Originally, the term “angioblast” was introduced by His [46], and the terms angiogenesis and vasculogenesis were used as synonyms that time with the general meaning of blood vessel development. The terms were then newly defined, unfortunately neglecting the existence of the second vascular system, the lymphatic vascular system. Thus, the blood vessels were the main focus, and vasculogenesis was defined as in situ differentiation of angioblasts or hemangioblasts into blood vessels [47, 48]. The growth of blood vessels from preexisting ones was called angiogenesis, but the two mechanisms cannot be clearly separated temporally and spatially [10, 49].

Already Sabin [17] proposed a subdivision of embryonic angiogenesis into three phases. Firstly, ECs are derived from precursor cells. These are either unipotent angioblasts (giving rise to endothelial cells) or bipotent hemangioblasts (giving rise to ECs and hematopoietic cells (HCs)) [18, 50]. The development of both ECs and HCs from a single precursor cell has not been shown unequivocally but was suggested because of their simultaneous development in blood islands. Evidence for the existence of hemangioblasts was provided mainly by in vitro studies in mice [51]. Here, it was shown that vascular endothelial growth factor receptor-2 (VEGFR-2)-positive embryonic stem (ES) cells are able to generate ECs as well as HCs. In accordance, VEGFR-2 null mice lack yolk sac blood islands and ECs [52]. In the chick, a single VEGFR-2⁺ cell can only give rise to either an EC or a HC colony [53]. Thereby, the emergence of ECs in vertebrates does not depend on normal mesoderm formation and can even be observed when gastrulation is prevented [54, 55], but, nevertheless, ECs and HCs are usually regarded as mesodermal cells. Secondly, endothelial cells are derived from precursor cells and simultaneously from ECs that are already integrated into the primary vascular plexus of the embryo. Thirdly, new endothelial cells solely emerge from already existing ones. This aspect has been challenged by the finding of circulating angioblasts in adults [56], which may, however, be of greater importance in pathologic angiogenesis than in embryonic angiogenesis.

There are two sources for blood vascular endothelial cells: mesodermal precursor cells (angioblasts and hemangioblasts) and proliferation-competent endothelial cells that exist as endothelial progenitor cells (EPCs) in vascular niches, which have not been defined unequivocally [57]. Angioblasts and EPCs possess high migratory potential. They migrate as single cells through the ECM or within the endothelial lining – preferentially against the blood stream [58, 59]. Angioblasts aggregate, flatten, and form tubes, so that the intercellular space becomes the lumen of the vessel [60]. However, intracellular lumen formation by confluence of vacuoles has also been described [61, 62], but live imaging studies in zebrafish again support the original view of lumen formation by expansion of the luminal cell membrane [63].

In earlier studies on embryonic angiogenesis, the term angiogenesis was used heterogeneously but was then commonly defined in a broad sense as the development of blood vessels from preexisting one, irrespective of the mechanisms involved in this process such as sprouting, intussusceptive vascular growth, splitting of vessels, and circumferential growth of vessels [13, 64]. In differentiated organisms, circumferential growth of arterioles and small arteries into larger arteries has been termed arteriogenesis [65]. In the early embryo, the initial (primitive) vascular plexus is made up of sinusoidal vessels lined solely by ECs. This early plexus has often been called “capillary plexus,” which is wrong because capillaries are organo-typically differentiated vessels with small lumen and often invested by a basal lamina and pericytes. Further growth of the initial vascular plexus is by integration of angioblasts and by a variety of mechanisms: mainly by sprouting and intussusceptive growth.

Soon after the formation of the initial vascular plexus, conduit vessels are formed. The first larger vessels are the aorta and the cardinal veins. In higher vertebrates and man, but not in fish, the aorta is initially formed from paired vessels, which fuse in the midline [80]. The specification of arterial and venous endothelium takes place in very early stages. The term arteriogenesis, however, has originally been proposed for the establishment of collateral arteries in ischemia [65]. The term “collateral formation” may be better suited to describe this process, and the terms arterio- and venogenesis seem to be more appropriate to describe the embryonic development of these vessel types.

The first markers that were found to differentiate between arterial and venous ECs were the juxtacrine signaling molecules EphrinB2 in arterial ECs and EphB4 in venous ECs [81]. Ephrins and Ehps are transmembrane molecules that bind to each other and induce signaling in both directions, forward and backward. EphrinB2 and EphB4 are expressed in arterial and venous ECs even before the onset of circulation but mainly appear to be functional during the remodeling of embryonic vascular networks [82]. The high plasticity of ECs has been demonstrated in grafting experiments that showed the capability of venous ECs to integrate into embryonic arteries during development, thereby adopting the arterial marker EphrinB2 [83]. Aberrant expression of the arterial marker EphrinB2 was observed in ECs of venous malformations [84]. Differentiation of embryonic arteries is accomplished by high expression of members of the Notch signaling pathway (Notch1, Notch4, Jag1, Jag2, Dll4, Hey2) and the upstream regulators sonic hedgehog (Shh) and VEGF-A.

Development of veins had been regarded as kind of a default pathway of vessel differentiation; however, the transcription factor COUP-TFII (nr2f2), an orphan nuclear receptor, was shown to induce venous specification. Disruption of its expression in murine ECs results in the upregulation of the arterial markers EphrinB2, Jag1, Notch1, and NP1 [80]. Development of veins is the prerequisite for the subsequent development of lymphatics.

Due to the pressure produced by the heart, 20–30 l of plasma leak each day from the blood vessels into the interstitium [85]. Ninety percent of the extravasated fluid is reabsorbed into the blood at the venous end of the capillaries and the post-capillary venules due to osmotic forces [86]. The remaining 10 % is taken up by initial lymphatics (lymphatic capillaries) and drained centripetally in lymphatic precollectors, collectors, and trunks [1]. Initial lymphatics are formed by only one cell type: lymphatic endothelial cells (LECs). A continuous basal lamina and pericytes are missing [87]. The diameter of initial lymphatics measures about 50 μm and is significantly larger than that of blood capillaries. LECs of initial lymphatics are interconnected by overlapping junctions, which, more appropriately, should be named “overlapping leaflets,” as they represent valves (first valve system), which allow the influx of fluid and hinder the efflux [88].

Precollectors measure about 100 μm in diameter. Their intermediate morphology still allows uptake of interstitial fluid, but occasionally they possess a muscular wall and a “second valve system.” This is made up of leaflets, formed by the intima in a comparable manner as in the veins. Next, collectors of up to 600 μm in width transfer lymph centrally. They possess numerous valves. Their wall is made up of three layers: intima, media, and adventitia. As a result of endogenous pacemaker cell activity and vegetative innervation, the nonstriated myocytes of the media contract spontaneously approximately 6–12 times per minute, which is of great importance for the lymph flow [2]. Collectors often drain the lymph over long distances (30–40 cm) into the first regional lymph node. Via lymphatic trunks and lymphatic ducts, such as the thoracic duct, the lymph is transported into the jugulo-subclavian junction [89].

In the human, the heart starts pumping around the 21st developmental day, when blood vessels are already present [90]. Development of lymphatics starts considerably later. Lymph sacs, which are the first lymphatic anlagen detectable by routine histology, have been found in embryos of 10–14 mm total length, corresponding to developmental week 6 or 7 [91]. However, with specific markers LECs can be detected in earlier stages of development. Studies on mammalian embryos have shown that there are three paired and two unpaired lymph sacs, which will later give rise to primary lymph nodes [92]. The consecutive development of BECs and LECs has led to the hypothesis that LECs are derived from BECs, specifically from segments of adjacent veins [93]. This has been supported on the basis of the expression pattern of the homeobox transcription factor Prox1, which is indispensible for LEC development [94]. Direct proof for the development of LECs from venous ECs came from in vivo cell lineage tracing studies in zebrafish embryos. Cells delaminating from a parachordal vein assembled into the thoracic duct-homolog lymphatic vessel [95]. Budding and migration of lymphangioblasts from multiple vessels was observed during development of facial lymphatics using similar techniques in zebrafish [96]. In mice, too, the major source for the developing lymphatics is the embryonic veins [97]. Delamination from the endothelial lining and migration of mesenchymal lymphangioblasts, in combination with sprouting phenomena, then give rise to early lymphatic networks [98, 99]. In larger animals, e.g., chicken, indications for direct differentiation of LECs from mesenchymal lymphangioblasts have been found [100]. In the human, circulating lymphangioblasts were found to contribute to inflammation-induced lymphangiogenesis in patients suffering from kidney transplant rejection [101], but it is unknown if integration of mesenchymal lymphangioblasts is an important mechanisms of embryonic lymphangiogenesis in the human. Growth and expansion of lymphatics is preferentially along embryonic arteries, which obviously secret lymphangiogenic growth factors effectively, as well as chemokines, which orchestrate the patterning of lymphatic trunks [102].

Commitment of LECs starts in specific segments of early embryonic veins, probably induced by adjacent mesenchymal cells and the regulation of retinoic acid concentrations [103]. Committed LECs express the transcription factor SOX18 [104], VEGFR-3, and the hyaluronan receptor Lyve-1 [105]. Upregulation of the homeobox transcription factor Prox1, which is a master regulator of LEC development, is a crucial step for the development of the lymphatic vascular network [94], and maintenance of Prox1 expression is found in lymphatics of aged humans [106]. Transcriptional profiling of LECs identified a large number of genes with high specificity for LECs, such as PDPN (podoplanin), RELN (reelin), WNT5A, and others [107–109]. High levels of VEGFR-3 expression were found in LECs of lymphangiomas [110], showing that fine tuning of the VEGF-C signaling is of major importance for tissue specific forming of lymphatics. This is underlined by the observation of lymphatic hyperplasia in mice after downregulation of soluble VEGFR-2 (esVEGFR-2), the endogenous inhibitor of VEGF-C signaling [111]. A soluble splicing variant of VEGFR-3 (esVEGFR-3) with anti-lymphangiogenic activity has been found recently [112]. The majority of familial cases of lymphedema seem to be due to mutations in the VEGFR-3 gene [113]. Molecules that regulate LEC development and the differentiation of lymphatic networks are shown in Fig. 1.6, taken from the review by Tammela and Alitalo [105]. For a detailed description of the molecular interactions, we recommend reading of this review. Not only cell-cell interactions but also the ECM controls lymphangiogenesis. Mutations of the CCBE1 protein (collagen and calcium-binding EGF domain-containing protein 1), which seems to control the composition of the ECM, cause lymphatic dysplasia in man, and functional analyses in animals have confirmed the importance of CCBE1 in embryonic lymphangiogenesis [114]. Important interactions of LECs with the ECM are mediated by integrin receptors, predominantly by integrin α9, which forms heterodimers with integrin β1. Inactivation of integrin α9 causes fatal chylothorax in mice [115].