Experimental analysis of blood vessel development in the avian wing bud.

Shortly after its appearance, the avian limb bud becomes populated by a rich plexus of vascular channels. Formation of this plexus occurs by angiogenesis, specifically the ingrowth of branches from the dorsal aorta or cardinal veins, and by differentiation of endogenous angioblasts within limb mesoderm. However, mesenchyme located immediately beneath the surface ectoderm of the limb is devoid of patent blood vessels. The objective of this research is to ascertain whether peripheral limb mesoderm lacks angioblasts at all stages or becomes avascular secondarily during limb development. Grafts of core or peripheral wing mesoderm, identified by the presence or absence of patent channels following systemic infusion with ink, were grafted from quail embryos at stages 16-26 into the head region of chick embryos at stages 9-10. Hosts were fixed 3-5 days later and sections treated with antibodies that recognize quail endothelial cells and their precursors. Labeled endothelial cells were found intercalated into normal craniofacial blood vessels both nearby and distant from the site of implantation following grafting of limb core mesoderm from any stage. Identical results were obtained following grafting of limb peripheral mesoderm at stages 16-21. However, peripheral mesoderm from donors older than stage 22 did not contain endothelial precursors. Thus at the onset of appendicular development angioblasts are present throughout the mesoderm of the limb bud. During the fourth day of incubation, these cells are lost from peripheral mesoderm, either through emigration or degeneration.     

Endogenous origin of the lymphatics in the avian chorioallantoic membrane.

The lymphatics of the intestinal organs have important functions in transporting chyle toward the jugulosubclavian junction, but the lymphangiogenic potential of the splanchnic mesoderm has not yet been tested. Therefore, we studied the allantoic bud of chick and quail embryos. It is made up of endoderm and splanchnic mesoderm and fuses with the chorion to form the chorioallantoic membrane (CAM) containing both blood vessels and lymphatics. In day 3 embryos (stage 18 of Hamburger and Hamilton [HH]), the allantoic mesoderm consists of mesenchymal cells that form blood islands during stage 19 (HH). The endothelial network of the allantoic bud, some intraluminal and some mesenchymal cells express the hemangiopoietic marker QH1. The QH1-positive endothelial cells also express the vascular endothelial growth factor receptor-3 (VEGFR-3), whereas the integrating angioblasts and the round hematopoietic cells are QH1-positive/VEGFR-3-negative. The ligand, VEGF-C, is expressed ubiquitously in the allantoic bud, and later predominantly in the allantoic epithelium and the wall of larger blood vessels. Allantoic buds of stage 17-18 (HH) quail embryos were grafted homotopically into chick embryos and reincubated until day 13. In the chimeric CAMs, quail endothelial cells are present in blood vessels and lymphatics, the latter being QH1 and VEGFR-3 double-positive. QH1-positive hematopoietic cells are found at many extra- and intraembryonic sites, whereas endothelial cells are confined to the grafting site. Our results show that the early allantoic bud contains hemangioblasts and lymphangioblasts. The latter can be identified with Prox1 antibodies and mRNA probes in the allantoic mesoderm of day 4 embryos (stage 21 HH). Prox1 is a specific marker of the lymphatic endothelium throughout CAM development.     

Angioblast differentiation is influenced by the local environment: FGF-2 induces angioblasts and patterns vessel formation in the quail embryo.

The embryonic vasculature forms by the segregation, migration, and assembly of angioblasts from mesoderm, a process termed vasculogenesis. The initial role of fibroblast growth factor 2 (FGF-2) in vascular development appears to be in the induction of endothelial precursors, angioblasts. Quail somites transplanted into chick embryos will give rise to angioblasts of quail origin. The number of angioblasts present within the chimera is dependent on the host environment. Angioblast induction can be demonstrated in vitro by the addition of FGF-2 to cultures of dissociated somitic mesoderm, as assessed by QH-1 epitope expression. Manipulation of FGF-2 concentration in the quail/chick chimeras by FGF-2 peptide or neutralizing antibody injections increases or decreases angioblast induction in the predicted manner. To better control growth factor release in vivo we have implanted beads that release FGF-2 into the embryonic environment. FGF-2 beads implanted into the somite induce angioblast differentiation in the epithelial somite; whereas, beads lateral to the somitic mesoderm induce the formation of ectopic vessels. These studies suggest that FGF-2 is important for both the induction of angioblasts and the assembly of angioblasts into the initial vasculature pattern.     

Experimental analysis of the origin of the wing musculature in avian embryos

Summary Interspecific grafts of somites, as well as parts of the somatic plate mesoderm, have been made between quail and chick embryos (stages 12–14 H.H.) at the level of the prospective wing bud in order to examine the relationship between somites and wing bud myogenesis. The stability of the natural quail nuclear labelling makes it possible to follow the developmental fate of grafted mesodermal cells in the host embryo. Embryos examined after subsequent incubation periods of 3–7 days show the following distribution of somatic and somitic cells within the wing bud: as soon as the three zones of different cell density within the mesoderm can be distinguished, cells of somitic origin are limited to the prospective myogenic are which is made up of a mixed population of somatic and somitic cells, whereas the prospective chondrogenic area as well as the subectodermal zone only consists of cells originated from the somatic plate mesoderm. After further incubation, single muscle blastema are present which were also seen to be a mixture of somatic and somitic cells. The cells of muscular bundles are of somitic origin, while the muscle connective tissue cells are derived from the somatic plate mesoderm. After grafting into the coelomic cavity or on the chorio-allantoic membrane, fragments of the somatic plate mesoderm previously isolated from quail embryos (stage 14 H.H.) at the level of the prospective wing bud exhibit well developed skeletal elements, but fail to differentiate any musculature. These experimental investigations support previous evidence for a somitic origin of wing bud myogenic cells. Histological and scanning electron microscopic studies of the brachial somites and the adjacent somatic plate mesoderm of chick embryo (stages 13–15 H.H.) reveal that migration of still undifferentiated somitic cells into the brachial somatic plate mesoderm begins to take place in embryos at stage 14.     

Study of the origin of connective tissue sheaths of peripheral nerves in the limb of avian embryos

Summary Quail-chick and chick-quail chimeras were constructed by grafting, isotopically, the limb bud of quail embryos into a chick of the same developmental stage and vice versa, prior to the entry of nerve fibres into the limb.After 5–14 days reincubation of the embryos, the components of the connective tissue sheaths of the peripheral nerves were observed by using Feulgen-Rossenbeck staining and light microscopy, in order to distinguish quail cells and chick cells.In all the chimeras studied, the connective tissue sheaths of peripheral nerves (the epineurium, perineurium, perineural septa and endoneural fibroblasts) were formed from the mesenchyme of the limb bud, while Schwann cells were of host origin. Also the outer and inner capsule of muscle spindles originated from the limb bud mesenchyme.These experiments suggest that the connective tissue sheaths of peripheral nerves (at least in the limb region of avian embryos) are not of neural crest origin, but are formed from limb bud mesenchyme.     

Distribution and migration of angiogenic cells from grafted avascular intraembryonic mesoderm

Summary The hemangiogenic potencies of initially avascular intra-embryonic mesoderm were studied in chick and quail embryos and in chick-quail chimeras. The prechordal mesoderm, primitive streak and primitive node of quail embryos were heterospecifically grafted into limb buds of chick embryos. Hemangiopoietic quail cells in the host limb were detected by immunohistological staining with the monoclonal anti-MB-1 antibody after 3–6 days of re-incubation. The antibody is specifically directed against quail hemangiopoietic cells and their derivatives. Quail endothelial cells were found in pure quail and in chimeric vessels, inside as well as outside the graft. The main artery of the limb and the vessels inside the graft were connected by chimeric arteries. Proximal to the graft, quail endothelial cells were located predominantly within the lining of the main artery, while distally they were found mainly in the veins and the marginal sinus. The results show that, as early as stage 3 (according to Hamburger and Hamilton 1951, HH) all parts of the avascular intraembryonic mesoderm tested, give rise to endothelial cells. Both mechanisms, angiogenesis and vasculogenesis, contribute to the vascularization of the limb. Immunocytological and scanning electron microscopic studies indicate that centrifugal and centripetal migration of angiogenic cells occurs outside the vessels as well as on the inner surface of the endothelium.Supported by the Deutsche Forschungsgemeinschaft (CH 44/9-1)     

Morphogenesis of blood vessels in the head muscles of avian embryo: spatial, temporal, and VEGF expression analyses.

Adult skeletal muscle is a highly vascularized tissue, but the development of intramuscular endothelial networks has not been well studied. In quail embryos, QH1-positive angioblasts are present and moving throughout myogenic head mesoderm before the onset of primary myotube formation. On day 5 of incubation, concurrent with early myotube formation and aggregation, angioblasts establish a transient vascular plexus surrounding the myogenic condensations. Between days 5 and 9, the intramuscular vessels form an irregular network of endothelial cords and patent channels and only later are the parallel arrays of capillaries characteristic of adult muscles established. Microinjections using India ink, QH1, and Mercox resin reveal that these intramuscular capillaries are typically not connected to systemic vessels of the head until day 10, which is near the end of primary myogenesis and corresponds to the onset of muscular function. Morphometric analyses performed during primary myogenesis stages show a decrease in muscle cell density but no significant changes in intramuscular vascular density between days 5 and 9. This finding was surprising, as it is generally assumed that muscle growth requires elevated oxygen and nutrient levels. Moreover, there are no significant morphometric differences in vascular supply to embryonic fast and slow muscles. Endothelial tissue density is similar in slow muscles (oculorotatory, e.g., lateral rectus), fast muscles (mandibular depressor), and mixed muscles, in which the fiber types can be interspersed (jaw adductors) or segregated (branchiomandibular). Vascular endothelial growth factor (VEGF) protein is abundant in myotubes but not endothelial cells within both fast and slow head muscles at days 7 and 9. However, in some mixed muscles, only a minority of myotubes, which do not correspond to one specific fiber type, express VEGF. These results document a dynamic set of intramuscular and perimuscular angiogenic reorganizations during avian head myogenesis. Thus far, no vasculogenic distinctions between fast and slow muscles have been observed, although muscle heterogeneity in VEGF expression is evident.     

Cytokinetic studies on the aortic endothelium and limb bud vascularization in avian embryos

Summary Cytokinetic studies on the aortic endothelium using the BrdU/anti-BrdU-method were carried out on 2.5– to 6-day chick and quail embryos. The mitotic activity of the aortic endothelium is related temporally to the age of the avian embryo and spatially to the embryonic region where the aorta originates. The mitotic activity of the aortic endothelium decreases with increasing age of the embryos. In the limb buds, however, the mitotic rate of the aortic endothelial cells increases independently of the age of the embryo. This increase in the mitotic activity of the aortic endothelium at the appropriate levels coincides with the vascularization of the outgrowing limb buds. We concluded therefore that the aortic endothelium probably supplies endothelial cells for the formation of limb vessels at this stage. Thus our results suggest that angiogenesis (sprouting of capillaries from pre-existing vessels) takes place during limb vascularization in avian embryos. On the other hand, immunohistochemical studies with QH-1 or MB-1 antibody show, beside a capillary network in the central core of the wing bud, individual immunolabelled cells of mesenchymal character within the primarily avascular subectodermal region from the onset of vascularization onwards. We suggest that these cells have partly to be regarded as endothelial precursor cells, which have differentiated in situ from the local limb mesenchyme, and which will contribute to the developing vascular plexus. This means that not only angiogenesis, but also vasculogenesis (in situ from mesenchymal precursors differentiated endothelial cells) appears to be involved in limb vessel formation.Supported by the Deutsche Forschungsgemeinschaft (Ch 44/9-1, Ch 44/9-2)This paper is dedicated to Prof. Dr. K.V. Hinrichsen on the occasion of his 65th birthday     

Antibodies to beta 1-integrins cause alterations of aortic vasculogenesis, in vivo.

Vasculogenesis is the de novo formation of blood vessels from mesoderm. This process occurs very early in development and provides a convenient system for studying morphogenesis in higher vertebrates. The cell-extracellular matrix (ECM) interactions that occur during dorsal aortic vasculogenesis were examined using the monoclonal antibody, CSAT, a reagent known to neutralize the ligand-binding activity of avian beta 1-integrins. We injected CSAT into quail embryos during a period of active vasculogenesis (4-10 somites). The CSAT antibodies, but not controls, had a marked and reproducible effect on aortic vessel formation. Vasculogenesis appeared to be arrested at the stage when slender cord-like assemblies of angioblasts rearrange to form tubules. Indeed, aortic primordia near the site of CSAT injection did not form patent vessels.     

The relationship between limb muscle and endothelial cells migrating from single somite

Somites contribute myogenic and endothelial precursor cells to the limb bud. Transplantations of single somites have shown the pattern of muscle cell distribution from individual somites to individual limb muscles. However, the pattern of the endothelial cell distribution from individual somites to the limb has not been characterized. We have mapped quail muscle and endothelial cell distribution in the distal part of the chick limb after single somite transplantation to determine if there is a spatial relationship between muscle and endothelial cells originating from the same somite. Single brachial somites from quail donor embryos were transplanted into chick embryos, and, following incubation, serial sections were stained with a quail-endothelial cell-specific monoclonal antibody (QH-1), an anti-quail antibody (QCPN) and an anti-desmin antibody to distinguish the quail endothelial and muscle cells from chick cells. Our results show that transplants of somite 16-21 each gave rise to quail endothelial cells in the wing. The anterioposterior position of the blood vessels formed by somitic endothelial cells corresponded to the craniocaudal position of the somite from which they have originated. Endothelial cells were located not only in the peri- and endomysium but also in the subcutaneous, intermuscular, perineural and periost tissues. There was no strict correlation between the distribution of muscle and endothelial cell from a single transplanted somite. Blood vessels formed by grafted quail endothelial cells could invade the muscle that did not contain any quail muscle cells, and conversely a muscle composed of numerous quail muscle cells was lacking any endothelial cells of quail origin. Furthermore, a chimeric limb with very little quail muscle cells was found to contain numerous quail endothelial cells and vice versa. These results suggest that muscle and endothelial cells derived from the same somite migrate on different routes in the developing limb bud.     

Early endocardial formation originates from precardiac mesoderm as revealed by QH-1 antibody staining

The formation of endocardial endothelium in quail embryos was investigated using in vivo and in vitro systems. At stage 7+ (2 somite), the initial emergence of endothelial cells within the bilateral heart forming region (HFR) was detected in quail embryos by immunohistochemistry with QH-1 (an anti-quail endothelial cell marker) and confocal microscopy. We consistently observed more QH-1 positive cells in the right HFR than the left. At stage 8 (4 somite), the HFR, including QH-1 positive cells, were located in the splanchnic mesoderm after formation of the coelom. During stage 8, the HFR migrated along the margin of anterior intestinal portal in association with the endoderm. By stage 8+ (5 somite), the two HFR had fused at the midline and formed a plexus of QH-1 positive endothelial precursor cells. The definitive endocardium developed as a single, hollow, tube within this plexus. Posteriorly, QH-1 positive cells of the HFR established vascular-like connections with QH-1 positive cells that had formed outside (peripheral to) the HFR. During migration and subsequent determination, the precardiac mesoderm is continuously associated with the basement membrane of the anterior endoderm. To determine the role of endoderm on endocardial endothelial cell formation and development, precardiac mesoderm from stage 5 embryos, which does not express QH-1 antigen, was explanted onto the surface of collagen gels. When co-cultured with endoderm, the outgrowth of free cells from the mesoderm was much more extensive, many of which invaded the gel and expressed the QH-1 antigen; mesoderm cultured without endoderm did not seed nor express QH-1 antigen. These findings suggest that the segregation of endothelial and myocardial lineages may occur by an endoderm-mediated, mesenchymal formation.     

A new model of vasculogenesis and angiogenesis in vitro as compared with vascular growth in the avian area vasculosa

In cultures of dissociated quail epiblast the basic constituents of the vascular system, blood cells and endothelial cells can be induced by basic fibroblast growth factor (Flamme and Risau, Development, 116:435–439, 1992). As we show here, in those cultures three types of vascular plexus differentiate spontaneously under different culture conditions: At the 3rd day a vascular plexus appears in situ closely resembling the vascular plexus of the quail area opaca vasculosa (vasculogenesis). Vascular sprouts are formed, extending long filopodia at their tips. Such filopodia are shown to build the first intervascular bridges in the growing vascular plexus of the area vasculosa at embryonic day 3. Connections of filopodia turn out to be precursors of new capillaries interconnecting pre-existing blood vessels (angiogenesis). Two further types of in vitro capillary plexus differentiate in long term endothelial cell cultures derived from induced angioblasts. Whereas one closely resembles so-called angiogenesis in vitro, the third type comprises mainly multinucleated giant endothelial cells lining loop like capillaries and represents a differentiation of aging endothelial cell culture. Thus, the present in vitro model is an approach to the sequence of angioblast induction, vasculogenesis, and angiogenesis. © 1993 Wiley-Liss, Inc.     

The lymphatic endothelium of the avian wing is of somitic origin.

It has recently been shown that there are lymphangioblasts in the early avian wing bud, but fate map studies on the origin of these cells have not yet been performed. The lymphatics in the wings of 10-day-old chick and quail embryos are characterized by both the position along with all major blood vascular routes and by the Vascular Endothelial Growth Factor Receptor-3 (VEGFR-3) expression. In the quail, the endothelium of both blood vessels and lymphatics can be marked with the QH1 antibody. We have grafted the dorsal halves of epithelial somites of 2-day-old quail embryos homotopically into chick embryos. The grafting was performed at the wing level and the host embryos were reincubated until day 10. The chimeric wings were studied with the QH1 antibody alone and with double staining consisting of VEGFR-3 in situ hybridization and QH1 immunofluorescence. Our results show that in the wing the endothelium of both the blood vessels and the lymphatics is derived from the somites. QH1-positive endothelial cells form the vasculature of the chimeric wings. Chimeric lymphatics of the wing can be identified because of their typical position and their VEGFR-3 and QH1 double-positivity. This shows that not only the blood vascular cells but also the lymphatic endothelial cells of the avian wing are born in the paraxial/somitic mesoderm.     

Critical role for retinol in the generation/differentiation of angioblasts required for embryonic blood vessel formation.

Numerous studies demonstrate that vitamin A (retinol) deficiency causes abnormal cardiovascular morphogenesis. We evaluated the impact of retinol deficiency on the regulation of the numbers of endothelial cells and angioblasts (endothelial progenitors) produced during embryonic quail development. At the one-somite stage, there were no discernible differences in the mean number of endothelial cells or angioblasts in normal and retinol-deficient embryos. However, retinol-deficient embryos at the three-somite stage had an increase in the mean number of endothelial cells but no difference in the mean number of angioblasts. By contrast, retinol-deficient embryos at the five-somite stage have 61% of the normal number of endothelial cells and 12% of the normal number of angioblasts. Similarly, retinol-deficient embryos at the 10-somite stage had 71% and 60% of normal numbers of endothelial cells in capillary-like networks and the sinuses venosus, respectively. Furthermore, we show that retinol deficiency did not elicit a global reduction in mesodermal cell numbers but was specific to cells of the endothelial lineage. Taken together, our findings suggest that vascular abnormalities observed under conditions of retinol deficiency are due to reduction in the number of angioblasts and consequently an insufficiency in the number of endothelial cells required to build complex vascular networks.     

Origin of spinal cord meninges, sheaths of peripheral nerves, and cutaneous receptors including Merkel cells

Summary The origin of cells covering the nervous system and the cutaneous receptors was studied using the quail-chick marking technique and light and electron microscopy. In the first experimental series the brachial neural tube of the quail was grafted in place of a corresponding neural tube segment of the chick embryo at HH-stages 10 to 14. In the second series the leg bud of quail embryos at HH-stages 18–20 was grafted in place of the leg bud of the chick embryos of the same stages and vice versa. It was found that all meningeal layers of the spinal cord, the perineurium and the endoneurium of peripheral nerves, as well as the capsular and inner space cells of Herbst sensory corpuscles, develop from the local mesenchymal cells. Schwann cells and cells of the inner core of sensory corpuscles are of neural crest origin. The precursors of Merkel cells migrate similarly to the Schwann cells into the limb bud where they later differentiate. This means that in addition to the Schwann cells and the melanocytes a further neural crest-derived subpopulation of cells enters the limb.     

Preservation of the ability of dissociated quail wing bud mesoderm to elicit a position-related differentiate response

Previous studies showed that grafting wedges of fresh or cultured anterior quail wing mesoderm into posterior slits in chick wing buds resulted in the formation of supernumerary cartilage in a high percentage of cases. When anterior quail mesoderm, which had been dissociated into single cells and pelleted by centrifugation, was grafted into posterior slits of host chick wing buds, supernumerary rods or nodules of cartilage formed in 74.3% of the cases. Few supernumerary skeletal structures formed following control operations in which pelleted dissociated anterior or posterior mesoderm was grafted into homologous locations in host chick wing buds. When pelleted, dissociated anterior mesoderm was cultured in vitro for 1 or 2 days prior to being implanted in posterior locations, the incidence of supernumerary cartilage formation increased to 95.5% and 93.8%, respectively. The incidence of supernumerary cartilage formation following control orthotopic grafts of cultured mesoderm was 11.8% for 1-day and 31% for 2-day cultured anterior mesoderm; for 1- and 2-day cultured posterior mesoderm, the incidence of supernumerary cartilage formation was 20% and 41.7%, respectively. Longer-term culture resulted in a substantial decrease in the percentage of supernumerary cartilage after anterior to posterior grafts and an increase in the incidence of supernumerary cartilage from control grafts. The results demonstrate that quail anterior wing bud mesodermal cells do not need to maintain constant contact with one another in order to retain the ability to form or stimulate the formation of supernumerary cartilage after being grafted into a posterior location in a host wing bud. This ability is retained when the pelleted dissociated mesoderm is cultured in vitro outside the limb field for at least 1 to 2 days.     

Localized vascular regression during limb morphogenesis in the chicken embryo: II. Morphological changes in the vasculature

The regression of blood vessels in the distal wing bud of chicken embryos from stages 19 to 31 was examined by light and electron microscopy. The vessels were double-labelled by an injection of Monastral blue B (MB) to label the regressing endothelial cells, followed 6-48 hours later with an injection of India ink which marked the lumens of patent vessels. Prior to stage 26 the vessels contained only India ink since the endothelial cells were not phagocytic at this stage. Vessels at stage 26 or later were often double-labelled, with MB sequestered in the endothelial cell cytoplasm and India ink in the vessel lumens. After stage 27 cells not associated with lumens, but labelled with MB, were observed in areas undergoing vascular regression. Ultrastructural changes in the endothelial cells as the vessels regressed included formation of luminal and abluminal processes, long complex junctions, and vacuoles containing MB. In many involuting vessels the endothelial cells appeared normal even though the lumens were collapsed. Occasionally, isolated pyknotic cells were observed in regions that had been previously vascularized. At stage 31 cells in the developing cartilage had vacuoles containing MB. Our study suggests that blood vessels may disappear from the prechondrogenic zone of the distal wing bud by several mechanisms. These could include a type of cell death that does not elicit a cellular infiltrate, migration of the endothelial cells away from vascularized regions, and/or transdifferentiation into cells that resembled chondrocytes.     

Pulmonary veins in the chick embryo: Origin as determined by radioautographic mapping

In the head process to early head fold stage chick embryo, the cells which will form the pulmonary veins are located in the mesoderm near the posteromedial edges of the heart-forming regions. At the 26 somite to early limb bud stage, the presumptive pulmonary vein cells have been folded to the midline of the embryo as part of the splanchnic mesoderm and form the endothelial plexus which courses through the dorsal mesentery of the sinoatrial region of the heart.     

Vascularization of embryonic adrenal gland grafted onto chorioallantoic membrane

 Vascularization and endothelial phenotype expression were analysed in embryonic adrenal tissue grafted onto chorioallantoic membrane (CAM), by means of routine light microscopy and immunocytochemical staining, and of electron microscopy. Adrenal gland tissue from chick or quail embryos (donors) was grafted onto CAMs of chick or quail embryos (host). Vessels of chick origin were discriminated from those of quail origin by monoclonal antibodies, anti-MB1, specific for quail endothelial and haemopoietic cells, and QCPN, which labels quail cell nuclei. Vessels of adrenal type were distinguished from those of CAM-type by their ultrastructural endothelial phenotype – porous in the former and continuous in the latter. The observations carried out 6 days after implantation indicate that the adrenal gland develops and differentiates according to a virtually normal histological pattern. As regards the adrenal and CAM vascularization, the grafting procedure elicits angiogenic events consisting in the formation of peripheral anastomoses between the graft and the CAM original microvasculature and in new-growth of vessels from the CAM into the grafted tissue and vice versa. As to the endothelial phenotype, the ultrastructural results demonstrate that besides its own native vasculature, the adrenal tissue contains vessels with continuous endothelium and the CAM mesenchyme is supplied by adrenal-type, fenestrated vessels. Accepted: 27 April 1998