Distal angiogenesis: a new concept for lung vascular morphogenesis

Marta Canis Parera, Marieke van Dooren, Marjon van Kempen, Ronald de Krijger, Frank Grosveld, Dick Tibboel, Robbert Rottier


Although several molecular players have been described that play a role during the early phases of lung development, it is still unknown how the vasculature develops in relation to the airways. Two opposing models describe development of lung vasculature: one suggests that both vasculogenesis and angiogenesis are involved, whereas the second describes vasculogenesis as the primary mechanism. Therefore, we examined the development of the murine pulmonary vasculature through a morphological analysis from the onset of lung development [9.5 days postcoital (dpc)] until the pseudoglandular stage (13.5 dpc). We analyzed fetal lungs of Tie2-LacZ transgenic mice as well as serial sections of wild-type lungs stained with endothelial-specific antibodies (Flk-1, Fli-1, and PECAM-1). Embryos were processed with intact blood circulation to maintain the integrity of the vasculature; hence individual vessels could be identified with accuracy through serial section analysis. Furthermore, circulating primitive erythrocytes, formed exclusively by the blood islands in the yolk sac, are trapped in vessels during fixation, which proves the connection with the embryonic circulation. We report that from the first morphological sign of lung development, a clear vascular network exists that is in contact with the embryonic circulation. We propose distal angiogenesis as a new concept for early pulmonary vascular morphogenesis. In this model, capillary networks surround the terminal buds and expand by formation of new capillaries from preexisting vessels as the lung bud grows. The fact that at an early embryonic stage a complete vascular network exists may be important for the general understanding of embryonic development.

  • mouse
  • bronchial and pulmonary system

the adult lung has a dual vascular system, the bronchial and the pulmonary system. The bronchial system oxygenates all nonrespiratory structures of the lung, whereas the pulmonary system transports deoxygenated blood to the alveoli for gas exchange. The pulmonary arteries arise from the pulmonary trunk of the heart and closely follow the bronchial tree, giving rise to the alveolar capillary plexus. These capillaries drain into the pulmonary veins, which run through the connective tissue septa back to the left atrium of the heart. Type I alveolar cells and endothelial cells form an air-blood barrier, which is mandatory for optimal gas exchange. Abnormalities in this delicate architecture may lead to inadequate function, as presented in congenital anomalies of the lung, such as congenital diaphragmatic hernia (32) and alveolar capillary dysplasia (1). The intimate anatomical structure suggests that the close interaction between the pulmonary vessels and the airways starts to be established early in development. However, the process by which the pulmonary vascular tree develops and the factors that control pulmonary vascular development are not completely understood.

Lung development is divided into distinctive stages with the earliest stages consisting of the embryonic stage, embryonic day (E) 9–10 in mice and 4–6 wk in humans, and the pseudoglandular stage, E10–16 in mice and 6–16 wk in humans. Models of pulmonary vascular morphogenesis at these early stages are derived from morphological data. Based on vascular casts and electron microscopy of murine lungs, deMello et al. (5) suggested that two processes are involved in the formation of the pulmonary vessels: the central vessels are formed by angiogenesis, defined as branching of new vessels from preexisting ones, and peripheral vessels by vasculogenesis, defined as development of blood lakes in the mesenchyme. A connection between the central and peripheral vascular lumen would be established through a lytic process around E13/14, and pulmonary circulation would start. A comparative analysis of serial sections of human embryos suggested that the same processes would also occur in humans (4). In addition, they concluded that pulmonary arteries and veins were dissociated in their timing and pattern of branching, since “distal veins are present throughout the mesenchyme and establish a central luminal connection with the main pulmonary vein before an airway or artery is present at the same level” (4).

Although it is generally accepted that the distal vasculature arises by vasculogenesis, recent morphological studies have questioned the basic mechanism of formation of the proximal pulmonary vessels. Using heterozygous mice with a targeted insertion of the bacterial lacZ gene into the flk-1 locus, Schachtner et al. (30) showed that the proximal and distal vascular structures were already connected at gestational age 10.5. However, only the proximal portion of the pulmonary artery contained a lumen (30). They also demonstrated that lung vessel development occurred at all stages and directly corresponded to overall lung growth. In another study, Hall et al. (11) used lungs from human embryos to stain serial sections with endothelial-specific antibodies and showed continuity of circulation between the heart and the distal lung vascular plexus from 38 days of gestation onward. They concluded that the intrapulmonary arteries originate from a continuous expansion and coalescence of the primary capillary plexus that would form by vasculogenesis during the pseudoglandular stage. In addition, they showed that the same mechanism takes place to form the pulmonary veins (10). In contrast to the definition of deMello and coworkers (4, 5), they defined vasculogenesis as differentiation of angioblasts from mesoderm to form primitive blood vessels, without the formation of hematopoietic lakes.

In spite of the lack of consensus of how the lung vasculature develops, many molecular players involved in blood vessel formation are already identified. Epithelial cells from the lung bud are suggested to induce the expansion of the capillary plexus through vascular growth factors (14). Three different growth factor systems have been described to act via endothelial cell-specific tyrosine kinases: VEGFs, angiopoietins and ephrins (37). VEGF is required for vasculogenesis and angiogenesis, and VEGF isoforms are expressed in lung epithelial (14, 18) and mesenchymal cells (8, 9). Furthermore, lung endothelial cells express Flk-1, the receptor for VEGF-A (8, 30), and in vitro experiments showed that VEGF has a potential role in lung vascular morphogenesis (14, 38). Tie2 and its ligand Ang-1 play a role in the regulation of angiogenesis (37). Gene disruption of either Tie2 or Ang-1 in mice leads to the formation of an abnormal vascular network with immature vessels that lack proper organization (28, 34). It is likely that these factors are involved in the stabilization of the network rather than its initial formation. Colen et al. (3) demonstrated expression of Ang-1 and Ang-2 in the mouse lung from E9.5 onward.

Because two opposing models on lung vessel development exist, we have performed an ontogenic morphological study of the lung vasculature in relation to the airways in mice ranging from E9.5, when the lung starts to become morphologically discernible, until the midphase of the pseudoglandular stage at E13.5. We performed analysis of fetal lungs from transgenic mice expressing the bacterial lacZ gene under the control of the Tie2 promoter (Tie2-LacZ) as well as serial section analysis of normal lungs using antibodies against platelet endothelial cell adhesion molecule (PECAM)-1, Friend leukemia integration site 1 (Fli-1), and α-smooth muscle (SM) actin to specifically identify endothelial cells and blood vessels. Mouse embryos were processed to keep the blood circulation intact, thereby maintaining the vascular tone and the integrity of the vasculature. Hence, individual vessels could be identified through serial section analysis, and their origin and connections could be described with accuracy. Furthermore, circulating cells, which are primitive erythrocytes formed by the blood islands in the yolk sac at the gestational age we investigated, are trapped in vessels during fixation, and this proves that these vessels are connected with the embryonic circulation. We report that even the earliest vessels formed in the lung are already connected with the heart vascular structures and thus are part of the embryonic circulation. Because our findings are not consistent with the current models, we propose distal angiogenesis as a new model for lung vascular morphogenesis.


Embryos from wild-type FVB and Tie2-LacZ mice (31) were isolated between gestational age E9.5 and E13.5, if the morning of the vaginal plug is considered as E0.5. All animal experiments were performed according to national and institutional guidelines. To avoid vessels from collapsing and to maintain circulation, we placed the embryos within an intact yolk sac and placenta in ice-cold PBS while dissecting the embryo for 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-gal) staining or for paraffin embedding. For X-gal staining, the partially dissected embryonic thoraxes (E9.5, E10.5, E11.5, E12.5, and E13.5) were rinsed in PBS, placed in fixative for 30 min [1% paraformaldehyde (PFA), 2 mM MgCl2, 5 mM EGTA in PBS] at room temperature, and rinsed twice in washing solution (2 mM MgCl2 and 1 mM EGTA in PBS). The thoraxes were stained overnight with 1 mg/ml X-gal in 5 mM K3Fe(CN)6, 5 mM K4Fe(CN)6·3H2O, and 2 mM MgCl2 at 4°C. After being rinsed in PBS, tissues were postfixed for 2 h in 4% PFA, followed by complete dissection of the lungs. After imaging of the whole mount lungs, the X-gal-stained transgenic lungs were embedded and sectioned to support the determination of the structures in the lung.

For paraffin embedding, the yolk sac and placenta were removed and the embryos were fixed in 4% PFA for 30 min (E9.5), 45 min (E10.5), 2 h (E11.5) at room temperature, or overnight (E12.5 and E13.5) at 4°C. After two PBS washes, the tissue was processed for paraffin embedding. Embryos were placed in coronal or transversal orientation, completely sectioned, and the 4-μm-thick sections were used for hematoxylin and eosin staining and immunohistochemistry with antibodies raised against PECAM-1 (rat monoclonal, MEC13.3, 1:50, BDPharMingen) and Fli-1 (rabbit polyclonal, C-19, 1:1,000, Santa Cruz Biotechnology) as endothelial cell markers. PECAM-1 is a 130-kDa transmembrane glycoprotein expressed on the surface of endothelial cells (6), and Fli-1 is a 50-kDa ETS domain-containing transcription factor expressed in hemangioblasts, angioblasts, and (early) endothelial cells (27). We also used an antibody against α-SM actin (mouse monoclonal, 1A4, 1:400, NeoMarkers) to assess the muscularization of the vascular and airway walls. Before incubation with the primary antibody, the sections were dewaxed and endogenous peroxidase was blocked by incubation in 3% hydrogen peroxidase in methanol for 20 min. Antigen unmasking was performed with trypsin treatment (1.25 mg/ml for 5 min at room temperature) for PECAM-1 and α-SM actin and with microwave treatment in 10 mM citric acid buffer (pH adjusted to 6.0, 9 min at 450 W) for Fli-1. Sections were blocked with 5% BSA in PBS for 30 min and incubated with primary antibody diluted in 5% BSA in PBS overnight at 4°C. As secondary antibody, we used rabbit anti-rat IgG-peroxidase (Dako) for PECAM-1, goat anti-rabbit IgG-peroxidase (Dako) for Fli-1, and goat anti-mouse IgG-peroxidase (Dako) for α-SM actin, all diluted 1:100 in 5% BSA in PBS for 2 h at room temperature. Antibody binding was detected by 3,3′-diaminobenzidine, and slides were counterstained with hematoxylin.


We divided the pulmonary vasculature in three components, the afferent, the effective, and the efferent. The afferent component comprises the vessels that are proximally continuous with the vascular outflow of the heart and distally with the capillary network, which we define as the effective component. This effective component is in close contact with the epithelium of the terminal bud of the airway. The pulmonary veins that transport the blood from the periphery of the lung back to the heart form the efferent component. The proximal part of the primitive pulmonary veins is continuous with the atrial structures and develops into the dorsal mesocardium. We identified specific structures and vessels throughout the lung with adequate precision by combining whole mount analysis of X-gal-stained transgenic lungs and serial sections stained with specific antibodies. Stained transgenic lungs were analyzed and subsequently sectioned to validate our conclusions based on the whole mount analysis. The expression of the bacterial lacZ gene under the control of the Tie2 promoter in the endothelial cells of the Tie2-LacZ transgenic strain phenocopies the endogenous Tie2 expression; therefore, these cells are considered to be Tie2 positive (31). Sections of wild-type murine embryonic lungs were stained with antibodies against antigens specific for endothelial cells (PECAM-1 and Fli-1) or SM cells (α-SM actin). Fli-1 is an important regulator of Tie2 expression in the mouse embryo, since the Tie2 promoter contains a cluster of ETS binding sites (29), and disruption of the Fli-1 gene in mice leads to a specific downregulation of the Tie2 gene (13). We have analyzed a minimum of five murine embryos for each gestational age, and representative pictures are displayed in the figures.

The primitive gut originates after the completion of gastrulation when a crescent layer of endodermal cells starts folding to form a tube. The foregut, which is the anterior part of this gut-like structure, gives rise to a number of organs, like the thyroid glands and lungs. Immediately caudal to the fourth pharyngeal arch, the first morphological sign of the lung starts to become visible at E9.5 when a cluster of cells buds from the ventral site of the foregut and invades the surrounding splanchnic mesenchyme. Subsequently, this lung bud grows and splits into the prospective left and right lobes, running alongside the future esophagus. The dorsal mesocardium, or heart stalk, connects the atrial myocardial wall with the splanchnic mesenchyme ventral to the foregut (36). We found that already at E9.5–10, the lung vasculature is part of the splanchnic plexus surrounding the developing esophagus and airways (Fig. 1, A and F). The maintenance of the blood circulation while fixing the embryos made it easy to define and trace the vessels throughout the serial sections, because they appeared as an open structure and contained circulating primitive erythrocytes. The afferent and efferent components form a Tie2-, PECAM-1-, and Fli-1-positive plexus of capillaries that surrounds the proximal foregut (Fig. 1, A–C) and is continuous with the developing aortic arches and dorsal aorta (Fig. 1A). A capillary network of Tie2 (Fig. 1A)-, PECAM-1 (Fig. 1B)-, and Fli-1 (Fig. 1C)-positive endothelial cells surrounds the two airway buds and is in close contact with the epithelial cells of the airway (Fig. 1, D and E). Primitive erythrocytes, exclusively produced by the blood islands of the yolk sac, are frequently observed within the main vessels, the dorsal aorta and heart structures (Fig. 1C), and the capillaries, indicating that even the smallest vessels are connected to the embryonic circulation (Fig. 1, D and E). The venous confluence of this network runs through the dorsal mesocardium (Fig. 1C) and forms an invagination at the entrance of the atrium, described as the pulmonary pit (36).

Fig. 1.

Representative pictures of embryonic lungs at day 9.5 (A–E), 10.5 (F–I), and 11.5 (J–O). A: Tie2-driven LacZ expression in whole mount transgenic lung shows continuity of the primitive lung vasculature with the aortic arch (arrowhead) and the dorsal aorta (arrow). The network of small capillaries surrounding the growing lung bud is encircled. B: platelet endothelial cell adhesion molecule (PECAM)-1 staining of a coronal section showing the lining of the vasculature. C: Fli-1 staining of a transversal section. The efferent vessels (EV) run through the dorsal mesocardium (DM) into the heart structure. The arrow indicates the pulmonary pit. Details of PECAM-1 staining (D) and Fli-1 staining (E) shows the presence of primitive erythrocytes (PE, arrow) within the vessels of the splanchnic plexus adjacent to the airway (A). F: Tie2-driven LacZ expression in whole mount transgenic lung shows continuity of the primitive lung vasculature with the sixth aortic arches (arrowheads) and the network of capillaries surrounding the lung bud (arrows). G: PECAM-1 staining of a coronal section showing vascular structures alongside the airways filled with primitive erythrocytes (arrows). The afferent vessels (AV) form 2 plexiform networks that coalesce alongside the trachea (F and G). H: Fli-1 staining of a transversal section shows a plexus of capillaries draining through the DM into the heart. I: α-smooth muscle (SM) actin staining of a transversal section. The vessels of the lung, the airway, and the esophagus are not yet muscularized, whereas the dorsal aorta and the DM are clearly muscularized. J: Tie2-driven LacZ expression in whole mount transgenic lung. The right and left pulmonary arteries can now be identified as 2 vascular tubes that run alongside the trachea. Inset: details of the right and left aortic arches. K–O: coronal sections stained with PECAM-1 (K and L), Fli-1 (M), and α-SM actin (N and O). K is a section at the level of the imaginary line between ML and B1 (in J), L between RA and LA (in J), M through LL (in J), and N between ML and AL (in J). The common pulmonary vein is formed by several EVs from the right and left lung. α-SM actin staining of a coronal section shows that the proximal arteries and airways are being muscularized (N). AL, accessory lobe; B1 and B2, branch 1 and 2 of the left lung; CDL, caudal lobe; CL, cranial lobe; CV, common vein; Dao, dorsal aorta; E, esophagus; FG, foregut; LA, left pulmonary artery; LL, left lung; LSA, left sixth aortic arch; ML, middle lobe; PV, pulmonary vein; RA, right pulmonary artery; RL, right lung; RSA, right sixth aortic arch; T, trachea; Vao, ventral aorta. Bars, 100 μm (inset in J, A–C, F–N), 50 μm (O), and 20 μm (D and E).

At E10.5, the first airway branches are formed and the trachea is separated from the esophagus, and both are surrounded by a contiguous mesenchyme. The left and right lungs are clearly formed and positioned slightly curved dorsally at each side of the esophagus (Fig. 1F). The afferent vessels are not a defined vascular tube yet but resemble two plexiform networks that coalesce alongside the trachea (Fig. 1, F and G) and are continuous with the sixth aortic arches (Fig. 1F). A capillary vascular network surrounds both primitive bronchi (Fig. 1, F and G) and is continuous with the larger vessels, as has been concluded from the examination of serial sections. The efferent vessels form a plexus of capillaries that drain through the dorsal mesocardium into the common atrium of the heart (Fig. 1, F and H). The heart in Fig. 1F is removed to better illustrate the point that a vascular connection exists between the efferent vessels and the dorsal mesocardium of the heart, as shown in Fig. 1H. All vascular structures have a clear and open lumen filled with primitive erythrocytes, indicating their connection with the total embryonic circulation (Fig. 1G). Although the endothelial cells of these vessels are positive for Tie2, PECAM-1, and Fli-1, they are not mature as shown by the lack of α-SM actin-positive staining. In addition, the airways and esophagus also lack muscularization, whereas the dorsal aorta and dorsal mesocardium clearly are muscularized (Fig. 1I).

At E11.5, lung asymmetry has become obvious: the left lung has one branch, whereas the right lung has four main branches, the primordia of the cranial, the middle, the caudal, and the accessory lobe (Fig. 1J). The proximal afferent vessels can now be clearly identified as two vascular tubes that run alongside the trachea, the right and left pulmonary arteries (Fig. 1, J and L). Proximally, the right pulmonary artery is still connected to the right sixth aortic arch but has been lateralized toward the left (Fig. 1J, inset). The left sixth aortic arch, which will form the ductus arteriosus, is more obvious than the right sixth aortic arch, which eventually degenerates (Fig. 1J, inset). The vessels that surround the lung buds are positive for Tie2, PECAM-1, and Fli-1 (Fig. 1, J, K, and M) and are filled with primitive erythrocytes. The efferent vessels form a vascular tube, the common pulmonary vein (Fig. 1, J and K). SM cells start to enfold the proximal parts of the arteries, airways (Fig. 1, N and O), and, to a lesser extent, veins (data not shown). However, the growing part of the distal airways and surrounding vessels are not yet muscularized.

At E12.5, the esophagus and trachea are only attached by mesothelial linings (Fig. 2A, inset). The lung has a single left lobe and four clearly distinguishable right lobes (Fig. 2A). Each lobe of the lung has undergone further branching, which occurs in an axial fashion, increasing length and number of generations from the periphery to the center. The airway terminal buds are completely surrounded by a polygonal, irregular capillary plexus (Fig. 2, A and inset), which contains a lumen filled with primitive erythrocytes (Fig. 2B), indicating a direct and closed connection with the embryonic circulation. The endothelial cells of these vessels form a Tie2 (Fig. 2A)-, PECAM-1 (Fig. 2, C and D)-, and Fli-1 (Fig. 2E)-positive network close to the epithelial cells of the terminal bud. The proximal airways and vessels as well as the trachea and dorsal aorta are clearly α-SM actin positive, indicative for a maturation process. In contrast, the immature distal airways and vasculature are not yet muscularized (Fig. 2F).

Fig. 2.

Representative pictures of embryonic lungs at day 12.5 (A–F) and 13.5 (G–K). Tie2-driven LacZ expression in whole mount transgenic lung. The Tie2-positive capillary network is the effective component that wraps the terminal buds. The arrow in A indicates the pulmonary artery of the right accessory lobe. The inset shows detail of a lobe, showing the mesenchymal layer (arrowhead), whereas the arrow shows a putative vascular sprout as part of the angiogenesis process. B: hematoxylin and eosin (H&E) staining shows entrapped primitive erythrocyte in the lumen of a capillary (arrows). PECAM-1 staining indicating the vasculature surrounding the growing lung buds (C and D), Fli-1-positive staining (E), and the α-SM actin staining (F and G) illustrates that the distal airways (DA) are not yet muscularized, in contrast to the more mature proximal airways (PA). G: Tie2-driven LacZ expression in whole mount transgenic lung displays a capillary network wrapping the terminal buds. The inset shows Tie2-positive vessels clearly surrounding the terminal lung buds. H: details of a cross section of the lacZ-stained lung showing the Tie2-positive endothelial cells around a growing lung bud. The arrows indicate primitive erythrocytes that fill the vasculature surrounding the airways. I: H&E staining shows the pulmonary veins are irregularly shaped and are a few cells' thickness away from the airway. The lumen of these veins is filled with erythrocytes (arrows). J and K: Fli-1 staining displays the pulmonary arteries form a straight vascular tube closely associated with the airway epithelium. Note the presence of primitive (PE) and definitive (DE) erythrocytes in the vascular lumen (K). TB, terminal bud. Bars, 200 μm (A and G), 100 μm (insets in A and G, and C, F, I, J), 50 μm (B, D, E, H), and 20 μm (K).

At E13.5, further branching of the airways results in a more complex bronchial tree (Fig. 2G), and the polygonal network of capillaries clearly encloses the growing lung bud (Fig. 2G, inset, and Fig. 2H). The proximal pulmonary arteries are no longer associated with the capillary plexus of the trachea and are connected with the developing pulmonary trunk and ductus arteriosus. Two proximal pulmonary veins exit the right and the left lung and form a common vein (Fig. 2G). The intrapulmonary arteries run in close proximity to the airway and form straight tube-like structures (Fig. 1K) that are muscularized more peripherally than the veins. The intrapulmonary veins run at some cell thickness away from the airway (Fig. 2I). The venous drainage is more irregular and has a wider lumen (Fig. 2, I and J). Because the site of hematopoiesis is shifted toward the liver, a mixture of primitive erythrocytes (nucleated, blood island origin) and definitive erythrocytes (enucleated, liver origin) is present in the lumen of the vessels (Fig. 2K). This clearly shows the existence of a vascular connection between the lung and other embryonic parts.


We describe the development of the pulmonary vasculature in the mouse from the first morphological sign of lung development (E9.5) until early pseudoglandular stage (E13.5) through the analysis of whole mount X-gal-stained fetal lungs of Tie2-LacZ transgenic mice. The transgenic strain expresses the bacterial lacZ gene under the control of the Tie2 promoter, and cells that convert the X-gal substrate are positive for the angiopoietin receptor Tie2 (31). We expanded our analysis of the pulmonary vasculature using immunohistochemistry on serial sections of wild-type mice with two distinct endothelial-specific cell markers, PECAM-1 and Fli-1. Furthermore, the embryos have been isolated and processed without disrupting the circulation to leave the vascular tone and integrity intact. This procedure prevents the collapse of vessels and the putative creation of artifacts in the sections. Therefore, we were able to follow individual structures throughout serial sections, which allowed us to identify vessels and airways with accuracy. More importantly, by fixing the tissue with the blood cells still in the vascular system, we could recognize primitive, and at later stages definitive, erythrocytes in the lumen of the lung vessels. Because primitive erythrocytes are exclusively produced by the blood islands of the yolk sac and definitive erythrocytes by the fetal liver (21, 22), our observations prove unequivocally the existence of a connection of the lung with the heart, yolk sac, and liver and thus the presence of a closed circulation from the earliest morphological signs of lung development. In addition, McGrath et al. (19) also used the distribution of primitive, nucleated red blood cells as proof for the existence of circulation.

Already at the first morphological sign of lung development, the vasculature consists of Tie2-, PECAM-1-, and Fli-1-positive endothelial cells that are part of a capillary network in the splanchnic mesoderm (Fig. 1, A, F, and G). The proximal and distal structures from this plexus are connected to each other and are continuous with the heart vascular structures (Fig. 1, C, H, and I). The presence of yolk sac-derived primitive erythrocytes within the vascular lumen proves that there is already blood circulation in the lung and connection with the embryonic circulation from the earliest point of lung development (Fig. 1, D, E, and G). Remodeling of this plexus forms the main trunks of the proximal vessels, which is especially notable in the case of the pulmonary artery that changes from a plexus alongside the trachea at E9.5 into two muscularized vascular tubes at E11.5 (Fig. 1, A, F, and J). Serial sections demonstrated that both the proximal and distal vessels of the afferent and efferent pulmonary vasculature were positive for both Fli-1 and PECAM-1 from E9.5 onward. The endothelial cells of the network of capillaries that surrounds the growing epithelial buds, the effective component, were also positive for both antigens during early lung development.

We clearly demonstrate for the first time that from the earliest morphological sign of lung development, a vascular network exists that is in contact with the embryonic circulation. Through the preparation of vascular casts, deMello et al. (5) investigated the development of the pulmonary vasculature in mouse and concluded that there was no connection between proximal and distal structures of the mouse lung before E13.5. This technique, although very valuable and informative, has considerable limitations when studying vascular networks that consist mainly of capillaries and very small vessels. Schachtner et al. (30) used heterozygous Flk-1-LacZ knockin mice to study the lung vasculature, starting their whole mount analysis at E11.5, 2 days after the initiation of lung development. They concluded that only the proximal part of the growing pulmonary artery contained a lumen through analysis of E10.5 mouse embryo sections (30). Our data are partly in line with the work of Hall et al. (10), who showed continuity from the proximal and distal structures of the lung using three-dimensional reconstruction of immunostained 34-day human lung. Hall et al. identified PECAM-1-positive cells in vessel walls, some of which had a narrow lumen (10, 11). We used a tissue processing procedure that keeps the blood circulation intact and prevents the collapsing of the vessels. It is clear that the lumen of the proximal arteries is narrower than the lumen of the proximal veins, due to differences in pressure and muscularization. This may explain why deMello et al. (4) observed that early veins were diffusely present throughout the mesenchyme, establishing a central luminal connection to the main pulmonary vein before airways or arteries were present at the same level, thus leading to the conclusion that veins and arteries are dissociated in their timing and pattern of branching. Schachtner et al. (30) named these venous drainages “lacunae,” which are very clear at E13.5. However, our analysis of serial sections revealed that these structures are, in fact, the pulmonary veins. Based on hematoxylin and eosin-stained sections, deMello et al. described the presence of “hematopoietic lakes” in E10 mouse lung (5) and 33-day human lung (4), but analysis of serial sections of embryonic lungs fixed with intact circulation did not reveal structures that could resemble these hematopoietic lakes. We conclude that the description of hematopoietic lakes in the lung is based on morphology, and these lakes most likely are collapsed vessels containing trapped primitive erythrocytes.

Vasculogenesis and hematopoiesis are intimately associated extraembryonically, and this led to the description of the existence of a common precursor cell, the hemangioblast (26). Poole and Coffin (24) used QH-1 antibodies as a label for angioblasts to study the major vessel primordia in quail chick chimeras. They defined vasculogenesis as the in situ formation of vessels from the aggregation of angioblasts into a cord that later acquires a lumen and angiogenesis as the formation of new vessels by sprouting of capillaries from existing ones. Later studies concluded that vasculogenesis gives rise to the heart and the first primitive vascular plexus, whereas angiogenesis is responsible for the remodeling and expansion of this primitive plexus (23). A vasculogenic study in early mouse embryos identified the angioblast as a mesodermal Tal-1+/Flk-1+/PECAM-1 cell (6). The presence of angioblasts in the lung has been described morphologically (4, 5) and as isolated PECAM-1-positive (10, 11) or Flk-1-positive (30) endothelial cells. Flk-1 is expressed by undifferentiated endothelial cells and angioblasts, but Schachtner et al. (30) showed that LacZ expression driven by the Flk-1 promoter overlaps with PECAM-1 expression in the lung. DeMello et al. (4, 5) suggested that the lung itself produces blood cells in hematopoietic lakes as part of a vasculogenic process. However, during our analysis of serial sections of the different embryonic stages of lung development, we have never encountered blood cells other than primitive and definitive erythrocytes. If the lung produced blood cells, it would be conceivable that intermediate hematopoietic precursor cells are observed in the sections, even if their frequency is very low. Furthermore, Medvinsky et al. (20) previously performed colony-forming unit spleen assays on several embryonic tissues and demonstrated that the embryonic lung does not contain hematopoietic progenitors; therefore, the lung is not recognized as an organ with the capacity to produce hematopoietic cells. Hematopoiesis has been described extraembryonically in the hematopoietic islands of the yolk sac as well as intraembryonically in the trunk intermediate cell mass and later in the aorta-gonad-mesonephros, liver, spleen, and bone marrow (7).

We hypothesize that the vasculature grows primarily by expansion of existing vessels, but we cannot exclude at this moment the possibility that a minor population of putative angioblast-like cells also contributes to the growth of the capillary network. Therefore, we have performed immunostaining with Flk-1 to investigate the existence of an angioblast-like cell in the lung mesenchyme that may contribute to the expanding vascular network. Positive Flk-1 staining was identified in the endothelial cells of the vessels in a pattern that resembled our presented Tie2 staining (data not shown), confirming the pattern of X-gal stainings reported by Schachtner et al. (30) with the Flk-1-LacZ knockin mice. However, we were unable to localize isolated Flk-1-positive cells that were not part of the vasculature that could serve as a putative angioblast. In addition, we have performed CD34 immunostainings, and the results support our data shown in this paper (data not shown). In contrast, Han et al. (12) describe CD34+ solitary cells, or groups forming a single layer, in the lung mesenchyme of 4-wk-old human embryonic lungs. Maeda et al. (18) also described endothelial cells in distal lung mesenchyme that are separated from the proximal vessels and that may be derived from mesenchymal cells. Although we formally cannot exclude that there may be scattered CD34+ cells in the mesenchyme not linked to the vasculature, we believe that these apparent single cells are part of a network that can only be revealed through careful analysis of serial sections.

Although we were unable to identify individual Flk-1+ or CD34+ cells, we cannot evocatively exclude the existence of angioblast-like cells in the lung mesenchyme. However, we reason that the contribution of putative intrapulmonary angioblast-like cells to the developing vasculature would be minimal. It remains interesting to speculate how the vasculature can expand so rapidly to keep up with the branching airways. Given the enormous expansion of the network required to match the growth of the airways in a short time, it is hard to understand that a small number of angioblasts would have a major contribution to the growth of the vascular network. It could be possible that circulating cells contribute to the expanding network of vessels, since the endothelial cell divisions and stretching may not fully explain the speed of growth (17). Although these authors refer to the process whereby circulating endothelial cell progenitors incorporate and differentiate into existing blood vessels as vasculogenesis, we would suggest calling this process angiogenesis since it is the expansion of an preexisting vessel.

Our observations of the gradual muscularization of the proximal vessels as part of their maturation process confirms the work described by Hall et al. (10, 11) in humans. They showed that veins acquired α-SM actin at 56 days, whereas arteries did at 38 days. In mice, we first detected α-SM actin-positive cells in the vascular wall of arteries and veins at gestational age E11.5 (Fig. 1, N and O). Two muscularized pulmonary arteries run alongside the trachea at E11.5, as also shown by a confocal microscopic study in mouse lungs (35). Two recent papers (12, 18) used double immunolabeling for PECAM-1 and α-SM actin in human fetal lungs with the conclusion that there are two populations of endothelial cells in the lung vascular system. One population of vascular plexi independent of α-SM actin-positive adjacent cells and another population of vascular plexi were juxtaposed to α-SM actin-positive cells. Both groups imply that vasculogenesis would rely on the endothelial cells that are independent from the α-SM actin, whereas angiogenesis would be based on the α-SM actin-dependent endothelial cells. In addition, the authors concluded that both plexi communicate at the midpseudoglandular stage. According to Maeda et al. (18), the intrapulmonary arteries make contact with a preexisting capillary network that develops independently. This would support the findings of deMello et al. (4, 5), but as we already discussed, our data do not support this hypothesis. We have clearly shown that already at the onset of lung development there is a clear vascular network surrounding the growing lung buds, and our data support the idea that angiogenesis is the major process by which the lung vasculature grows and expands. In addition, our results demonstrate that the vessels forming in the distal mesenchyme are in contact with those in the proximal mesenchyme. Our analysis of serial sections revealed that there was just one vascular network and that proximal vessels undergo muscularization as part of the wall maturation as they accommodate the increasing blood flow and pressure. Muscularization is part of the gradual maturation process of airways and vessels. We have shown that the proximal structures in the lung are α-SM actin positive before the immature distal structures, which confirms the findings by Hall et al. (10, 11).

Distal angiogenesis as a new concept for lung vascular morphogenesis.

Pulmonary vascular morphogenesis has been described to occur either by a combination of central angiogenesis and vasculogenesis with the formation of hematopoietic lakes (Fig. 3A) (4, 5, 33) or just by vasculogenesis with formation of new vessels from endothelial precursor cells (Fig. 3A) (10, 11, 15). However, our data do not support the existence of central angiogenesis or distal vasculogenesis in lung development. Therefore, we propose “distal angiogenesis” as a model for pulmonary vascular morphogenesis (Fig. 3, A and B). Distal angiogenesis is the formation of new capillaries from preexisting ones at the periphery of the lung. On the basis of extensive and detailed morphological observations, we define the concept of the “tip zone” as the distal part of the branching airway that lacks the layer of SM cells. It is wrapped by a polygonal meshwork of capillaries, the effective component, that expands by distal angiogenesis as the lung bud grows, finally leading to the alveolar capillary plexus. We hypothesize that epithelial-endothelial interactions are decisive, inducing angiogenesis at the tip zone, which ensures the coordinate expansion of the vascular network as the branching proceeds (Fig. 3B). Newly formed vessels remodel dynamically, as they form part of the afferent or efferent component. This vascular remodeling implies that some vessels will grow and fuse with neighboring vessels, whereas others will remain small or degenerate (Fig. 3B).

Fig. 3.

A: lung vascular morphogenesis models. Model 1, proposed by deMello et al. (5), describes 2 mechanisms of lung vascular morphogenesis, central angiogenesis (sprouting of arteries and veins from central vascular trunks) and distal vasculogenesis (development of hematopoietic lakes in the mesenchyme). Connection between the 2 vascular beds would occur at embryonic day (E)13/14. Model 2, proposed by Hall et al. (11), proposes distal vasculogenesis (development of new vessels from endothelial cell precursors) as the mechanism of lung vascularization. Model 3 explains our newly proposed mechanism of distal angiogenesis as the process to develop lung vasculature (formation of new capillaries from preexisting ones). HL, hematopoietic lakes; ECP, endothelial cell precursor. B: distal angiogenesis model. In the model we propose, the formation of new capillaries from preexisting vessels takes place at the “tip zone,” where airway and capillary network expand in a coordinated way through epithelial-endothelial interactions. Newly formed vessels remodel dynamically as they form part of the afferent or efferent component. BA, airway branch A; BA.1, BA1.1, and BA.2, daughter branches from branch A.

Two basic mechanisms of embryonic angiogenesis by which the capillary network can expand have been proposed: sprouting and nonsprouting angiogenesis (25). Sprouting angiogenesis involves the expansion of the capillary network by the formation of vascular sprouts from opposite preexisting capillaries; sprouts meet each other by filopodia and form a solid strand that later acquires a lumen and splits the intervascular space (26). In the nonsprouting angiogenesis or intussusception, a solid mesenchymal pillar grows into a capillary, subsequently enlarges, and forms a new intervascular space (26). Intussusceptive angiogenesis has been described to be responsible for the postnatal growth of the lung capillary bed (2). At present, the precise mechanism of angiogenic expansion of the vascular network in our model is under investigation, but the arrow in the inset of Fig. 2A points at a putative sprouting vascular network. In addition, preliminary data using confocal microscopy would support this finding.

In summary, we performed a detailed ontogenic morphological analysis of the pulmonary vasculature from the earliest embryonic stage onward. The present study describes the development of the different vascular components (arteries, capillaries, and veins) of the lung in relation with the developing airways. We conclude that the vasculature is part of the embryonic circulation from the moment the lung starts to develop. This implies that the presence of blood vessels could be more important for the development of the lung than previously anticipated. The endothelial cells of the splanchnic mesoderm may be involved in the prepatterning of the presumptive lung region, like it has been shown for liver and pancreas development where endothelial cells are involved in the induction of these organs (16). Our observations led us to propose distal angiogenesis as a new concept for lung vascular development. We defined the concept of the tip zone, where the epithelial-endothelial interactions are crucial to determine the expansion of the lung vascular network. Based on our observations, we propose that angiogenesis already starts at the embryonic phase of lung development and is the major blood vessel-forming process.


This work was supported by the Sophia Foundation for Medical Research (SSWO project number 412).


The authors thank Prof. Martin Post for providing the Tie2-LacZ transgenic mice, the Animal Care Facility for animal husbandry, and Marta Santos for technical assistance with some of the immunohistochemical experiments.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract