide 3-kinaseregulates membrane Type 1-matrix met-alloproteinase (MMP) and MMP-2 activ-ity during melanoma cell vasculogenicmimicry. Cancer Res 63:4757–4762.

Hess AR, Postovit LM, Margaryan NV,Seftor EA, Schneider GB, Seftor RE,Nickoloff BJ, Hendrix MJ. 2005. Focaladhesion kinase promotes the aggressivemelanoma phenotype. Cancer Res 65:9851–9860.

Hess AR, Seftor EA, Gruman LM, KinchMS, Seftor RE, Hendrix MJ. 2006. VE-Cadherin Regulates EphA2 in Aggres-sive Melanoma Cells Through a NovelSignaling Pathway: Implications forVasculogenic Mimicry. Cancer Biol Ther5:228–233.

Hunter SG, Zhuang G, Brantley-Sieders D,Swat W, Cowan CW, Chen J. 2006. Es-sential role of Vav family guanine nucle-otide exchange factors in EphA receptor-mediated angiogenesis. Mol Cell Biol 26:4830–4842.

Jones EA, le NF, Eichmann A. 2006. Whatdetermines blood vessel structure? Ge-

netic prespecification vs. hemodynamics.Physiology (Bethesda ) 21:388–395.

Kataoka H, Igarashi H, Kanamori M,Ihara M, Wang JD, Wang YJ, Li ZY,Shimamura T, Kobayashi T, MaruyamaK, Nakamura T, Arai H, Kajimura M,Hanai H, Tanaka M, Sugimura H. 2004.Correlation of EPHA2 overexpressionwith high microvessel count in humanprimary colorectal cancer. Cancer Sci 95:136–141.

Kinch MS, Moore MB, Harpole DH, Jr.2003. Predictive value of the EphA2 re-ceptor tyrosine kinase in lung cancer re-currence and survival. Clin Cancer Res9:613–618.

Kobayashi H, Shirakawa K, Kawamoto S,Saga T, Sato N, Hiraga A, Watanabe I,Heike Y, Togashi K, Konishi J, BrechbielMW, Wakasugi H. 2002. Rapid accumu-lation and internalization of radiolabeledherceptin in an inflammatory breast can-cer xenograft with vasculogenic mimicrypredicted by the contrast-enhanced dy-namic MRI with the macromolecularcontrast agent G6-(1B4M-Gd)(256). Can-cer Res 62:860–866.

Krull CE, Lansford R, Gale NW, Collazo A,Marcelle C, Yancopoulos GD, Fraser SE,Bronner-Fraser M. 1997. Interactions ofEph-related receptors and ligands conferrostrocaudal pattern to trunk neuralcrest migration. Curr Biol 7:571–580.

Landen CN Jr., Chavez-Reyes A, BucanaC, Schmandt R, Deavers MT, Lopez-Be-restein G, Sood AK. 2005. TherapeuticEphA2 gene targeting in vivo using neu-tral liposomal small interfering RNA de-livery. Cancer Res 65:6910–6918.

Landen CN Jr., Lu C, Han LY, CoffmanKT, Bruckheimer E, Halder J, MangalaLS, Merritt WM, Lin YG, Gao C,Schmandt R, Kamat AA, Li Y, Thaker P,Gershenson DM, Parikh NU, GallickGE, Kinch MS, Sood AK. 2006. Efficacyand antivascular effects of EphA2 reduc-tion with an agonistic antibody in ovar-ian cancer. J Natl Cancer Inst 98:1558–1570.

Lazarova P, Wu Q, Kvalheim G, Suo Z,Haakenstad KW, Metodiev K, NeslandJM. 2006. Growth factor receptors in he-matopoietic stem cells: EPH family ex-pression in CD34� and CD133� cellpopulations from mobilized peripheralblood. Int J Immunopathol Pharmacol19:49–56.

Lickliter JD, Smith FM, Olsson JE, Mack-well KL, Boyd AW. 1996. Embryonicstem cells express multiple Eph-subfam-ily receptor tyrosine kinases. Proc NatlAcad Sci USA 93:145–150.

Lin YG, Han LY, Kamat AA, Merritt WM,Landen CN, Deavers MT, Fletcher MS,Urbauer DL, Kinch MS, Sood AK. 2007.EphA2 overexpression is associated withangiogenesis in ovarian cancer. Cancer109:332–340.

Liu C, Huang H, Donate F, Dickinson C,Santucci R, El-Sheikh A, Vessella R,Edgington TS. 2002. Prostate-specificmembrane antigen directed selectivethrombotic infarction of tumors. CancerRes 62:5470–5475.

3294 HESS ET AL.

Liu DP, Wang Y, Koeffler HP, Xie D. 2007.Ephrin-A1 is a negative regulator in gli-oma through down-regulation of EphA2and FAK. Int J Oncol 30:865–871.

Macrae M, Neve RM, Rodriguez-Viciana P,Haqq C, Yeh J, Chen C, Gray JW, Mc-Cormick F. 2005. A conditional feedbackloop regulates Ras activity throughEphA2. Cancer Cell 8:111–118.

Maniotis AJ, Folberg R, Hess A, Seftor EA,Gardner LM, Pe’er J, Trent JM, MeltzerPS, Hendrix MJ. 1999. Vascular channelformation by human melanoma cells invivo and in vitro: vasculogenic mimicry.Am J Pathol 155:739–752.

Mansour SJ, Matten WT, Hermann AS,Candia JM, Rong S, Fukasawa K, VandeWoude GF, Ahn NG. 1994. Transforma-tion of mammalian cells by constitu-tively active MAP kinase kinase. Science265:966–970.

Maung K, Easty DJ, Hill SP, Bennett DC.1999. Requirement for focal adhesion ki-nase in tumor cell adhesion. Oncogene18:6824–6828.

McBride JL, Ruiz JC. 1998. Ephrin-A1 isexpressed at sites of vascular develop-ment in the mouse. Mech Dev 77:201–204.

Mellitzer G, Xu Q, Wilkinson DG. 1999.Eph receptors and ephrins restrict cellintermingling and communication. Na-ture 400:77–81.

Miao H, Burnett E, Kinch M, Simon E,Wang B. 2000. Activation of EphA2 ki-nase suppresses integrin function andcauses focal-adhesion-kinase dephos-phorylation. Nat Cell Biol 2:62–69.

Miao H, Wei BR, Peehl DM, Li Q, Alexan-drou T, Schelling JR, Rhim JS, Sedor JR,Burnett E, Wang B. 2001. Activation ofEphA receptor tyrosine kinase inhibitsthe Ras/MAPK pathway. Nat Cell Biol3:527–530.

Mudali SV, Fu B, Lakkur SS, Luo M, Em-buscado EE, Iacobuzio-Donahue CA.2006. Patterns of EphA2 protein expres-sion in primary and metastatic pancre-atic carcinoma and correlation with ge-netic status. Clin Exp Metast 23:357–365.

Nguyen DH, Webb DJ, Catling AD, Song Q,Dhakephalkar A, Weber MJ, Ravichand-ran KS, Gonias SL. 2000. Urokinase-typeplasminogen activator stimulates the Ras/Extracellular signal-regulated kinase(ERK) signaling pathway and MCF-7 cellmigration by a mechanism that requiresfocal adhesion kinase, Src, and Shc. Rapiddissociation of GRB2/Sps-Shc complex isassociated with the transient phosphory-lation of ERK in urokinase-treated cells.J Biol Chem 275:19382–19388.

Ogawa K, Pasqualini R, Lindberg RA, KainR, Freeman AL, Pasquale EB. 2000. Theephrin-A1 ligand and its receptor,EphA2, are expressed during tumorneovascularization. Oncogene 19:6043–6052.

Ojima T, Takagi H, Suzuma K, Oh H, Su-zuma I, Ohashi H, Watanabe D, Suga-nami E, Murakami T, Kurimoto M,Honda Y, Yoshimura N. 2006. EphrinA1inhibits vascular endothelial growth fac-

tor-induced intracellular signaling andsuppresses retinal neovascularizationand blood-retinal barrier breakdown.Am J Pathol 168:331–339.

Orioli D, Henkemeyer M, Lemke G, KleinR, Pawson T. 1996. Sek4 and Nuk recep-tors cooperate in guidance of commis-sural axons and in palate formation.EMBO J 15:6035–6049.

Orsulic S, Kemler R. 2000. Expression ofEph receptors and ephrins is differen-tially regulated by E-cadherin. J Cell Sci113:1793–1802.

Owens LV, Xu L, Dent GA, Yang X, SturgeGC, Craven RJ, Cance WG. 1996. Focaladhesion kinase as a marker of invasivepotential in differentiated human thy-roid cancer. Ann Surg Oncol 3:100–105.

Pandey A, Lazar DF, Saltiel AR, Dixit VM.1994. Activation of the Eck receptor pro-tein tyrosine kinase stimulates phospha-tidylinositol 3-kinase activity. J BiolChem 269:30154–30157.

Pandey A, Shao H, Marks RM, PolveriniPJ, Dixit VM. 1995. Role of B61, the li-gand for the Eck receptor tyrosine ki-nase, in TNF-alpha-induced angiogene-sis. Science 268:567–569.

Park S, Frisen J, Barbacid M. 1997. Aber-rant axonal projections in mice lackingEphA8 (Eek) tyrosine protein kinase re-ceptors. EMBO J 16:3106–3114.

Passalidou E, Trivella M, Singh N, Fergu-son M, Hu J, Cesario A, Granone P, Ni-cholson AG, Goldstraw P, Ratcliffe C,Tetlow M, Leigh I, Harris AL, GatterKC, Pezzella F. 2002. Vascular pheno-type in angiogenic and non-angiogeniclung non-small cell carcinomas. Br JCancer 86:244–249.

Pratt RL, Kinch MS. 2002. Activation ofthe EphA2 tyrosine kinase stimulatesthe MAP/ERK kinase signaling cascade.Oncogene 21:7690–7699.

Pratt RL, Kinch MS. 2003. Ligand bindingup-regulates EphA2 messenger RNAthrough the mitogen-activated protein/extracellular signal-regulated kinasepathway. Mol Cancer Res 1:1070–1076.

Qi JH, Claesson-Welsh L. 2001. VEGF-in-duced activation of phosphoinositide3-kinase is dependent on focal adhesionkinase. Exp Cell Res 263:173–182.

Ruf W, Seftor EA, Petrovan RJ, Weiss RM,Gruman LM, Margaryan NV, Seftor RE,Miyagi Y, Hendrix MJ. 2003. Differen-tial role of tissue factor pathway inhibi-tors 1 and 2 in melanoma vasculogenicmimicry. Cancer Res 63:5381–5389.

Schlaepfer DD, Hauck CR, Sieg DJ. 1999.Signaling through focal adhesion kinase.Prog Biophys Mol Biol 71:435–478.

Schneider GB, Kurago Z, Zaharias R, Gru-man LM, Schaller MD, Hendrix MJ.2002. Elevated focal adhesion kinase ex-pression facilitates oral tumor cell inva-sion. Cancer 95:2508–2515.

Seftor EA, Meltzer PS, Schatteman GC,Gruman LM, Hess AR, Kirschmann DA,Seftor RE, Hendrix MJ. 2002. Expres-sion of multiple molecular phenotypes byaggressive melanoma tumor cells: role invasculogenic mimicry. Crit Rev OncolHematol 44:17–27.

Seftor RE, Seftor EA, Koshikawa N, Melt-zer PS, Gardner LM, Bilban M, Stetler-Stevenson WG, Quaranta V, HendrixMJ. 2001. Cooperative interactions oflaminin 5 gamma2 chain, matrix metal-loproteinase-2, and membrane type-1-matrix/metalloproteinase are requiredfor mimicry of embryonic vasculogenesisby aggressive melanoma. Cancer Res 61:6322–6327.

Sharma N, Seftor RE, Seftor EA, GrumanLM, Heidger PM, Jr., Cohen MB,Lubaroff DM, Hendrix MJ. 2002. Pros-tatic tumor cell plasticity involvescooperative interactions of distinct phe-notypic subpopulations: role in vasculo-genic mimicry. Prostate 50:189–201.

Shen TL, Park AY, Alcaraz A, Peng X,Jang I, Koni P, Flavell RA, Gu H, GuanJL. 2005. Conditional knockout of focaladhesion kinase in endothelial cells re-veals its role in angiogenesis and vascu-lar development in late embryogenesis.J Cell Biol 169:941–952.

Shin D, Garcia-Cardena G, Hayashi S,Gerety S, Asahara T, Stavrakis G, IsnerJ, Folkman J, Gimbrone MA, Jr., Ander-son DJ. 2001. Expression of ephrinB2identifies a stable genetic difference be-tween arterial and venous vascularsmooth muscle as well as endothelialcells, and marks subsets of microvesselsat sites of adult neovascularization. DevBiol 230:139–150.

Shirakawa K, Wakasugi H, Heike Y, Wa-tanabe I, Yamada S, Saito K, Konishi F.2002. Vasculogenic mimicry and pseudo-comedo formation in breast cancer. Int JCancer 99:821–828.

Smith A, Robinson V, Patel K, WilkinsonDG. 1997. The EphA4 and EphB1 recep-tor tyrosine kinases and ephrin-B2 li-gand regulate targeted migration ofbranchial neural crest cells. Curr Biol7:561–570.

Sood AK, Fletcher MS, Zahn CM, GrumanLM, Coffin JE, Seftor EA, Hendrix MJ.2002. The clinical significance of tumorcell-lined vasculature in ovarian carci-noma: implications for anti-vasculogenictherapy. Cancer Biol Ther 1:661–664.

Sood AK, Coffin JE, Schneider GB,Fletcher MS, DeYoung BR, Gruman LM,Gershenson DM, Schaller MD, HendrixMJC. 2004. Biological signifiance of FAKin ovarian cancer: Role in migration andinvasion. Am J Pathol 165:1087–1095.

Straume O, Akslen LA. 2002. Importanceof vascular phenotype by basic fibroblastgrowth factor, and influence of the angio-genic factors basic fibroblast growth fac-tor/fibroblast growth factor receptor-1and ephrin-A1/EphA2 on melanoma pro-gression. Am J Pathol 160:1009–1019.

Thaker PH, Deavers M, Celestino J,Thornton A, Fletcher MS, Landen CN,Kinch MS, Kiener PA, Sood AK. 2004.EphA2 expression is associated with ag-gressive features in ovarian carcinoma.Clin Cancer Res 10:5145–5150.

Tremblay L, Hauck W, Aprikian AG, BeginLR, Chapdelaine A, Chevalier S. 1996.Focal adhesion kinase (pp125FAK) ex-pression, activation and association with

EPH RECEPTOR SIGNALING AND MELANOMA VM 3295

paxillin and p50CSK in human meta-static prostate carcinoma. Int J Cancer68:164–171.

van der Schaft DW, Hillen F, Pauwels P,Kirschmann DA, Castermans K, Eg-brink MG, Tran MG, Sciot R, Hauben E,Hogendoorn PC, Delattre O, MaxwellPH, Hendrix MJ, Griffioen AW. 2005.Tumor cell plasticity in Ewing sarcoma,an alternative circulatory system stimu-lated by hypoxia. Cancer Res 65:11520–11528.

Vanhaesebroeck B, Waterfield MD. 1999.Signaling by distinct classes of phospho-inositide 3-kinases. Exp Cell Res 253:239–254.

Walker-Daniels J, Coffman K, Azimi M,Rhim JS, Bostwick DG, Snyder P, KernsBJ, Waters DJ, Kinch MS. 1999. Overex-pression of the EphA2 tyrosine kinase inprostate cancer. Prostate 41:275–280.

Wang HU, Anderson DJ. 1997. Eph familytransmembrane ligands can mediate re-pulsive guidance of trunk neural crestmigration and motor axon outgrowth.Neuron 18:383–396.

Wang HU, Chen ZF, Anderson DJ. 1998.Molecular distinction and angiogenic in-teraction between embryonic arteriesand veins revealed by ephrin-B2 and itsreceptor Eph-B4. Cell 93:741–753.

Winslow JW, Moran P, Valverde J, Shih A,Yuan JQ, Wong SC, Tsai SP, Goddard A,Henzel WJ, Hefti F,. 1995. Cloning ofAL-1, a ligand for an Eph-related ty-rosine kinase receptor involved in axonbundle formation. Neuron 14:973–981.

Zantek ND, Azimi M, Fedor-Chaiken M,Wang B, Brackenbury R, Kinch MS.1999. E-cadherin regulates the functionof the EphA2 receptor tyrosine kinase.Cell Growth Differ 10:629–638.

Zelinski DP, Zantek ND, Stewart JC,Irizarry AR, Kinch MS. 2001. EphA2overexpression causes tumorigenesis ofmammary epithelial cells. Cancer Res 61:2301–2306.

Zeng G, Hu Z, Kinch MS, Pan CX, Flock-hart DA, Kao C, Gardner TA, Zhang S,Li L, Baldridge LA, Koch MO, UlbrightTM, Eble JN, Cheng L. 2003. High-level expression of EphA2 receptor ty-rosine kinase in prostatic intraepithe-lial neoplasia. Am J Pathol 163:2271–2276.

Zhang S, Zhang D, Sun B. 2007. Vasculo-genic mimicry: Current status and fu-ture prospects. Cancer Lett. [in press]

Zhong TP, Childs S, Leu JP, Fishman MC.2001. Gridlock signalling pathway fash-ions the first embryonic artery. Nature414:216–220.

3296 HESS ET AL.