Vasculogenesis represents de novo vessel formation through differentiation of angioblasts, while angiogenesis requires pre-existing vessels, from which new vessels are formed through proliferation of endothelial cells (Poole and Coffin, 1989; Risau, 1997). Vasculogenesis occurs in developing embryos, but can also occur during vascular repair in adults; the latter is accomplished through differentiation of endothelial progenitor cells (Asahara et al., 2011). Stepwise development of vessels is undertaken during embryogenesis. The yolk sac and the embryo proper undergo distinct vasculogenesis. These steps of vasculogenesis are precisely regulated by transcriptional programmes and cell–cell and cell–tissue interactions (Marcelo et al., 2013; Park et al., 2013). The role of midkine in vasculogenesis has not yet been studied extensively; however, there are no reports in the literature showing that midkine-deficient mice would exhibit any gross abnormality of vascular formation. In contrast, midkine is involved in angiogenesis.
While vasculogenesis is the principal mechanism of vessel formation, angiogenesis is the predominant means of vascularization for all organs. In addition, angiogenesis occurs in a variety of pathological settings, such as cancer and inflammation, and often plays critical roles in their pathogenesis. The relationship between angiogenesis and midkine was first demonstrated by the promotion of proliferation of HUVECs by midkine purified through a heparin-affinity column from midkine-transfected MCF-7 breast cancer cells (Choudhuri et al., 1997). Midkine-expressing MCF-7 cells promoted angiogenesis in the rabbit corneal assay, compared with non-expressing MCF-7 cells. Midkine overexpression in MCF-7 cells also enhanced not only tumour growth, but also angiogenesis in a subcutaneous xenograft model (Choudhuri et al., 1997). Furthermore, Weckbach et al. have recently reported that hypoxia induced midkine production by HUVECs (Weckbach et al., 2012)and that exogenous midkine induced neovascularization in a chorioallantoic membrane assay (Weckbach et al., 2012). Although it is well known that endothelial cells and resident tissue cells are essential for angiogenesis, inflammatory cells, such as PMNs and monocytes/macrophages, also play important roles (Murdoch et al., 2008). For example, M2-polarized macrophages are essential for angiogenesis (Mantovani et al., 2013) and PMNs secrete molecules important for angiogenesis, such as VEGF, oncostatin M and MMP9 (Tazzyman et al., 2009). In this context, it should be noted that midkine acts as an chemoattractant for inflammatory cells in a variety of conditions involving inflammation (Horiba et al., 2000; Sato et al., 2001; Banno et al., 2006).
Midkine inhibits cardiac remodelling
Midkine expression was progressively increased after myocardial infarction in a mouse model of ligation of the left coronary artery (Takenaka et al., 2009). Midkine-deficient mice showed a higher mortality compared with wild-type mice. Exogenous administration of midkine improved survival and left ventricular function in both wild-type and midkine-deficient mice. Notably, this treatment enhanced angiogenesis in the peri-infarct zone. Similar results were obtained in a rat model of chronic cardiac infarction (Fukui et al., 2008). In this study, recombinant midkine was injected into hearts, 2 weeks after induction of myocardial infarction. Six weeks later, cardiac remodelling was significantly and dose-dependently attenuated by midkine treatment. Midkine treatment facilitated angiogenesis in the infarcted area, and the viable muscle area after myocardial infarction dose-dependently increased. Despite this increase of viable muscle area, midkine-treated hearts showed significantly less cardiomyocyte hypertrophy than vehicle-treated hearts. These results show that midkine prevents cardiac remodelling, at least in part, through its angiogenic activity.
Tumourigenic effect of midkine
Anti-angiogenic therapy for cancer was proposed by Judah Folkman in 1971 (Folkman, 1971), more than 10 years before the vascular permeability factor, now known as VEGF, was isolated in 1983 (Senger et al., 1983). The N-terminal amino acid sequence of VEGF was determined in 1989 (Ferrara and Henzel, 1989). VEGF and its receptor VEGFR2 are major regulators of angiogenesis. Besides VEGF, many other growth factors, such as fibroblast growth factor (FGF) and PDGF, are involved in tumour angiogenesis (Claesson-Welsh, 2012). Most types of cells in the tumour microenvironment produce VEGF which promotes vascular growth and sprouting by accelerating endothelial cell proliferation. PDGF is produced by endothelial cells, and attracts pericytes to support the vasculature, while FGF is expressed by tumour cells and enhances endothelial cell growth in tumours. Tumour blood vessels, however, are incomplete and leaky compared with normal blood vessels. This is in part due to incomplete perivascular support and high expression of VEGF, which stimulates endothelial cell proliferation, rounds up cells, breaks cell–cell junctions and consequently increases vascular permeability (Baluk et al., 2005; Fukumura and Jain, 2007). The leaky tumour vasculature increases interstitial pressure, which then becomes a barrier against anti-tumour drug delivery to tumour tissues. Nevertheless, tumour blood vessels do transport nutrients and oxygen to tumour tissue, and thus, support growth of tumour cells.
The anti-VEGF antibody bevacizumab was the first anti-angiogenic agent drug to be approved for cancer therapy by the Food and Drug Administration (FDA) in 2004 (Grothey and Galanis, 2009; Van Meter and Kim, 2010; Kieran et al., 2012). Although most VEGF blocking therapies require adjuvant chemotherapy, as demonstrated in the first trial of bevacizumab for metastatic colon cancer (Hurwitz et al., 2004), anti-angiogenic therapy is effective. However, the majority of patients eventually succumb to their disease. Resistance to anti-angiogenic therapy relies on tumour plasticity. Thus, sensitivity to anti-angiogenic therapy decreases with progression of disease, probably because of changes in the characteristics of tumour cells with accumulated mutations or intra-tumour heterogeneity (Bergers et al., 1999; Gerlinger et al., 2012). Tumour cells may also acquire vasculogenic mimicry, with tumour cells differentiating to endothelial cell-like cells and/or pericyte-like cells (Ricci-Vitiani et al., 2010; Cheng et al., 2013). Furthermore, tumour cells are switched to respond more to compensatory growth factors such as FGF and PDGF rather than to VEGF if the VEGF axis is blocked (Casanovas et al., 2005); (Orimo et al., 2005; Crawford et al., 2009). Therefore, identification of these angiogenic factors and verification of the mechanisms of their actions are important for further development of cancer therapy.
The degree of midkine expression correlates with microvessel density in salivary gland tumours (Ota et al., 2010). Midkine expression is high in human neurofibromas, schwannomas and various nervous system tumours associated with neurofibromatosis type 1 or 2 suppressor gene, where midkine expression can be detected in endothelial cells of tumour blood vessels, but not in normal blood vessels (Mashour et al., 2001). Finally, midkine stimulates proliferation of human systemic and brain endothelial cells in vitro (Mashour et al., 2001). These observations suggest that midkine is involved in cancer progression through its angiogenic activity. Consistent with this idea, a high expression of midkine correlates with a poor prognosis in patients with invasive bladder cancers (O'Brien et al., 1996). Levels of midkine expression are significantly correlated with microvessel density, tumour size, clinical stage and prognosis in oral squamous cell carcinoma (Ruan et al., 2007). Midkine overexpression also enhances tumour growth and microvessel density of human UM-UC-3 bladder cancer cells in both subcutaneous and orthotropic xenograft models. It also increases their sensitivity to anti-angiogenic therapy (Muramaki et al., 2003). Moreover, midkine antisense oligonucleotides inhibit not only proliferation of HUVECs, but also angiogenesis induced by HEPG2 human hepatocellular cancer cells in chorioallantoic membranes (Dai et al., 2007).
Mice deficient in midkine are useful tools in the investigation of the role of host midkine in tumour progression. Salama et al. reported that lung metastasis of Lewis lung carcinoma cells was less in midkine-deficient mice (Salama et al., 2006). Because the Lewis lung carcinoma cells do not significantly express midkine, this result suggests that midkine is also a host factor regulating metastasis. Furthermore, Kishida et al., have recently reported that midkine-deficient mice show significantly less tumourigenesis of neuroblastoma in MYCN transgenic mice, which spontaneously develop neuroblastoma (Kishida et al., 2013). Although angiogenesis has not been examined in these models, it would be an intriguing subject of future studies.
Midkine induces vascular stenosis
Atherosclerosis is the primary cause of life-threatening events such as stroke and heart attack. Low-density lipoprotein (LDL) diffuses passively through the endothelial cell junction and enters the space between the endothelium and internal elastic lamina, the so-called intima. High serum concentrations of LDL in blood therefore increase the chance of LDL entering the intima, where it is trapped by proteoglycans in the extracellular matrix and undergoes oxidation through interaction with reactive oxygen species, including hydroperoxyeicosatetraenoic acids, the products of 12/15 lipoxygenase (Lusis, 2000).
Minimally oxidized LDL stimulates endothelium to produce chemoattractants such as CCL2 (MCP-1) and M-CSF, and as a result, recruits monocytes to the intima. Monocytes recruited to the intima become activated macrophages that produce M-CSF, other cytokines and proteoglycans, laying the groundwork for further inflammation. Highly oxidized LDL forms aggregates that are engulfed by macrophages via scavenger receptors such as scavenger receptor A and CD36 (Lusis, 2000; Moore and Tabas, 2011). Engulfment of oxidized LDL turns macrophages into foam cells. Foam cells undergo apoptosis and secondary necrosis, leading to formation of a necrotic core rich in extracellular lipids. On the other hand, activated macrophages and infiltrated T-cells produce cytokines and growth factors, which stimulate proliferation of smooth muscle cells (SMCs). The SMCs migrate from their original space underneath the internal elastic lamina to the intima. As a result, the intimal region significantly expands and becomes so-called neointima.
Finally, fibrous plaques consisting of extracellular lipids, SMCs and SMC-derived extracellular matrix are formed. Thrombosis associated with fibrous plaques is the major cause of acute coronary events, as well as of stroke. It is triggered by the rupture of a plaque. In addition, new vessel formation occurs in the plaque. This angiogenesis is induced by local ischaemia and inflammation, and is promoted by monocytes recruited to the arterial wall (Jaipersad et al., 2013).
Midkine expression is increased in the rat common carotid artery after intraluminal balloon injury (Horiba et al., 2000). In addition, ligation of the common carotid artery induces neointima at the site of ligation in wild-type mice, but this neointimal formation is diminished in midkine-deficient mice. Exogenous midkine restores neointima formation in midkine-deficient mice. Midkine promotes macrophage migration in vitro and, consistent with this, leukocytes are less recruited to the vascular wall in midkine-deficient mice. Midkine also promotes migration of SMCs in vitro. These data suggest that midkine plays a pivotal role in neointima formation (Horiba et al., 2000).
Midkine antisense oligodeoxynucleotides transfected by means of lipofection to the vascular wall suppressed neointima formation after the rabbit carotid artery balloon injury (Hayashi et al., 2005). Increased midkine expression is also found in jugular vein-to-carotid artery interposition vein grafts in rabbits (Figure 3) (Banno et al., 2006). Controlled release of siRNA to rabbit midkine, which is accomplished by wrapping the grafted vein with atelocollagen containing the siRNA, markedly suppressed inflammatory cell infiltration and SMC proliferation, and consequently suppressed neointima formation. Indeed, this method of perivascular application of siRNA using atelocollagen efficiently delivers siRNA to the vascular wall (Figure 3) (Banno et al., 2006). The same animal model was used to evaluate the effect of statin in vascular stenosis, with pitavastatin suppressing midkine expression, and consequently, neointima formation (Fujita et al., 2008).
Figure 3. Suppression of neointima formation by knockdown of midkine. Increased midkine expression was found in jugular vein-to-carotid artery interposition vein grafts in rabbits (Banno et al., 2006). To accomplish a controlled release of siRNA to the grafted vein, the drug delivery system of atelocollagen mixed with siRNA was put around the vein. Because atelocollagen is solidified at around 37°C, the vein was wrapped with this mixture. Midkine siRNA (MKsiRNA), but not control siRNA (SCRsiRNA), markedly suppressed neointima formation. This figure was modified from Banno et al., 2006. with permission. CCA, common carotid artery; ECA, external carotid artery; ICA, internal carotid artery; SCR, scramble; VG, vein graft. Arrows indicate internal elastic lamina.
Download figure to PowerPoint
Compared with balloon injury, stenting induces more prolonged inflammation and more macrophage infiltration in the vascular wall. Midkine expression is also increased in the neointima when induced by a bare metal stent, which is implanted in the atheromatous lesion of hypercholesterolemic rabbits. The main source of midkine expression is macrophages in this model (Narita et al., 2008). These data suggest that midkine is important for pathogenesis of vascular restenosis not only after ballooning and vein grafting, but also after stenting, and can be a target of therapy for these conditions.