Tumor cells release angiogenic factors to initiate the process of angiogenesis (Sun and Tang,2004). Endothelial cells express a variety of receptors that bind to angiogenic molecules secreted by the tumor cells. It has been postulated for many years that neoangiogenesis originates from migration of endothelial cells in the pre-existing vessels in tumor-surrounding tissues. Accumulated evidence suggests that some of the tumor neovessels may be derived from circulating bone marrow-derived endothelial progenitor cells (EPCs; Shi et al.,1998; Asahara et al.,1999), although some other studies propose a limited role of EPCs in tumor neovascularization (Yung et al.,2004; Peters et al.,2005).
In addition to tumor cells, other types of cells, such as endothelial cells and infiltrating lymphocytes, can also secrete angiogenic factors. On the other hand, endogenous presence of antiangiogenic factors has also been identified. Therefore, tumor angiogenesis is a dynamic process regulated by the balance between angiogenic and antiangiogenic pathways. In our previous review, we have listed some of the relatively well-characterized angiogenic and antiangiogenic factors, with a focus on their potential roles in HCC (Pang and Poon,2006). The roles of a few more important factors are briefly described as follows, including updated information from the recent literature.
Vascular Endothelial Growth Factor
Vascular endothelial growth factor (VEGF) is the most well-studied angiogenic factor in HCC. The angiogenic effects of VEGF are achieved by binding to its receptors, VEGFR1 (Flt-1) and VEGFR2 (Flk-1), expressed on endothelial cells. Interaction between VEGF and VEGFR initiate several signaling pathways, such as Akt/PI3K/MAPK pathway, resulting in proliferation, migration, and invasion of endothelial cells (Gerber et al.,1998; Ahmad et al.,2006; Vogel et al.,2007). In addition to endothelial cells, recent findings have demonstrated the expression of VEGFR in tumor cells, proposing an autocrine loop of VEGF/VEGFR in tumor cells (Masood et al.,2001; Weigand et al.,2005).
Mise et al. (1996) first reported the expression of VEGF in HCC and detected significantly higher levels of VEGF mRNA and protein in the tumor tissues than the nontumorous counterpart. Subsequent studies reported similar findings of high VEGF expression at both mRNA and protein levels in HCC (Torimura et al.,2004; Yasuda et al.,2004). The levels of VEGF expression during the development of HCC have been shown to be associated with the increase of unpaired arteries and sinusoidal capillarization (Park et al.,2000). Increasing evidence of up-regulated expression of VEGF in small HCC suggests the importance of VEGF not only in tumor progression but also in hepatocarcinogenesis (Iavarone et al.,2007). This hypothesis has been proven in a diethylnitrosamine-induced murine hepatocarcinogenesis model (Yoshiji et al.,2004). The expression of VEGF in HCC has been found to be associated with the clinicpathological features of HCC patients in several studies. Torimura et al. (1998) showed a significant association between higher levels of VEGF protein by immunostaining and poorly differentiated HCC. A higher level of VEGF expression in HCC is also associated with a higher proliferation index, poor encapsulation of tumors, and venous tumor emboli or portal vein thrombosis (Chow et al.,1997; Li et al.,1998). In addition, circulating VEGF levels before treatment could predict disease outcome after a variety of treatments, including hepatic resection, radiofrequency ablation, transcatheter arterial chemoembolization (TACE; Poon et al.,2003a,2004a,2007). Parallel to VEGF, VEGFR expression is also associated with the clinical outcome of HCC patients, although only the level of VEGFR1 mRNA significantly correlated with the VEGF mRNA level in tumor tissues and intrahepatic metastasis in HCC (Ng et al.,2001). Another study also reported that VEGFR1 expression in tumor tissues correlated with CD31 expression, a marker for tumor microvessels (Dhar et al.,2002). The above findings suggest that VEGFR1 may play a more important role in VEGF/VEGFR pathway in mediating angiogenesis of HCC.
In addition to VEGFR1 and VEGFR2, VEGF also interacts with the members of the neuropilin family, neuropilin-1 and neuropilin-2, to regulate angiogenesis. Moreover, these neuropilin members can act together with VEGFR1 and VEGFR2 to mediate signal transduction (Staton et al.,2007). Little evidence has shown how neuropilins participate in the angiogenic process of HCC. However, some studies on other types of cancers reported that neuropilins that are expressed on tumor cells could directly modulate cell survival, proliferation, and invasion (Hong et al.,2007; Fukasawa et al.,2007). In endothelial cells, neuropilin-1 has been identified to be a coreceptor of VEGFR2 and function on angiogenic behavior by means of an enhancement of the VEGFR-2 phosphorylation threshold (Favier et al.,2006). Hence, exploring the potential role of the neuropilin family on HCC, either alone or combined with other VEGFRs, is warranted, and may shed light on a novel antiangiogenic therapy for this malignancy.
The VEGF promoter contains a hypoxia-responsive element, which is regulated by hypoxia inducible factor-1α (HIF-1α), a transcription factor. Activated HIF-1α under stress, such as hypoxia, can bind to hypoxia-responsive elements expressed in several pro-survival and pro-proliferation genes, leading to up-regulation of HIF-1α downstream proteins (Harris,2002; Brown and Wilson,2004). HIF-1α–mediated up-regulation of VEGF is pivotal for hypoxia-induced angiogenesis (von Marschall et al.,2001; Yang et al.,2004), and a higher level of HIF-1α is also associated with a poorer prognosis of HCC patients (Huang et al.,2005). Transcriptional activation of HIF-1α can also be induced by hepatitis B virus X protein (Yoo et al.,2003), indicating the potential interaction among viral infection, HIF-1α and VEGF up-regulation, and hepatocarcinogenesis.
VEGF can directly function on HCC tumor cells, as VEGFRs are also expressed by tumor cells. Our previous study demonstrated that stimulation of HCC cell lines by exogenous VEGF recombinant protein could stimulate tumor cell proliferation (Liu Y et al.,2005), suggesting the presence of an autocrine loop in HCC tumor cells. Moreover, VEGF may also have an angiogenesis-independent effect on HCC cell invasion. A study by Schmitt et al. (2004) showed that VEGF induced a marked loss of pseudocanaliculi and disruption of occludin-delineated tight junctions in tumor cells. Sections from surgically removed tumor specimens showed VEGF expression in the tumor and occludin disassembly in normal liver parenchyma next to the tumor, suggesting that VEGF may enhance invasion of HCC cells into normal liver parenchyma.
VEGF has been reported to be overexpressed in cirrhosis, a well-recognized precancerous lesion of HCC. Therefore, VEGF is also postulated to play an important role in the development of liver cirrhosis (Deli et al.,2005; Medina et al.,2005). However, there is no solid evidence showing the relationship among VEGF expression, cirrhosis and the development of HCC, although a recent study demonstrated that a higher level of VEGF165 mRNA in noncancerous liver tissue correlated significantly with a higher risk of HCC recurrence (Sheen et al.,2005). The significance of angiogenesis in cirrhosis and its relationship with hepatocarcinogenesis in cirrhotic liver are areas that need further investigation.
Angiopoietins constitute another group of angiogenic molecules. Different members in the angiopoietin family play different roles in the process of angiogenesis. For instance, angiopoietin-1 provides survival signals to endothelial cells and promotes recruitment of pericytes and smooth muscle cells to form mature blood vessels, whereas angiopoietin-2 functions as a natural antagonist of angiopoietin-1 (Tanaka et al.,1999). The presence of angiopoietin-2 enhances the sensitivity of endothelial cells to other angiogenic factors, such as VEGF. Higher levels of angiopoietin-2 expression have been detected in the tumor tissues than in the nontumorous tissues in HCC. The level of angiopoietin-2 was found to be associated with the clinicopathological parameters of HCC patients, and the ratio between angiopoietin-2 and angiopoietin-1 indicated the status of tumor angiogenesis (Zhang et al.,2006). A study by Torimura et al. (2004) also showed a positive correlation between angiopoietin-1 and angiopoietin-2 expression with tumor dedifferentiation. By using immunohistochemistry, angiopoietin-1 and angiopoietin-2 were detected in HCC cells, hepatic stellate cells, and smooth muscle cells, whereas their receptor Tie-2 was detected in endothelial cells, hepatic stellate cells and smooth muscle cells, suggesting that multiple cell types are involved in the angiopoietin/Tie-2 signaling pathways to mediate tumor angiogenesis. Direct interaction between angiopoietin-2 and VEGF has been identified using a murine HCC model (Yoshiji et al.,2005), suggesting a synergistic effect between angiopoietin-2 and VEGF. Based on the above findings, the expression of angiopoietins might be a potent parameter to predict the outcome of HCC patients. Indeed, Mitsuhashi et al. (2003) reported that HCC patients harboring tumors with high angiopoietin-2/1 mRNA ratio had a shorter survival than those with tumors having low angiopoietin-2/1 mRNA ratio. A study by Wada et al. (2006) also revealed that a higher level of angiopoietin-2 in the tumors was associated with a shorter disease-free survival in HCC patients with hepatic resection.
However, the mechanism responsible for the regulation of angiopoietin-2 expression appears to be different from that of VEGF in HCC. Unlike VEGF, hypoxia does not regulate the expression of angiopoietin-2 in HCC (Sugimachi et al.,2003). In addition, the role of angiopoietin-1 in HCC is controversial. Sugimachi et al. (2003) showed that angiopoietin-1 was more frequently expressed in normal liver, whereas angiopoietin-2 was more frequently expressed in HCC. Hence, a balance between angiopoietin-1 and angiopoietin-2 may play a more important role in mediating angiogenesis in HCC.
Epidermal Growth Factor
The epidermal growth factor (EGF) family contains six ligands, that is, EGF, tumor growth factor (TGF) -α, amphiregulin, betacellulin (BTC), heparin-binding EGF-like growth factor and epiregulin, and four receptors, that is, EGFR, ErbB2, ErbB3, and ErbB4 (Salomon et al.,1995; Riese and Stern,1998). All the EGF family members share common features including an extracellular ligand binding domain, a transmembrane domain, and an intracellular tyrosine kinase domain. The interaction between any of the six ligands to EGFR activates cell signaling pathways that lead to cell proliferation and differentiation. HCC tumor cells secrete EGF, which may act on multiple EGFRs expressed by tumor cells, and thus achieve an autocrine cascade (Zhang et al.,2004).
Recent evidence has shown the potential role of EGF family in angiogenesis of HCC. Moon et al. (2006) demonstrated that BTC mRNA was prominently expressed by tumor cells in human HCC specimens, whereas EGFR was mainly expressed by sinusoidal endothelial cells. In addition, a strong correlation between BTC expression in tumor cells and EGFR expression in tumor endothelial cells was identified. Furthermore, BTC was secreted by HCC cell lines, suggesting the potential interaction between BTC and EGFR in tumor angiogenesis. Targeting EGFR signaling has been recently demonstrated to be effective in blockade of angiogenesis in animal models (Ueda et al.,2006).
Platelet-derived Endothelial Cell Growth Factor
Platelet-derived endothelial cell growth factor (PD-ECGF) is another angiogenic molecule that can promote endothelial cell migration, but it has little effect on cell proliferation. Its angiogenic effect is mediated by the release of 2-deoxy-d-ribose as a result of breakdown of thymidine by its thymidine phosphorylase activity (Griffiths and Stratford,1997). Up-regulation of PD-ECGF has been identified in HCC tumor tissues compared with the adjacent nontumorous tissues, suggesting the potential role of PD-ECGF in tumor angiogenesis. However, in a study, the expression of PD-ECGF in tumor tissues did not correlate with the outcome of HCC patients (Morinaga et al.,2003). Interestingly, another study revealed that a higher level of thymidine phosphorylase activity in nontumorous tissues adjacent to tumor tissues was associated with an earlier tumor recurrence after hepatic resection, suggesting a prognostic value of thymidine phosphorylase activity in nontumorous livers (Ezaki et al.,2005). Another two studies have reported a positive correlation between PD-ECGF expression in the tumor tissue and portal vein tumor thrombosis in HCC (Zhou et al.,2000; Guo et al.,2001).
Other Angiogenic Factors
Several other factors secreted by tumor cells have also been reported to participate in the angiogenic process. Fibroblast growth factors, a family of heparin-binding growth factors, exhibit potent angiogenic properties. Basic fibroblast growth factor (bFGF) could be secreted by both tumor cells and endothelial cells, suggesting the presence of both autocrine and paracrine cascades in bFGF-mediated angiogenic activities (Schweigerer et al.,1987). El-Assal et al. (2001) demonstrated that the expression of bFGF protein in human HCC specimens correlated with tumor angiogenesis. HCC cells also secrete platelet-derived growth factor (PDGF), another angiogenic factor, to function on endothelial cells which express PDGFR, resulting in an increased metastatic potential (Zhang et al.,2005). A recent report demonstrated that PDGF links TGF-beta signaling to nuclear beta-catenin accumulation for HCC progression (Fischer et al.,2007). Inducible nitric oxide synthase (iNOS) is also regarded to be an angiogenic factor, supported by the evidence that its expression was positively correlated with MVD and tumor recurrence after hepatic resection, and its angiogenic effects seemed to be mediated by matrix metalloproteinase related pathways (Sun et al.,2005). However, detailed mechanism studies are still required to investigate the role of inducible nitric oxide synthase in HCC angiogenesis.
Our previous study showed that the expression of tissue factor, a hemostatic protein, correlated with tumor angiogenesis and tumor invasiveness (Poon et al.,2003b), suggesting that clotting-dependent induction of tumor angiogenesis is mediated by tissue factor-induced generation of thrombin and subsequent deposition of a cross-linked fibrin network, which provides a proangiogenic matrix that facilitates blood vessel infiltration.
Interleukin-8 (IL-8) is another potential angiogenic factor that is expressed by various types of tumor cells including HCC (Shin et al.,2002; Azenshtein et al.,2005). However, controversial findings have been reported about the angiogenic property of IL-8 in HCC. One study demonstrated that increased expression of IL-8 in human HCC specimens was associated with venous and bile duct invasion, and IL-8 produced by HCC cell lines stimulated proliferation of human umbilical vein cells (Akiba et al.,2001). In contrast, a more recent study reported that the expression of IL-8 in human HCC was related to vascular invasion but not tumor angiogenesis as assessed by MVD (Kubo et al.,2005).
Cyclo-oxygenase (COX-2) contributes to several physiological and pathological processes. COX-2 expression has been found to be associated with VEGF expression and MVD in HCC tumor tissue (Cheng et al.,2004). Our data also showed that tumor cytosolic COX-2 levels in HCC correlated with VEGF levels and features of invasiveness such as venous invasion and microsatellite lesions (Tang et al.,2005). These findings suggest an angiogenic effect of COX-2 in HCC.
Other angiogenic factors, such as macrophage migration inhibitory factor (Ren et al.,2003; Hisai et al.,2003) and thrombospondin-1 (Poon et al.,2004b), might also play important roles in mediating tumor angiogenesis in HCC, though further studies are needed to clarify the molecular pathways of these molecules in angiogenesis. With the development of gene and protein expression profiling technologies, novel angiogenic factors, such as pituitary tumor transforming gene 1 (PTTG1), brain-derived neurotrophic factor (BDNF) and stromal-derived factor-1 (SDF-1), have been identified. Fujii et al. (2006) reported that PTTG1 was remarkably overexpressed in HCC tumor tissues, as compared with nontumorous tissues. In addition, PTTG1 mRNA expression significantly correlated with MVD. Finally, PTTG1 mRNA was an independent prognostic factor for disease-free survival of HCC patients. These findings suggest that PTTG1 might be involved in the development of HCC through promotion of angiogenesis.
Our previous animal study has detected an increasing production of BDNF by tumor cells during HCC tumor development (Yang et al.,2005). As endothelial cells and bone marrow-derived progenitor cells express BDNF receptor, tyrosine kinase receptor B (TrkB), it is reasonable to hypothesize that BDNF might exhibit angiogenic effects through recruitment of endothelial cells and bone marrow-derived progenitors. In fact, some studies have shown the presence of autocrine and paracrine loops of BDNF in endothelial cells (Kim et al.,2004), and BDNF could stimulate mobilization of TrkB+ progenitor cells to the site of hypoxia to promote revascularization (Kermani et al.,2005). In addition, BDNF/TrkB interaction could up-regulate HIF-1α and VEGF expression in tumor cells (Nakamura et al.,2006), providing another clue that BDNF might be a tumor angiogenic factor. Therefore, further studies are warranted to explore the expression of BDNF in relation to angiogenic parameters in HCC.
SDF-1 is a recently identified angiogenic factor that could recruit endothelial progenitors to the site where neovasculature is formed (Aghi et al.,2006). Because bone marrow-derived progenitors (CXCR4+Flt-1+ hemagiocytes) express CXCR4, a receptor of SDF-1, the chemotaxic role of SDF-1 seems to be more prominent in these endothelial progenitors. However, available studies about SDF-1 and CXCR4 in HCC mainly focused on CXCR4+ tumor cell migration and metastasis (Schimanski et al.,2006; Sutton et al.,2007). The role of SDF-1/CXCR4 interaction in angiogenesis of HCC deserves further studies.