During embryogenesis, the vascular system is the first organ system that develops through the assembly of endothelial/mesodermal precursor cells (angioblasts) by a process referred to as vasculogenesis [25, 26]. In this context, VEGF signaling is crucial for the formation of a primary vascular plexus (E8.5) [27, 28] (see also Chap. 1). At later stages (E9.5–12.5), the embryonic vasculature expands by sprouting angiogenesis and maturates through the recruitment of pericytes and smooth muscle cells (SMCs). This process is essentially governed by angiopoietin/Tie2 signaling [5, 6, 27] . Angiopoietins are secreted growth factors that exert downstream signaling through the Tie2 receptor tyrosine kinase. Tie2 is predominantly expressed in endothelial cells. Consequently, deletion of the Tie2 gene leads to embryonic lethality around E10.5–12.5 due to cardiac defects and defective remodeling of the poorly organized primary vascular plexus [11, 12]. This primitive plexus contains fewer endothelial cells that display disrupted interactions with perivascular support cells. Further studies also identified irregular hematopoiesis in the Tie2-null animals [29]. The Tie2 homolog Tie1 receptor tyrosine kinase Tie1 is still considered orphan with no apparent ligand. Ang1 has been described to bind and cluster Tie1 at cell–cell contacts only in the presence of Tie2 [30, 31]. Tie1 deficiency in mice leads to embryonic lethality between E13.5 and birth due to defects in endothelial cell integrity and the formation of pulmonary edema [12, 32]. In addition, deletion of Tie1 affects lymphangiogenesis [33, 34]. However, defects in the lymphatic vasculature were manifested prior to vascular alterations around E12.5. In addition, and in contrast to Tie2, Tie1 deficiency does not effect definitive hematopoiesis [35].

Among the identified Tie receptor ligands, Ang1 was soon assigned as the activating binding partner due to the nearly identical embryonic lethal phenotypes of Ang1 and Tie2 mutant mice [7, 8, 11, 12]. Ang1-null embryos display defects in the endocard and die by E12.5 due to improper integrity of endothelial and perivascular support cells [8]. Ang2 has thereafter been identified as an Ang1 antagonist as specific expression of Ang2 in endothelial cells leads to an embryonic lethal phenotype that phenocopies Ang1 and Tie2-deficient mouse mutants [9]. The agonist-antagonist concept of Tie2 activation has been approved in vivo in a gain-of-function (GOF) mouse model [36]. However, context-dependent agonistic functions of Ang2 have also been reported [10, 37, 38]. Notably, Ang2-deficiency (loss-of-function [LOF]) is compatible with life, depending on the genetic background, and leads to defects in some vascular beds and in the lymphatic vasculature [10].

Among the four Angiopoietin ligands , Ang1 and Ang2 are characterized best, while Ang3 and Ang4 represent less well-studied orthologs in humans and mice [5, 6, 28]. Ang1 is expressed in numerous cell types, including perivascular cells (SMCs and pericytes), fibroblasts, osteoblasts, or tumor cells, and acts in a paracrine manner on Tie2 [5, 7, 16, 39]. In contrast, Ang-2 is predominantly expressed in endothelial cells and only upregulated at sites of angiogenesis, i.e. during development or pathological angiogenesis [9, 10, 15, 16]. Ang2 exerts its functions on the Tie2 receptor tyrosine kinase in an autocrine manner. Interestingly, Ang2 is not only regulated at the transcriptional level upon hypoxia or by cytokine activation but is also additionally stored in Weibel–Palade bodies (WPB) [5, 40]. Storage in this particular cellular compartment, which is special to endothelial cells, allows the rapid release of its contents (cytokines, growth factors, adhesion receptors). In addition, it elicits immediate responses to stress factors such as histamin, thrombin, etc., and provides a link to the inflammatory cascade [18–20, 40] (Fig. 13.2). Ang1 and Ang2 show high sequence and structure homologies [5, 9, 41]. Angiopoietins consist of an amino-terminal (N) domain, a coiled-coil (C) domain, and a carboxy-terminal fibrinogen-homology (F) domain [41]. While the F domain is responsible for the interaction with the Tie2 receptor, the C-terminal and N-terminal domains are essential for clustering of the ligand. Angiopoietins primarily exist as tetramers and higher order oligomers, and structural analysis revealed that at least one tetramer of Ang1 is necessary to activate endogenous Tie2 [41–43]. Interestingly, Ang2 occurs in the same higher-order multimers as Ang1 but fails to induce receptor activation [41, 42, 44]. Originally, it was identified that Ang1 and Ang2 bind to the same domains of Tie2 with similar kinetics (first Ig-like domain and the EGF-like repeats) [45, 46]. However, more recent evidence suggests the second Ig domain as the primary angiopoietin binding site [42]. Hitherto, it is not entirely resolved how Ang1 and Ang2 achieve the different agonistic versus antagonistic functions, although they form remarkably similar complexes with Tie2 [42, 46, 47]. Recent evidence indicated that specific molecular surface structures are important for the interaction with Tie2 [47]. In particular, three critical residues within the angiopoietin fibrinogen domain are necessary to confer to the differential ligand activities [47].

Tie2 is expressed on vascular and lymphatic endothelial cells and is also present on circulating hematopoietic cells, including megakaryocytes and neutrophils [5, 48]. Additionally, a subpopulation of tumor-infiltrating monocytes has been shown to express Tie2 [49–51]. Tie1 is specifically expressed in endothelial cells but does not directly bind to angiopoietins [5, 6]. Tie1 is abundantly expressed in the embryonic vasculature but its expression subsides with vessel maturation. In the adult, Tie1 is induced in malignant melanoma [6, 52] or in areas of disturbed flow in atherogenic vascular niches [6, 53]. Tie1 and Tie2 form complexes at the endothelial cell surface, which are mainly mediated by electrostatic forces [54]. The association of Tie1 with Tie2 has an inhibitory effect on the receptor, which is independent on the Tie1 kinase domain. Interestingly, stimulation with Ang1, but not Ang2, was able to dissociate Tie1 from Tie2, leading to receptor phosphorylation [54]. Upon stimulation with Ang1, Tie2 not only dissociates from Tie1 but also induces receptor translocation and assembly in distinct signaling complexes [31, 55]. In confluent resting endothelial cells, Ang1 induces the receptor translocation to cell–cell junctions and mediates clustering of Tie2 in trans with neighboring endothelial cells. This shifts intracellular signals to increased Akt signaling and results in endothelial cell quiescence. In contrast, in mobile endothelial cells the interaction of matrix bound Ang1 to Tie2 mediates adhesion and induces a promigratory phenotype mediated by Erk and Dok-R signaling [31, 55]. Interestingly, Ang2 is additionally able to elicit signaling via integrins in Tie2-negative/silenced endothelial cells and thus differentially regulates angiogenesis versus destabilization, depending on the presence of endothelial Tie2 [56, 57].

The deployment of genetic mouse models (GOF/LOF) helped to further understand angiopoietin signaling in the adult. Ang1 signaling is thus dispensable in quiescent vessels in the adult but necessary to modulate vascular response after injury [58]. This finding challenges the concept of Ang1 acting as a survival factor [43]. Nonetheless, Ang1-mediated constitutive Tie2 phosphorylation leads to vessel stabilization in the adult vasculature [13, 36]. This signal is antagonized by Ang2 in vivo and thus leads to vessel destabilization [36, 59]. Detailed mechanistic insights on how endothelial cell stabilization versus destabilization are maintained by angiopoietin signaling are rare. Evidence for vascular stabilization and leakage resistance mediated by Ang1 was deducted from studies in GOF mice [60, 61]. Mice overexpressing Ang1 in the skin display more vessels with larger volume, indicative for sprouting [62]. In addition, these mice are leakage-resistant against permeability-inducing agents such as mustard oil or are unreponsive against VEGF-induced permeability [60, 61]. Mechanistically, this effect has been pinpointed to the crosstalk between Ang1 and vascular endothelial (VE)-cadherin signaling via the small protein kinase Src [63, 64]. In detail, Ang1 stimulation of endothelial cells prevents VEGF-induced endothelial cell permeability by sequestering Src through mammalian diaphanous (mDia) [63, 64] (Fig. 13.2). Ang2 has been implicated to antagonize these functions in vivo (Fig. 13.2) [for details see also Chaps. 6 and 8]. Intradermal and tracheal permeability was largely attenuated in an Ang2 LOF model [65]. A further study demonstrated direct evidence for the involvement of Ang2 in vascular permeability in the skin of Ang2 GOF animals [66]. Hitherto, the permeability changes induced by Ang2 were demonstrated merely in peripheral vessels. The question arises whether specialized endothelial cells of the brain are similarly affected by Ang2 and whether the blood–brain barrier (BBB) easily opens upon challenge with Ang2. Until now, evidence in vivo is rare. However, one study evidenced that Ang2-mediated BBB impairment was antagonized by an Ang2 inhibitory antibody in a brain metastasis model [67].

Angiopoietin/Tie signaling has also been implicated to drive tumor progression [22, 68]. Ang2 has been identified to be expressed early during tumor growth similar to the onset of VEGF expression [14, 16, 17]. Ang2 is not expressed in normal vessels and thus appears to be critical for tumor initiation (i.e. by vessel cooption) and tissue remodeling and has been shown to promote new vessel growth in concert with other growth factors [14, 16, 22]. Consequently, high Ang2 levels have been identified in the serum of patients with different neoplasias and Ang2 has thus also been implicated as a biomarker for cancer and other pathologies that involve neovessel growth or vessel permeability [5, 24]. Numerous studies with genetically engineered tumor cells expressing Ang1 or Ang2 provided evidence that both molecules act as agonists/antagonists to induce or inhibit Tie2 downstream signaling [5, 68, 69]. Although outcome on tumor growth upon Ang1 or Ang2 expression was sometimes controversial (see Reiss [68] for review), the net outcome on tumor angiogenesis was similar to findings during developmental angiogenesis , i.e. vessel stabilization mediated by Ang1 versus vessel destabilization mediated by Ang2 (Fig. 13.2) [5, 6]. Ang1 expression leads to an improved vasculature with more pericytes [5, 6, 68]. As a consequence, perfusion of the normally chaotic tumor vasculature is improved upon Ang1 expression, a process denominated as ‘vascular normalization’ [70–72]. In contrast, Ang2 expression leads to more, but smaller, vessels, which are devoid of mural cells [5, 6, 68]. This finding is indicative of an immature vascular phenotype, which results in improper perfusion. Our own investigations at the ultrastructural level in both mouse glioma and breast cancer models showed similar vascular changes, indicative of vascular stabilization versus destabilization phenotypes upon Ang1 and Ang2 expression [59, 73]. Moreover, studies in transgenic animals supported the finding obtained in angiopoietin-expressing tumor cell lines. Ang2 LOF identifed the rate-limiting role for Ang2 during the early phases of tumor growth, possibly also regulating the angiogenic switch [74]. Ang2 GOF in subcutanous tumors (Lewis lung carcinoma) or in the Rip1Tag2 transgenic mouse model further supported the proangiogenic and destabilizing functions of Ang2 [75, 76].

Aside from the vascular functions, angiopoietins have been implicated as gatekeepers of immediate endothelial responses such as permeability, hemostasis (coagulation), and inflammation [5, 20]. Along with other hemostasis-regulating proteins (e.g. von Willebrand factor [vWF], P-selectin, interleukin-8), Ang2 is stored in WPB [40]. Mediators contained in WPB can be deployed rapidly in response to signaling molecules and mechanical stress [5, 40]. Opposite to this, Ang1 prevents endothelial cell activation and inflammation when overexpressed in the skin of transgenic mice [60]. Moreover, Ang1 was capable of counteracting VEGF-mediated skin permeability and inflammation when coexpressed with VEGF in transgenic mice [61]. Those initial genetic studies indicated that Ang1 is able to circumvent inflammatory reactions by stabilization of the vasculature. In fact, Ang1 has been shown to interfere with nuclear factor kappa B (NFkB) signaling through recruitment of the adapter protein A20-binding inhibitor of NF-kB 2 (ABIN-2) [77]. Our own studies demonstrated that the absence of Ang2 delayed the onset of short-term inflammatory reactions in LOF mice, which was restored by administration of recombinant Ang2 [18]. This implicated antagonistic roles of Ang1 and Ang2 in immediate vascular responses, including inflammation (Fig. 13.2) [19]. Mechanistically, the delayed onset of inflammatory responses was attributed to extended rolling and the defective leukocyte adhesion as evidenced by intravital fluorescence video microscopy [18]. Furthermore, by employing a GOF model, i.e. Ang2 expression in endothelial cells, we demonstrated that Ang2 on its own was effective in inducing the recruitment of innate immune cells and thus served as an instigator of inflammation (Fig. 13.2) [78, 79]. In these mice, short-term inflammation was augmented as a consquence of prolonged myeloid cell adhesion [78]. Similarly, Ang2 has been demonstrated to link vascular remodeling and inflammation in the model of Mycoplasma pulmonis-induced airway inflammation in a Tie2-dependent manner [80]. Selective targeting of Ang2 was able to restore Tie2 phosphorylation and reduce disease severity [80].

The recruitment of immune cells from the blood is a key cellular response to tissue damage and inflammation [81]. The immigration of immune cells from blood into tissues is a crucial process that not only applies during inflammatory conditions but also in neoplastic diseases as an inflammatory microenvironment is existent in all tumors [82]. The link between inflammation and cancer was recognized by Virchow in the 19th century [83]. Immune cells are recruited to sites of tumor progression where they are able to exert anti- or protumorigenic functions the latter also accounting for therapeutic resistance [82, 84–86]. The most frequent subsets of immune cells within the tumor microenvironment are tumor-associated macrophages (TAMs) and myeloid-derived suppressor cells [82, 84]. Importantly, growth factors or cytokines that are secreted by tumor-infiltrating immune cells activate key inflammatory transcription factors, e.g. NF-kB [82, 87]. Tumor-infiltrating myeloid cells, also termed accessory cells, have been recognized as major contributors to tumor angiogenesis through the secretion of numerous cytokines or their education by growth factors [86, 88, 89]. High numbers of TAMs have been associated with poor prognosis in numerous cancer entities [90]. They secrete a plethora of cytokines (among which are growth factors (e.g. VEGF), chemokines, and matrix metalloproteinases) that are able to drive angiogenesis in tumors [84, 86]. A subpopulation of these cells expressing the Tie2 receptor tyrosine kinase (Tie2-expressing monocytes [TEM]) has been identified to be crucial for tumor progression [49, 50, 91]. Selective depletion of these TEMs in tumor-bearing mice inhibited tumor angiogenesis and growth, suggesting that they might regulate angiogenic processes in tumors by providing paracrine signals to newly formed blood vessels [49]. Our own studies identified that Ang2 was directly responsible for the recruitment of proangiogenic macrophages in subcutaneous tumors [75] and in settings of inflammation [78]. Furthermore, Ang2 stimulation increased the inherent angiogenic properties of TEMs and led to the upregulation of genes associated with the M2-polarized phenotype of macrophages [75]. In addition, Ang2 transgene expression promoted the expansion of T regulatory cells and thus further contributed to immune suppression [92]. As such, the Ang2–TEM/TAM axis may represent a new dual target for antiangiogenic/anti-immunosuppressive cancer therapy.

Antiangiogenic therapy has become a valuable clinical target since the approval of bevacizumab (Avastin^®) in 2004 for the treatment of colorectal cancer [93] and thereafter for other cancer entities [72, 94]. However, efficacy of anti-VEGF therapy is rather dismal due to escape mechanisms that involve the presence of proangiogenic innate immune cells or increased invasiveness [84]. Several tumor types, including glioblastoma, are not as responsive or have been shown to develop escape mechansims that lead to further tumor progression despite antiangiogenic therapy [95]. Unresponsiveness to antiangiogenic therapy has, in part, been attributed to the infiltration of tumor-associated macrophages that promote tumor growth by the secretion of proangiogenic cytokines [96–98]. In this regard, novel therapeutic strategies are being explored by pharmaceutical companies. Ang2 emerged as a valuable clinical target as it is solely upregulated under angiogenic conditions, where it is also coexpressed with VEGF [14, 16, 17]. As such, targeting the angiopoietin/Tie signaling pathway opens new avenues for therapeutic inhibition of tumor growth [22–24]. A peptibody targeting Ang1 and Ang2 (Trebananib^®/AMG386) [99, 100] is currently being explored in a phase III clinical trial and has met the endpoint of improved progression-free survival [101, 102]. Numerous reagents targeting the angiopoietin/Tie pathway have been developed and were tested in preclinical or clinical models [23, 24, 99, 100, 103–108]. The new drug regimen appears very successful, particularly in combination treatment with drugs that target the VEGF/VEGFR signaling pathway [23, 24]. The combination of VEGF- and angiopoietin-targeting drugs appears to be superior to targeting either pathway alone [23, 24, 38, 109–113] .

In this regard, it is of interest to note that VEGF blockade has been associated with increased invasiveness and even increased metastasis of the treated tumor. Mechanistically, this phenomenon has been linked to a hypoxia-mediated, c-Met-dependent activation of VEGFR-2 [114]. Although this phenomenon is currently confined to preclinical models as confirmatory data from human cancer are largely missing, the potential risk of increased invasiveness/metastasis remains a major concern. In contrast to VEGF blocklade, inhibition of Ang2 has not been associated with increased metastasis in preclinical cancer models. Blockade of Ang2 resulted in decreased metastasis in a number of preclinical models [105, 106, 113, 115]. The reason why inhibition of one antiangiogenic agent (VEGF) potentially induces invasion/metastasis whereas blocking of another (Ang2) blocks metastasis is not entirely clear but may be related to the extent of therapy-induced hypoxia, as well as to decreased vascular permeability and increased vascular stability as a consequence of Ang2 blockade. Ang1 binding to the Tie2 tyrosine kinase leads to receptor phosphorylation and junctional stabilization in a RhoA kinase-dependent manner [63, 64]. Vice versa, Ang2 is able to block Ang1-induced Tie2 phosphorylation and can reduce cell-matrix interactions, with the consequence of destabilizing vessels [8, 9, 36]. Ang2 inhibitors may therefore counteract Ang2-induced vascular stabilization and thus diminish metastasis.

The dual blockade of VEGF and Ang2 affects numerous cellular compartments, i.e. vascular, perivascular (see Table 13.1), and immune cells (see below), which may explain additive therapeutic results. Myeloid cells have long been known to be responsible for mediating tumor refractoriness [84, 90, 97, 98]. They are associated with poor prognosis and thus their depletion has a beneficial outcome on survival [84,90,96]. Continuous expression of Ang2 in endothelial cells (Ang2 GOF mouse model) led to an increased number of tumor-associated myeloid cells that exhibited a proangiogenic/M2 gene signature [75]. In addition, Ang2 GOF increased the frequency of T regulatory cells, which have immunosuppressive capacities and are thus also able to negatively contribute to tumor progression [92]. In addition, high Ang2 levels have recently been associated with therapy resistance [113]. Clearly, these findings are in favor of dual targeting of angiopoietin and VEGF signaling pathways. Interestingly M2-polarized proangiogenic macrophages still reside after Ang2/VEGF combination therapy in a syngeneic glioblastoma model (own unpublished observations). These results demand for additional targeting of the innate immune cell compartment for example by using antibodies against colony-stimulating factor 1 receptor (CSF1R). Proof of principle with drugs targeting tumor-associated macrophages has been established in preclinical models [116, 117]. Moreover, immune cell activation may be an additional option for enhancing efficacy of cancer therapies as an increased number of T lymphocytes/cytotoxic T cells following anti-VEGF or anti-CSF1R therapy have been reported [98, 118]. New avenues targeting the immune system (activation) by the usage of checkpoint inhibitors are currently also being explored [119] and results from preclinical studies will show their efficacy and demonstrate whether they may be useful for future cancer therapies [120, 121].