The Hemostatic System and Angiogenesis in Malignancy

Abstract

Coagulopathy and angiogenesis are among the most consistent host responses associated with cancer. These two respective processes, hitherto viewed as distinct, may in fact be functionally inseparable as blood coagulation and fibrinolysis, in their own right, influence tumor angiogenesis and thereby contribute to malignant growth. In addition, tumor angiogenesis appears to be controlled through both standard and non-standard functions of such elements of the hemostatic system as tissue factor, thrombin, fibrin, plasminogen activators, plasminogen, and platelets. “Cryptic” domains can be released from hemostatic proteins through proteolytic cleavage, and act systemically as angiogenesis inhibitors (e.g., angiostatin, antiangiogenic antithrombin III aaATIII). Various components of the hemostatic system either promote or inhibit angiogenesis and likely act by changing the net angiogenic balance. However, their complex influences are far from being fully understood. Targeted pharmacological and/or genetic inhibition of pro-angiogenic activities of the hemostatic system and exploitation of endogenous angiogenesis inhibitors of the angiostatin and aaATIII variety are under study as prospective anti-cancer treatments.

Hemostasis, Angiogenesis, and Tumor Progression

The hemostatic system is anatomically and functionally inseparable from the vasculature. It is therefore somewhat ironic that in the context of cancer, these respective host elements have, until recently, been studied as virtually independent entities. This article is intended to illustrate the ongoing revision of this historical view and to summarize the therapeutic potential of targeting elements of the hemostatic system as a strategy to inhibit tumor angiogenesis.

Close association between malignant disorders and various perturbations in blood coagulation has been recognized for over 130 years [1]. Coagulation dysfunctions of different nature and magnitude, ranging from subtle laboratory abnormalities to overt thromboembolism, thrombophlebitis, and disseminated intravascular coagulation, are routinely found in cancer patients [2–8]. Up to 50% of all patients with malignant disease and up to 90% of those with metastatic lesions demonstrate abnormalities in hemostatic parameters [5]. In this regard, pancreatic cancer, breast cancer, and particularly acute promyelocytic leukemia are the best-described examples [7,9]. However, cancer-related hemostatic complications are usually heterogenous in nature and their pathogenesis is often poorly understood. This is why they are often collectively referred to as “cancer coagulopathy” or “paraneoplastic syndrome” [2], as their manifestations are found at both the systemic level (deregulation of blood coagulation) and locally at the tumor site (crosslinked extravascular fibrin and fibrinogen) [10–13]. There are ample data suggesting that these respective changes are not merely an epiphenomenon of the disease, but rather represent an integral part of the pathobiology of tumor growth and dissemination [2,14–16]. In this regard, the interrelationship between cancer coagulopathy and the onset of tumor angiogenesis is of particular interest.

It is widely accepted that most primary tumors and metastatic lesions cannot grow beyond 2 to 3 mm in size in the absence of vascularization [17]. Regardless of whether the latter is secured by occasional “cooption” of preexisting capillaries [18] or by active recruitment of their new extensions (angiogenesis) [17,19,20], tumor-associated vasculature is essential not only to ensure continued metabolite and oxygen exchange but also as a source of important “paracrine stimulation” [21] through endothelial cell-derived extracellular matrix (ECM) [22], proteases [23] and cytokines, regulating tumor cell growth [24], survival [25], invasion [26], and metastasis [27].

It is now thought that the onset of tumor neovascularization (“angiogenic switch”) results from a shift in balance between angiogenesis stimulators and inhibitors released by both tumor parenchyma and “activated” host stromal cells [28–32]. Among the latter, stromal fibroblasts [33,34], mast cells [35], resident macrophages [36], blood-borne mononuclear leukocytes [15,37,38], and platelets [39,40] are considered the main sources of angiogenesis regulators.

Operationally, pro-angiogenic conditions may be triggered by a gain-of-function and/or a loss-of-function event [28,32,41]. In the former case, angiogenesis stimulators such as vascular endothelial growth factor (VEGF), members of the fibroblast growth factor family (e.g., bFGF, aFGF), hepatocyte growth factor (HGF), or other similarly acting entities are induced or upregulated in the tumor microenvironment, evoking responses of normally quiescent capillary endothelial cells [42,43]. Conversely, such pro-angiogenic state may result from downregulation of constitutively expressed angiogenesis inhibitors acting either locally (e.g., thrombospondin-1 and -2 (TSP-1, -2), maspin, brain-specific angiogenesis inhibitor 1 (BAI1), pigment epithelium-derived factor (PEDF), interferons α, β, and γ, Meth-1, -2) [32,44–47] or systemically (agiostatin, endostatin, vasostatin, aaATIII) [48–51]. The combined, cumulative impact of those various types of influences is believed to be responsible for initiation and maintenance of tumor angiogenesis and resulting escape from a dormant avascular state [28].

In the context of cancer, the imbalance between angiogenesis stimulators and inhibitors can be ultimately traced to direct or indirect influences exerted by genetic events underlying the disease progression [32,52,53]. Thus, loss-of-function mutations in several tumor suppressor genes (e.g., p53) can trigger changes in expression of both angiogenesis inhibitors (e.g., TSP-1, BAI-1) [32] and stimulators (e.g., VEGF) [54–56]. Likewise, gain-of-function genetic events, such as activation of dominant transforming oncogenes in tumor cells (e.g., ras, myc, EGFR, or HER-2), affect expression of multiple molecular effectors of angiogenesis [41]. In both cases, the impact of genetic lesions can be greatly amplified by epigenetic influences such as hypoxia [57–60], cell-cell contact [53,61], paracrine growth factors, and inflammatory cytokines present in the tumor microenvironment [53,62]. Factors produced by transformed cells can also attract, activate, and induce angiogenic phenotype in host stromal cells [35], thereby promoting the process in an “indirect manner.” It is interesting that in a similar fashion, consequences of cellular transformation may also participate in deregulation of the hemostatic system with resulting modulation of vascular responses [63]. In this regard, Rak and Klement [63] have recently proposed that transforming genetic changes can play a causative role in “cancer coagulopathy” by changing the expression of tissue factor (TF), proteases, VEGF, and other mediators. Such genetic influence would likely exert at least an “indirect” effect on tumor angiogenesis. While many scenarios can be envisaged how such genetically driven tumor angiogenesis-hemostasis crosstalk can be realized, it may be helpful to begin by considering some of the relevant consequences of the commonly observed upregulation of VEGF in tumors, an event for which a genetic cause is relatively well defined [41].

In 1979, Dvorak et al. [64] suggested that tumor neovascularization can be explained by apparent capillary hyperpermeability at the tumor site and resulting leakage of fibrinogen into extravascular spaces followed by formation of the pro-angiogenic fibrin matrix. He argued that it is the tumor vascular permeability factor (VPF), rather than then hypothetical tumor angiogenesis factor (TAF) [17,65], that causes formation of the intratumoral capillary network [64]. In 1983, VPF was biochemically identified [66] and in 1989 cloned and found to be identical with the newly discovered endothelial mitogen named VEGF [67,68]. VEGF/VPF (VEGF-A) is now known to be the central endothelial cell-specific growth factor, angiogenesis inducer, survival factor, and permeability regulator, the expression of which is upregulated in the vast majority of solid tumors and leukemias [69]. The corresponding gene (VEGF-A) contains a classical hypoxia-responsive element in its 5′ untranslated region and is normally under physiological control of the hypoxia-inducible factor 1 (HIF-1), a transcription factor regulated by an oxygen-sensing cellular mechanism [70]. However, it is now believed that upregulation of VEGF in many tumor cells is not solely a result of hypoxia, but rather is linked to genetic changes associated with malignant transformation [41]. Some of these changes (expression of activated src oncogene, loss of p53 tumor suppressor gene) act by mimicking the intracellular signals normally induced by hypoxic conditions [70], while others (e.g., mutational activation of ras oncogenes) can amplify the effect of hypoxia [57,62,70] or act in a constitutive manner to stimulate VEGF expression [71]. Once released from tumor cells, VEGF acts specifically on at least five types of endothelial cell receptors, namely, VEGFR-1/flt-1, VEGFR-2/flk-1/KDR, VEGFR-3/flt-4, and non-kinase co-receptors neuropilins-1 and 2 (NP-1, 2) [68,72–74]. The vascular permeability-inducing effects of VEGF/VPF, which are thought to promote fibrinogen extravasation and extravascular clotting, are known to be regulated by a concerted action of VEGFR-2, c-src kinase, focal adhesion kinase, and αvβ5 integrin [75,76]. While the case of VEGF is very instructive, it merely exemplifies the more extensive and complex tripartite interactions that, in different pathological contexts, may occur among tumor parenchyma, microvasculature, and the hemostatic system.

Angiogenic Activities of the Coagulation System

Several constituents of both intrinsic and extrinsic coagulation cascades have been found to possess overt or cryptic angiogenesis-regulating properties [40]. They include, e.g., angiogenesis inhibitory activity expressed by kininostatin, i.e., domain 5 of the high-molecular-weight kininogen (HK) [77], elastase/thrombin-cleaved fragment of antithrombin III [49], and possibly factor XIII [78]. Activities that may stimulate or modulate tumor angiogenesis are linked even more strongly with such key regulators of coagulation as TF, thrombin, and fibrinogen/fibrin, an issue which deserves a more thorough discussion.

Tissue Factor

TF is the principal initiator of the extrinsic pathway of coagulation traditionally recognized to play an important role in “cancer coagulopathy” [79,80]. Increased expression of TF has been observed in many tumor types, including small cell carcinoma, bronchoalveolar carcinoma and large cell carcinoma of the lung, colon adenocarcinoma, head and neck cancer, malignant gliomas, bladder cancer, and prostate cancer [10,81–85]. Tumor cells express TF constitutively [80,81] and possibly trigger production of TF by adjacent host cells, including monocytes and endothelial cells [86].

The functional linkage between TF and angiogenesis has been implicated by a number of studies, which invoke existence of several possible complementary mechanisms of action, such as: 1) generation of pericellular thrombin and pro-angiogenic thrombin signaling; 2) extravascular coagulation and formation of pro-angiogenic fibrin matrix; and 3) changes in expression of angiogenic growth factors by autonomous TF-mediated intracellular signals [85,87–93] (Figure 1).

Figure 1.

The “vicious circle” of coagulation and angiogenesis in cancer [63]. Tumor cells are thought to acquire angiogenic properties under the influence of transforming changes, i.e., expression of oncogenes and/or tumor suppressor genes (TSG). Resulting onset of VEGF production, along with imbalance in expression of other angiogenesis stimulators and inhibitors, triggers several endothelial cell responses, such as new blood vessel formation, expression of tissue factor (TF), and vascular permeability. This complex reaction allows extensive contact (and activation) of coagulation factors (VII, X, II) in plasma with pro-coagulant extravascular environment (tumor parenchyma, stroma, ECM) and TF-positive tumor endothelium. TF- and PARs-mediated intracellular signals induce further increase in expression of angiogenic factors (e.g., VEGF) by tumor cells. TF signaling also sensitizes endothelial cells to further VEGF stimulation (see text for details).

Studies involving enforced TF downregulation provide compelling evidence for the causative role of this transmembrane receptor protein in developmental and pathological blood vessel formation. Thus, mice harboring the homozygous TF null mutation die in utero with/of apparent abnormalities in their yolk sack vasculature [94,95]. Expression of the antisense TF mRNA in Meth-A sarcoma led to decreased transcription of the VEGF gene, increase in expression of TSP-1, and decline in tumor growth and vascularity [87]. This apparent interrelationship between expression of TF and VEGF is of particular interest because co-localization of these proteins has been documented in several instances, such as human melanoma cells grown as xenografts in immunodeficient mice, human breast cancer, human glioma, and human adenocarcinoma of the lung [83,88,96].

The expression of both TF [97,98] and VEGF [99–101] is controlled by hypoxia. Pentoxifylline treatment downregulates TF expression by monocytes and endothelial cells, and inhibits hypoxia-induced synthesis of both TF and VEGF by three different malignant cell lines [86]. However, while under low oxygen conditions, VEGF promoter responds mainly to HIF-1-induced stimulation [70,102]; TF expression is increased by a mechanism independent of HIF-1 activity, as it readily occurs in the absence of the β-subunit of HIF-1 [103]. Experiments with mice homozygous for a null mutation of the early growth response (Egr-1) gene indicated that this transcription factor is responsible for upregulation of TF in hypoxic lung tissue [97,98]. Interestingly, Egr-1 also appears to be responsible for VEGF-and TNF α-dependent upregulation of TF by endothelial cells [104]. TF is considered to be an early response gene, the expression of which is regulated by serum, growth factors, and cytokines [105].

Structurally, TF gene product is composed of three functionally distinct domains: namely, the extracellular — i.e., the factor VIIa — binding region, the transmembrane domain, and the short cytoplasmatic tail [106]. The relative contribution of these submolecular regions to VEGF upregulation and angiogenesis has been recently a subject of intensive experimental exploration, but also of some debate. Thus, re-expression of the extracellular domain alone was shown to rescue the TF-/- phenotype in mice and prevent both embryonal lethality and vascular defects [107]. This may indirectly imply that, if VEGF is the main mediator of the TF-dependent prenatal angiogenesis, the extracellular domain of TF may be both necessary and sufficient to control VEGF levels. Contrary to this result, Abe et al. [92] demonstrated that transfection of human melanoma cells with the cDNA encoding TF cytoplasmic domain alone resulted in elevated VEGF expression. Furthermore, the cytoplasmatic domain of TF has been recently implicated in triggering protein kinase C-dependent signalling, a mechanism known to mediate VEGF upregulation by a variety of stimuli [108–110]. Although intriguing in itself, this postulated exclusive role of the cytoplasmatic domain in proangiogenic, TF-dependent cellular signalling is probably an exception rather than the rule, if the following observations are taken collectively: 1) aforementioned “rescue” of the TF-/- phenotype in mice by expression of the extracellular TF domain; 2) dependence of TF-induced Ca²⁺ oscillations and changes in gene expression in several cell types upon binding of fVIIa (to the extracellular domain of TF) [109,111–113]; 3) activation of MAPK pathway, a major inducer of VEGF expression [53,114], by fVIIa/TF interactions [115,116]; 4) fVIIa/TF complex-dependent activation of phosphatidyl inositol 3′-OH kinase (PI3K) and members of src family of kinases, both involved in angiogenesis [117]; and 5) TF-mediated VEGF upregulation in human fibroblasts driven by activation of factor VII-associated procoagulant activity and resulting generation of factor Xa and thrombin [93,118].

The catalytic activity of the TF/VIIa complex triggers activation of factor X and thrombin, both of which can change expression of various genes (likely including angiogenesis regulators), notably through activation of so-called protease-activated receptors (PARs) on the cell surface [107]. This effect can be tamed by the action of the TF pathway inhibitor (TFPI), synthesized by vascular endothelial cells and macrophages [119]. TF can also form a quaternary complex with factor VIIa, factor Xa (fXa), and cell-bound TFPI — an interaction which results in cellular transport of TF to small plasmalemmal vesicles (caveolae), downregulation of its proteolytic activity, and placing it in the proximity of such elements of intracellular signaling as G-protein-coupled receptors, non-receptor tyrosine kinases, and the plasma membrane Ca²⁺ pumps [8,120–122]. The consequences of these latter events for angiogenesis remain to be elucidated.

Endothelial cells stimulated with VEGF upregulate the expression of TF on their surfaces [104,106,123] — an event likely to induce procoagulant conditions within the tumor microvasculature. This is consistent with the observation that TF is often upregulated on endothelial cells associated with VEGF-positive tumors [88]. It is tempting to speculate that the mutual interrelationship between VEGF and TF expression may constitute a centerpiece of the proangiogenic feedback loop in tumors (compare Figure 1). However, VEGF can also stimulate expression of the elements of the fibrinolytic cascade, including: plasminogen activators (PAs; u-PA and t-PA), urokinase receptor (u-PAR) [124], and plasminogen activator inhibitor 1 (PAI-1) [125]. Collectively, these are not merely uncontrolled changes in proteolytic activities of the host hemostatic and fibrinolytic systems, but rather manifestations of a new equilibrium that favors tumor vascularization invasion and metastasis [20,106,126].

Thrombin

Activation of TF-dependent blood coagulation leads to generation of thrombin, the presence of which has been detected in several tumor types (small cell lung cancer, renal cancer, malignant melanoma, ovarian cancer, laryngeal cancer, and gastric cancer [10–12,127–130]). Thrombin per se is believed to contribute to cancer progression [131,132] through increase in tumor cell adhesiveness, metastatic potential, and tumor cell-induced platelet aggregation [131]. Thrombin increases TF synthesis in different cells and thus facilitates activation of blood coagulation in a self-perpetuating fashion [86]. However, by activating tissue-type plasminogen-activator inhibitor-1 (PAI-1) and thrombin-activatable fibrinolysis inhibitor (TAFI) thrombin also inhibits fibrinolysis [133,134]. Thrombin is pro-migratory [135] and mitogenic for tumor cells, but is also known to amplify the effects of other mitogens [136,137].

Thrombin is a potent inducer of various processes involved in angiogenesis (Figure 2) [96,138–144]. For example, its action can induce vascular permeability [133] via a VE-cadherin and protein-tyrosine phosphatase SHP2-dependent mechanism [145]. Using the chick chorion allantoic membrane (CAM) assay, Maragoudakis and Tsopanoglou [141] and Tsopanoglou et al. [143] demonstrated that thrombin directly promotes angiogenesis in vivo. This was executed by a proteolytic mechanism independent of fibrin formation because both α-thrombin (containing both the catalytic site and the anionic exosite) and γ-thrombin (catalytically active but lacking the anion-binding exosite required for clotting activity) were active in this model [143]. It is not clear whether the same activity is involved in spontaneous tubule formation by thrombin-treated endothelial cells cultured in Matrigel, an effect dependent on PKC activity [146]. Furthermore, thrombin was shown to be directly mitogenic for endothelial cells [147–149] owing to both proteolytic (i.e., thrombin receptor-dependent) and non-proteolytic (via B loop of the β-thrombin chain) signalling pathways [149].

Figure 2.

Modulation of angiogenesis by thrombin. Several facets of the angiogenic reaction can be influenced by thrombin in a direct or indirect manner, e.g., by formation of provisional fibrin matrix (see text). Proteolysis of prothrombin and antithrombin III gives rise to angiogenesis inhibitors (fragments 1 and 1.2, aaATIII, respectively). Thrombin-activated platelets release a number actual or potential angiogenesis effectors, including VEGF, PF4, TGFα and -β, or FGF. Various direct effects of thrombin on endothelial cells and other cell types that could be involved in the angiogenic process are likely mediated by signals generated from PARs (PAR-1, PAR-3, PAR-4) [247].

Thrombin exerts an indirect pro-angiogenic effect by upregulating VEGF expression in fibroblasts [118], as well as by increasing levels of VEGF receptors on vascular endothelium [142]. Both VEGFR-1/flt-1 and VEGFR-2/KDR/flk-1 were shown to be upregulated in thrombintreated HUVEC cells, through a PKC- and MAPK-dependent mechanism [142]. Thrombin has also been reported to increase the release of VEGF from platelets [150], expression of pro-angiogenic bFGF by endothelial cells [148], and release of bFGF from the ECM stores [151].

In endothelial cell cultures, thrombin causes release of tissue-type plasminogen activator (t-PA), plasminogen activator inhibitor-1 (PAI-1) [152], and activation of progelatinase A (MMP-2) [153]. It has also been proposed that thrombin may affect angiogenesis by promoting degradation of the angiogenesis inhibitor TSP-1 [154].

While active thrombin is, for the most part, believed to be pro-angiogenic, cryptic angiogenesis inhibitory domains do exist within various domains of prothrombin [40]. Thus, prothrombin kringle-2 domain (fragment 2) was shown to inhibit angiogenesis in the CAM assay [155]. Likewise, human prothrombin fragment 1+2 was documented to inhibit bFGF-induced bovine capillary endothelial cell growth and angiogenesis in the CAM [156].

In the presence of heparin, antithrombin III (AT-III) inhibits proteolytic action of 60th, thrombin and activated factor X (fXa) in plasma. Thrombin and neutrophil elastase can cleave the thrombin-binding site of AT-III, thereby generating the anti-angiogenic form of AT-III (aaAT-III), which inhibits bFGF- and VEGF-induced endothelial cell proliferation in vitro and in vivo [49].

Fibrinogen/Fibrin

One of the most important consequences of thrombin activation is the cleavage of plasma fibrinogen to fibrin [7]. Shortened half-life of the plasma fibrinogen and its increased turnover is often observed in cancer patients [7,157]. Tumors are frequently surrounded by fibrin(ogen) matrix also present throughout the tumor stroma (both fibrin I and II) [7,10–14,90,96,127–129,158]. This observation, in conjunction with the extravascular presence of prothrombin fragment 1+2, constitutes the evidence for activation of the extravascular coagulation in the tumor microenvironment [7,13].

Crosslinked fibrin is believed to provide a provisional matrix that promotes angiogenesis [7,96,158,159] by supporting endothelial cell adhesion, migration, and survival [158]. Fibrinogen interacts directly with integrin αvβ3 [160], an effect which likely provides endothelial cells with survival signals [161] and is mediated through β572–574 region of the human fibrinogen chain [162]. Also, factor XIII, which participates in fibrin crosslinking, can serve as a ligand for this anti-apoptotic integrin [78]. Stabilization and pro-angiogenic action of bFGF, and possibly other soluble growth factors, may be promoted by binding to fibrinogen and fibrin [163–166]. In vitro, fibrin induces expression of TF by HUVECs [167], expression of pro-angiogenic chemokine IL-8 in calf pulmonary artery endothelial cells [159], and expression of PAs and TSP-1 in corneal endothelial cells [167]. Likewise, fibrin degradation products — mainly fragment E — may promote angiogenesis [168]. It has been shown that fibrinopeptide B cleavage and exposure of the β15–42 region of the fibrin molecule by thrombin are necessary for various biological effects of fibrin such as stimulation of cellular mitogenesis (in both HUVECs and human skin fibroblasts) [169] and endothelial cell spreading [170]. However, it is noteworthy that in mice deficient for fibrinogen (Fib-/-), growth of B16 and LLC tumors after subcutaneous injection was found to be uninhibited, which may suggest that all the various functions ascribed to fibrin during tumor angiogenesis could be redundant [16].

Components of the Fibrinolysis System as Regulators of Tumor Angiogenesis

Numerous studies have demonstrated enhanced activity of u-PA in breast, gastric, colon, lung, prostate, and ovarian cancer, as well as in malignant melanoma and brain tumors, mostly in association with high grade, clinical malignancy, and poor prognosis [171–176]. High expression of t-PA in highly malignant gliomas and breast cancers often correlates with unfavorable prognosis as well [177,178]. Paradoxically, elevated PAI-1 also indicates poor prognosis in several types of cancers [171,176,177], a notion suggesting that change in balance, rather than utter deregulation and exuberance of the pericellular proteolysis, is compatible with aggressive tumor growth and angiogenesis [126].

Proangiogenic growth factors, such as VEGF and bFGF, are known to increase the expression of PAs, their receptors (e.g., u-PAR), and PAI-1 by endothelial cells [124,125, 179–181]. VEGF and bFGF synergize in stimulating both angiogenesis [182,183] and expression of u-PA and its receptor, u-PAR, in bovine adrenal cortex-derived microvascular endothelial cells [184]. These results are consistent with co-expression of VEGF and tPA transcripts [185] and co-localization of t-PA, u-PA, PAI-1, and VEGF in rat gliomas [186]. Moreover, co-expression of VEGF and u-PA was found in human colorectal cancer [187].

The ability of endothelial cells to express fibrinolytic activity is largely attributed to plasminogen and PAs: t-PA and u-PA. PAs convert a zymogen-plasminogen to plasmin, an enzyme capable of degrading ECM both directly and indirectly (e.g., through the activation of latent matrix metalloproteinases) [152,188]. Regulation of u-PA activity involves binding to the cell surface receptor (u-PAR) [124]. The single-chain proenzyme u-PA (sc-u-PA) can be activated more efficiently in the context of the u-PAR. Active two-chain u-PA is quickly inhibited by PAI-1. The u-PA/PAI-1 complex is internalized together with u-PAR and degraded in lysosomes, after which u-PAR is recycled to the cell surface (see Refs. [189,190] for review). The role of u-PA/u-PAR system in tumor progression is supported by the observation that expression of u-PAR is localized to the leading front of migrating monocytes and invading tumor cells [124] and that tumor growth and metastasis can be blocked by anti-u-PA antibodies [191].

Plasmin and PAs have been implicated in angiogenesis-promoting phenomenon of “pericellular fibrinolysis” [192]. This process, which facilitates endothelial cells invasion into the fibrin matrix, can be antagonized by plasmin inhibitors, α₂-antiplasmin and α₂-macroglobulin [192]. Recent study suggests, however, that PA/plasminogen system is not essential for endothelial cell penetration into fibrin gels because such function could be performed by the membrane type-1 matrix metalloproteinase [193]. Again, it should be born in mind that both inhibition of the PA/Plg/MMP-dependent proteolysis and its supraoptimal stimulation (in PAI-1 deficiency) are incompatible with effective angiogenesis [106,126,194].

PA system acts in a complex and pleiotropic fashion so that it can simultaneously influence, in many ways, endothelial cell adhesion, proteolysis, and signaling (Figure 3). Cell surface u-PAR binds to vitronectin in a PAI-1 inhibitable manner — an interaction which could clearly promote cell adhesion and angiogenesis [195]. Moreover, domain 2/3 of u-PAR expressed on endothelial cells serves as a receptor for high-molecular-weight kininogen (HK) and this binding is, in turn, inhibited by interaction with vitronectin [196]. Formation of the HK/kallikrein/u-PAR complex can further promote fibrinolytic activity and possibly angiogenic competence of endothelial cells [196]. This action can be antagonized by domain 5 of HK (kininostatin), which is known to downregulate endothelial cell proliferation and migration and thereby inhibit angiogenesis [77]. In a recent study, two-chain form of HK (HKa), HK domain 6, or corresponding peptides, but not the intact HK, were shown to directly bind to and inhibit endothelial cell proliferation, survival, and angiogenesis in a manner independent of u-PAR or vitronectin binding [197]. The latter effects should not be viewed in isolation from the proangiogenic effects exerted by the entire kinin system. Thus, release of bradykinin from HK and enforced overexpression of kallikrein were both found to be pro-angiogenic under certain conditions [198–200]. Moreover, interaction between HK and sc-u-PA with or without factor XII contribution may lead to plasmin generation on endothelial cell surfaces [201] and further barrage of complex proteolytic influences on the progress of angiogenesis. Finally, binding of u-PA to its receptor triggers intracellular signals [202], which could modulate angiogenesis in a proteolysis-dependent or -independent manner [189].

Figure 3.

Pro-and anti-angiogenic activities encoded within the fibrinolytic system. A multitude of effects on angiogenesis has been attributed to various effectors of fibrinolysis. Additional detailed studies are warranted on some of the more recently uncovered modulators of this system, e.g., TAFI [248] or endostatin [221]. The view emerging from “gene knockout” studies is that both excessive activation (PAI-1 deficiency) as well as excessive inhibition of fibrinolysis (PA deficiency) are incompatible with effective angiogenic response [20].

Another activity associated with the PA system, which is relevant to angiogenesis regulation, is related to the proteolytic processing of soluble growth factors and cytokines. For example, u-PA can activate pro-HGF, while plasmin cleaves bFGF, activates latent TGFβ (transforming growth factor β) [203,204], and can liberate the membrane-bound isoforms of VEGF (VEGF₁₈₉) [68]. HGF is known to possess intrinsic pro-angiogenic properties, but also the ability to stimulate VEGF production [205,206], and to promote motility and cellular invasiveness. TGFβ, while growth-inhibitory for endothelial cells in vitro, is thought to promote angiogenesis in vivo by a variety of mechanisms, such as regulation of ECM formation and proteolysis [171], participation in blood vessel maturation [20,207,208], stimulation of VEGF release [209], and inhibition of angiostatin generation [210].

Discovery of angiostatin is one of the most spectacular events that illustrate the link between the PA system and regulation of angiogenesis [211]. Angiostatin, a 38-kDa cleavage product of mature plasminogen, made up of the first four of the five highly homologous kringle domains, has been detected in tumor-bearing mice as a systemically acting, circulating angiogenesis inhibitor [211]. At least two sources of proteases, i.e., matrix metalloelastases (MME) released from tumor-infiltrating macrophages and serine proteases produced directly by tumor cells, catalyze plasminogen conversion to angiostatin or its isoforms [212]. Several such isoforms have been identified. For example, the 52-kDa isoform of angiostatin containing kringles 1 through 4 and a portion of kringle 5 — so-called “angiostatin 4.5” (apparently generated by plasmin) — was detected in brain tumors and in malignant ovarian ascites [213]. Another 55-kDa angiogenesis inhibitor was obtained by digestion of plasminogen with urokinase-activated plasmin, and contains intact kringles 1–4 and most of the kringle 5 (denoted as K1–5) [214]. In this case, the endothelial cell-specific inhibitory effect appears to be 50-fold more potent than that of angiostatin [214]. Also kringle 5 of plasminogen, in itself, exerts a negative effect on angiogenesis [212]. Recently, a recombinant protein composed of kringles 1–3 (rPK1–3) has been reported to inhibit growth of human glioma xenografts in nude mice [215]. Several studies confirmed the anti-angiogenic effect of angiostatin or its derivatives in various systems [216], albeit to a different degree and with various modifications of the drug delivery. Also, gene therapy approaches utilizing liposomes complexed to plasmids encoding angiostatin were found to be effective against breast cancer in nude mice [217]. Interestingly, radiation therapy combined with angiostatin treatment can, in some cases, act synergistically to inhibit expansion of the tumor vasculature [218].

The molecular mechanism, by which angiostatin inhibits angiogenesis, remains unclear. In this regard, it is known that angiostatin binds to the α/β-subunits of the ATP synthase on the surface of endothelial cells, potentially inducing H⁺ cytoplasmic influx and cytolysis [219]. Furthermore, angiostatin may inhibit endothelial cell invasion via complex formation with t-PA and resulting blockade of the PA/plasminogen system [220]. In this context, it is thought-provoking that another inhibitor of angiogenesis, endostatin, apparently acts through an opposite mechanism, i.e., by stimulation of plasmin generation and removal of proangiogenic fibrin matrix [221]. Clearly, further studies are warranted to elucidate and predict the net effect of the various fibrinolytic and procoagulant proteases implicated in angiogenesis.

The Role of Platelets in Tumor Angiogenesis

Analysis of platelets in cancer patients often reveals quantitative and qualitative abnormalities [39,222,223]. Among those, thrombocytosis; increased, reduced, or spontaneous aggregation; impaired adhesion; and hypersensitivity to different agonists are the most frequently cited examples [39,223]. Various types of tumor cells can activate platelets in vitro by virtue of direct contact, release of ADP, production of thromboxane A₂ or cancer procoagulant, generation of thrombin, or activation of the tumor-associated proteinases [223]. In the presence of VEGF, endothelial cells promote platelet activation [224]. Adhesion and aggregation of activated platelets are accompanied by the release (mainly from α-granules) of many potential angiogenesis regulators, such as: VEGF-A [145], VEGF-C [225], bFGF [226], HGF [227], insulin-like growth factor-1 and -2 [228,229], epidermal growth factor [230], and platelet-derived endothelial cell growth factor [231]. These sorts of observations led Pinedo et al. [39] to hypothesize that platelets may play an active and causative role in tumor angiogenesis.

Platelets are the source of angiogenesis stimulators, but also of angiogenesis inhibitors (Table 1) [39,40,232]. With regard to the latter, platelet factor-4 (PF4) was the first hemostatic protein demonstrated to be an inhibitor of angiogenesis in vivo [40]. PF4 interacts with surface heparin-like glycosaminoglycans on endothelial cells, thereby blocking the binding sites for heparin-binding endothelial growth factors [233–235]. PF4 also directly inhibits bFGF dimerization and activity [236]. In addition, PF4 inhibits the endothelial stimulatory activity of bFGF, EGF, and VEGF₁₂₁ by a mechanism independent of their interactions with heparin sulfate proteoglycans [235]. This is consistent with the observation that the analogue of PF4, which lacks the heparin-binding capacity (rPF4-241), is also able to inhibit tumor angiogenesis [237]. This notion was explored further by using various peptides derived from PF4 [238]. Thus, the peptide derived from the C-terminus of PF4 (residues 47–70) interferes with biological functions of both bFGF and VEGF. It is puzzling that a peptide derived from the central PF4 region (17–58), containing a potential heparin-binding domain, apparently does not interfere with endothelial cell responses to these respective angiogenic ligands (i.e., bFGF or VEGF). However, this peptide inhibits heparin-dependent interactions of these growth factors with their high-affinity receptors, when these receptors are expressed “ectopically”, i.e., in non-endothelial cells (e.g., CHO cells) [239]. More recently, gene therapy approaches have been developed to use modified PF-4 as an anti-tumor agent [240].

Table 1.

Examples of Platelet-Derived Angiogenesis Stimulators and Inhibitors.

Platelet-Derived Angiogenesis	Stimulators Platelet-Derived Angiogenesis Inhibitors
VEGF-A [150,225], VEGF-C [225], bFGF [226], HGF/SF [227], IGF-1 [228], IGF-2 [229], EGF [230], PD-ECGF [249], Ang-1 [250]	PF-4 and PF-4 fragments [235,236,238], HGF domains NK1 and NK2 [245,251–253], TSP-1 [241,254,255]

TSP-1, a potent angiogenesis inhibitor, is also a constituent of platelet α granules [241]. The central stalk of TSP-1 interacts with CD36 receptor on the surface of endothelial cells, resulting in angiogenesis inhibition [241,242]. However, very high concentrations of TSP-1 can also stimulate angiogenesis via interaction with the integrin-activating protein IAP/CD37 [241]. TSP-1 is also involved in activation of the latent TGF-β, thereby triggering angiogenesis-regulating effects of this cytokine [243]. Platelet-derived HGF can serve as another example of this “one-mediator-two-functions” paradigm. While essentially pro-angiogenic, HGF contains cryptic anti-angiogenic subdomains in its α-chain [40,244]. Thus, alternative splicing of the HGF mRNA can result in expression of the first kringle domain inhibitor (NK1) or first two kringle domain (NK2) inhibitor, both of which suppress HGF-induced angiogenesis [233], as does the recombinant HGF/NK4 variant [245].

Activated platelets stimulate endothelial cells to express TF [246] and to form tubular networks in Matrigel, an effect apparently independent of platelet aggregation or release processes [232]. It was argued that adhesion of platelets to endothelium via surface glycoproteins may be responsible for this morphogenetic response [232]. Collectively, as with other components of the hemostatic system, the role of platelets in various aspects, stages, and forms of tumor angiogenesis is complex and not fully amenable to simple generalizations and predictions.

Summary

Processes regulating vascular expansion, homeostasis and blood clotting intersect at many critical points. In the context of cancer, this anatomical and functional mutual interdependence is manifested by consistent co-incidence of “cancer coagulopathy” and activation of tumor angiogenesis. It is an open question to what extent one is the cause or the consequence of the other. Nevertheless, the explosion of findings implicating various hemostatic mechanisms in tumor growth and neovascularization is a foundation of at least two novel therapeutic approaches to treat cancer, namely: 1) derivation of hemostatic proteins as angiogenesis inhibitors, and 2) using antithrombotic pharmacotherapy (e.g., heparin or low-molecular-weight heparin) to control blood vessel formation. However, the great complexity of the hemostatic system and the multitude of pro- and anti-angiogenic activities it encodes (even within individual constituent proteins) suggest that for each therapeutic action, a proper clinical context should be precisely defined and validated. In this regard, we have previously proposed that both the molecular nature of “cancer coagulopathy” [63] and the operating angiogenic mechanism [21] likely vary as a function of tumor progression (angiogenesis progression) as well as genetic and epigenetic variables characteristic of each tumor type. We believe that it is important to consider such tumor specific, heterogenous and evolving nature of these respective host responses (i.e., coagulopathy and angiogenesis), while designing anti-angiogenic cancer therapies based on interference with the hemostatic system.