Molecular regulation of vasculogenic mimicry in tumors and potential tumor-target therapy

Abstract

“Vasculogenic mimicry (VM)”, is a term that describes the unique ability of highly aggressive tumor cells to express a multipotent, stem cell-like phenotype, and form a pattern of vasculogenic-like networks in three-dimensional culture. As an angiogenesis-independent pathway, VM and/or periodic acid-schiff-positive patterns are associated with poor prognosis in tumor patients. Moreover, VM is resistant to angiogenesis inhibitors. Here, we will review the advances in research on biochemical and molecular signaling pathways of VM in tumors and on potential anti-VM therapy strategy.

Key Words: Tumor-target therapy, Signaling pathways, High aggressive tumor, Molecular regulation, Prognosis, Vasculogenic mimicry

INTRODUCTION

Tumors require a blood supply for growth and hematogenous metastasis. Most attention has focused on the role of angiogenesis[1]. However, Maniotis et al[2] have described an angiogenesis-independent pathway called “vasculogenic mimicry (VM)”, a novel phenomenon in which highly aggressive human melanoma cells mimic endothelial cells and form vascular channel-like structures to convey blood plasma and red blood cells without the participation of endothelial cells. Currently, two distinctive types of VM have been described; tube and patterned matrix types[3]. VM channels are revealed as periodic acid-schiff (PAS)-positive patterns. VM consists of three elements: the plasticity of malignant tumor cells, remodeling of the extracellular matrix (ECM), and the connection of the VM channels to the host microcirculation system[4,5]. Different approaches have suggested that these channels provide a mechanism of perfusion and a dissemination route within the tumor that functions either independently of or simultaneously with angiogenesis. Several papers have evidenced the VM channel functional role in tumor circulation by using microinjection method, Doppler ultrasonography, MRI technique and laser scanning confocal angiography[6-11]. VM and/or PAS-positive patterns are also associated with a poor prognosis, worse survival and the highest risk of cancer recurrence for patients with melanoma[2,12], cell renal cell carcinoma[13], breast cancer[14], ovarian carcinoma[15], primary gallbladder carcinoma[16], malignant esophageal stromal carcinoma[17], mesothelial sarcomas and alveolar rhabdomyosarcomas[18], hepatocellular carcinoma[19-22]. In addition, tumor cell plasticity has been demonstrated in prostatic carcinoma[23], bladder carcinoma[24], osteosarcoma[25], astrocytoma[26] and pheochromocytoma[27].

The detailed mechanism of tumor VM remains to be further elucidated. At present, novel signaling pathways are discussed, involving factors which promote cell migration, invasion and matrix remodeling. These include vascular endothelial-cadherin (VE-cad)[28,29], epithelial cell kinase (EphA2)[30-32], focal adhesion kinase (FAK)[33,34], phosphoinositide 3-kinase (PI3-K), matrix metalloproteinase (MMPs), laminin 5 (Ln-5) γ² chain[35-38], tissue factor (TF), TF pathway inhibitor (TFPI)[11], Vascular endothelial growth factor-a (VEGF-a)[39]. Therefore, understanding the key molecular mechanisms that regulate VM would serve as an important target for new cancer therapies.

MOLECULE MECHANISMS OF TOMOR VM

VM describes the unique ability of highly aggressive tumor cells to express endothelial cell-associated genes (such as EphA2 and VE-cad) and form ECM-rich, patterned tubular networks when cultured on a three-dimensional matrix. However, the exact mechanism underlying VM still needs to be unraveled. Up to now, several molecules have been identified which have a functional role (Figure 1).

Figure 1 Hypothetical model for the molecular signaling pathways of vasculogenic mimicry in tumors and anti-vasculogenic mimicry (VM) therapy strategies. (1) The down-regulation of BMP-4 activity can lead to reduce expression of EphA2 and VE-cad; (2) TEL regulate VM by synergizing with signaling pathways downstream of RAS; (3) Celecoxib, COX-2 inhibition, may inhibit vascular channel formation, which is abrogated by addition of PGE2; (4) cAMP inhibits VM by activating Epac/Rap1 which produces Rap1-GTP; (5) Activation of Nodal signaling supports VM and expression of the VE-cad. Inhibition of the Nodal signaling pathway results in a reduction in keratin and VE-cad; (6) Knockdown of Id2 expression can inhibit VE-cad expression and abrogate the formation of tubular networks; (7) Gal-3 silencing can result decrease of VE-cadherin activities due to enhanced recruitment of EGR-1; (8) The decreasing in Gal-1 expression can provoke a marked decrease in BEX2, which impairs vasculogenic mimicry channel formation; (9) Over-expression of Mig-7 increased γ2 chain domain III fragments. Laminin 5 is the only laminin that contains the γ² chain, which following cleavage into promigratory fragments, the domain III region, causes increased levels of MMP-2, and MT1-MMP cooperate to cleave γ² chain into fragments that promote tumor cell VM; (10) Hypoxia→VEGF→EphA2→MMPs→VM; (11) VE-cadherin can promote the interaction between FAK and EphA2, phosphorylated EphA2 can form an interaction with FAK, which would lead to phosphorylation and activation of FAK. The signal transduction pathways activated through VE-cad and EphA2 can converge, resulting in activation of PI3K; (12) PI3K regulates MT1-MMP activity, which promotes the conversion of pro-MMP into its active conformation through an interaction with TIMP-2. Both enzymatically active MT1-MMP and MMP-2 may then promote the cleavage of the laminin 5(Ln-5) γ² chain in pro-migratory γ^2, and γ^2x fragments, the deposition of these fragments into tumor extracellular milieu may result in VM formation; (13) Blockade of TFPI-2 is able to suppress MMP-2 activation and prevent VM formation. Moreover, TFPI-1 has anticoagulant function of relevance for perfusion of VM; (14) Several drugs express specific anti-VM effects. Genistein inhibits VE-cad expression; COL-3 inhibits VE-cad, MMP-2 and MT1-MMP expression respectively. Doxycycline inhibits MMP-2 expression. Thalidomide inhibits MMP-2 and VEGF expression. In addition, forskolin inhibits VM formation through MAPK and PI3K pathway. VM: Vasculogenic mimicry; COX: Cyclooxygenase; VE-cad: Vascular endothelial-cadherin; MMPs: Matrix metalloproteinase; EphA2: Epithelial cell kinase; FAK: Focal adhesion kinase; VEGF: Endothelial growth factor.

PI3-K, MMPs and Ln-5 γ² chain

PI3-K is a lipid kinase that phosphorylates phosphatidylinositol or its derivatives on the 3-hydroxyl of the inositol head group. PI3-Ks are made up of four different 110-kDa catalytic subunits (p110a, p110b, p110g, and p110d) and a smaller regulatory subunit. The main product of PI3-K activity, PI(3,4,5)-P3 acts as a binding site for many intracellular proteins that include pleckstrin homology (PH) domains with selectivity for this lipid. The PI3-K signaling pathway plays an integral role in many normal cellular processes, including survival, proliferation, differentiation, metabolism and motility, in a variety of cell types[40].

In highly aggressive melanoma tumor cells, a recently published paper has indicated that PI3-K is an important adjustor of VM directly affecting the cooperative interactions of membrane type 1 (MT1)-MMP and matrix metalloproteinase-2 (MMP-2) activity. PI3-K regulates MT1-MMP activity, which promotes the conversion of pro-MMP into its active conformation through an interaction with TIMP-2. Both enzymatically active MT1-MMP and MMP-2 may then promote the cleavage of the Ln-5 γ² chain into pro-migratory γ^2, and γ^2x fragments. The deposition of these fragments into the tumor extracellular milieu may result in increased migration, invasion and VM formation. Poorly aggressive melanoma cells seeded on collagen matrices, preconditioned by aggressive tumor cells, formed tubular networks along the Ln-5 γ² chain-enriched tracks deposited by the aggressive cells. These observations indicate that the Ln-5 γ² chain in the ECM is able to promote VM formation[35,36]. Another observation showed that highly aggressive melanoma tumor cells can secrete the Ln-5 γ² chain and that the γ^2, and γ^2x chains, antisense oligonucleotides to the Ln-5 γ² chain and antibodies to MMP-2 or MT1-MMP may inhibit VM formation. Special inhibitors of PI3-K may impair VM formation and decrease MT1-MMP and MMP-2 activity. Furthermore, inhibition of PI3-K blocked the cleavage of Ln-5 γ² chain, resulting in decreased levels of the γ^2, and γ^2x pro-migratory fragments[37]. So, PI3-K may represent a predominant target for cancer therapy.

Similarly, in aggressive ovarian tumor cells, MMP-2 or MT1-MMP seems to play an important role in the VM channel. Human ovarian cancers with MMP over-expression are more likely to have tumor cell-lined vasculature[38].

Protein tyrosine kinases, EphA2, FAK and VE-cad

Protein tyrosine kinases (PTKs) have been shown to play important and diverse roles in regulating cell adhesion, migration and invasion[30]. Highly invasive malignant melanomas appear to express higher levels of PTKs with their phosphorylation centered specifically in the area where VM channels formed[30,41]. Thus, it could be concluded PTKs were pivotal factors of VM. Additionally, one of the receptor PTKs that was up-regulated in the aggressive melanoma cells was EphA2[30]. EphA2, a receptor tyrosine kinase and a member of the Eph (ephrin-receptor) family of PTKs, has been found to play an important role in angiogenesis[42,43]. Eph is a large family containing 14 members. The binding of EphA2 to its ligand ephrin-A1 results in the phosphorylation of EphA2. Other potential binding partners of EphA2 may be PI3-K and FAK[44]. FAK, a non-receptor protein tyrosine kinase, is a 125-kDa cytoplasmic tyrosine kinase associated with focal adhesions and is the major protein to become tyrosine phosphorylated after integrin activation. It localizes to regions of the cell that attach to the ECM, the focal adhesions. FAK localization is predominantly cytoplasmic in proliferating cells with punctate areas located at the periphery, presumably at the cell membrane, indicating integration in focal contacts. VE-cad is an adhesive protein, known to be expressed exclusively by endothelial cells, which belongs to the cadherin family of transmembrane proteins which promote homotypic cell-to cell interaction.

Microarray analyses revealed that EphA2 and VE-cad were dramatically over-expressed in aggressive human cutaneous and uveal melanoma cells, although not in poorly aggressive melanoma cells. Transient knockout of EphA2 and VE-cad in vitro abrogated the ability of highly aggressive melanoma cells to form the vasculogenic-like networks[28,30]. VE-cad and EphA2 are co-localized at sites of cell-cell adhesion. Additionally, knockdown of EphA2 expression does not affect the localization of VE-cad at sites of cell-cell adhesion, but does result in a redistribution of EphA2 on the cell-membrane, and an inability of cells to form vasculogenic structures. Collectively, association between VE-cad molecules on adjacent cells facilitates the organization of EphA2, either by interacting directly or indirectly with EphA2 on the cell membrane. When organized on the cell membrane, EphA2 is capable of binding to its ligand EphA1, resulting in the phosphorylation of the receptor, i.e. phosphorylated EphA2. VE-cad and EphA2 may converge to activate the PI3-K pathway leading to the activation MMP-2, and consequent cleavage of Ln-5γ^2[28-32]. Also, similar to VE-cad localization, the localization of EphA2 in aggressive human melanoma tissues is associated with areas containing and patterned, vasculogenic-like networks.

Recently, researches have demonstrated FAK to be an important key mediator of the aggressive melanoma phenotype, including VM[33,34]. FAK is phosphorylated on Tyr³⁹⁷and Tyr⁵⁷⁶ in aggressive human cutaneous and uveal melanoma cells cultured on a three-dimensional type 1 collagen matrix in vitro, as well as in radial and vertical growth phase melanomas in situ. Furthermore, expression FAK-related non-kinase in melanoma cells, which acts to disrupt FAK signaling, directly results in the inhibition of the aggressive phenotype, as demonstrated by decreased invasion, migration and VM potential. FAK signaling regulates invasion, migration and VM through two different signaling pathways. Firstly, FAK signals through Erk1/2 to increase the levels of urokinase activity, thus regulating invasion of the aggressive melanoma cells.Additionally, FAK seems to signal through unknown downstream effectors to promote migration in aggressive melanoma cells that may contribute to an increase of VM potential. Secondly Erk1/2 regulates MMP-2 and MT1-MMP activity, thus promoting melanoma invasion and VM[33,34]. Collectively, these observations implicate FAK as a promoter of the aggressive melanoma phenotype, thereby identifying it as a rational target for therapeutic intervention of malignant melanoma.

In conclusion, VE-cad appears to promote the interaction between FAK and EphA2 through regulation of EphA2’s ability to translocate to the membrane. Interaction between EphA2 and its membrane-bound ligand results in phosphorylation of EphA2. Phosphorylated EphA2 then forms an interaction with FAK, which leads to phosphorylation and activation of FAK. The signal transduction pathways activated through VE-cad and EphA2 converge, resulting in activation of PI3-K which then leads to VM via activation of MMP-2, finally resulting in cleavage of the Ln-5γ2 chain[30,32]. These results suggest that VE-cad, EphA2 and FAK act in a coordinated manner as a key regulatory element in the process of melanoma VM and illustrate a novel signaling pathway that could be potentially exploited for therapeutic intervention.

TF and TFPI-1, TFPI-2

TF, which is expressed in endothelial cells, macrophages, smooth muscle cells, and a variety of solid tumors and tumor cell lines[45-47], is a 47-kDa transmembrane protein that binds plasma factor VII/VIIa[48]. This bimolecular complex initiates blood coagulation by activating both factor X and factor IX, which leads to the generation of thrombin, fibrin deposition and platelet activation. In addition, TF is also involved in vascular development and is induced in angiogenic endothelial cells[49,50]. The TF/factor VIIa (TF-VIIa) complex is inhibited by a Kunitz-type protease inhibitor called TFPI, which is typically associated with glycosyl-phosphatidyl inositol-anchored receptors on the cell surface. TFPI type 1 (TFPI-1) consists of three TFPI Kunitz-type inhibitory domains and a proteoglycan-binding COOH terminus TFPI-1 locks TF into an inactive TF-VII-X a-TFPI-1 complex by binding simultaneously to factors VII a and X a. TFPI type 2 (TFPI-2) is a 32-kDa member of the Kunitz-type family of serine protease inhibitors with strong homology to TFPI type 1 (TFPI-1). As an important factor associated with coagulation, TFPI-2 exhibits inhibitory activity toward a broad spectrum of proteases including the TF/factor VIIa catalytic complex, plasmin and plasma kallikrein[51,52]. TFPI-2 also participates in the regulation of ECM remodeling and pericellular proteolysis through plasmin-dependent manner and the action of MMPs[53,54]. However, TFPI-2 expression has also been shown to enhance the migration of certain tumor cells[55]. In addition, TFPI-2 is synthesized by endothelial cells and supports their firm adhesion by mechanisms that are independent of inhibition of plasmin. Recent data from several sources suggest that matrix-associated TFPI-2 can regulate adhesion and migration of endothelial cells and tumor cells in a context-dependent manner[55].

A recent study reported that aggressive melanoma cells in vivo over-expressed TF, TFPI-1 and TFPI-2. TFPI-1 has an anticoagulant function which is of relevance for perfusion of VM. In conclusion, the over-expression of TFPI-1 by aggressive melanoma cells might help to explain the possible dynamic conduction of blood through a VM tumor cell-lined meshwork. TFPI-2 associated with a three-dimensional collagen matrix can induce the VM phenotype in poorly aggressive melanoma cells. Blockage of TFPI-2 is able to suppress MMP-2 activation and prevent VM formation. Therefore, TFPI-2 appears to regulate an essential pathway of VM[11].

Vascular endothelial growth factor, hypoxia and hypoxia-inducible factor-1α

Vascular endothelial growth factor (VEGF-a/VEGF) secreted by tumor cells and fibroblasts plays a crucial role in tumor angiogenesis, lymphangiogenesis and VM formation[56-60]. Moreover, VEGF is the most potent endothelial-specific mitogen; it directly participates in angiogenesis by recruiting endothelial cells into hypoxic and avascular areas and stimulating their proliferation[61,62]. Recently, it has been reported that the expression of VEGF in bi-phase differential malignant tumor with VM is less than that in those without VM proving that VM can sustain tumor blood supply[58]. Inhibition of VEGF expression by sequence-specific siRNA can suppress VM formation in osteosarcoma cells and hence, VEGF appears to be crucial for formation of VM.

It is believed that hypoxia is able to induce VM channel formation directly to enhance the ability of tumor to metastasize[63]. VEGF has been shown to increase with hypoxia challenge, a response which seems to depend on hypoxia regulated in the 5 and 3 regions of the VEGF gene. The hypoxia-inducible protein complex hypoxia-inducible factor1-α (HIF-1α) binds to the enhancer sequences of the VEGF gene, and both transcription and RNA stability are enhanced[64,65]. Su et al[65] have shown that the HIF-1α inhibitor, rapamycin, could prevent VM and phenotype transformation of human ovarian cancer cells, HIF-1α protein expression correlated with CD31 and Factor VIII protein expression. These findings indicate that VM might be associated with HIF-1α. EphA2 or/and VE-cad regulate the activity of MMPs, which promote the cleavage of Ln-5γ² chain into promigratory γ², and γ^2x fragments. The release of these fragments into the tumor microenvironment can increase the VM formation of aggressive melanoma[35]. It has been indicated that VEGF may significantly stimulate EphA2 and VE-cad expression at the protein and mRNA levels in ovarian tumor cells, MMP-2 and MMP-9 which act as effecter molecules, induced by EphA2, are controlled by VEGF. Through this process, aggressive human ovarian tumor cells enhance their capacity for migration, invasion and VM formation. Moreover, the down-regulation of EphA2 following VEGF stimulation can also decrease VM formation in human ovarian tumor cell 3D cultures after EphA2 knockdown, whereas there is no significant change in VE-cad[39]. Additionally, hypoxic activation of HIF-1α might be involved in driving VM in Ewing sarcoma tumors. However, the relationship between tube formation by Ewing sarcoma tumor cells and VEGF regulated by hypoxia could not be identified[66].

The amount of VM channels and gene expression of HIF-1α, MMP-2, MMP-9, and VEGF was increased significantly in the ischemic group than that in non- ischemic group of melanoma tumors[67]. It has been proposed that thalidomide inhibits VM channel and mosaic vessel formation in melanoma through inhibiting VEGF, MMP-2 and MMP-9 expression[68]. The VEGF expression and reactive oxygen species (ROS) level are key requirements for formation of capillary-like structures (CLS) formation. Antioxidants (AOs) may induce a significant decrease of VEGF expression in melanoma cells. The reduction of ROS generation in melanoma cells by AOs may completely abolish CLS[69]. Recent studies have reported that CLS formation requires apoptotic cell death through activation of caspase-dependent mechanisms. Apoptosis occurs before CLS but not after CLS assembly and the formation of CLS is related to the ROS levels[70,71].

Migration-inducing protein 7

Migration-inducing protein 7 (Mig-7) is the cysteine-rich protein found in cell membranes and the cytoplasm of carcinoma cells. Mig-7 is also an early marker of migration and circulation in carcinoma cells. Several tumor cells that form vessel-like structures and embryonic cytotrophoblast cells can masquerade as endothelial cells by expressing VE-cadherin and Factor VIII-associated antigen[4,30,34,66]. RTK c-Met activation of the hepatocyte growth factor/scatter factor (HGF/SF) receptor can induce Mig-7 expression[72,73]. Integrin avβ⁵ ligation is required in cross-talk signaling with RTK c-Met to initiate Mig-7 expression[72]. Furthermore, receptor tyrosine kinase ligands, such as HGF or epidermal growth factor (EGF) and avβ5 Integrin have been reported to induce expression of Mig-7 in carcinoma cells. Petty et al[74] observed that Mig-7 protein over-expression was found in aggressive invasive melanoma cells capable of VM but not in poorly invasive cells that do not form the tumor-lined structure, Mig-7 protein was primarily co-localized with VM markers VE-cad, Factor VIII-associated antigen and Ln-5γ² chain domain III fragment in lymphnode metastases. Over-expression of Mig-7 increased γ² chain domain III fragments that are known to contain EGF-like repeats that can activate EGF receptor. EGF can also induce Mig-7 expression. Ln-5 is the only laminin that contains the γ² chain the domain III region of which following cleavage into promigratory fragments, causes increased levels of MMP-2. MMP-2 and MT1-MMP cooperate to cleave γ² chain into fragments that promote melanoma cell invasion and VM.

Galectin-3, galectin-1 and brain-expressed X-linked gene 2

Galectin-3 (Gal-3) is a 31 kDa member of the galectin family which consists of three distinct structural domains: (1) a short NH₂-terminal domain that controls its cellular targeting; (2) a repetitive collagen-like sequence rich in glycine and proline, which serves as a substrate for MMPs; and (3) a COOH-terminal domain which a globular structure that encompasses the carbohydrate-binding site[75-77]. Gal-3 has pleiotropic biological functions which depending on its subcellular location. Extracellular Gal-3 mediates cell migration, cell adhesion and cell-to-cell interaction[78]. Intracellular Gal-3 inhibits Fas-induced T-cell apoptosis[79]. Mourad-Zeidan AA and colleagues[80] proposed that Gal-3 is a important upstream regulator of interleukin-8 (IL-8) and MMP-2 expression in melanoma and a key gene in VM formation. Gal-3 silencing could result in a decrease of VE-cad and IL-8 promoter activities due to enhanced recruitment of early growth response-1 (EGR-1). Gal-3 silencing could also inhibit melanoma cell invasion capability through Matrigel-coated filters and VM formation. EGR-1, described as a tumor suppressor, acts as a negative regulator of the VE-cad and IL-8 promoters. Thus the over-expression of EGR-1 results in the inhibition of VE-cad and IL-8 expression and of their promoter activities, and Gal-3 acts upstream to prevent EGR-1 binding. In addition, Gal-3 causes tumor angiogenesis and melanoma VM by inducing the expression of fibronectin-1 and endothelial differentiation sphingolipid G-1 (EDG-1) genes.

Gal-1 is a 14 kDa β-galactoside binding protein, capable of forming lattice-like structures with glycans of cellular glycoconjugates and inducing intracellular signaling. Gal-1 is present both inside and outside cells. As an extracellular effecter, it can bind to cell-surface glycoconjugates that contain suitable galactose-containing oligosaccharides, acting as a homobifunctional cross-linker. It also binds to some of the glycoproteins in the ECM, such as laminin, fibronectin and elastin. As an intracellular effecter, Gal-1 shuttles between the nucleus and the cytoplasm. It is active in processes that are essential for basic cellular functions, like pre-mRNA splicing, cell growth, apoptosis and cell cycle regulation[81].

Recent observations have indicated that the decrease in Gal-1 expression in Hs683 cells through targeted small interfering RNA could provoke a marked decrease in brain-expressed X-linked gene 2 (BEX2) expression. BEX2, described as a tumor suppressor gene in astro gliomas, has been implicated in apoptotic features of breast cancer. Decreasing BEX2 expression impairs VM channel formation in vitro and angiogenesis in vivo by inducing the up or down-regulation of a number of genes involved in migration including MMP-2, plexin C1, integrin β6 and SWAP70[82].

Inhibitor of DNA binding 2

The inhibitors of DNA binding (Ids) proteins are a family of helix-loop-helix (HLH) proteins that lack the basic domain necessary for DNA binding. Id proteins, including Id1-Id4, are inhibitors of basic helix-loop-helix (bHLH) transcription factors. Id2 is involved in the regulation of cell differentiation, proliferation, development, cell cycle regulation, myogenesis, tumorigenesis and neurogenesis[83-85]. Recently, it has been believed that knockdown of Id2 expression can significantly inhibit VE-cad expression and abrogate the formation of tubular networks in highly aggressive uveal melanoma cells in 3D culture.VE-cad may be the target molecule of Id2 during VM formation[86].

Cyclic AMP and nodal

Cyclic AMP, a second messenger controlling many cellular processes with idiosyncratic responses depending on cell type, is produced by adenylyl cyclase after binding of physiologic ligands (α MSH, VIP and ADR) to G protein-coupled receptor (GPCR). Cyclic AMP phosphorylates several substrates involved in signal transduction pathways after binding to PI3K-dependent protein kinase Akt (PKA). In particular, the mitogen-activated protein kinase (MAPK) cascade consists of small GTP-binding protein Ras/B- and/or C-Raf kinase 1/2 (MEK1/2) and ERK1/2[87]. Recent observation indicates that cyclic AMP (Epac) Epac1 and Epac2 can mediate PKA-independent cell responses. Epac1 and Epac2 are unique exchange factors of small the GTPases, Ras-associated proteins (Rap) 1 and 2, which become activated upon releasing GDP and binding GTP[88,89]. However, Epac responses can cooperate with PKA responses[88,90]. Numerous studies have suggested the link between PKA activation and Epac. Akt is a major effecter of the PI3K signal and the PI3K/Akt pathway is frequently altered in cancers[91]. It has been reported that cyclic AMP could mediate a physical association between Epac, Rap1 and phosphorylated Akt[92,93].

Cyclic AMP can inhibit VM formation through multiple signaling pathways in aggressive melanoma cells in vitro. Firstly, cyclic AMP results in the inhibition of VM mediated by activation of Epac/Rap1 producing Rap1-GTP; PKA is not mediator of VM. Secondly, forskolin inhibits VM formation through inhibiting PI3K/Akt signals (dephosphorylation of Akt-activating kinase domain, pAkt) but not through GPCR ligands α MSH or VIP; whereas, GPCR ligands only act as stimulators. Finally, forskolin also inhibits ERK1/2 activity-related phosphorylation (pERK1/2) independently of Epac and PKA, consequently inhibiting VM formation through the MAPK pathway[94].

The human Nodal gene, containing three exons, is located on Chromosome 10q22.1. Nodal is a member of the transformation growth factor β (TGF-β) superfamily and is pivotal in inhibition of human embryonic stem cells (hESCs) differention. Indeed, Nodal has been shown to maintain the pluripotency of ESCs and is one of the first genes to be down-regulated as totipotent hESCs differentiate during embryoid body formation.Moreover, as a melanoma plasticity biomarker, Nodal plays an instrumental role in the maintenance of melanoma cell plasticity and tumorigenicity. Nodal expression is positively associated with melanoma tumor progression; tumorigenic melanoma cells lines express high levels of Nodal. There are four known mammalian Notch receptors (Notch1-4) and five ligands. Recent studies have proven that Nodal is up-regulated by Notch-4 in aggressive melanoma cells[95]. Activation of Nodal signaling supports VM and expression of the VM plasticity marker VE-cad in aggressive melanoma cells. Inhibition of the Nodal signaling pathway results in a reduction in keratin and VE-cad. Inhibition of the Nodal signaling pathway can also decrease melanoma cell invasiveness and impair the ability of aggressive melanoma cells to form vascular networks on a three-dimensional collagen matrix by down-regulation the expression of keratin and VE-cad[96]. Furthermore, through an Epac/Rap1 signaling, the cyclic AMP can reinforce endothelial barrier properties via the redistribution of VE-cad and strengthened cell adhesion. The cyclic AMP -mediated activation of Nodal signal may result in the inhibition of VM formation[94].

Cyclo-oxygenase-2

Cyclooxygenase (COX) is the enzyme catalyzing the rate-limiting step in prostaglandin synthesis. It has two isoforms, COX-1 and COX-2. COX-1 is constitutively expressed in normal cells and is involved in homeostasis. However, COX-2 is not found in normal conditions but is induced by a variety of pathophysiological factors, such as growth factors, inflammatory stimuli, oncogenes and tumor promoters. COX-2 has been shown to promote cell survival, proliferation, and angiogenesis and prohibit apoptosis, all process influencing cancer development[97].

Recent reports have revealed that highly invasive human breast cancer cells that exhibit higher COX-2 expression can develop vascular channels when cultured on three-dimensional Matrigel, whereas non-invasive cell lines that express low levels of COX-2 cannot develop such channels. Moreover, human high-grade invasive tumor specimens that expressed high levels of COX-2 proteins had detectable vascular channels. Low-grade tumors with no or low COX-2 expression showed little evidence of VM[98]. COX-2 inhibition by celecoxib or specific siRNA may inhibit vascular channel formation in human breast cancer cells. Vascular channel formation was abrogated by addition of exogenous prostaglandin E₂ (PGE₂). Therefore, the effect of celecoxib in inhibiting vascular channels is probably related to the dependence on PGE₂[99].

Translocation-Ets-Leukemia/ETV6

Translocation-Ets-Leukemia (TEL) or ETV6 is a member of the ETS family of transcription factors, and is frequently a target of chromosomal translocations in several forms of acute leukemia. A translocation between region 12p13, which contains the ETV6 gene, and 21q22, which contains the RUNX1 gene, creates an ETV6/RUNX1 chimeric gene. The fusion protein, ETV6/RUNX1, contains amino acids 1-336 of ETV6 linked to residues 21-480 of RUNX1[100]. ETV6 functions as a transcriptional repressor by recruiting the co-repressors mSin3A (N-CoR) and HDAC3 to the promoters of target genes. Several studies have suggested that cytogenetic abnormalities of chromosome 12p13 involving the TEL/ETV6 gene exist in a variety of hematopoietic neoplasms including acute leukemias, myelodysplastic syndromes, and myeloproliferative disorders[101]. Furthermore, TEL is also expressed in endothelial cells in large mature blood vessels during normal and tumor angiogenesis.

A study demonstrated that TEL-transduced NIH3T3-UCLA cells exhibit hollow cellular cords with a diameter of several cell bodies (15-25 um), which would be wide enough to allow the transport of fluids and red blood cells. In addition, TEL can act in synergy with RAS expression to induce aggregation in MS1 (endothelial cell line) and NIH3T3 cells. Dominant-negative (DN)-RAS expression can also induce slower growth in parallel with effects of TEL in these cell lines. Thus TEL might play a key role in aggregation and VM by synergizing with signaling pathways downstream of RAS[102].

Bone morphogenetic protein-4

Bone morphogenetic proteins (BMPs) are members of the TGF-b superfamily. The more than 30 members in the BMP family are described either as BMPs, osteogenic proteins, growth/differentiation factors, or cartilage-derived morphogenetic proteins[103]. They exert their biological activities by binding to a complex of serine/threonine kinase receptors type I (i.e. BMPR-IA, BMPR-IB, or ALK2) and type II (i.e. BMPR-II)[104]. BMP-4, a representative member of the BMP family, is required for several different processes in early development beginning with gastrulation and mesoderm formation.

In a study by Rothhammer et al[105], BMP-4 was identified as an important molecule in melanoma migration and invasion. Further study revealed that melanoma cells with reduced BMP-4 activity were not able to form tube-like structures at all. In addition, melanoma and endothelial cells were able to form chord-like networks in a cooperative manner on Matrigel. Co-cultures of endothelial cells and different melanoma cell clones revealed the formation of tubular structures only in those cell clones with unchanged BMP levels. The antisense BMP-4 and chordin-overexpressing cell clones were not able to form networks and even prevented tube formation by endothelial cells. Furthermore, the down-regulation of BMP-4 activity could lead to reduced expression of genes involved in VM, including EphA2 and VE-cad. On the basis of these studies, it is considered that BMP-4 is a pivotal factor of VM[106].

VM AND CANCER THERAPEUTICS

Tumor growth and metastasis require a blood supply for survival and aggressive cancers use several mechanisms to increase tumor perfusion. Most of the anti-vascular therapy in tumors is antiangiogenic, aimed at blocking the function of specific growth factors or receptors. The endothelial cells isolated from tumors can grow independently of the presence of endothelial growth factors, suggesting that tumor endothelial cells can acquire some characteristics that make them less sensitive to anti-angiogenic therapy. Not all tumors are dependent on some specific factor or receptor in order to be vascularized. VM, an angiogenesis-independent novel pathway, is thought to provide a mechanism of perfusion and a dissemination route within the tumor that functions either independently of or simultaneously with angiogenesis. VM was also reported to be resistant to angiogenesis inhibitors such as endostatin and TPN-470 in melanoma tumor cells and the B16F10 murine melanoma model[73]. So, therapies that target angiogenesis must not be the only strategies that target the tumor microcirculation.Additional tumor-target therapy for non-angiogenic pathways of tumor perfusion and metastasis should be considered. VM, an independent risk factor of prognosis, and/or PAS-positive patterns was associated with poor prognosis in tumor patients[2,12-22]. Hence, it would be prudent, and possibly essential, to target VM in developing strategies for tumors therapy.

Until now, a variety of genes have been investigated for their role in tubular network formation in tumor cells. An option for therapy is the use of monoclonal antibodies and antisenseoligonucleotides to these molecular for drug targeting. It has been reported that MMP inhibitor, PI3K inhibitor, PSMA (prostate-specific membrane antigen) inhibitor, a knockout EphA2 gene, down-regulation VE-cad, and an antibody against Ln-5 γ² chain antisense oligonucleotides have an effect in the inhibition of VM. A number of recent papers demonstrated that several drugs demonstrate specific anti-VM effects. Genistein, a predominant isoflavone in soybeans, was able to inhibit VM formation of uveal melanoma through down-regulation of VE-cad in vitro. The ectopic model study showed that VM in uveal melanoma specimens were significantly reduced by Genistein in vivo[107]. Zhang et al[68] have discovered thalidomide, which was used to treat morning sickness during pregnancy in the 1960s but was banned for its side effects (caused phocomelus), could inhibit VM through the regulation of vasculogenic factors. Doxycycline may inhibit the growth of engrafted melanoma and result in reduced expression of MMP-2, MMP-9 and VM formation. However, VEGF expression in the tumors increased in the doxycycline-treated animals. VM channel and endothelium-dependent vessels were reduced, suggesting that microcirculation patterns were inhibited by doxycycline administration, aggravating hypoxia, and increasing VEGF expression[108]. It was recently reported that Rapamycin, a HIF-1α inhibitor, could inhibit VM and phenotype transformation of SKOV3ip[65].

6-demethyl-6-deoxy-4-dedimethylamino-tetracylcine (COL-3) was used to evaluate for treatment of patients with refractory solid tumors. The administration of COL-3 to aggressive melanoma cells in three-dimensional culture inhibited MMP-2, MMP-9, MT1-MMP, and VE-cad expression. In addition, Ln-5 γ² chain was inhibited and decreased vascular network formation was observed[36]. Celecoxib is a highly selective COX-2 inhibitor, which may inhibit vascular channel formation in human breast cancer cells through PGE₂ pathway[99]. Imatinib, an inhibitor of PTKs, has been extensively used in the clinical treatment of gastrointestinal stromal tumors (GIST). It is primarily being targeted toward inhibiting continuous activation of Kit PTK caused by the mutation of oncogene c-kit in GIST. The capacity of inhibition of PTK to decrease VM could be one of the reasons why such drugs could be successfully used in the treatment of GIST[17]. In aggressive human melanoma cells in vitro, forskolin was shown to inhibit VM channel formation through MAPK and PI3K pathway[94]. However, the study of VM in tumors is still at early stage, VM makes tumor growth inhibition even more complex and it is difficult to propose a precise anti-VM therapy strategy. An efficient anti-VM therapy should focus on three aspects: remodeling of the ECM and tumor microenvironment, blocking biochemical and molecular signaling pathways of VM, inhibiting plasticity of tumor cells. However, there is still a long way to go to completely elucidate mechanisms of VM and to translate tumor angiogenesis into efficient tumor therapies which can been applied to human cancer treatments.

CONCLUSION

Tumor vascularization can be explained by angiogenesis, mosaic blood vessel formation and VM. As a novel and functional tumor microcirculation, VM has been described in more and more tumors. VM and the level of VM density are associated with shorter survival and poor prognosis. To date, relatively little is known about the ability of tumor cells to form highly patterned vascular channels at the molecular level.

Tumor cell plasticity underlies VM and aggressive tumor cells may revert to an undifferentiated stem cell-like phenotype. A study by Monzani et al[109] showed that a stem cell population which could increase the melanoma progression was found in melanoma biopsy. Therefore, the cancer stem cells (CSCs) subpopulation inside the tumor can organize VM and a mosaic network, depending on the environmental conditions. At least for melanoma, VM or mosaic vessel formation is due to the trans-differentiative capacity of the CSCs subpopulation. The evidence of such a subpopulation, of course, opens a new perspective for the treatment of melanoma. CSCs afford an opportunity to investigate tumor cell plasticity characteristics and a new therapeutic perspective. In addition, recent reports have indicated that bone marrow macrophages and dendritic cells have trans-differentiative capacity and that macrophages contributed to building neovessels in active multiple myeloma through VM[110]. Further research about VM has demonstrated that Mast cells (MCs) contributed to neovascularization; MCs were located in the vessel wall that connect with endothelial cells (Ecs) to line the vessel lumina. This behavior of MCs could be regarded as an example of VM like that of melanoma or other tumor cells, which themselves form vascular channels[111]. Hence, there further research required to explore the mechanisms how the tumor microenvironment and ECM influence both tumor cells and normal cells.

The relationships between VM channels, mosaic blood vessels, endothelium-dependent blood vessels, and lymphatic tubes need to be elucidated. Mosaic blood vessels constitute 4% of the total surface area of tumor microcirculation. Some researchers believe that Mosaic blood vessels are not related to VM channels and are a mosaicism of endothelial and tumor cells. Mosaic blood vessels are a transient phenomenon and a three-stage phenomenon including VM channels, mosaic blood vessels, endothelium-dependent blood vessels was proposed by Zhang et al[4]. During the early stages of transplanted melanoma, all three patterns are observed in tumor tissues, especially VM. With the tumor growing, more mosaic blood vessels, endothelium-dependent blood vessels are observed, and in the later stages endothelium-dependent blood vessels become the main blood supply pattern. Hence, it is very important to choose the right time node for anti-VM therapy.

Results from a recent study showed that aggressive melanoma tumor cells could express VEGF-C and lymphatic-vessel endothelial hyaluronan receptor 1[112]. These results raise the intriguing possibility that the fluid-conducting meshwork could mimic a lymphatic-like network. VM might be associated with lymphatic circulation, and a lymphatic-like network probably exists in tumor tissues.

The full range of molecular players and their specific roles in VM are still to be elucidated. The exploration of drugs targeted at molecular signaling pathways in VM is a field filled with challenges and hope. Therapies targeting VM have only been attempted in experimental research till date. VM provides an avenue to investigate the interrelationships between the genetically dysregulated invasive tumor cell, tumor microenvironment, and the malignant switch. Perhaps, the combination of anti-VM with other therapy strategies, such as chemotherapy, anti-angiogenesis, and anti-lymphangiogenesis will prove to be promising option.