Endothelial EphB4 overexpression induces resistance against antiangiogenic therapy in SF126 glioma by altering vascular morphogenesis shifting the microvascular environment towards large, therapy-resistant microvessels. Alterations in microvascular blood flow as well as changes in cellular proliferation, apoptosis rate, and pericyte-endothelial cell interactions represent important mechanisms which are involved in this process.
Selection of therapy-resistant tumor vessels represents a key feature of evasive resistance mechanisms against antiangiogenic therapy [
3]. Therapy-resistant vasculature in glioma has been characterized by distinct changes in morphogenesis. Combined temozolomide and sunitinib treatment leads to selection of large, high-flow, therapy-resistant glioma vessels by aggravating pericyte-mediated resistance mechanisms
via angiopoetin–Tie 2 and delta-like-4 (Dll4)/notch signaling [
4]. Dll4 expressing gliomas have been shown to be resistant towards anti-VEGF therapy by activating different angiogenesis-related pathways leading to large, resistant microvessels [
7]. Pericyte–endothelial cell interactions are important regulators of vascular morphogenesis and represent key players in vascular resistance mechanisms [
3]. Endothelial EphB4 overexpression has been shown to regulate glioma microvascular morphogenesis by aggravating pericyte–endothelial cell interactions [
10]. We demonstrate that endothelial EphB4 overexpression leads to maintenance of pericyte–endothelial interactions despite effective anti-VEGF and anti-PDGF treatment leading to resistant glioma vasculature in a subcutaneous SF126 glioma model. In the orthotopic microenvironment, EphB4 overexpression induced significant alterations of vascular morphogenesis with a significant increase in microvascular diameter without altering the number of pericyte–endothelial cell interactions. In this regard, Erber et al. demonstrated that EphB4 overexpression leads to a change in the quality of pericyte–endothelial cell interactions with a tight pericyte–endothelial cell interface as compared to loose pericyte–endothelial cell interactions in controls without altering the number of pericyte–endothelial cell interactions [
12].
Reduced pericyte–endothelial cell interactions may be the result of the high sunitinib dose used in the orthotopic experiments. The dose was applied purposely as this approach was used in the previous experiments to induce a relevant antiglioma effect in orthrotopically implanted experimental glioma [
12]. Despite reduction of pericyte–endothelial cell interactions, therapy resistance was maintained. Similar findings were described by Li et al., showing that Dll4 expressing gliomas are characterized by large, bevacizumab-resistant microvessels despite significantly reduced pericyte–endothelial cell interactions [
7]. The authors demonstrated, in detail, that inhibition of EphB4 abolished vascular resistance to bevacizumab [
7]. One hypothesis explaining pericyte-independent vascular resistance mechanisms focuses on large and high-flow tumor vessels (as a result of EphB4 overexpression) providing superior supply of oxygen and nutrients which in turn leads to reduced expression of HIF1α and diminished VEGF dependence of the tumor with reduced endothelial expression of VEGFR [
13]. Observations of improved microvascular hemodynamics were described previously with significantly increased microvascular delivery of chemotherapy in sunitinib-resistant glioma vasculature [
13]. Other potential pericyte-independent mechanisms may include activation of other angiogenesis-related molecules that maintain angiogenesis signaling despite anti-VEGF therapy [
3]. Apart from vascular resistance mechanisms, endothelial EphB4 overexpression induced a reduction of proapoptotic and antiproliferative effects of sunitinib indicating resistance mechanisms in glioma cells. Interactions between endothelial cells and glioma cells define the perivascular niche in glioma development [
12]. In this regard, EphB4 overexpression reduced glioma growth in orthotopically implanted glioma in our study by leading to reduced proliferation. These observations are opposed by studies from Chen et al. demonstrating increased glioma growth in response to EphB4 upregulation mediated by increased epidermal growth factor receptor activity (EGFR) [
14]. In human glioblastoma multiforme patients, high EphB4 expression correlates with decreased progression-free survival underlining a more aggressive phenotype [
15]. However, ephrinB2-EphB4-mediated (onco)biological effects are extremely variable depending on tumor biology, microenvironment, presence or absence of ligand-dependent and ligand-independent signaling, in addition to forward and backward signaling mechanisms. In contrast to the above-named studies, which investigated the effects of EphB4 expression in tumor cells, we analyzed the effects of endothelial EphB4 overexpression. Moreover, discrepancies may further be explained by different ligand-independent and ligand-dependent effects of EphB4. In this regard, EphB4 has been shown to exert tumor progressive effects in the case of EphB4 overproduction in a ligand-independent mechanism, while ligand-dependent stimulation acts tumor suppressive [
16]. Recent experiments depicted reverse signaling, mediated
via ephrinB2, to act tumor promoting despite destabilizing tumor vascularization following EphB4 overexpression in malignant melanoma [
17]. In human glioblastoma multiforme, a large heterogeneity exists in the expression of EphB4 and ephrinB2 depending on the biological characteristics of glioma cells and the associated microenvironment [
18]. This heterogeneity is additionally complicated by the current changes in neuropathological classification of brain tumors [
19]. It becomes clear that brain tumors are classified beyond the WHO classification system based on a molecular fingerprint (e.g., methylation status) which explains the very variable pathological and clinical courses observed in clinical treatment of the disease [
19]. Therefore, it may be speculated that different expression profiles of ephrinB2 and EphB4 and the associated equilibrium between receptor and ligand account for the reported different effects of EphB4 overexpression in glioma biology. To address these issues, we used one glioma cell line and analyzed the effects of endothelial EphB4 overexpression on vascular resistance using defined hetero- and orthotopic experimental models. Further studies will have to focus on EphB4 mediated effects using different glioma cell lines and loss of function approaches to characterize EphB4 signaling depending on the oncobiological context to support clinical translation of EphB4 as a potential target in glioma resistance.