Cellular and molecular mechanisms of VEGF-induced dose-dependent angiogenesis

VEGF is the fundamental regulator of angiogenesis both in development and in adult tissues. We have previously found that over-expression of murine VEGF164, both in normal and ischemic muscle, can induce either normal or aberrant angiogenesis depending strictly on its microenvironmental dose in vivo. The best characterized mode of angiogenesis is sprouting and relies on a balanced formation of tip cells, which migrate towards the VEGF gradient, and stalk cells, which instead proliferate behind the tip. These initial morphogenic events require the generation of an alternate pattern of Dll4 expression on tip cells and consequent Notch1 activation on neighbouring endothelial cells, which become stalk. Here, we firstly sought to determine how VEGF controls the initial stages of vascular morphogenesis leading to normal and aberrant angiogenesis in skeletal muscle and then how VEGF dose-dependent angiogenesis can be controlled in a therapeutically compliant lentiviral platform. Primary myoblasts were retrovirally or lentivirally transduced to express mouse VEGF164 or human VEGF165 and individual clones were isolated, so that every cell expressed the same level. Clonal populations expressing low (60 ng/106 cells/day) or high VEGF levels (120 ng/106 cells/day) were implanted in the auricularis (ear) and tibialis anterior (leg) muscles of SCID mice.
We demonstrated that both VEGF doses cause a similar initial response 4 days after implantation, with homogeneous enlargement of pre-exisiting microvessels. By 7 days, enlarged vessels remodeled into normal pericyte-covered capillary networks with low VEGF levels or into aberrant angioma-like structures with high levels. Corrosion casts analysis of implanted leg muscles showed that VEGF-overexpression in skeletal muscle induces normal or aberrant angiogenesis by splitting rather than by sprouting. Therefore, we decided to investigate how the Dll4/Notch1 pathway is activated and regulated in splitting angiogenesis. Four days after implantation, we found that Dll4 was expressed on long stretches of several contiguous endothelial cells, which also had activated Notch1 localized in the nucleus, both with low and high VEGF levels. After 7 days, neither the normal nor the aberrant vascular structures that were generated showed any Dll4 expression or activated Notch1. Pharmacological Notch inhibition by systemic DAPT treatment disrupted the initial vascular enlargement induced by either VEGF dose and led to disorganized aggregates of endothelial cells. In summary, these data indicate that: 1) both normal and aberrant vascular structures induced by over-expression of different VEGF doses in skeletal muscle are generated by a first stage of vessel enlargement, followed by intussusceptive remodeling, rather than sprouting; 2) the initial vascular enlargement depends on the simultaneous activation of Notch1 on contiguous endothelial cells, leading to an all-stalk phenotype, rather than an alternate pattern of tip and stalk cells; 3) the remodeling to either normal or aberrant vascular structures by intussusception is not regulated by differential Notch1 activation.
In the second part of the project, in order to translate our myoblast-based VEGF delivery in a potential clinical application, we decided to develop a compliant lentiviral delivery platform. Here, we aimed to study how VEGF expression could be modulated in vivo by using a different viral vector rather than retroviruses which have been banished from the clinical practice due to insertional mutagenesis events occurred in some patients. Copy number analysis revealed that lentiviruses integrate in multiple copies in target cells leading to higher and wider VEGF expression levels in vitro. Furthermore, we demonstrated that the kinetics of VEGF expression in vivo is dramatically different using lentiviruses. While retroviruses caused a drastic decrease in VEGF expression per cell already 7 days after implantation, lentiviruses allowed a sustained VEGF expression per cell up to two weeks. In conclusion, lentiviral-based VEGF expression leads to features that are predicted to make the control of VEGF dose in vivo more difficult. These results highlight the importance of our FACS-sorting based technology that allows to purify specific VEGF-expressing populations and that will help in defining the VEGF angiogenic potential in rigorous dose-escalation phase I clinical trials.