Blood vessel formation is essential for tumor growth, as it depends on a proper nutrient and oxygen supply. VEGF is not only a master regulator of angiogenesis [10] and the immune response [73] but can also stimulate tumor cell proliferation in breast cancer [74] and acute myeloid leukemia models [75]. It remains poorly understood whether endothelial-secreted VEGF is involved in these processes. However, there is evidence that the endothelial expression of VEGF-C promotes leukemic cell survival and proliferation by activating VEGF receptor-3 on cancer cells [69]. Endothelin is an angiocrine factor which regulates the vascular tone under physiological conditions. However, it can also promote ovarian carcinoma cell proliferation [35,36]. Moreover, endothelial Delta-like 4 (DLL4), which is important for controlling blood vessels formation through activating Notch1 receptors [76], can bind to Notch3 receptors on tumor cells to support their survival [32]. Interestingly, ECs can also release angiocrine factors such as Slit2, which inhibit cancer cell proliferation and motility [64]. Future work will have to assess to which extent the release of angiocrine factors contributes to cancer cell proliferation and whether this could be targeted to enhance chemotherapy.

Stem cells are characterized by the capacity to self-renew and the ability to differentiate into diverse specialized cell types. It is assumed that a subpopulation of stem-like cells within tumors, known as cancer stem cells (CSCs), which exhibit characteristics of both stem cells and cancer cells, have the ability to seed tumors when transplanted into an animal host. There is increasing evidence suggesting that CSCs are resistant to conventional chemotherapy and radiation treatment. The adoption of the CSC phenotypes appears to depend on signals from neighboring cells which are capable of forming a stem cell niche [77]. Intriguingly, under physiological conditions, ECs are responsible for the self-renewal and repopulation of hematopoietic cells, for instance through Notch signaling, fibroblast growth factor-4 (FGF-4) and CXCL12 cytokine expression [78,79], indicating that at least in the bone marrow ECs are capable of forming stem cell niches. It is therefore not surprising that ECs, through their membrane-bound or angiocrine secreted factors, also play a role in the acquisition of the CSC phenotype. Indeed, tumor ECs can induce the expression of genes involved in the CSC phenotype [80]. Similar to its functions within hematopoietic stem cell niches, the expression of Notch ligands such as Jagged-1 on ECs promotes the CSC phenotype in colorectal [49] and breast cancer [50,51] by activating Notch receptors. Moreover, ECs promote a stem-like phenotype of glioma cells through the secretion of Shh and activating the Hedgehog pathway in cancer cells [63] or by secreting the basic fibroblast growth factor (bFGF) [38]. VEGF can be released by ECs to promote the CSC phenotype through its receptor Neuropilin-1 on skin cancer cells [70]. There is also evidence that EC-derived cytokines are involved in adopting a CSC phenotype. The most prominent examples are IL-8 and CXCL12 in glioblastoma [30] and gastric cancer [31]. Furthermore, IL-6 secreted by tumor ECs is responsible for the generation of a small sub-population of CSCs in head and neck squamous cell carcinomas [43].

CSCs are associated with chemoresistance [77], and in mouse models, cancer treatment with chemotherapy upregulated IGF1 expression in tumor ECs, which activated IGF1 receptors on cancer cells, making them resistant to chemotherapy [42]. As such, there is evidence that angiocrine factors contribute to the adoption of the CSC phenotype and potentially also to chemoresistance. Future studies are needed to elucidate whether these factors are suitable for drug targeting and how this would affect cancer progression.

The epithelial to mesenchymal transition (EMT) of cancer cells describes a highly dynamic and reversible process in which cancer cells of epithelial origin lose some of their typical features, like cell-cell and cell-matrix adhesion, and gain migratory and invasive properties which are typical of mesenchymal cells. This is associated with profound changes in gene transcription. The EMT of cancer cells increases their metastatic potential and resistance toward chemotherapy [81]. Tumor ECs are involved in providing factors that influence EMT. For example, the EC-secreted EGF induces EMT transition in head and neck cancer cells [34]. Furthermore, ECs enhance EMT, breast cancer cell migration, invasion and metastasis by the release of the plasminogen activator inhibitor-1 (PAI-1) and the chemokine CCL5 [26]. At this moment, it is not yet clear whether angiocrine factors are involved in the EMT of a variety of cancer cells or only under certain conditions and cancer entities.

Cancer invasion relies on the detachment from the basal membrane, remodeling of cell-cell and cell-matrix adhesions and remodeling of the extracellular matrix. These processes are also required for invasion into blood vessels (intravasation), which is a crucial step in the metastatic cascade. Invasion and metastasis are facilitated by the EMT of cancer cells [82]. Tumor ECs are involved in EMT (as described above) and play an active role in different aspects of invasion and metastasis.

When tumor cells travel through the blood stream, they first interact with ECs during colonization at distant sites. There is increasing evidence that tumor-derived signals change transcriptional programs in ECs and immune cells at distant sites to facilitate later metastatic spreading [105]. This concept of a pre-metastatic niche is still under debate. However, there are clear hints that ECs contribute to the homing and survival of circulating tumor cells [106,107,108]. For instance, disseminated breast tumor cells reside in close proximity to ECs at distant sites and endothelial-derived thrombospondin-1 (TSP-1) maintain breast cancer cell quiescence for a long period. However, upon activation of these ECs and the subsequent angiogenic sprouting, this angiocrine effect is suppressed, allowing the formation of tumor nodules [68]. In addition, the presence of a subcutaneous tumor in mice can over-activate Notch1 signaling, not only in ECs within the tumor, but also at sites as distant as the lungs. This increases endothelial VCAM1 expression, which facilitates cancer cell homing and the formation of secondary tumors [52]. During ageing, platelet-derived growth factor (PDGF)-B expression levels increase, regulating the quiescence of the dormant disseminated tumor cells and influencing therapy resistance in bone metastasis. Therefore, bone metastasis probability increases with age [109].

These studies indicate the importance of the ECs also for the preconditioning of the metastatic niche. However, much more needs to be learnt about the factors by which a primary tumor communicates with ECs at distant sites.

Myeloid derived suppressor cells (MDSCs) comprise a heterogeneous group of immunosuppressive immature myeloid cells which inhibit T cell and natural killer (NK) cell activity. In physiological conditions, myeloid progenitor cells differentiate into mature myeloid cells like macrophages and dendritic cells (DC), whereas, in pathological conditions like cancer, the differentiation of myeloid progenitor cells is impaired, leading to the formation of MDSCs [113]. The majority of tumor-associated macrophages (TAMs) also has an anti-inflammatory phenotype which promotes tumor progression [114]. Several chemotactic cytokines are known to recruit MDSCs and TAMs into the microenvironment [114,115,116]. Here, we only focus on angiocrine factors and how these contribute to immune cell infiltration into tumors.

Angiopoietin-2 can be secreted by tumor ECs, which leads to an autocrine activation of STAT3 signaling with secretion of CCL2 and higher expression of ICAM1 [23]. These factors promote the infiltration of CCR2-expressing monocytes into the TME [115,117]. Moreover, sustained Notch1 signaling in ECs, induced by cancer cells, drives cytokine and VCAM1 expression and the infiltration with myeloid cells, which facilitates metastasis [52]. Furthermore, E-selectin expression on ECs is required for myeloid cell infiltration in different mouse tumor models [62]. Again, this is promoted by the interaction of tumor cells with ECs, which leads to a higher E-selectin expression on tumor endothelium [62,118].

Lastly, there is increasing evidence that ECs can also impact on the polarization of recruited immune cells. In glioblastoma, the tumor endothelium was shown to be the main source of IL-6 secretion, which contributes to the adoption of an anti-inflammatory and pro-tumorigenic macrophage phenotype [44]. As such, there is still little knowledge on how ECs affect the polarization of tumor-infiltrating myeloid cells and how this contributes to immunosuppression.

The cytotoxic function of tumor-infiltrating lymphocytes (TILs) is often impaired, for example by the activation of programmed cell death-1 (PD-1) and cytotoxic T-lymphocyte-associated protein-4 (CTLA4) receptors on T cells [53,54,55]. The expression of the respective ligands not only on cancer cells but also on other cells of the TME contributes to immunosuppression [119]. The role of ECs in this regard is still poorly understood. However, pre-clinal models [120] as well as recent clinical trials in hepatocellular carcinoma and non-small cell lung carcinoma demonstrated that antiangiogenic therapy targeting VEGF synergizes with immunotherapy targeting the PD-1/PD-L1 axis [121,122]. It will be of outstanding importance to unravel the mechanism responsible for this. In addition, there is evidence for the angiocrine control of T cell activity in tumors. Tumor ECs can secrete Inf-γ [45] and VEGF-A [71], which influence T cell responses. In addition, the induction of T cell immunoglobulin and mucin domain-containing molecule 3 (TIM-3) expression on ECs in lymphoma inhibits CD4+ T helper cell activity by the activation of the IL-6/STAT3 signaling pathway [66,67]. As such, there is some evidence that ECs play a role in adapting immune responses in cancer.