IOVS
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

(Investigative Ophthalmology and Visual Science. 2004;45:2020-2032.)
© 2004 by The Association for Research in Vision and Ophthalmology, Inc.
DOI:  10.1167/iovs.03-1169
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (21)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Chan-Ling, T.
Right arrow Articles by Lutty, G. A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Chan-Ling, T.
Right arrow Articles by Lutty, G. A.

Astrocyte–Endothelial Cell Relationships during Human Retinal Vascular Development

Tailoi Chan-Ling,1 D. Scott McLeod,2 Suzanne Hughes,1 Louise Baxter,1 Yi Chu,1 Takuya Hasegawa,2 and Gerard A. Lutty2

1From the Department of Anatomy and Histology and Institute for Biomedical Research, University of Sydney, Sydney, New South Wales, Australia; and 2Wilmer Ophthalmological Institute, Johns Hopkins University School of Medicine, Baltimore, Maryland.


   Abstract
Top
 Abstract
Methods
 Results
 Discussion
 References
 
PURPOSE. To evaluate evidence for the presence of vascular precursor cells (angioblasts) and astrocyte precursor cells (APCs) in the developing human retina and determine their relationship.

METHODS. Pax-2/GFAP/CD-34 triple-label immunohistochemistry was applied to four retinas aged 12, 14, 16, and 20 weeks of gestation (WG) to label APCs, astrocytes, and patent blood vessels. APCs are Pax-2+/GFAP–, whereas astrocytes are Pax-2+/GFAP+. Adenosine diphosphatase (ADPase) enzyme histochemistry, which identifies endothelial cells and vascular precursors, was applied to human retinas aged 12, 16, 17, and 19 WG. Nissl stain, a nonspecific cell soma marker, was applied to 14.5-, 18-, and 21-WG retinas. Established blood vessels were visualized with CD34 and ADPase.

RESULTS. Topographical analysis of the distribution of Nissl-stained spindle cells and ADPase+ vascular cells showed that these two populations have similar distributions at corresponding ages. ADPase+ vascular precursor cells preceded the leading edge of patent vessels by more than 1 millimeter. In contrast, Pax-2+/GFAP– APCs preceded the leading edge of CD34+ blood vessels by a very small margin, and committed astrocytes (Pax-2+/GFAP+) were associated with formed vessels and nerve fiber bundles. Two populations of ADPase+ cells were evident, a spindle-shaped population located superficially and a deeper spherical population. The outer limits of these populations remain static with maturation.

CONCLUSIONS. A combination of Pax-2/GFAP/CD34 immunohistochemistry, Nissl staining, and ADPase histochemistry showed that the vascular precursor cells (angioblasts), identified using ADPase and Nissl, represent a population distinct from Pax-2+/GFAP– APCs in the human retina. These results lead to the conclusion that formation of the initial human retinal vasculature takes place through vasculogenesis from the prior invasion of vascular precursor cells.

The retina consists of a single neuroblastic layer when the optic fissure closes. At 4 to 5 weeks of gestation (WG) in humans, the hyaloid artery and associated mesenchyme invade the area that was previously the fissure.1 The hyaloid vascular system provides nutrients for subsequent retinal and lens development. The neuroblastic layer rapidly differentiates into two nuclear layers. Although the time course for appearance of the layers and structures was documented many years ago by Mann2 the interactions and relationships between cell types in humans and the origins of these cell types are not fully understood.3 4 5 6 For example, the heart (endocardium),7 aorta and aortic arches,8 and vasculature in some organ systems, such as lung, pancreas, and spleen, develop by the process of vasculogenesis, differentiation, aggregation of vascular precursors (angioblasts) to form cords, and canalization of the cords.9 10 11 12 In other organ systems such as the kidney, the vasculature is said to develop by the process of angiogenesis: migration and proliferation of endothelial cells from existing blood vessels.13 Formation of superficial primordial vessels of the retina in some species are mediated through vasculogenesis,5 14 although angioblasts have not been identified in human retina; whereas, angiogenesis is responsible for increasing vascular density and vascularization of the peripheral and inner retina in humans5 and deep plexus in dogs.14 15 Therefore, the mechanism of retinal vascularization appears similar to that observed during brain development. The primordial vascular bed on the surface of the neuroepithelium is derived from migratory vascular precursor cells,16 presumably by vasculogenesis. New vessel segments sprout from these preexisting vessels and grow tangentially by angiogenesis into the neuroepithelium.17 18

In vitro19 and immunohistochemical studies in several species support the view that astrocytes are not generated in the retina, but are immigrants from the optic nerve.20 21 22 23 Further evidence in support of this conclusion comes from retroviral lineage-tracing studies that show that, although clones containing various types of retinal neurons and Müller cells are frequently found, clones containing astrocytes are seldom observed.24 25 However, our recent studies have provided evidence for a second source of astrocytes within a small peripapillary rim of the undifferentiated neuroepithelium of the human retina.3 Astrocytes are intimately associated with the retinal vasculature during development of the mammalian retina,26 and both endothelial cells and astrocytes have been reported to proliferate during retinal vascular development.4 27 More recently, in vitro, vascular endothelial cells have been shown to be capable of inducing glial fibrillary acidic protein (GFAP) expression in astrocyte precursor cells (APCs) derived from rat optic nerve. This induction is likely to be mediated through leukemia inhibitory factor (LIF).28 Others have shown that astrocytes play a role in endothelial cell differentiation29 and blood–retinal barrier function.30 These observations point to a need for further elucidation of the relationship between the astrocytic and vascular lineage cells during differentiation of the human retina.

Many studies have reported the prior invasion of vascular precursor cells in the formation of the mammalian retinal vasculature, including studies in the human,5 6 cat,31 dog,14 32 and rat.33 A recent matter of contention (reviewed by Provis34 ) is whether the vascular precursor cells identified by earlier workers using Nissl staining,5 PAS staining of glycogen granules,14 35 alkaline phosphatase activity,36 and adenosine diphosphatase (ADPase) enzyme histochemistry37 could in fact have been APCs3 38 39 or whether they are angioblasts, as in other developing organs such as lung, pancreas, and spleen.9 10 40 Further, the failure of supposed vascular precursor markers in other systems such as flk-1 and flt-1 to identify these cells has suggested that vasculogenesis does not occur in retinal development, at least not in the mouse.39

This study focuses on the relationship between the astrocytic and vascular lineages during human retinal development. Pax-2/GFAP/CD-34 triple-label immunohistochemistry was used to determine the relationship between APCs, astrocytes, and patent blood vessels (CD34+).3 5 ADPase enzyme histochemistry, which identifies endothelial cells and vascular precursors37 and Nissl stain, a nonspecific cell soma marker,5 31 were applied to human fetal retina to evaluate further the evidence for the presence of vascular precursor cells in the developing human retina. ADPase is an ectoenzyme on the luminal surface of endothelial cells responsible for degrading extracellular ADP, preventing platelet aggregation. ADPase is exclusively in the vasculature in normal adult human retina37 41 42 and has been used to identify vascular precursors (angioblasts) in the neonatal canine retina.32 43 44


   Methods
Top
 Abstract
 Methods
Results
 Discussion
 References
 
Collection, Age Determination, and Preparation of Human Fetal Eyes
Fifteen human fetal eyes, ranging in age from 12 to 21 WG, were collected in accordance with the guidelines set forth in the Declaration of Helsinki and with the approval of the Human Ethics Committee of the University of Sydney. Fetuses were obtained after prostaglandin-induced abortions. The age of each fetus was determined from the date of last menstruation. Preparation and analysis of the tissue was also approved by the Joint Commission for Clinical Research at Johns Hopkins University School of Medicine.

For flatmount immunohistochemistry, the anterior segment and vitreous were removed, and eyecups were fixed at 4°C for 1 hour with 4% paraformaldehyde in 0.1 M sodium phosphate buffer (pH 7.4). The eyecup was then subjected to radial incisions, and the retina was dissected while floating in 0.1 M phosphate-buffered saline (pH 7.4; PBS) as described previously.45

Anti-CD34 Flatmount Immunohistochemistry and Nissl Staining
The vasculature was visualized by immunohistochemistry with a monoclonal antibody (QBEND/10, 1:50 dilution; Serotec, Oxford, UK) to CD34, a single-chain transmembrane glycoprotein with a molecular mass of 110 kDa that binds L-selectin and is selectively expressed on human lymphoid and myeloid hematopoietic progenitor cells as well as on vascular endothelial cells,46 as described previously.5 Immunohistochemical labeling was performed as previously described.5 45 For light microscopy, biotinylated antibodies to mouse immunoglobulin (1:50 dilution; Amersham Biosciences, Piscataway, NJ) were used as secondary antibodies, followed by horseradish peroxidase-conjugated avidin (Extravidin, 1:100 dilution; Sigma-Aldrich, St. Louis, MO). For fluorescence microscopy, Texas red–conjugated antibody to mouse IgG1 (1:60 dilution; Southern Biotechnology Associates, Birmingham, AL) was used as the secondary antibody. Wholemount Nissl preparations of the retina at 14.5, 18, and 21 WG were stained with 1% cresyl fast violet as described.47

Adenosine Diphosphatase Enzyme Histochemistry and Flat-Embedding
Retinas from 12, 16, 17, and 19 WG fetuses were dissected from the RPE, fixed, and incubated for ADPase activity, as described previously.37 The retina was dissected in 5% sucrose in 0.1 M sodium cacodylate buffer (pH 7.2) at 4°C, fixed with 2% paraformaldehyde in 0.1 M sodium cacodylate buffer for 20 hours at 4°C, and washed with 5% sucrose in 0.1 M sodium cacodylate buffer at 4°C. The retinas were incubated for the histochemical demonstration of ADPase activity.37 They were then fixed flat in 25% Karnovsky’s fixative and embedded flat in glycol methacrylate (JB-4 kit; Polysciences, Warrington, PA), as described previously.37 The lead ADPase reaction product was visualized and photographed en bloc under dark-field illumination. Areas of interest were serially sectioned with a ultramicrotome (model MT-2B; Sorvall, Newtown, CT) with a dry, glass knife. The 2.5-µm-thick sections were stained with McManus’ periodic acid-Schiff and hematoxylin counterstain. The ADPase activity was developed in some sections with ammonium sulfide, resulting in a dark brown ADPase reaction product, and then they were counterstained with 1% toluidine blue or 1% thionin.

Immunohistochemistry on Cryopreserved Tissue
Culottes of eyecup from a 16-WG fetus were cryopreserved as reported previously.48 Streptavidin alkaline phosphatase immunohistochemistry was performed on cryopreserved tissue sections using a nitroblue tetrazolium (NBT) system recently developed by Bhutto et al.49 In brief, 8-µm-thick cryosections were permeabilized with absolute methanol, blocked with 2% normal goat serum in Tris-buffered saline (TBS [pH 7.4] with 1% BSA), and then incubated overnight at 4°C with one of the following primary antibodies: mouse anti-human CD-34 (1:800; Signet Laboratory, Dedham, MA); mouse anti-human VEGF-R2 (1:1200 dilution, clone 6.64; ImClone Systems, New York, NY); and anti-CD-39 (1:400; Chemicon, Temecula, CA). CD39 is now known to be ecto-ADPase,50 which is the enzyme that is used for our ADPase enzyme histochemical labeling of the retinal vasculature and angioblasts. After washing in TBS, sections were incubated at room temperature for 30 minutes with goat anti-mouse biotinylated antibody diluted 1:500 (Kirkegaard and Perry, Gaithersburg, MD). Finally, sections were incubated with streptavidin alkaline phosphatase (1:500; Kirkegaard and Perry), and alkaline phosphatase activity was developed with a 5-bromo-4-chloro-3-indoyl phosphate/nitroblue tetrazolium (BCIP-NBT) kit (Vector Laboratories, Inc., Burlingame, CA), yielding a blue reaction product.

Melanin in RPE and choroidal melanocytes was bleached by a technique recently developed by Bhutto et al.49 Sections were fixed in 4% paraformaldehyde overnight at 4°C immediately after streptavidin alkaline phosphatase immunohistochemistry. Slides were washed in distilled water at room temperature, immersed in a 0.05% potassium permanganate solution (Aldrich Chemical Co., Inc., Milwaukee, WI) for 25 minutes, and rinsed in distilled water for 5 minutes. Sections were covered with 35% peracetic acid (FMC Corp., Philadelphia, PA) in a humidified container for 20 minutes at room temperature followed by washing in distilled water for 10 minutes twice. Finally, coverslips were mounted with Kaiser’s aqueous mounting medium without counterstaining.

Pax-2/GFAP/CD34 Triple-Label Immunohistochemistry
Pax 2 is a member of the Pax family of transcription factors, all of which are DNA-binding proteins that contain a paired-box domain. Each member of the Pax family is expressed in a spatially and temporally restricted manner, which suggests that these proteins contribute to the control of tissue morphogenesis and pattern formation. Pax 2 is implicated in organogenesis of the eye, ear, kidney, and the central nervous system (CNS).51 Pax2 is currently the earliest available in vivo marker of the astrocytic lineage.3 To examine the distribution of APCs (Pax2+/GFAP–) and astrocytes (Pax2+/GFAP+) with respect to the developing vasculature, 14-, 16-, and 20-WG retinas were immunoreacted for 2 to 3 days at 4°C with a mixture of three primary antibodies: antibodies against CD34 to identify the vasculature as just described, antibodies against GFAP (mouse monoclonal GA5, 1:100 dilution, Sigma-Aldrich), and antibodies against Pax2 (rabbit polyclonal, 1:100 dilution; Babco, Richmond, CA). A mixture of Cy3-conjugated antibodies to rabbit IgG (1:200 dilution; Jackson ImmunoResearch, West Grove, PA) and FITC-conjugated antibodies to mouse immunoglobulins (1:50 dilution; Amersham) were used to detect CD34 and GFAP labeling.3

Microscopy
Light microscopy was performed with Nomarski optics on a research microscope (Lietz DMBRE; Leica Microsystems, Wetzlar, Germany) or a photomicroscope (Photomicroscope II; Carl Zeiss Meditec, Meiden, Germany). Confocal microscopy was performed with an argon-krypton laser (Leica Microsystems) mounted on an epifluorescence photomicroscope (DMRBE; Leica Microscope Systems). FITC and Cy3 fluorescence was excited at 488 and 588 nm, respectively. Images were collected at a resolution of 300 pixels per inch and processed with image-analysis software (Photoshop 5.0; Adobe Systems, Mountain View, CA). A montage of a sector of the triple-labeled 14-WG retina was prepared to document the location of the APCs and astrocytes in relation to the developing vasculature, as previously described.5 The vasculature and the location of APCs and astrocytes were traced from the montage by hand. The tracing was then scanned and, from the scan, a color representation of the vasculature, APCs, and astrocytes was produced (Photoshop 5.0; Adobe Systems).

Mapping of the Outer Limit of Spindle-Shaped Cells and Determination of the Area of the Retinal Vasculature
From four specimens aged 14.5, 18, and 21 WG that were subjected to Nissl staining, the outer limit of spindle-shaped vascular precursor cells and the outer limit of solid vascular cords were determined as previously published by Hughes et al.5 In addition, one eye from each fetus aged 12 and 16 WG was subjected to CD34 immunohistochemistry, and the fellow eye was incubated for ADPase enzyme histochemistry.

Mapping the ADPase Positive Cells and Blood Vessels
Calibrated dark-field images of ADPase-incubated, flat-embedded retinas (12 and 16 WG) were used for manual mapping of ADPase+ cells with spindle-shaped and spherical morphologies in addition to formed retinal vessels. During mapping, the superficial and deep retina were inspected by adjusting focus up and down in each field.


   Results
Top
 Abstract
 Methods
 Results
Discussion
 References
 
Role of Vascular Precursor Cells in the Formation of the Primordial Vessels of Human Retina: Evidence from ADPase+ Enzyme Histochemistry
ADPase enzyme histochemistry was applied to determine the relationship between vascular precursor cells, vascular chords, and the leading edge of patent vessels. At 12 WG, there was a sparse ADPase+ vasculature in the inner retina (Fig. 1A) that was similar to the CD-34+ vasculature observed in the fellow eye (Fig. 1B) . There were also numerous single ADPase+ cells well in advance of the formed blood vessels (Fig. 1C) . At higher magnification, the single cells had two distinct morphologic shapes: spherical and spindle-shaped (Figs. 1D 1E) . Focusing exclusively on the superficial retina or on deeper retina suggested that the spindle-shaped cells were more superficial in the retina than were the spherical cells, which appeared to be deeper (Figs. 1D 1E) . This observation was confirmed by serial sectioning of the retinas after they were embedded in glycol methacrylate. Most of the spindle-shaped cells were in the superficial retina, and there appeared to be more spindle-shaped cells just in advance of the formed vasculature (Figs. 1F 1G) . Most of the spherical ADPase+ cells were deeper in the retina at the level of the ganglion cell layer. In some peripheral areas, there were also a few ADPase+ cells in the neuroblastic layer. When the round and spindle-shaped single ADPase+ cells were mapped in the flat perspective (Fig. 2C) , it was apparent that most of the cells distant from the formed vessels were spherical, and most spindle-shaped cells were immediately in advance of the formed vessels (Fig. 2E) .



View larger version (141K):
[in this window]
[in a new window]
  		FIGURE 1. ADPase-incubated retina from a 12-week gestation (WG) fetus. (A) Complete retina viewed as a flatmount before it was embedded in JB-4. The ADPase+ vasculature appeared white with darkfield illumination of the retina. The vasculature is butterfly-shaped and is limited to the peripapillary retina. The area peripheral to the vasculature indicated by the double arrow is also shown in (C). (B) Map of CD34+ blood vessels in the fellow retina to that shown in (A) and (C). The pattern is similar but less extensive than the pattern of formed vessels shown in (A) and (C). Half of the retina shown in (A) after it was flat embedded in JB-4 is shown in (C). At this magnification, ADPase+ cells were apparent more than 1 millimeter in advance of the vasculature (double arrow). (D) Most ADPase+ cells were spindle shaped (arrows) when focus was limited to the superficial retina in area with a double arrow in (A) and (C). (E) When the focus was deeper in the retina, most ADPase+ cells were spherical in shape (arrowheads). (F) Nomarski illumination of a section through same area shows that spindle-shaped, ADPase+ cells (arrows) were more superficial to spherical, ADPase+ cells (arrowheads). (G) Same section in (F) after the ADPase reaction product was developed with ammonium sulfide and counterstained with toluidine blue. The disposition of the two populations of ADPase+ vascular precursors (spindle, arrows; spherical, arrowheads) is obvious.

 


View larger version (53K):
[in this window]
[in a new window]
  		FIGURE 2. Maps of Nissl-stained cells and vasculature (A, B) and ADPase+ cells (C–E) and blood vessels in wholemounts from fetuses of different ages. (A) Map of the outer limits of Nissl-stained vascular cords and spindle cells in a 14.5-WG fetal retina. (B) Map of the outer limits of Nissl-stained vascular cords and spindle cells in an 18-WG fetal retina. (C) Map indicating the area occupied by all ADPase+ cells (red stippling) in the 12-WG fetus shown in Figure 1 . (D) Map indicating the area occupied by all ADPase+ cells (red stippling) in the 16-WG fetus shown in Figure 3 . (E) In this map of the 12-WG retina (Fig. 1A) , the ADPase+ formed vasculature is shown in green, the spindle-shaped ADPase+ cells in blue and the spherical ADPase+ cells in red.

 
At 16 WG, the formed vasculature had expanded substantially, in that the area vascularized had enlarged, and the four lobes, or arcades, of the vasculature had become more distinct (Fig. 3A) . Single ADPase+ cells were still present in advance of the vasculature by more than a millimeter (Fig. 2D) . The two populations of ADPase+ cells, spherical and spindle-shaped, were still present. Most of the spherical cells were at the level of the ganglion cell layer and the spindle-shaped cells in the more superficial retina (Fig. 3) . After they were mapped, the outer limits of the two populations appeared to remain static with maturation; however, the relative number of round cells declined with maturation. There were also individual ADPase+ cells posterior to and between the formed vessels (Fig. 3B) . There were both spherical and spindle-shaped cells in this population.



View larger version (138K):
[in this window]
[in a new window]
  		FIGURE 3. ADPase incubated retina from a 16-WG fetus. (A) When the retina was viewed as a flatmount before embedding in JB-4, the ADPase+ vasculature at this age was more extensive, and the four arcades were already defined. (B) A higher magnification of the area shown with a double arrow in (A). ADPase-positive vascular precursors were apparent in advance of the vasculature (double arrow) and between formed blood vessels. Single arrows: areas shown in sections (D–F). (C) Higher magnification of area at the edge of the forming vasculature shows an apparent angioblast aggregate that is contiguous with the other vascular elements only by a thin ADPase+ process (E, F, arrows). (D) Section through the area shows spindle-shaped angioblasts (arrows) in advance of and superficial to a vascular cord. Spherical angioblasts (arrowheads) were deeper in the retina at the level of the ganglion cell layer. (E, F) In the area indicated E&F (arrow) in (B) and (C) where an isolated aggregate was present, cell-free spaces formed by Müller cells were apparent, and spindle-shaped angioblasts (arrows) were more superficial to spherical angioblast (arrowheads). (D–F) Ammonium-sulfide–developed ADPase reaction (brown) with toluidine blue counterstain.

 
In both the 12- and 16-WG ADPase-incubated retinas, there appeared to be cells in transition between the two morphologic types (short spindle-shaped cells) between the ganglion cell layer and superficial retina in cross sections (Fig. 4) . This was most apparent in extracellular spaces created by the inner Müller cell processes within the inner retina in areas devoid of nerve fibers (Fig. 5) . The progression from single ADPase+ cells to ADPase+ endothelial cells was illustrated by sectioning from the midperipheral retina to the optic nerve head. The individual cells assembled into small aggregates at first (Fig. 4D) . As more cells were recruited to the aggregates, vascular cords were formed (Figs. 4E 4F 4G 4H) . Some cords remained unattached to canalized blood vessels (Fig. 4A) .



View larger version (149K):
[in this window]
[in a new window]
  		FIGURE 4. Edge of the ADPase+ vasculature in the 16-WG fetal retina. (A, B) Dark-field micrographs of the edge of the vasculature and surrounding retina. The area that is shown in semiserial sections (C–K) is enclosed in the box in (B) and includes single angioblasts aggregating (thin arrow) to the canalization of the vascular cord (thick arrow). As the structure was sectioned from periphery to disc, it appeared first as individual angioblasts (C, thin arrow; B) that assembled into aggregates (D) and eventually a cord (E–G). Cells appeared to join the cord in sections more central (H–J). Eventually, the cord or primordial blood vessel canalized (K) at the position indicated by the double arrowheads in (B). (K) The vascular channel as shown in (B, thick arrow). Thin arrows: indicate individual ADPase+ angioblasts.

 


View larger version (85K):
[in this window]
[in a new window]
  		FIGURE 5. Angioblasts (arrows) near the edge of vasculature in a 16-WG fetal human. ADPase+ angioblasts that occupied a position along the inner plexiform layer, were round, and had little cytoplasm (A). These cells increased in cytoplasmic volume and developed pseudopodia as they migrated within the cell-free space of the immature retina (B, D, E). Migration is accompanied by nuclear shape changes from rounded to spindlelike and the appearance of nucleolus-like bodies (arrow) within their cytoplasm (C). (F, arrow) An angioblast that reached the top of the space formed by the inner Müller cell processes.

 
Role of Vascular Precursor Cells in the Formation of the Primordial Vessels of Human Retina: Evidence from Nissl-Stained Wholemounts
Nissl-stained wholemount preparations of human fetal retina at 14.5, 18, and 21 WG revealed a substantial number of spindle-shaped cells (Figs. 2A 2B ; 6A ) well in advance of vascular cords, which were continuous with patent blood vessels. Nissl stained all cells in the retina including the vascular cells, which were identified based on location and morphology, but CD34 stained only established blood vessels and clearly demonstrated blood vessels with canalized lumens (Figs. 1B 6B 6E) . Focusing through the specimen showed that spindle-shaped Nissl stained cells were present superficially with their long axis oriented predominantly along the nerve fiber bundles. At 14.5 WG, the spindle cells showed a four-lobed topography, extending farthest temporally and superiorly (Fig. 2A) . The peripheral limit of the vascular cords and patent vessels also showed a four-lobed topography, as reported previously.5 36 52 With increasing maturity, the outer limit of the vascular cords expanded markedly, whereas that of the vascular precursor cells did not (Fig. 2) . With ADPase and Nissl staining (Figs. 2 6A 6C 6D) , large numbers of spindle cells were present well in advance of the vascular cords, their apparent aggregation to form cords and the development of patent vessels were consistent with the formation of primordial vessels in human retina through the mechanism of vasculogenesis.



View larger version (124K):
[in this window]
[in a new window]
  		FIGURE 6. Appearance of the edge of vasculature using different staining techniques. Nissl-stained preparation from an 18-WG specimen (A), a CD34-stained retina from an 18-WG specimen (B, E), and an ADPase-stained retina from a 12 WG-specimen (C, D). A section at the edge of the formed vasculature in (C) shows superficial spindle-shaped (arrows) and deeper round (arrowheads) ADPase+ angioblasts after development of the ADPase reaction product with ammonium sulfide (D). Whereas Nissl staining and ADPase histochemistry showed spindle-shaped cells in advance of the vasculature, CD34 labeled only formed blood vessels. Red blood cells were apparent in some canalized lumens of CD34+ blood vessels and filopodial extensions were apparent from the endothelial cells at the edge of the formed blood vessels (E, arrows).

 
The relationship between vascular precursor cells (angioblasts), vascular cords, and the leading edge of patent vessels was consistent for both Nissl staining and ADPase enzyme histochemistry (Figs. 2 6) ; however, absolute comparisons were not possible, because the ages of the specimens were not identical.

Relationship between Ecto-ADPase and VEGF-R2
Immunohistochemistry was used to determine whether ADPase+ cells had VEGF-R2, a marker for vascular precursors in some other tissues.53 54 CD39 was recently identified as ecto-ADPase, which is responsible for degrading luminal ADP and preventing platelet aggregation.50 CD39 was localized to formed blood vessels (CD34+) and to individual cells in advance of the vasculature (Fig. 7) , similar to enzyme histochemical labeling of ADPase (Figs. 1 3 4) . Formed blood vessels and individual cells in advance of the vasculature were also positive for VEGF-R2. All three markers labeled the choroidal vasculature (Fig. 7) .



View larger version (114K):
[in this window]
[in a new window]
  		FIGURE 7. Bleached full-thickness eye wall cryosections from a 16-WG fetal human showing CD34 (A, B), CD39 (C, D), and VEGFR-2 (E, F) immunolabeling in the avascular retina, peripheral to formed blood vessels (A, C, E), and at the border of vascularized retina (B, D, F). CD34 staining was absent in the avascular retina but clearly labeled the formed retinal vessels more posteriorly (B, long arrow) and the more mature choriocapillaris (A, B, arrowheads) and larger choroidal vessels. CD39 labeled angioblasts were in avascular retina and at the edge of formed retinal vessels (C, D, short arrows). In addition, formed retinal vessels (D, long arrow), choriocapillaris (C, D, arrowheads), and large choroidal vessels are also stained. VEGFR-2 labels angioblasts (E, F, short arrows) in inner avascular retina (E) and in advance of formed blood vessels (F, long arrow). Choriocapillaris (E, F, arrowheads) and large choroidal vessels are also positive for VEGF-R2.

 
Relationship between CD34+ Patent Vasculature and Pax-2+/GFAP– APCs
The relationship between formed blood vessels and Pax2+/GFAP– APCs and Pax2+/GFAP+ astrocytes was investigated by triple-antibody immunohistochemistry with CD34 (formed blood vessels), GFAP, and Pax2. CD34 immunohistochemistry showed the outer limit of patent canalized vessels as evidenced by the presence of red blood cells at the peripheral limits (Fig. 6E) . Further, CD34 visualized filopodial extensions of endothelial cells within formed blood vessels (Fig. 6E) , as reported previously.5

A recent report39 concluded that there is an absence of vascular precursor cells during formation of the mouse retinal vasculature and that earlier reports mistakenly identified APCs as vascular precursor cells. To examine this question more critically, we applied Pax-2/GFAP immunohistochemistry to several human fetal specimens, because Pax2+/GFAP– APCs are the earliest stage of the astrocytic lineage identified in vivo to date.3 Figure 8 shows the presence of a significant population of APCs just beyond the outer limit of CD34+ blood vessels at 14 and 20 WG. Pax-2+/GFAP–APCs were evident at 14 and 20 WG. Our earlier report showed that the density of APCs was reduced by 32 WG.3 Thus, there is a window during development when the number of APCs peaks and then diminishes significantly.



View larger version (157K):
[in this window]
[in a new window]
  		FIGURE 8. Human fetal retinal wholemounts triple labeled with Pax2/GFAP/CD34 at 14 weeks gestation (A–C). (A) At 14 WG, Pax2+(TR)/GFAP– (FITC) APCs extended in advance of the leading edge of CD34+ (FITC) blood vessels by a small but distinct margin. Arrows: some neonatal astrocytes that were just starting to express GFAP. (B) Representative field of view from midretina. Arrows: Pax2+/GFAP– APCs. Most cells in this area were Pax2+/GFAP+ immature astrocytes. CD34+ blood vessels (FITC) were also clearly evident. (C) A region near the optic nerve head (ONH) of a 14 WG fetus. Arrows: Pax2+/GFAP– APCs. Most of cells in this area were Pax2+/GFAP+ immature astrocytes. (D–F) Human fetal wholemounts quadruple-labeled with Pax2/GFAP/CD34/vimentin at 20 weeks’ gestation. (D) By 20 WG, Pax2+(TR)/GFAP– (FITC)/vimentin+ (Cy5) APCs extended in advance of the leading edge of CD34+ (FITC) blood vessels by a very small margin only. Arrow: neonatal astrocytes that were just starting to express GFAP. (E) By 20 WG, most cells have differentiated to become Pax2+/GFAP+/vimentin– mature perinatal astrocytes. (F) A region near the optic nerve head (ONH) of a 20-WG fetus. Most of cells in this area were Pax2+/GFAP+/vimentin– mature perinatal astrocytes. A small number of PAX2+/GFAP+/vimentin+ cells with the antigenic characteristics of immature perinatal astrocytes were still evident at this age, with only a few cells (arrows) still expressing vimentin (Cy-5 conjugated).

 
Whereas the margin between the leading edge of CD34 blood vessels and the leading edge of APCs decreased with maturation, Pax-2+ APCs preceded the leading edge of patent blood vessels by only a small margin, which varied between 20 and 110 µm even at 14 WG (Fig. 8) . The map in Figure 9 is a representation of the exact soma location of each APC and astrocyte and the CD34 vascular tree within a segment of a 14-WG human retina. Pax2+/GFAP– APCs (red somas, Fig. 9 ) were concentrated in a small rim just ahead of the leading edge of CD34+ vessel formation and persisted at lower density in more central regions of the retina. Pax2+/GFAP+ astrocytes (yellow somas, Fig. 9 ) with simple bipolar morphology were found at very low numbers preceding the leading edge of vessel formation. Astrocytes (Pax2+/GFAP+) were observed throughout the vascularized regions of the retina, frequently in close proximity to CD34+ formed blood vessels (Fig. 9) . This map confirms that the margin between APCs and CD34+ blood vessels is never greater than 120 µm. The margin between the leading edge of APCs and CD34+ blood vessels in human fetal retina demonstrated using Pax-2/GFAP immunohistochemistry differed markedly from the margin between ADPase+- and Nissl-stained vascular precursor cells and patent blood vessels demonstrated with ADPase and CD34.



View larger version (63K):
[in this window]
[in a new window]
  		FIGURE 9. Map of the 14-WG retina shown in Figure 7A 7B 7C . Red: Pax2+/GFAP– APCs; yellow: Pax2+/GFAP+ immature astrocytes; green: CD34+ blood vessels.

 
At 20 WG, the APCs were still slightly in advance of the formed blood vessels and dispersed over the central vascularized regions of retina. Few astrocytes were present at the peripheral edge of the formed blood vessels, whereas numerous astrocytes were associated with blood vessels and nerve fiber bundles in the central retina.

Evidence for Vascular Precursor Cells
Topographical analysis of spindle-shaped, Nissl-stained cells in superficial retina showed a typical four-lobed topography well in advance of the patent vascular cords (Figs. 2A 2B) , as reported previously.5 In contrast, ADPase was present only in vascular elements, cords, and formed blood vessels, as has been reported in normal retina of many species.32 43 55 56 57 ADPase was also present in vascular precursors (angioblasts), which were more than a millimeter in advance of formed blood vessels in the fetal human retina, which is in agreement with our observations in neonatal dogs.32 43 Application of the earliest available marker of the astrocyte lineage3 showed Pax2+/GFAP– APCs were never more than 120 µm in advance of the CD34+ formed blood vessels. Of the cell-specific markers used in this study, only ADPase was present in cells well in advance of the vasculature.

In summary, ADPase+ vascular precursor cells (angioblasts) preceded the leading edge of patent vessels by more than 1 millimeter at 12, 16, 17, and 19 WG. The most distant cells from formed vessels appeared mostly spherical, whereas most spindle-shaped ADPase+ cells were closer to the edge of the forming blood vessels. In contrast, Pax-2+/GFAP– APCs preceded the leading edge of CD34+ blood vessels by only a few micrometers, to a maximum of 120 µm. Astrocytes were almost exclusively observed within retina with formed vasculature. Furthermore, examination of wholemount ADPase preparations at high magnification revealed continuity between ADPase+ cells and the patent vasculature. Thus, using multiple-marker immunohistochemistry to detect cells of the astrocytic lineage, Nissl stain, ADPase, and CD34 immunohistochemistry, we have shown that APCs are a population of cells that are distinct from vascular precursor cells in the human fetal retina.


   Discussion
Top
 Abstract
 Methods
 Results
 Discussion
References
 
Evidence in Support of Vasculogenesis during Formation of the Primordial Human Retinal Vasculature
In the current study, a combination of Nissl staining, ADPase enzyme histochemistry, and triple-label immunohistochemistry for APCs showed that precursors for both endothelial cells and astrocytes were present in human fetal retina. ADPase+ and Nissl-stained spindle cells preceded the leading edge of patent vessels by up to 1 millimeter, whereas Pax-2+/GFAP– APCs preceded the leading edge of patent blood vessels by approximately 0.1 mm. This confirms earlier reports28 29 that the differentiation of astrocytes occurs in association with formed blood vessels.

The first suggestions of vascular precursors in retina came from nonspecific staining such as Nissl staining of cell somas, which demonstrated a spindle-shaped population of cells located superficially in advance of the vasculature.5 6 31 33 ADPase enzyme histochemistry is selective for blood vessels and vascular precursors or angioblasts.32 37 43 ADPase has been shown to be a vasculature-specific stain in normal retinas from mouse, rat, cat, dog, and humans.32 37 41 43 44 55 56 57 58 ADPase+ precursors have been observed in advance of the developing vasculature in both human and canine retinas32 37 and in rat and mouse (Ash J, McLeod DS, Lutty GA, unpublished data, 2004). In monkey, Gariano et al.59 reported that spindle cells were found 300 to 750 µm peripheral to the advancing vessels at fetal day 70.

In the human retina, vascular precursors or angioblasts had two morphologic profiles—spherical, and more superficial spindle-shaped—as we have observed in the dog.14 In dogs and humans, the transition in morphology from round to spindle-shaped was apparent within the cell-free spaces formed by the Müller cell processes in the inner retina, which we demonstrated in dog to be rich in low-sulfated glycosaminoglycans.14 The transition in cells from round to spindle-shaped in both dogs and humans appears to represent part of the angioblast differentiation process, and only the spindle-shaped ADPase-positive cells coalesce at the tips of forming cords and more centrally in vascular tubes.

VEGF-R2 is considered to be a marker for vascular precursors in some organs.53 54 We found VEGF-R2 in formed blood vessels and in individual cells present in avascular retina, peripheral to formed vessels, Gogat et al. 60 recently reported a similar localization of VEGF-R2 in human fetal retina.60 The localization of VEGF-R2 was similar to the ADPase-positive cells in the flat-embedded preparations. CD39, which is now considered to be ecto-ADPase,50 was also in cells in advance of formed vessels, according to immunohistochemistry.

The relative number of individual ADPase+ precursors between 12 and 16 WG was similar, suggesting that the pool of precursors is not sufficient to create a complete retinal vasculature. We have shown in an earlier report5 that the process of angiogenesis is responsible for increasing capillary density within the region of the primordial vessels formed by vasculogenesis, as well as the formation of the remaining peripheral retinal vessels and the outer vascular plexus. Because the initial vasculature forms by aggregation and canalization of ADPase+ angioblasts, vasculogenesis appears to be the primary mechanism for the formation of the primordial vessels. Because the pool of labeled angioblasts are restricted to the central two thirds of the retina and appears too few in number to complete formation of the superficial vascular network, observations from the present study support our earlier conclusion that angiogenesis is involved in the formation of some of the remaining retinal vessels in the human fetal retina.5 This progression of vasculogenesis forming the initial vasculature and angiogenesis then expanding the vascular network has been postulated in the brain.10

Evidence in Support of Angiogenesis during Formation of the Primordial Retinal Vasculature
Recently, investigators have concluded that vascular precursor cells identified by earlier workers in retina using markers including Nissl stain, GS ioslectin B4, and ADPase enzyme histochemistry in fact are not vascular precursor cells, but are APCs4 39 61 62 63 and that angiogenesis (formation of the retinal blood vessels from preexisting blood vessels) takes place without the prior invasion of vascular precursor cells in the mouse,39 ferret,63 monkey,61 and humans.4 Evidence in support of this conclusion has come from several sources. Attempts have been made to label vascular precursors in the retina with markers for vascular precursors in other tissues. Vascular precursors were not observed with antibodies against the VEGF receptors Flt-1 and Flk-1 in mouse39 and KDR in canine retinas.64 The absence of labeling for either receptor on cells in advance of the vasculature and the observance of these markers only in endothelial cells of formed tubes has led some investigators to conclude that vascular precursors do not exist in the retina. Therefore, at least in mice, they conclude that the retinal vasculature forms by angiogenesis, not vasculogenesis.39 Furthermore, spindle cells do not express factor VIII, CD31, CD34,5 61 65 or NADPH diaphorase,52 markers used to label mature endothelial cells.

Further evidence in support of the retinal vasculature’s forming by angiogenesis has come from GFAP-GFP transgenic mice, in which transgenic mice overexpress green fluorescent protein (GFP) under the control of the astrocyte-specific GFAP promoter.38 66 From findings in those studies, GFP-containing astrocytes in transgenic mice are purported to cover the entire mouse retina at birth, before any retinal vessels appear. This earlier appearance of GFP+ astrocytes can be interpreted as evidence that astrocytes in fact are the spindle cells in advance of the vasculature in mouse, which were identified using other markers in other species, although the GFAP-GFP cells are not spindle shaped. Dorrell et al.38 suggested that the GFAP-GFP astrocyte template serves as the template on which the retinal vasculature forms by angiogenesis.38 Otani et al.67 demonstrated that intravitreally administered hematopoietic stem cells target this astrocyte template and get incorporated into formed blood vessels. Other investigators have demonstrated that vascular endothelial cells can induce GFAP expression by astrocytes.28 Alternatively, it could be interpreted to mean that spindle-shaped cells (presumed angioblasts) do not exist in mice. Similar evidence is available from a comparative study of the ferret at birth that demonstrated strongly GFAP+ astrocytes extending over 22% of the radius of the retina, whereas weakly GFAP+ processes were already extending to the edge of the retina.63

The present study confirms our earlier report3 that there are two populations of astrocytes in developing human retina between 12 and 20 WG, Pax-2+/GFAP– APCs and Pax-2+/GFAP+ neonatal astrocytes. Pax-2+/GFAP– being the accepted antigenic expression that characterizes APCs in vivo3 and in vitro.67 Both populations are intimately associated with forming vasculature and, unlike the angioblast population, do not extend substantially beyond formed blood vessels. One possible limitation to this observation could be that Pax-2 does not identify the earliest astrocyte precursors. Until the availability of a marker expressed earlier along the astrocytic differentiation pathway is identified, our present interpretation of the data is well substantiated. Further, the process of angiogenesis by definition takes place through budding from existing vessels. If, as other investigators suggest, no vascular precursors invade the mammalian retina, then further studies are needed to determine the vessels from which angiogenesis originates.

In our study, ADPase+ angioblasts that were present far in advance of formed blood vessels were neither astrocytes nor astrocyte precursors, but the study did not address directly their relationship to microglia. However, extensive studies by Diaz-Araya et al.68 69 and Penfold et al.70 demonstrate that there are many differences in the timing, topography, morphology, and pattern of distribution between microglia and angioblasts.68 69 70 71 72 Microglia in fetal human retina are found in three layers of the retina (superficial, middle and deep) at 10 to 25 WG,68 whereas ADPase+ angioblasts are found predominantly in the ganglion cell and nerve fiber layers from 12 to 19 WG. Microglia somas are regularly spaced and their processes do not make contact with neighboring cells.68 Microglia are not present in large numbers near the tips of forming blood vessels in the ages studied, have a morphology that is mostly ramified, and have never been reported in cordlike structures.68 ADPase+ angioblasts are far more numerous near the tips of forming blood vessels, have a round or spindle-shaped morphology, and form queues as they align to aggregate into cords near the edge of the developing vasculature.

Species Differences
In our observations in the human fetal retina, the outer limit of vascular precursor cells displayed a four-lobed topography that never extended beyond the inner two-thirds of the retina. This is similar to the fetal feline retina, where the outer limit of vascular precursor cells showed a three-lobed topography during embryonic development; however, vascular precursor cells reached the ora serrata by the second postnatal week.31 In the dog, where the inner vasculature covers approximately 60% of the retina at birth, vascular precursor cells (ADPase+/ATPase+/M{alpha}GPDH+) extend to the ora serrata.14 32 37 73 Their transition from spherical to spindle-shaped occurs in cell-free spaces made by inner Müller cell processes, similar to our observations in human fetus. In monkey, Gariano et al.59 reported that "spindle cells were found 300 to 750 µm peripheral to the advancing vessels" at fetal day 70. Henkind and DeOliveira33 observed mesenchymal precursors well in advance of the formed blood vessels in the rat. They suggested that the mesenchymal precursors differentiates into endothelial cells and forms blood vessels, the current definition of vasculogenesis. Jiang et al.29 similarly reported the presence of lectin+ endothelial precursor cells ahead of the leading edge of vessel formation in rat. Compelling evidence for the existence of vascular precursor cells in the mouse retina is still not published, but we have observed ADPase+ cells in advance of the forming vasculature in P-3 mice (Ash J, McLeod DS, Lutty GA, unpublished data, 2004).

By immunohistochemical detection, GFAP+ retinal astrocytes have been observed to precede the formation of patent vessels by a short distance in humans,3 5 cats,19 rats,21 and mice.22 The dog differs slightly, as GFAP+ astrocytes were evident posterior to the tips of the newly formed blood vessels.74 In contrast, when astrocytes are detected in mice expressing GFP under control of the GFAP promoter, they showed an outer limit of GFP+ astrocyte distribution with a near circular topography, which reaches almost the periphery of the retina at birth.38 Further, using in situ hybridization for PDGF-{alpha}, Frutigger39 similarly showed a near circular topography for presumed astrocytes that almost reached the edge of the retina at birth in the mouse. The marked differences in the outer limit of astrocytes in the mouse could be due to a difference in the sensitivity of methods of detection or could truly reflect species differences in which the mode of vascular formation and astrocyte–vascular interactions during development is not representative of the primate or human retina.

The mouse retina has been suggested as a suitable model for the study of angiogenesis and for furthering our understanding of pathogenetic mechanisms of neovascularizing conditions of the retina because of the availability of transgenic and knockout mice. A review of the literature and the present study have reported the presence of spindle cells and ADPase+ vascular precursor cells in all species except mice, leading to the conclusion that the retinal vasculature is formed by both vasculogenesis and angiogenesis in the human, primate, cat, dog and rat, whereas the retinal vasculature in the mouse is formed by angiogenesis alone. If this significant species difference does exist between mice and man in the basic mechanism of retinal vascularization and in the relationship between the vasculature and astrocytes, caution must be exercised when interpolating observations made in neonatal mice.

The controversy over the mechanism by which the retinal vasculature forms—vasculogenesis and angiogenesis or angiogenesis alone—has important implications for the understanding of neovascularization of the retina. If in fact the human retina is vascularized through two distinct pathways and yet, in pathologic conditions, it is generally agreed that new vessels form by angiogenesis alone, the existence of a second pathway with distinct inhibitory and stimulatory signals could provide greater scope for intervention. Further studies are needed to elucidate both pathways more clearly.


   Footnotes
 
Supported by the National Health and Medical Research Council, Australia (TC-L), the Baxter Charitable Trust and J. S. Love Trust (TC-L), and National Eye Institute Grants R01-EY09357 (GAL) and EY01765 (Wilmer Ophthalmological Institute).

Submitted for publication October 23, 2003; revised February 16, 2004; accepted March 1, 2004.

Disclosure: T. Chan-Ling, None; D.S. McLeod, None; S. Hughes, None; L. Baxter, None; Y. Chu, None; T. Hasegawa, None; G.A. Lutty, None

The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Corresponding author: Gerard A. Lutty, 170 Woods Research Building, Johns Hopkins University School of Medicine, 600 North Wolfe St., Baltimore, MD 21287-9115; glutty{at}jhmi.edu.


   References
Top
 Abstract
 Methods
 Results
 Discussion
 References
 

  1. Jakobiec FA. Ocular Anatomy, Embryology, and Teratology. 1982; Harper and Row Publishers, Inc. Philadelphia.
  2. Mann IC. The Development of the Human Eye. 1928; Cambridge: University Press Cambridge, UK:.
  3. Chu Y, Hughes S, Chan-Ling T. Differentiation and migration of astrocyte precursor cells and astrocytes in human fetal retina: relevance to optic nerve coloboma. FASEB J. 2001;15:2013–2015.[Abstract/Free Full Text]
  4. Sandercoe TM, Madigan MC, Billson FA, Penfold PL, Provis JM. Astrocyte proliferation during development of the human retinal vasculature. Exp Eye Res. 1999;69:511–523.[CrossRef][ISI][Medline][Order article via Infotrieve]
  5. Hughes S, Yang H, Chan-Ling T. Vascularization of the human fetal retina: roles of vasculogenesis and angiogenesis. Invest Ophthalmol Vis Sci. 2000;41:1217–1228.[Abstract/Free Full Text]
  6. Ashton N. Retinal angiogenesis in the human embryo. Br Med Bull. 1970;26:103–106.[Free Full Text]
  7. Sugi Y, Markwald RR. Formation and early morphogenesis of endocardial endothelial precursor cells and the role of endoderm. Dev Biol. 1996;175:66–83.[CrossRef][ISI][Medline][Order article via Infotrieve]
  8. Coffin JD, Poole TJ. Embryonic vascular development: immunohistochemical identification of the origin and subsequent morphogenesis of the major vessel primordia in quail embryos. Development. 1988;102:1–14.[Free Full Text]
  9. Beck L, D’Amore PA. Vascular development: cellular and molecular regulation. FASEB J. 1997;11:365–373.[Abstract]
  10. Risau W. Development and differentiation of endothelium. Kidney Int. 1998;54(suppl.)S3–S6.[CrossRef]
  11. Risau W, Flamme I. Vasculogenesis. Ann Rev Cell Dev Biol. 1995;11:73–91.[CrossRef][ISI][Medline][Order article via Infotrieve]
  12. Risau W. Mechanisms of angiogenesis. Nature. 1997;386:671–674.[CrossRef][Medline][Order article via Infotrieve]
  13. Ekblum P, Sariola H, Karkiknen M, Saxen L. The origin of the glomerular endothelium. Cell Differ. 1982;11:35–39.[CrossRef][ISI][Medline][Order article via Infotrieve]
  14. McLeod DS, Lutty GA, Wajer SD, Flower RW. Visualization of a developing vasculature. Microvasc Res. 1987;33:257–269.[CrossRef][ISI][Medline][Order article via Infotrieve]
  15. Flower RW, McLeod DS, Lutty GA, Goldberg B, Wajer SD. Postnatal retinal vascular development of the puppy. Invest Ophthalmol Vis Sci. 1985;26:957–968.[Abstract/Free Full Text]
  16. Noden DM. Development of craniofacial blood vessels. Feinberg RN Sherer GK Auerbach R eds. The Development of the Vascular System. 1991;1–24. Karger New York.
  17. Duckett S. The establishment of internal vascularization in the human telencephalon. Acta Anat. 1971;80:107–113.[ISI][Medline][Order article via Infotrieve]
  18. Herken R, Gotz W, Wattjes KH. Initial development of capillaries in the neuroepithelium of the mouse. J Anat. 1989;164:85–92.[ISI][Medline][Order article via Infotrieve]
  19. Watanabe T, Raff MC. Retinal astrocytes are immigrants from the optic nerve. Nature. 1988;332:834–837.[CrossRef][Medline][Order article via Infotrieve]
  20. Ling T, Stone J. The development of astrocytes in the cat: evidence of migration from the optic nerve. Dev Brain Res. 1988;44:73–85.[CrossRef][Medline][Order article via Infotrieve]
  21. Ling T, Mitrofanis J, Stone J. Origin of retinal astrocytes in the rat: evidence of migration from the optic nerve. J Comp Neurol. 1989;286:345–352.[CrossRef][ISI][Medline][Order article via Infotrieve]
  22. Huxlin KR, Sefton AJ, Furby JH. The origin and development of retinal astrocytes in the mouse. J Neurocytology. 1992;21:530–544.[CrossRef][ISI][Medline][Order article via Infotrieve]
  23. Sarthy PV, Fu M, Huang J. Developmental expression of the glial fibrillary acidic protein (GFAP) gene in the mouse retina. Cell Mol Neurobiol. 1991;11:623–637.[CrossRef][ISI][Medline][Order article via Infotrieve]
  24. Turner DL, Cepko CL. A common progenitor for neurons and glia persists in rat retina late in development. Nature. 1987;328:131–136.[CrossRef][Medline][Order article via Infotrieve]
  25. Turner DL, Snyder EY, Cepko CL. Lineage-independent determination of cell type in the embryonic mouse retina. Neuron. 1990;4:833–845.[CrossRef][ISI][Medline][Order article via Infotrieve]
  26. Chan-Ling T. Glial, neuronal and vascular interactions in the mammalian retina. Prog Retin Res. 1994;13:357–389.[CrossRef]
  27. Chan-Ling T, Gock B, Stone J. The effect of oxygen on vasoformative cell division: evidence that "physiological hypoxia" is the stimulus for normal retinal vasculogenesis. Invest Ophthalmol Vis Sci. 1995;36:1201–1214.[Abstract/Free Full Text]
  28. Mi H, Haeberle H, Barres B. Induction of astrocyte differentiation by endothelial cells. J Neurosci. 2001;21:1538–1547.[Abstract/Free Full Text]
  29. Jiang B, Bezhadian MA, Caldwell RB. Astrocytes modulate retinal vasculogenesis: effects on endothelial cell differentiation. Glia. 1995;15:1–10.[CrossRef][ISI][Medline][Order article via Infotrieve]
  30. Chan-Ling T, Stone J. Degeneration of astrocytes in feline retinopathy of prematurity causes failure of the blood-retinal barrier. Invest Ophthalmol Vis Sci. 1992;33:2148–2159.[Abstract/Free Full Text]
  31. Chan-Ling T, Halasz P, Stone J. Development of retinal vasculature in the cat: processes and mechanisms. Curr Eye Res. 1990;9:459–477.[ISI][Medline][Order article via Infotrieve]
  32. McLeod DS, Crone SN, Lutty GA. Vasoproliferation in the neonatal dog model of oxygen-induced retinopathy. Invest Ophthalmol Vis Sci. 1996;37:1322–1333.[Abstract/Free Full Text]
  33. Henkind P, DeOliveira L. Development of retinal vessels in the rat. Invest Ophthalmol Vis Sci. 1967;6:530–520.
  34. Provis JM. Development of the primate retina. Prog Retin Eye Res. 2001;20:799–821.[CrossRef][ISI][Medline][Order article via Infotrieve]
  35. Serpell G. Polysaccharide granules in association with developing retinal vessels and with retrolental fibroplasia. Br J Ophthalmol. 1954;38:460–471.
  36. Nilhausen K. The vasoformative tissue in foetal retina with particular reference to the histochemical demonstration of its alkaline phosphatase activity. Acta Ophthalmol. 1958;36:65–70.[Medline][Order article via Infotrieve]
  37. Lutty GA, McLeod DS. A new technique for visualization of the human retinal vasculature. Arch Ophthalmol. 1992;110:267–276.[Abstract]
  38. Dorrell MI, Aguilar E, Friedlander M. Retinal vascular development is mediated by endothelial filopodia, a preexisting astrocytic template and specific R-cadherin adhesion. Invest Ophthalmol Vis Sci. 2002;43:3500–3510.[Abstract/Free Full Text]
  39. Fruttiger M. Development of the mouse retinal vasculature: angiogenesis versus vasculogenesis. Invest Ophthalmol Vis Sci. 2002;43:522–527.[Abstract/Free Full Text]
  40. Pardanaud L, Yassine F, Dieterlen-Lievre F. Relationship between vasculogenesis, angiogenesis and hemopoiesis during avian ontogeny. Development. 1989;105:473–485.[Abstract/Free Full Text]
  41. McLeod D, Goldberg M, Lutty G. Dual perspective analysis of vascular formations in sickle cell retinopathy. Arch Ophthalmol. 1993;111:1234–1245.[Abstract]
  42. McLeod DS, Merges C, Fukushima A, Goldberg MF, Lutty GA. Histopathological features of neovascularization in sickle cell retinopathy. Am J Ophthalmol. 1997;124:473–487.[ISI][Medline][Order article via Infotrieve]
  43. McLeod DS, Brownstein R, Lutty GA. Vaso-obliteration in the canine model of oxygen-induced retinopathy. Invest Ophthalmol Vis Sci. 1996;37:300–311.[Abstract/Free Full Text]
  44. McLeod DS, D’Anna SA, Lutty GA. Clinical and histopathologic features of canine oxygen-induced proliferative retinopathy. Invest Ophthalmol Vis Sci. 1998;39:1918–1932.[Abstract/Free Full Text]
  45. Chan-Ling T. Glial, vascular, and neuronal cytogenesis in whole-mounted cat retina. Microsc Res Tech. 1997;36:1–16.[CrossRef][ISI][Medline][Order article via Infotrieve]
  46. Schlingemann RO, Rietveld FJ, de Waal RM, et al. Leukocyte antigen CD34 is expressed by a subset of cultured endothelial cells and on endothelial abluminal microprocesses in the tumor stroma. Lab Invest. 1990;62:690–696.[ISI][Medline][Order article via Infotrieve]
  47. Provis JM, Van Driel D, Billson FA, Russell P. Development of the human retina: patterns of cell distribution and redistribution in the ganglion cell layer. J Comp Neurol. 1985;233:429–451.[CrossRef][ISI][Medline][Order article via Infotrieve]
  48. Lutty GA, Merges C, Threlkeld AB, Crone S, McLeod DS. Heterogeneity in localization of isoforms of TGF-b in human retina, vitreous, and choroid. Invest Ophthalmol Vis Sci. 1993;34:477–487.[Abstract/Free Full Text]
  49. Bhutto IA, Kim SY, McLeod DS, et al. Retinal and choroidal localization of collagen XVIII and the endostatin portion of collagen XVIII in aged human control and in age-related macular degeneration subjects. Invest Ophthalmol Vis Sci. .In press
  50. Marcus AJ, Broekman MJ, Drosopoulos JH, et al. Metabolic control of excessive extracellular nucleotide accumulation by CD39/ecto-nucleotidase-1: implications for ischemic vascular diseases. J Pharmacol Exp Ther. 2003;305:9–16.[Abstract/Free Full Text]
  51. Nornes HO, Dressler GR, Knapik EW, Deutsch U, Gruss P. Spatially and temporally restricted expression of Pax2 during murine neurogenesis. Development. 1990;109:797–809.[Abstract/Free Full Text]
  52. Provis JM, Leech J, Diaz C, Penfold PL, Stone J, Keshet E. Development of the human retinal vasculature: cellular relations and VEGF expression. Exp Eye Res. 1997;65:555–568.[CrossRef][ISI][Medline][Order article via Infotrieve]
  53. Yamaguchi TP, Dumont DJ, Conlon RA, Breitman ML, Rossant J. Flk-1, an flt-rated receptor tyrosine kinase is an early marker for endothelial cell precursors. Development. 1993;118:489–498.[Abstract]
  54. Millauer B, Witzigmann-Voos S, Schnuerch H, et al. High affinity VEGF binding and developmental expression suggest Flk-1 as a major regulator of vasculogenesis and angiogenesis. Cell. 1993;72:835–846.[CrossRef][ISI][Medline][Order article via Info