In this study, we used the developing mouse retina and the aortic ring model to address the role of microglial cells in angiogenesis. The retina is an organ where too many or to few vessels are associated with pathology. The retina is also subject to pharmacological application of anti-VEGF therapy, which is used to counteract the edema that compromises vision in age-dependent macula degeneration 
. This clinical relevance combined with the many advantages of the retina for experimental studies of angiogenesis makes it an ideal location to study the effect of angiogenic modulators. Accordingly, the retina is also a suitable location to study the influence on angiogenesis of non-vascular cell types such as microglial cells. The aortic ring model reproduces angiogenic sprouting in culture in three-dimensional biomatrix gels 
. The vessel outgrowths produced by aortic rings consist of endothelial cells in interaction with mural cells as well as other types of mesenchymal cells, such as fibroblasts and macrophages 
. Because the aortic ring model is intermediate between simpler in vitro
models of angiogenesis and complex in vivo
models, the aortic ring model has become attractive as a reproducible and relatively high-throughput assay for the study of angiogenesis. Hence it has been broadly used for the study of basic mechanisms of angiogenesis, and to test the effects on angiogenesis of diverse components, such as growth factors and cytokines, immune regulatory molecules, proangiogenic or antiangiogenic compounds, protease inhibitors, extracellular matrix components and their receptors, and different cell types (reviewed in 
Our observations in vivo
suggest that microglial cells exert a stimulatory effect on angiogenesis. These observations are consistent with those of Fantin and colleagues 
, that tissue macrophages/microglia are associated with angiogenesis in different developmental organs. However, the in vivo
studies fail to show if the effect of microglial cells on angiogenesis is direct or indirect. Our in vitro
findings demonstrate that microglial cells directly promote vessel sprouting in the aortic ring model, and are consistent with a general stimulatory role of microglia on angiogenesis. Our observations in the developing retina also suggest that microglia affect the number and direction of filopodial extensions from tip-cells at the vascular front and within the forming retinal vessel plexus. In the presence of microglia, endothelial tip-cell filopodia protruded both in a forward (radial) orientation - towards the avascular retinal periphery - and sideways towards microglial cells and other tip-cells, resulting in filopodial brushes with obtuse angle. In the absence of microglia the tip-cell filopodial protrusions were mainly oriented forward. These observations are consistent with a model in which microglial cells exert their angiogenic effect by promoting the protrusion and/or stabilization of endothelial tip-cell filopodia.
Although the VEGF/Notch signaling loop likely constitutes the basic signaling machinery that ensures formation of the nascent vascular network, it is clear that there must be other modulators of the system. Fantin and coworkers reported that the microglia-derived angiogenic activity acts parallel to VEGF-A, since the effects of microglia and VEGF-A appeared additive when studied in genetic mouse models 
. In the aortic ring model, addition of microglia promoted formation of a fine network of branches with one to two cells at the branch circumference. In contrast, addition of VEGF-A promoted formation of thicker branches with multiple cells at the branch circumference. Importantly, while addition of the VEGF inhibitor could reverse the effect of VEGF-A, it had no profound effect on microglial-induced vessel branching. Thus, our findings in the aortic ring model are consistent with the in vivo
observations reported herein and previously.
In this context it is of interest that resident macrophages present in the aortic ring have been reported to play a permissive role in the angiogenic response from the ring explants; macrophage depletion inhibits angiogenic responses in the rings 
. The same group also reported that a subset of immature immune cells grown out from aortic ring cultures stimulate angiogenesis in freshly cut rings 
. However, in contrast to the microglial cells used in this study, those cells produced significant quantities of VEGF-A, suggesting that different sources and phenotypes of leukocytes may affect angiogenic responses via different mechanisms.
A attractive explanation for the in vivo observations that microglia localize near sites of endothelial tip-cells, would be that microglia act as guideposts for anastomosis formation during development of the vascular network that initially covers the retina. This model would imply (although not necessarily demand) direct contact between microglial cells and endothelial tip-cells as a critical step in the anastomosis process. Intriguingly, however, using the aortic ring model, we found that microglia induced sprouting in the absence of a direct contact with the vessels. Moreover, conditioned medium from microglia could partly mimic the effect of ectopically added microglia. Thus, at least some of the effect of microglia on angiogenesis seems to rely on a secreted factor(s). Together, our in vivo and in vitro observations would suggest that microglia provide a signal(s) for filopodial protrusion from endothelial tip-cells, a signal that competes with the VEGF-A produced by retinal astrocytes. Inflammation is also known to activate endothelial cells and promote vessel branching. However, when assaying a panel of inflammatory cytokines, we failed to detect any up-regulation of such cytokines in media form aortic rings cultured with microglia compared to media from control aorta ring cultures (SR and CB, unpublished), and further studies will be required to identify and characterize the microglia-derived signal(s).
Our results using aortic ring explants further suggest that the close association of microglia with the vascular front and endothelial tip cells observed in vivo may depend on vessel-derived secretion of microglial attractant(s). Furthermore, the angiogenic effect of microglial cells was significantly stronger when they were exposed to aortic rings; plain microglial conditioned medium was angiogenic but the effect was weaker compared to the co-culture situation. Together, these observations suggest a two-way communication between the vasculature and the microglial cells, in which the vasculature attracts the microglial cells and promotes their release of angiogenic factor(s).