Although many of the cellular and molecular mechanisms of angiogenesis have been intensely studied [1], little is known about the processes that underlie vascular anastomosis. We have generated transgenic fish lines expressing an EGFP-tagged version of the junctional protein zona occludens 1 (ZO1) to visualize individual cell behaviors that occur during vessel fusion and lumen formation in vivo. These life observations show that endothelial cells (ECs) use two distinct morphogenetic mechanisms, cell membrane invagination and cord hollowing to generate different types of vascular tubes. During initial steps of anastomosis, cell junctions that have formed at the initial site of cell contacts expand into rings, generating a cellular interface of apical membrane compartments, as defined by the localization of the apical marker podocalyxin-2 (Pdxl2). During the cord hollowing process, these apical membrane compartments are brought together via cell rearrangements and extensive junctional remodeling, resulting in lumen coalescence and formation of a multicellular tube. Vessel fusion by membrane invagination occurs adjacent to a preexisting lumen in a proximal to distal direction and is blood-flow dependent. Here, the invaginating inner cell membrane undergoes concomitant apicobasal polarization and the vascular lumen is formed by the extension of a transcellular lumen through the EC, which forms a unicellular or seamless tube.

Organ formation and growth requires cells to organize into properly patterned three-dimensional architectures. Network formation within the vertebrate vascular system is driven by fusion events between nascent sprouts or between sprouts and pre-existing blood vessels. Here, we describe the cellular activities that occur during blood vessel anastomosis in the cranial vasculature of the zebrafish embryo. We show that the early steps of the fusion process involve endothelial cell recognition, de novo polarization of endothelial cells, and apical membrane invagination and fusion. These processes generate a unicellular tube, which is then transformed into a multicellular tube via cell rearrangements and cell splitting. This stereotypic series of morphogenetic events is typical for anastomosis in perfused sprouts. Vascular endothelial-cadherin plays an important role early in the anastomosis process and is required for filopodial tip cell interactions and efficient formation of a single contact site.

Organ morphogenesis requires the coordination of cell behaviors. Here, we have analyzed dynamic endothelial cell behaviors underlying sprouting angiogenesis in vivo. Two different mechanisms contribute to sprout outgrowth: tip cells show strong migratory behavior, whereas extension of the stalk is dependent upon cell elongation. To investigate the function of Cdh5 in sprout outgrowth, we generated null mutations in the zebrafish cdh5 gene, and we found that junctional remodeling and cell elongation are impaired in mutant embryos. The defects are associated with a disorganization of the actin cytoskeleton and cannot be rescued by expression of a truncated version of Cdh5. Finally, the defects in junctional remodeling can be phenocopied by pharmacological inhibition of actin polymerization, but not by inhibiting actin-myosin contractility. Taken together, our results support a model in which Cdh5 organizes junctional and cortical actin cytoskeletons, as well as provides structural support for polymerizing F-actin cables during endothelial cell elongation.

Vascular networks are formed and maintained through a multitude of angiogenic processes, such as sprouting, anastomosis and pruning. Only recently has it become possible to study the behavior of the endothelial cells that contribute to these networks at a single-cell level in vivo. This Review summarizes what is known about endothelial cell behavior during developmental angiogenesis, focusing on the morphogenetic changes that these cells undergo.

During embryonic development, the vertebrate vasculature is undergoing vast growth and remodeling. Blood vessels can be formed by a wide spectrum of different morphogenetic mechanisms, such as budding, cord hollowing, cell hollowing, cell wrapping and intussusception. Here, we describe the vascular morphogenesis that occurs in the early zebrafish embryo. We discuss the diversity of morphogenetic mechanisms that contribute to vessel assembly, angiogenic sprouting and tube formation in different blood vessels and how some of these complex cell behaviors are regulated by molecular pathways.

During embryonic development, vascular networks remodel to meet the increasing demand of growing tissues for oxygen and nutrients. This is achieved by the pruning of redundant blood vessel segments, which then allows more efficient blood flow patterns. Because of the lack of an in vivo system suitable for high-resolution live imaging, the dynamics of the pruning process have not been described in detail. Here, we present the subintestinal vein (SIV) plexus of the zebrafish embryo as a novel model to study pruning at the cellular level. We show that blood vessel regression is a coordinated process of cell rearrangements involving lumen collapse and cell–cell contact resolution. Interestingly, the cellular rearrangements during pruning resemble endothelial cell behavior during vessel fusion in a reversed order. In pruning segments, endothelial cells first migrate toward opposing sides where they join the parental vascular branches, thus remodeling the multicellular segment into a unicellular connection. Often, the lumen is maintained throughout this process, and transient unicellular tubes form through cell self-fusion. In a second step, the unicellular connection is resolved unilaterally, and the pruning cell rejoins the opposing branch. Thus, we show for the first time that various cellular activities are coordinated to achieve blood vessel pruning and define two different morphogenetic pathways, which are selected by the flow environment.

After the initial formation of a highly branched vascular plexus, blood vessel pruning generates a hierarchically structured network with improved flow characteristics. We report here on the cellular events that occur during the pruning of a defined blood vessel in the eye of developing zebrafish embryos. Time-lapse imaging reveals that the connection of a new blood vessel sprout with a previously perfused multicellular endothelial tube leads to the formation of a branched, Y-shaped structure. Subsequently, endothelial cells in parts of the previously perfused branch rearrange from a multicellular into a unicellular tube, followed by blood vessel detachment. This process is accompanied by endothelial cell death. Finally, we show that differences in blood flow between neighboring vessels are important for the completion of the pruning process. Our data suggest that flow induced changes in tubular architecture ensure proper blood vessel pruning.

Protein–protein interactions are crucial for cellular homeostasis and play important roles in the dynamic execution of biological processes. While antibodies represent a well-established tool to study protein interactions of extracellular domains and secreted proteins, as well as in fixed and permeabilized cells, they usually cannot be functionally expressed in the cytoplasm of living cells. Non-immunoglobulin protein-binding scaffolds have been identified that also function intracellularly and are now being engineered for synthetic biology applications. Here we used the Designed Ankyrin Repeat Protein (DARPin) scaffold to generate binders to fluorescent proteins and used them to modify biological systems directly at the protein level. DARPins binding to GFP or mCherry were selected by ribosome display. For GFP, binders with KD as low as 160 pM were obtained, while for mCherry the best affinity was 6 nM. We then verified in cell culture their specific binding in a complex cellular environment and found an affinity cut-off in the mid-nanomolar region, above which binding is no longer detectable in the cell. Next, their binding properties were employed to change the localization of the respective fluorescent proteins within cells. Finally, we performed experiments in Drosophila melanogaster and Danio rerio and utilized these DARPins to either degrade or delocalize fluorescently tagged fusion proteins in developing organisms, and to phenocopy loss-of-function mutations. Specific protein binders can thus be selected in vitro and used to reprogram developmental systems in vivo directly at the protein level, thereby bypassing some limitations of approaches that function at the DNA or the RNA level.