Its Role in Embryonic Vascular Network Formation
Abstract—Intussusceptive angiogenesis is a novel mode of blood vessel formation and remodeling, which occurs by internal division of the preexisting capillary plexus without sprouting. In this study, the process is demonstrated in developing chicken eye vasculature and in the chorioallantoic membrane by methylmethacrylate (Mercox) casting, transmission electron microscopy, and in vivo observation. In a first step of intussusceptive angiogenesis, the capillary plexus expands by insertion of numerous transcapillary tissue pillars, ie, by intussusceptive microvascular growth. In a subsequent step, a vascular tree arises from the primitive capillary plexus as a result of intussusceptive pillar formation and pillar fusions, a process we termed “intussusceptive arborization.” On the basis of the morphological observations, a 4-step model for intussusceptive arborization is proposed, as follows: phase I, numerous circular pillars are formed in rows, thus demarcating future vessels; phase II, formation of narrow tissue septa by pillar reshaping and pillar fusions; phase III, delineation, segregation, growth, and extraction of the new vascular entity by merging of septa; and phase IV, formation of new branching generations by successively repeating the process, complemented by growth and maturation of all components. In contrast to sprouting, intussusceptive angiogenesis does not require intense local endothelial cell proliferation; it is implemented primarily by rearrangement and attenuation of the endothelial cell plates. In summary, transcapillary pillar formation, ie, intussusception, is a central and probably widespread process, which plays a role not only in capillary network growth and expansion (intussusceptive microvascular growth), but also in vascular plexus remodeling and tree formation (intussusceptive arborization).
According to the present conception, the formation of new blood vessels is achieved by 2 distinct processes: vasculogenesis, which is found in early stages of embryonic development, and angiogenesis, which occurs during the whole life of an organism. Vasculogenesis is the de novo formation of blood vessels from angioblastic precursor cells,1 2 whereas angiogenesis is characterized by vessel formation from preexisting vessels (eg, by sprouting). The process of sprouting angiogenesis was investigated in numerous physiological and pathological processes such as embryogenesis, postnatal development and growth, wound healing, regeneration, inflammation, and carcinogenesis. Today it is well established and widely accepted.3 4 5 6
More recently, another type of blood vessel formation called intussusceptive microvascular growth (IMG) was repeatedly reported under normal and pathological conditions. Looking at the microvasculature of rabbit lungs in the light microscope, Short7 proposed as early as 1950 that capillary network growth could occur by formation of new meshes. Unfortunately, this statement, which was not backed by further investigations, went unnoticed. A similar idea was put forward again in 1986 by Caduff et al.8 Investigating the postnatal maturation of rat lung microvasculature in casts by scanning electron microscopy, they observed the appearance in the capillary network of tiny pillars. They proposed that the lung capillary network expanded by insertion of slender transcapillary tissue pillars (for the sake of convenience simply called pillars) and coined the term “intussusceptional,” which was later changed to IMG.8
Later, the same group described 4 consecutive steps in pillar formation, as follows: phase I, creation of a zone of contact between opposite capillary walls (formation of a transcapillary interendothelial bridge); phase II, reorganization of the intercellular junctions of the endothelium, with central perforation of the endothelial bilayer; phase III, formation of an interstitial pillar core, invaded successively by cytoplasmatic extensions of myofibroblasts and pericytes and finally by interstitial fibers; and phase IV, growth of the slender pillar to a capillary mesh (pillar with diameter >2.5 μm). These findings provided the first morphological corroboration for the new concept of IMG.9
In a recent review article, Risau10 stated that both types of angiogenesis—sprouting and intussusception—were involved in embryonic blood vessel formation, but that the processes of primary plexus formation and of growth and maturation of vascular networks were still not very well understood.
In the present study, we have addressed the following 2 questions: (1) By what means do rapidly growing embryonic organs establish capillarization? (2) What are the structural alterations transforming a primitive capillary plexus into a complex 3-dimensional meshwork with large supplying and draining blood vessels?
Our findings indicate that the process of intussusception plays an important role both in early capillarization and in network remodeling and formation of larger vessels.
Materials and Methods
Culture of Chicken Embryos
All experiments were performed on chicken embryos grown by the shell-free culture method.11
After 3 days of incubation, Brown Leghorn eggs were opened, and their contents were carefully poured into a plastic Petri dish 80 mm in diameter. The embryos were incubated for several days at 37°C in humid atmosphere.
Methylmethacrylate (Mercox) Casting
Mercox (Japan Vilene Co) vessel casts were prepared from various embryonic organs. Investigations were then concentrated on the choroid lamina of the eye and on the chorioallantoic membrane (CAM) between stages 27 and 45, according to Hamburger and Hamilton.12
The vasculature was perfused with a solution of 0.9% sodium chloride containing 1% heparin (Liquemine; Roche Pharma AG, Reinach, Switzerland) and 1% procaine followed by a freshly prepared solution of Mercox containing 0.1 mL accelerator per 5 mL resin.8
After 2 to 4 weeks of dissolution of tissues in 15% KOH, the casts were dehydrated in ethanol and dried in a vacuum desiccator. Samples were glued onto stubs with carbon and spattered with gold. They were examined in a Philips XL 30 FEG scanning electron microscope.
In Vivo Microscopy
Fluorescence microscopy was performed with a Polyvar-Reichert microscope using ×10 and ×25 objectives. The microscope was equipped with a custom-built heating table to maintain the temperature of the specimens at 37°C. The microvascular growth and blood flow were monitored in vivo over periods of up to 10 hours using an LE-470 charge-coupled device Optronics video camera (Visitron Systems) and a digital video recorder (Sony DHR-1000 VC). Observations were performed after IV injections of 0.1 mL 2.5% FITC-dextran 500 000 molecular weight (Sigma).13
Tissue samples of chicken eye and CAM were obtained at various ages and fixed in 2.5% glutaraldehyde in 0.03 mol/L potassium phosphate buffer (pH 7.4, 370 mOsm). They were postfixed in 0.1 mol/L sodium cacodylate (pH 7.4, 340 mOsm)–buffered 1% OsO4, dehydrated in ethanol, and embedded in Epon 812 according to Burri et al.14
Using glass knives, 1-μm-thick sections were obtained, stained with toluidine blue, and analyzed in the light microscope. Sections of 80- to 90-nm thickness were cut, picked up on Formvar-coated (polyvinyl formal; Fluka Chemie AG, Buchs, Switzerland) copper grids, double stained with lead citrate and uranyl acetate, and viewed in a Philips EM 400 electron microscope.15
Three-Dimensional Reconstruction Based on EM Pictures
Electron micrographs from several series of serial sections (up to 450 sections per series) were recorded and positive copies produced. Appropriate locations containing a site of transcapillary interen-dothelial contact (“kissing” junction), a pillar or a septum, were followed through the whole stack of sections. Border traces of the serial images were traced by hand on semitransparent paper and then digitized using a Hewlett Packard flatbed scanner. The whole set of image files was restacked with dedicated software (IMARIS, Bitplane AG) on an Indigo 2 workstation (Silicon Graphics), where the 3D surface rendering based on a triangulation algorithm was performed. With the Scene Viewer software (Silicon Graphics), light and color effects were added.
The chicken eye choroid vasculature consists of a cuplike 2-dimensional capillary plexus with supplying and collecting vessels on its convex side. The analysis of microvascular cast morphology at various ages allowed us to establish the following picture of choroid vessel development.
At the early stage 27, the vasculature of the choroid presents as a simple, slightly curved capillary meshwork, connected to a few larger vessels with short branches. The network contains numerous pillars and meshes of various sizes, represented in the casts by holes of varying dimensions (Figure 1a⇓). Tiny holes in casts have been recognized to represent a characteristic feature of IMG.9
In the rapidly growing eye cup (from a diameter of 2 mm at stage 27 to 10 mm at stage 44, which corresponds to a 25-fold increase in overall surface area), new pillars are continuously inserted in the capillary meshwork, which results in increasing pillar density. By this process we observed that just distally to the supplying vessels and proximally to the collecting vessels, pillars were arranged in lines in the direction of blood flow. The rows of pillars seemed to preform the main paths of the future larger blood vessels (Figure 1a⇑). Although the pillars were at first round when they appeared, they later changed shape, becoming a septum-like structure appearing as slits in the casts. Alternatively, 2 pillars could become connected by a longitudinal, fold-like invagination of the capillary wall, which in the end would result also in a septum-like structure (Figure 1b⇑). Comparing earlier and later stages, it appears that the septa elongated and merged with each other along the axis of blood flow, whereas new pillars were formed distally to the old ones and thus extended the delineation of the vessel (Figure 1a⇑ through 1c). Six days later, no more circular posts could be detected in the vicinity of the big vessels; they had all been replaced by lines of long tissue septa, which, as by merging, were going to separate the adjacent vessels from each other (Figure 1d⇑). With growth of the eye, the plexus surface increased further and the segregated vessels now stood out from the capillary plane. This means that a bilayered structure had emerged from the original 2-dimensional capillary network, as follows: the choriocapillaris at the inner side and the future vascular lamina of the choroid at the outer side. Thereby the axes of the pillars along these vessels changed their orientation from perpendicular (Figure 1a⇑ through 1c) to oblique and nearly horizontal (ie, parallel to the capillary meshwork) (Figure 1e⇑ and 1f⇑). The number of channels between the feed vessels and the capillary plexus declined with time, probably because of pillar fusions. The remaining connections grew in diameter and transformed to larger supplying and collecting branches. By these means more and more larger branches arose, were lifted, and shifted to the vascular lamina of the choroid (Figure 1e⇑ and 1f⇑). Simultaneously, further pillars were produced in the capillary layer to allow for the growth of the eye (Figure 1e⇑). Finally, the choroid presented as a rather complex bilayered structure with an external vascular lamina with arteries and veins and an internal cuplike capillary network, the choriocapillaris (Figures 1g⇑ and 1h⇑). Our observations indicate that feed vessel formation was primarily based on the mechanism of intussusception in the sequence, as follows: pillar formation, development of septa, delineation, segregation, and finally extraction of the supplying or collecting vessels. For this reason, we have called this process intussusceptive arborization (IAR). Both processes, IMG (see Introduction) and IAR, run in parallel and are 2 facets of intussusceptive angiogenesis (IA), leading to the development of all vascular tree components.
In Vivo Investigation
In top view, the fluorescent capillary network represents a bright sheet with numerous dark islands of varying dimensions and shapes, the smallest of them being barely detectable pillars. The flow in this labyrinthic network is visualized by the erythrocytes moving across the field of vision as dark shadows. They can be seen accelerating, decelerating, changing direction, bumping against obstacles, and also sometimes wrapping around the slender pillars and hanging there for a few seconds or minutes. Over an observation period of up to 10 hours, it was repeatedly possible to detect in the day 10 CAM (stage 35) important alterations in the vascular network structure, as depicted in the sequence of video still frames of Figure 2⇓. In an area with unimpeded flow, pillar formation was signaled by flow disturbances with erythrocytes decelerating or even stopping at a given place. Later, a tiny dark spot became visible, indicating that a new pillar was born (Figure 2⇓). In parallel to the de novo formation of pillars, pillars and meshes were observed to increase in size, merge, or elongate to septa. The process of septum formation appears to be initiated sometimes by development of a longitudinal fold, bulge, or cushion between 2 neighboring pillars. The thin fold arises from the lower or upper part of the capillary wall (perpendicularly to the capillary plexus) and is visible only by focusing the microscope to a given plane. Subsequently, the fold grows until complete separation of the vessel lumen is achieved. This merging of pillars and meshes may occur immediately when the pillar appears or many hours later. Frequently, new pillars were seen arising distally and in line with older ones. As in the casts, they appeared to delineate future larger vessels and thus to contribute to vascular remodeling.
Electron Microscopy and 3D Reconstruction
To investigate the morphology of IMG and IAR in developing organs, transmission electron micrographs were recorded from serial sections running perpendicularly to the capillary lamina of the chicken choroid at different stages of embryogenesis. Screening the serial sections, we found the following 3 types of tissue bridges across the capillary lumen (Figures 3 through 7⇓⇓⇓⇓⇓):
1. Kissing contacts: very small spots of interendothelial contacts between opposite capillary walls (Figures 3⇑ and 6⇑). They had a disk-like shape, and the joined cell membranes sometimes presented an increased contrast reminiscent of junctional complexes.
2. Pillars: numerous slender transcapillary tissue pillars with diameters ≤2.5 μm. These consisted of a core of connective tissue enwrapped by endothelial cells. As a rule, interendothelial junctions were seen to run around the pillar; they formed either simple end-to-end contacts or, more often, complex interdigitations (Figure 4⇑). Some of the pillars traversed the lumen obliquely, causing protrusions of the capillary wall at the post basis (Figure 6⇑).
3. Meshes (round) and septa (elongated) (Figure 6⇑). These usually contained extensions of interstitial cells and collagen fibrils (Figure 5⇑). Some of the cells found in the meshes and septa showed morphological features of endothelial cells, similar to the endothelial-like cells described in the CAM.16
Only rarely we observed mitoses in the growing endothelium of the choroid, suggesting a limited proliferative activity of these cells from stage 29 onward, but no actual counts were made. Comparisons between capillaries from early and late stages of chick development showed, however, a dramatic reduction of endothelial cell thickness and endothelial cell nuclei per section area (data not shown).
As a result of sprouting angiogenesis with fusion of vascular buds into a primary capillary meshwork, the developing chick eye and CAM are connected and integrated at an early stage into the circulatory system. We propose here that with increasing metabolic demands of those organs, growth of the capillary networks switches from sprouting to IA.
Previously, the morphometric analysis of Schlatter et al17 showed that CAM angiogenesis underwent 3 phases of development. The early phase (days 5 to 7) was characterized by sprouting angiogenesis with multiple capillary sprouts invading the mesenchyme. During the second (or intermediate) phase (days 8 to 12), the sprouts could no longer be observed; concurrently, small transcapillary tissue pillars became prominent and reached a peak in frequency on day 11. During the late phase (day 13 and older), pillars were observed only occasionally; these were replaced by intercapillary meshes >2.5 μm in diameter.17
These results and interpretations are in agreement with previously obtained pictures in microvascular corrosion casts of the CAM by Burton and Palmer.18
Our findings in the CAM, but also in the choroid of the eye, confirm that IMG is indubitably the prevalent angiogenic mechanism during middle and late embryogenesis in these capillary beds. But they furthermore demonstrate that the formation of new tissue pillars is involved not only in increasing vascular network complexity (what corresponds to IMG), but also in the process of vascular tree formation (IAR). In the prospective trunk region, either arterial or venous, a cascade of forming pillars delineates the longitudinal contours of 1 or several prospective larger vessels and initiates the vascular remodeling. As seen in the casts, transformation of pillars to septa by reshaping and fold formation followed by septa merging successively isolates trunk vessels and their branches from the capillary plexus. The increased blood flow may further contribute to enlarge the selected vessels, whereas the number of connecting vessels to the primary meshwork decreases rapidly. By these means, a complex 3-dimensional treelike structure develops with trunk, boughs, and many filial branches of different size. These morphological changes are illustrated in the drawing of Figure 7⇑.
It is evident that the results documented in Mercox casts represent static snapshots of particular stages of vascular development. It could be argued that the sequence of events, as described here, is a construct that might not necessarily correspond to facts. Therefore, the in vivo observations made in the CAM are important. They provide the missing links in the interpolated interpretation of the cast findings. A provisional reservation, however, has still to be made. Both the choroid and the CAM capillaries may represent a “special” situation in that both networks are nearly 2-dimensional. The question whether IA has similar functions in 3-dimensional networks and in particular in tumor growth has to remain open, although we strongly suspect that it does, given that IMG has been reported to occur in tumors.19
Endothelial cell proliferation is a basic mechanism in sprouting angiogenesis. Regulation of endothelial cell division has therefore attracted a major interest in regulation of angiogenesis, in particular regarding antiangiogenic therapy in tumor growth. By way of contrast, several reports suggest that in IA, endothelial proliferation plays only a secondary role. It appears that pillar formation and capillary expansion primarily occur by rearrangement and thinning out of the endothelial cells. The presence of bromodeoxyuridine-labeled endothelial cells in the growing chicken CAM from day 6 to day 15 has been carefully investigated by Kurz et al.20 The authors report a significant loss in proliferative activity (>50%) at day 10 of incubation (intermediate phase) in comparison with day 6 (sprouting phase). After day 10, proliferative activity decreased further, and at days 14 and 15 (late phase), dividing cells represented <10% of the value of day 6.20 We could confirm this in our own bromodeoxyuridine labeling experiments (data not shown). In earlier work, DeFouw et al21 reported rapid extension of the CAM surface from 6 cm2 at day 6 to ≈65 cm2 at day 14. During the same period, the number of feed vessels increased quickly from day 6 to day 14 to 2.5- and 5-fold for precapillary and postcapillary vessels, respectively. This was predominantly due to the growth and remodeling after day 10 of incubation.21 These results indicate that endothelial cell proliferation drops dramatically during the phase of IMG and during IAR.
The total endothelial cell volume is clearly redistributed during network growth by thinning out of the endothelial cells. Morphometric analyses of the thickness of CAM endothelium confirmed this hypothesis. Rizzo and DeFouw22 found the CAM capillary endothelium to attenuate up to 50% from day 10 to day 14.
In electron micrographs of the choroid capillaries, we also observed a marked decrease in endothelial thickness with ongoing development.
Furthermore, in the lung, where the intussusceptive principle of microvascular growth was documented for the first time, morphometric and autoradiographic investigations reported that the total number of endothelial cells remained constant after the second postnatal week and that after the third postnatal week practically no dividing endothelial cells could be found,23 despite the fact that the capillary surface area increased 20 times and the capillary volume >35 times between birth and adult age.
All of these findings suggest that proliferation is more associated with sprouting than with IA. At the moment, it is poorly understood which factors are involved in the regulation of these processes and how they act. In suggesting that both types of angiogenesis—sprouting and IA—were taking place in the embryonic blood vessel formation, Risau10 proposed that vascular endothelial growth factor (VEGF) was probably an essential regulatory factor for both types of microvascular growth. Indeed, the external addition of VEGF165 protein on 13-day-old CAMs induced brushlike vessel formation in precapillary arterioles, but also the formation of sinusoidal or lacunar vessels in the venous parts of the vascular system.24 This could signify that VEGF may have different effects depending on location and timing.
Recently, Tie1 and Tie2 receptor tyrosine kinases have been proposed as possible effectors of IMG.25
On the basis of our observations, we suspect that hemodynamic factors are heavily involved in the regulation of this process. Arising of new pillars occurs regularly in areas of enlarged capillaries or vessels, sometimes also along flow pathways. Differences and changes in speed of flow can induce changes in static vascular pressure, and the latter again can be responsible for either direct morphological effects or indirect structural changes due to the triggering of growth factor expression by stretch and shear forces.
In summary, we propose in this report that the concept of intussusception, ie, the formation of transcapillary tissue pillars, plays an important role in embryonic vascular development, both in growth of capillary networks (classic IMG) and in formation of the larger vessels, ie, of the vascular tree (=IAR). The existence of the latter process is a new postulate. According to our observations it comprises the following steps:
Formation of numerous circular pillars in rows, thus demarcating future vessels
Formation of narrow tissue septa by pillar reshaping and pillar fusions
Delineation, segregation, growth, and extraction of the new vascular entity by merging of septa
Production of new branching generations by successively repeating the process, complemented by growth and maturation of all components
Both processes, IMG and IAR, run hand in hand and represent 2 facets of IA, which largely contributes to the formation of vascular components in middle and late embryogenesis of the organs investigated.
This research was supported by the Swiss National Science Foundation (Grant 31-45831.95), the Swiss Cancer League (Grant SKL 131-7-1995), and the Bernese Cancer League. We thank B. Haenni, B. de Breuyn, K. Babl, and B. Krieger for technical assistance, and E. de Peyer for typing of the manuscript and in particular for the art work.
- Received July 21, 1999.
- Accepted December 13, 1999.
- © 2000 American Heart Association, Inc.