Encyclopedia
Please note this is an old version of this entry, which may differ significantly from the current revision.
Subjects:
Cell Biology
Angiogenesis has long been implicated as a crucial process in GBM growth and progression. GBM can adopt several strategies to build up its abundant and aberrant vasculature. Targeting GBM angiogenesis has gained more and more attention in anti-cancer therapy, and many strategies have been developed to interfere with this hallmark. However, recent findings reveal that the effects of anti-angiogenic treatments are temporally limited and that tumors become refractory to therapy and more aggressive.
- glioblastoma
- anti-angiogenic therapy resistance
- glioblastoma stem-like cells
1. Introduction and Overview
Glioblastoma (GBM) is the most common and lethal primary brain tumor in adults, with a high rate of recurrence and mortality. The standard multimodal therapy for GBM, including aggressive surgery, radiotherapy, and chemotherapy, has remained unchanged for more than two decades [1,2,3]. Meanwhile, many GBM patients receiving the standard of care treatment experience a short progression-free survival (PFS), and due to the lack of effective therapies for recurrent disease, the clinical outcome is very poor. GBM is among the most vascularized of all solid tumors, and vascular proliferation is a pathological hallmark of GBM [4]. GBM vasculature is typically composed of abnormal glomeruloid vascular structures consisting of complex aggregates of newly formed microchannels, lined with hyperplastic endothelial cells characterized by an altered morphological phenotype and supported by basal lamina and pericytes [5]. These vessels are structurally and functionally abnormal, contributing to a hostile microenvironment, and their presence is the key histopathological characteristic that discriminates GBM from lower-grade gliomas contributing to the generation of a more malignant phenotype with increased morbidity and mortality [5]. Hence, the introduction of the recombinant humanized monoclonal antibody anti-VEGF (vascular endothelial growth factor), bevacizumab, approved in 2009 in the United States as an anti-angiogenic adjuvant therapeutic strategy for recurrent GBM, raised initial enthusiasm among clinicians, but unfortunately, it was rapidly attenuated due to its limited survival advantage compared with experimental therapy at recurrence [6]. Indeed, despite notable radiographic responses and improved PFS, the effects of anti-angiogenic therapy are unfortunately not long-lasting. Currently, several alternative cellular and molecular mechanisms of vessel recruitment and tumor angiogenesis have been described in GBM that contribute to tumors escaping from anti-angiogenic therapy and even evolving into a more aggressive phenotype.
Traditionally, new vessel formation, within the tumor mass, has been ascribed to the sole process of neoangiogenesis and simply described as capillary sprouting from pre-existing capillaries. Through this process, the growing solid tumors generate an increased blood supply to satisfy their increasing nutrient and oxygen demand. It is now recognized that tumor angiogenesis is a multi-step, finely tuned process in which a highly complex spectrum of events occurs.
Along with the most studied mechanisms of neovascularization, i.e., angiogenesis and vasculogenesis, at least three distinct mechanisms of neovascularization in GBMs have been identified: (i) vessel co-option, (ii) vasculogenic mimicry (VM), and (iii) GBM stem-like cell (GSC) transdifferentiation [7]. These mechanisms are not independent but are rather closely interconnected and triggered by hypoxia as a major stimulator of angiogenesis in GBM [8,9]. Temporally, vessel co-option, the process whereby tumors utilize native brain vessels to recruit blood supply, is the first mechanism by which gliomas achieve their vasculature, followed by angiogenesis, the process by which new vessels are obtained from pre-existing ones. The weight of vasculogenesis, i.e., the differentiation of the circulating bone-marrow-derived cells, in tumor neovascularization is still controversial, and animal studies of tumor-associated vasculogenesis produced discordant conclusions [10,11]. Differently from vasculogenesis, VM, defined as the ability of tumor cells to form functional vessel-like networks, has been described first in human melanoma models [12] and later in human astrocytoma samples [13]. The most recently described mechanism of GBM neovascularization consists of the direct transdifferentiation of GSCs [14,15]. This mechanism is complementary to VM, but even though no lines of evidence exist for a direct relationship between hypoxia and VM, hypoxia is the main driver for transdifferentiation, as hypoxia and the hypoxia-inducible factors (HIFs) play critical roles in maintaining and supporting the survival and self-renewal of GSCs [16]. Despite the significant hypoxia-induced production and secretion of VEGF by GSCs, the transdifferentiation process is VEGF-independent.
2. GBM-Associated Mechanisms of Neovascularization
2.1. Vessel Co-Option
Vessel co-option, also known as perivascular migration, is a non-angiogenic process whereby cancer cells co-opt and utilize the pre-existing vessels of the surrounding normal tissue for their growth and spread. Vessel co-option is independent of the classic angiogenic switch and occurs in the absence of angiogenic growth factors [17]. To meet their metabolic demands, tumor cells migrate along the abluminal surface of pre-existing vessels and/or infiltrate the tissue space between them, leading to the incorporation of the pre-existing vasculature into the tumor [18]. Initially observed in lung cancers with alveolar non-angiogenic growth patterns [19], vessel co-option was further described in several other cancer types including gliomas, in which the tumor cells organize themselves into cuffs around normal microvessels [7]. In GBM, vessel co-option was first described by Holash et al., following the injection of rat C6 glioma cells into the rat striatum. They suggested the term “vessel co-option” to indicate this non-angiogenic phenomenon in which the co-opted vessels were characterized for the expression of angiopoietin-2 (ANG-2) [20]. The brain is a highly vascularized tissue; thus, vessel co-option represents one of the different invasion pathways used by glioma cells for the infiltration of the surrounding brain tissue [21]. Different patterns of tumor growth involving the vessel co-option mechanism have been described in the brain. In the perivascular pattern, tumor cells migrate in the perivascular space and adhere to the surface of the pre-existing vessels. In the diffusely infiltrating pattern, cancer cells infiltrate the stromal space by moving in between host vessels without creating permanent contact with them [22].
Distinctive morphological features are associated with vessel co-option and can be used to distinguish this process from angiogenesis through histological examination. The maintenance of the vascular architecture of the normal brain tissue in the infiltrated areas is one of the clearest signs of vessel co-option, in contrast to the abnormal and chaotic vasculature typical of tumor angiogenesis [18]. Moreover, vessel co-option generates histological pictures, such as satellites and finger-like extensions, depending on the cutting plane [23]. In terms of immunohistochemical markers, Baker et al. examined human GBM patient biopsies containing the infiltrative tumor margin for the presence of vimentin and the von Willebrand factor (vWF). The immunolabeling of these two markers showed perivascular-associated vimentin+ tumor cells at the transition zone between tumor and normal tissues [24]. However, there are no clinically validated techniques and biomarkers that can be used to undoubtedly identify vessel co-option in GBM patients and to argue its predictive and/or prognostic significance.
In gliomas, the process of tumor vascularization is spatially and temporally dynamic in terms of glioma cell–vascular interactions and the mechanisms utilized for tumor growth. Temporally, it has been shown that vascular co-option is the first mechanism by which gliomas achieve their vasculature and precedes angiogenesis by up to 4 weeks [7]. In particular, it appears to be the dominant mechanism of vascularization in early-stage rat C6 gliomas, followed by the rapid development of angiogenesis due to vessel regression and hypoxia [20]. However, in mouse GL26 and rat CNS-1 orthotopic gliomas, vessel co-option has been described as an iterative growth process that occurs throughout the entire course of brain tumor progression [24]. Likewise, different ways of glioma cell-vasculature interactions have been described using different orthotopic glioma models. By using the U87MG brain xenograft mouse model of GBM, Watkins et al. examined glioma–vascular interactions at a much earlier disease stage, showing that the invading glioma cells colonize the perivascular area of pre-existing vessels, wrap themselves around the abluminal surface, and displace astrocytic endfeet from endothelial or mural cells. This causes a local breach of the blood–brain barrier (BBB), followed by abnormal permeability and impaired functionality [25]. Recently, in a patient-derived GSC xenograft model, Pacioni et al. observed a variety of interactions between glioma cells and perivascular astrocytes, most of which do not imply the disruption of the BBB. In particular, they showed that tumor cells invading perivascular spaces are spatially associated with vessels showing preserved BBB [26].
2.2. Vasculogenic Mimicry
Vasculogenic mimicry (VM) is a new mechanism of tumor neovascularization in which highly invasive and genetically dysregulated tumor cells acquire vascular cell features or function, forming de novo vascular-like structures. These structures mimic the function of blood vessels, thus providing an adequate blood supply for tumor growth and metastasis [53]. The functional role of VM structures in tumor circulation has been demonstrated by different approaches, such as the microinjection method, Doppler ultrasonography, magnetic resonance imaging, and laser scanning confocal angiography [54,55,56,57,58,59]. VM was firstly discovered in uveal melanoma as the formation of a circulatory system by dedifferentiating tumor cells [12]. VM structures were characterized by the abundance of matrix proteins (proteoglycans, laminin, collagen IV, and VI), negative staining for endothelial markers (CD31 and CD34), positive staining for periodic acid–Schiff (PAS), and the presence of blood components (such as erythrocytes, platelets, and hemoglobin) in their lumen [12,53,60].
Subsequently, VM was described in other solid tumors. The first evidence for VM in gliomas was reported by Yue and Chen in 2005. The authors examined 45 cases of WHO II-IV grade astrocytoma tissues by the dual staining of CD34 and PAS to investigate whether VM existed in these tumors. They reported that the tumor microvasculature was mainly composed of endothelium-lined vessels that stained positively for PAS, laminin, and endothelial markers. However, they found that 2 out of the 45 astrocytoma tissues had PAS+/CD34− vessels containing red blood cells [13]. Liu et al. found an association between microvascular density (MVD) and VM in gliomas. VM-positive gliomas had low MVD compared with VM-negative gliomas, suggesting that VM structures could represent a complementary mechanism to sustain blood supply, particularly in areas with low MVD [61]. Furthermore, the authors observed a positive correlation between VM and the WHO grade of glioma. The patients with VM-positive gliomas had shorter overall survival than those with glioma without VM [61]. As described in other tumors, VM might be of two types in GBM: a patterned matrix type of secreting matrix proteins and a tubular type characterized by tumor cells lining the vessel-like structures [62].
In GBM, the involvement of GSCs in VM has been reported by several studies [62,63,64]. Chiao et al. demonstrated that CD133+ GSCs contributed to forming VM in tumor xenografts, particularly the CD133+ GSC-derived xenografts showed vessel-like structures negative for CD31 staining and positive for PAS and α-smooth muscle actin (α-SMA), suggesting that these cells may contribute to forming vessel-like structures by transdifferentiating in vascular smooth muscle-like cells. Likewise, other studies reported that GSCs might differentiate into vascular smooth muscle-like cells or vascular mural-like cells to induce VM of the tubular type in GBM [62,64].
The molecular mechanisms underlying VM formation involve a complex network of signaling pathways. Hypoxia seems to be one of the main inducers of this process, and through its main effector, HIF-1α, it directly regulates several VM-related effectors such as VE-cadherin (CDH5) and MMPs [65,66]. MMP-14, MMP-9, and MMP-2 induce matrix remodeling, which promotes VM in glioma [67,68]. MMPs are also regulated by upstream regulators such as the transforming growth factor-β (TGF-β), which has a critical role in VM formation (Ling et al. 2011 [69]). When activated by upstream effectors, MMP-14 activates MMP-2, inducing the cleavage of the laminin subunit-γ2 (LAMC2) chain into promigratory γ2′ and γ2, which in turn promote VM formation [67,69]. TGF-β promotes the expression of other adhesion molecules such as CDH5 [70]. It has been reported that CDH5 and erythropoietin-producing human hepatocellular receptor A2 (EphA2) are highly expressed in VM-positive glioma compared with VM-negative glioma, and their expression is required for VM formation [65,71]. CDH5 modulates the EphA2 activity, which in turn regulates p85, the regulatory subunit of phosphoinositide 3-kinase (PI3K), promoting the loss of intercellular adhesion and facilitating migration and infiltration [72,73]. It has been demonstrated that the EGFR/PI3K/AKT/mammalian target of the rapamycin (mTOR) pathway is closely related to VM formation, as well as other signaling pathways, including the CXCR4/AKT, insulin-like growth factor-binding protein 2 (IGFB2), VEGF/VEGFR-2, and interleukin 8 (IL-8)/CXCR2 pathways [74,75,76,77,78,79,80]. Notably, VEGF/VEGFR-2 can stimulate VM, inducing EphA2 to enhance MMP expression, thus favoring extracellular matrix remodeling [67,76,81]. As expected, with such a complex process, the regulatory roles of non-coding RNAs (ncRNAs), particularly miRNAs and lncRNAs, have widely been reported in the induction of VM in glioma [68,71,82,83,84,85,86,87,88,89].
2.3. Cell Transdifferentiation
Cell transdifferentiation is a process whereby GBM cells have the ability to acquire an endothelial and/or pericyte phenotype, contributing to the formation of the tumor vasculature. The hypothesis of the endothelial transdifferentiation of tumor cells was first described in human cutaneous melanoma models [102,103]. Then, it was demonstrated that primary GBM cell lines and GSCs can be induced to differentiate in the cell types of the mesenchymal lineage [104,105]. Similar to normal neural stem cells (NSCs), able to generate glial and neuronal lineages, pluripotency is a typical feature of GSCs. The GBM–endothelial cell transdifferentiation represents one of the most recent GBM-associated neovascularization mechanisms that has been described. In 2010, two groups independently demonstrated the transdifferentiation of GSCs into ECs in vitro and the role of GSCs in tumor endothelium [14,15]. Specifically, Ricci-Vitiani et al. demonstrated that a proportion of CD31+ endothelial cells shared the same chromosomal alterations as the tumor cells within GBM specimens. Moreover, they showed that a significant fraction of GFAP+ microvascular cells displayed an aberrant glial/endothelial phenotype. Interestingly, in mouse GSC xenografts, about 70% of the CD31+ cells from the inner part of the tumor were of human origin [14]. Wang et al. reported that the fraction of CD105+ endothelial cells harboring the amplification of EGFR and the centromeric portion of chromosome 7 was similar to that of the tumor cells themselves. Furthermore, following the experiments with dissociated human GBM specimens, the authors postulated that the CD144+/CD133+ double-positive population represents the endothelial progenitor cells (EPCs) that arise from the CD133+ population and can differentiate into an endothelial phenotype [15]. Under specific culture conditions that promote endothelial differentiation, GSC-derived endothelial cells (GdECs) show a typical flagstone vascular endothelial cell morphology and the ability to form tubular-like structures when cultured in Matrigel [106,107]. GdECs express specific vascular endothelial cell markers, such as CD31, Tie-2, VEGFR-2, vWF, but also EPC-specific markers, such as CD34 [14,15,108]. Interestingly, an analysis of surgical GBM specimens showed tumor vessels co-expressing markers of early vascular endothelial cells (CD34) and GSCs (ABCG2 and nestin), suggesting the existence of interim cells during the transdifferentiation process [107]. This hypothesis might be supported by the findings that, under proangiogenic conditions, the glioma cells incorporated into the tumor vasculature lost GFAP expression and gained CD133 expression, shifting to a more stem/progenitor phenotype [109]. Intriguingly, a recent study reported that the tumor xenografts originating from CD34high-expressing GdECs showed a more undifferentiated phenotype compared with CD34-/low expressing cells [108].
It has been demonstrated that GBM cells are also able to transdifferentiate into pericyte-like cells [110]. Cheng et al. reported that GSCs have the capacity to acquire a pericyte lineage phenotype in vitro, characterized by the expression of typical pericyte markers such as CD146, α-SMA, neuron–glial antigen 2 (NG2), and CD248. Analyzing human GBM specimens, they also showed that the vast majority of pericytes were of neoplastic origin, carrying the same genetic alterations as cancer cells [110].
The tumor microenvironment plays a remarkable role in the transdifferentiation process. The interaction between tumor cells and the endothelium is bidirectional, and their roles seem to be interchangeable, depending on the microenvironment demands [39,111]. A study on human GBM tissues reported the existence of a perivascular niche in which GSCs are close to CD34+ endothelial cells [112]. This has been described as a favorable environment that allows for the crosstalk between tumor cells and endothelial cells, thus promoting GSC maintenance and vascular development [113]. A recent study by Zheng at al simulated a perivascular niche through a hypoxic co-culture system in vitro, showing that GSCs transdifferentiate in nestin+/CD31+ cells, whose frequency in the histological samples of GBM correlated with a poor prognosis [114]. The distribution of tumor-derived ECs does not appear to be homogeneous within the tumor, but these cells were more abundant in the core rather than in the tumor periphery [111]. Interestingly, this distribution correlated with the high density of GSCs found in the hypoxic core of the tumor [115]. Indeed, it has been proven that hypoxia represents the main driver for GSC transdifferentiation [106,116], controlling stem cell self-renewal and plasticity, promoting the formation of pseudocapillary structures laid on a matrix structure and GdECs [117]. Moreover, endothelial transdifferentiation has been described as a VEGF-independent mechanism [111]. Several studies reported the involvement of the NOTCH signaling pathway in this process [114,118,119]. In particular, Zheng et al. reported that nestin+/CD31+ cells of the hypoxic perivascular niche showed a high expression of NOTCH–ligands JAG1 and DLL4, also suggesting the role of these factors in mediating the interaction with GSCs [114]. Furthermore, Hu et al. demonstrated that GSC transdifferentiation into GdECs is mediated by the epigenetic activation of Wnt5a, through Akt signaling, and promotes host EC recruitment to create a vascular niche sustaining GSC growth and survival [120]. Recently, it has been reported that the P4HA1/COL6A1 signal axis can drive the transdifferentiation of GSCs into GdECs, promoting the expression of the endothelial marker CD31, thus contributing to the neovascularization process in response to the hypoxic microenvironment [121]. The authors also showed the co-expression of P4HA1 and CD31 in endothelial cells within blood vessels in human glioma specimens, other than a positive correlation between P4HA1 and the blood vessel density [121].
The transdifferentiation of GSCs into pericyte-like cells has been described to rely on the recruitment of GSCs by tumor endothelial cells via SDF-1/CXCR4 signaling and the generation of pericytes through the activation of the TGF-β pathway [110]. It has been also reported that GSC-derived pericytes depend on VEGFR-2 expression [64] or on the activation of the NOTCH signaling pathway [122]. Recently, it has been suggested that EGFR and NF-kB signaling are involved in GSC transdifferentiation to pericytes [123]. Two distinct vascular phenotypes associated with different statuses of the EGFR gene have been described, characterized by GSC-derived pericytes closely related to ECs or delocalized with the disruption of the BBB [123,124]. Interestingly, an emerging regulatory mechanism of this transdifferentiation process involves netrin-1, a protein recently postulated as a non-canonical angiogenic ligand [125]. In particular, it has been hypothesized that the netrin-1 contained in exosomes secreted in the microenvironment could interact with the receptor UNC5, trigger the activation of NF-kB, and regulate the transdifferentiation to pericyte [125].
This entry is adapted from the peer-reviewed paper 10.3390/jpm12101625
This entry is offline, you can click here to edit this entry!
