Norrie Gene Product Is Necessary for Regression of Hyaloid Vessels

Abstract

purpose. To investigate the nature and origin of the vitreous membranes in mice with knock-out of the Norrie gene product (ND mice).

methods. Eighty-two eyes of ND mice of different age groups (postnatal day [P]0–13 months) and 95 age-matched wild-type control mice were investigated. In vitreoretinal wholemounts and in sagittal sections, vessels and free cells were visualized by labeling for lectin. In addition, staining with a marker for macrophages (F4/80) and collagen XVIII/endostatin known to be involved in regression of hyaloid vessels was performed for light and electron microscopic investigations. Endostatin expression was confirmed by Western blot analysis.

results. Wild-type controls showed the typical pattern of hyaloid vessels, their regression and concomitantly retinal vasculogenesis and angiogenesis. Hyaloid vessels all stained for endostatin, whereas retinal vessels remained unstained. In ND mice, 1 to 5 days after birth, the hyaloid and retinal vasculatures were comparable to that in control mice. The hyaloid vessels also stained for endostatin. Numerous F4/80-positive cells were present adjacent to the vessels. With increasing age, only a few connecting branches of the hyaloid vessels regressed. Even in old mice most of the hyaloid vessels persisted. The vessels still stained for endostatin. Retinal angiogenesis was impaired.

conclusions. Retrolental membranes in ND mice consist of persistent hyaloid vessels, indicating that the ND gene product is important for the process of regression of these vessels. The ND gene product neither influences endostatin expression nor the presence of macrophages.

Norrie disease (ND) is a rare X-linked recessive disorder characterized by congenital blindness due to malformed retinas. ^{1 2 3 4 5} The eye disorder is characterized by bilateral retinal degeneration and extensive vitreous membranes. Previous studies have shown that ND is caused by mutations affecting the ND gene, which is located at on the short arm of chromosome X at position 11.4 and encodes a protein consisting of 133 amino acids. ^{6 7} Although the precise function of the ND protein remains unknown, protein structure and sequence comparison analysis suggest that the ND protein is a secreted protein rich in cysteines. ^{8 9}  

Berger et al. ¹⁰ were able to establish a mouse with a targeted disruption of the Norrie gene (ND mouse). In these mice retinal degeneration and vitreoretinal membranes develop similar to those that develop in patients with Norrie disease. ¹¹  

In the ND mice, retinal changes are accompanied by a striking impairment of retinal capillarization. ¹² Different kinds of vascular changes are observed. There is a lack of vessels in the outer retinal layers, but an increase in the number of vessels in the inner retinal layers. Some of the capillaries in the inner retinal layers show fenestrations and are in contact with vessels in the adjacent vitreous. No correlation between vascular abnormalities and retinal changes has been found. ¹²  

The nature of the large vitreous membranes observed in ND mice is not yet clear. 

The presence of retrolental membranes has been described in patients with retinopathy of prematurity (ROP), ^{13 14 15 16} familial exudative vitreoretinopathy (FEVR), ^{17 18 19 20 21 22 23} and persistent hyperplastic primary vitreous (PHPV). ^{24 25 26 27 28 29} In ROP and FEVR, most of the retrolental membranes presumably derive from proliferating retinal vessels, ^{14 15} whereas in PHPV the fetal vasculature persists in the vitreous body. ^{24 25 26 27 28 29}  

In an experimental study, vitreoretinal membranes consisting of persistent hyaloid vessels have been found in mice lacking collagen XVIII and endostatin. ³⁰ Endostatin is a 20-kDa protein that was extracted by O’Reilly et al. ³¹ in 1998 from a hemangioendothelioma cell line. The microsequence analysis of endostatin revealed identity to a C-terminal NC1 fragment of collagen XVIII, a well-known basal lamina heparan sulfate proteoglycan. Its tissue form, recognized by anti-endostatin antibodies, has a molecular mass of 38 kDa. ³¹ In vitro endostatin inhibits endothelial cell proliferation ³² and migration. ³³ It is also known that endostatin induces endothelial cell apoptosis. ³⁴ In vivo tumor growth is arrested by endostatin. The functional significance of endostatin in vivo is not yet fully understood. The findings in endostatin knockout mice indicate that the presence of this molecule is important for regression of hyaloid vessels. 

The purpose of this study was to analyze whether the vitreous membranes in ND mice consist of proliferating retinal vessels or of persistent hyaloid vessels and whether the lack of the Norrie disease gene product (NDP) influences expression of endostatin and/or macrophages normally phagocytosing the regressing hyaloid vessels. 

Methods

Animals

ND mice were generated by Berger et al. ¹⁰ using the gene-targeting technology. We received the mice as a gift from Wolfgang Berger (University of Zurich, Switzerland), and they were bred in Erlangen, as described by Richter et al. ¹²  

All mice were treated in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. The animals were kept under dim cyclic light (approximately 15 lux 12 h/d) with food and water ad libitum. 

Norrie gene product deficient mice (ND mice) were screened by isolating genomic DNA from tail biopsy specimens and tested for transgenic sequences by the use of polymerase chain reaction (PCR). ¹⁰ Male mice with the wild-type allele of the ND gene are referred to as normal mice (controls) and male mice with the defective gene copy as ND or mutant mice. 

For morphologic investigations of the eyes, the animals were anesthetized with ether and killed by cervical dislocation. 

Morphology

Preparation of Vitreoretinal Wholemounts.

After enucleation, the eyes were fixed in paraformaldehyde 4% for 4 hours. The eyes were then rinsed in phosphate-buffered saline (PBS; pH 7.4), and the posterior segment dissected along the ora serrata with fine scissors. Lens and anterior segments were carefully removed avoiding traction on the vitreous. The posterior segment was then incised in the periphery in full thickness between each quadrant. PBS was trickled carefully onto the retina, which detached itself from the retinal pigmented epithelium. With two pins, the retina was easily removed together with the optic disc and separated from the remaining tissues. 

Sagittal Sections.

The eyes were fixed in paraformaldehyde 4% for 4 hours and deep frozen in liquid nitrogen. Serial frozen 14-μm thick sagittal sections were cut through the entire globe. Midsagittal sections including lens and optic nerve head were used for immunohistochemical staining. 

Griffonia (Bandeiraea) simplicifolia Lectin I Isolectin B4.

Griffonia simplicifolia lectins bind to endothelial cells ³⁵ and can be used to visualize the entire vascular bed in wholemount preparations. ³⁶ They also label phagocytosing cells like macrophages and microglial cells. ³⁷  

Wholemounts of 16 eyes of ND mice and 18 of control mice aged P0 to P12 and 11 eyes of ND mice aged 1 month to 13 months and 14 age-matched controls were prepared and pretreated with Triton X+Tween-20+BSA overnight. After the specimens were rinsed in PBS three times, they were incubated with dry milk solution (Blotto; Santa Cruz Biotechnology, Santa Cruz, CA) for 1 hour and then overnight with biotinylated G. (Bandeiraea) simplicifolia lectin I isolectin B4 (Vector Laboratories-Linaris, Wertheim, Germany) diluted in TBS+BSA 2%+Triton X 1:100. Wholemounts were then washed in PBS three times, and Cy2-conjugated streptavidin antibody (1:2000; Jackson/Dianova, Hamburg, Germany) was applied. After they were rinsed in PBS, wholemounts were mounted in Kaiser glycerin jelly and viewed with a fluorescence microscope (Leica, Bensheim, Germany). In addition, sagittal section of eight eyes of ND mice and six eyes of control mice were stained. The sections were not pretreated with Triton X+Tween-20+BSA; otherwise the staining procedure was the same as described for the wholemounts. For the negative control, wholemounts were treated similarly, but without the primary antibody. 

Endostatin.

Wholemounts of 15 eyes of ND mice, 15 eyes of age-matched controls aged P0 to P12, 10 eyes of ND mice, and 8 eyes of wild-type controls aged P13 to 1 year were studied. Wholemounts were preincubated in 1 mL Triton-X +1 mL Tween-20+1 mL TBS overnight. After they were washed three times in PBS and incubated with dry milk solution (Blotto; Santa Cruz Biotechnology) for 60 minutes, the wholemounts were incubated with rabbit anti-endostatin (mouse) antibody (Chemicon International, Hofheim, Germany) at a dilution 1:250 in TBS and Triton-X and BSA 2% overnight at room temperature. After they were again washed in PBS three times, the specimens were incubated with the secondary goat anti-rabbit antibody AlexaFluor 488 (dilution 1:2000; Molecular Probes, Leiden, The Netherlands) for 2 hours at room temperature. After another rinse in PBS, wholemounts were mounted in Kaiser glycerin jelly and viewed with a fluorescence microscope (Leica). 

For the negative control, the wholemounts were treated as just described but without the primary antibody. Absorption control was performed using human recombinant endostatin (Calbiochem, Bad Soden, Germany) together with the endostatin antibody. 

In addition, sagittal sections of 8 eyes of ND mice (P0–P12) and 8 age-matched control mice, as well as 16 eyes of ND mice aged P13 to 1 year and 10 age-matched controls, were stained for endostatin in the same way. For lectin and endostatin double staining, lectin labeling was performed first. For visualization of endostatin, a Cy3-conjugated goat anti rabbit antibody (1:2000; Jackson/Dianova, Hamburg, Germany) was used. 

F4/80.

F4/80 is an antibody that is expressed on bone marrow–derived cells and especially on macrophages and dendritic cells. To characterize further the vitreous-free cells, retinal wholemounts of 6 eyes of ND mice and 10 eyes of control mice aged P0 to P12 were prepared and pretreated with Triton-X+Tween-20+BSA as described earlier. After incubation with dry milk solution (Blotto; Santa Cruz Biotechnology) for 1 hour, the specimens were incubated overnight with biotinylated F4/80 (dilution 1:200; Linaris, Wertheim, Bettingen, Germany). As a secondary antibody, streptavidin-Cy3 (1:2000; Jackson/Dianova) was used. For double staining, incubation with lectin was performed as described earlier and incubation for F4/80 followed. Negative controls were performed with the same protocol but without the primary antibody. 

Electron Microscopy and Immunogold Labeling.

Eight eyes of ND mice and eight age-matched control animals (P0 to 13 months) were fixed in solution containing paraformaldehyde 2.5%, glutaraldehyde 2.5%, and picric acid 0.05% in cacodylate buffer (pH 7.3) ³⁸ for at least 24 hours. After they were rinsed in cacodylate buffer (pH 7.2), the eyes were bisected sagittally and the lenses removed carefully with fine forceps, guided by a stereoscopic dissecting microscope (Carl Zeiss Meditec, Oberkochen, Germany). For electron microscopy, the specimens were postfixed in 1% OsO₄, dehydrated in an ascending series of alcohol, and embedded in Epon or methacrylate (Technovit 7100; Heraeus Kulzer, Wehrheim, Germany) according to standard methods. 

Semithin sections were cut with a microtome (Ultracut OmU3; Reichert, Vienna, Austria) and stained with hematoxylin-eosin or toluidine blue. Ultrathin sections were stained with uranyl acetate and lead citrate and viewed with an electron microscope (EM 902; Carl Zeiss Meditec). 

For immunoelectron microscopy, cryosections were mounted on sheets coated with sheets with poly-l-lysine (Thermanox; Electron Microscopy Sciences, Fort Washington, PA). After preincubation with dry milk solution (Blotto; Santa Cruz Biotechnology) for 30 minutes, the sections were incubated overnight with endostatin antibody (Chemicon International, Hofheim, Germany) diluted 1:250 in PBS and 2% BSA. After the sections were washed with PBS and 2% BSA, they were incubated overnight with secondary antibody conjugated with ultrasmall gold (1:100). The sections were then postfixed with 2.5% glutaraldehyde for 2 hours and washed in PBS and distilled water. Silver enhancement followed for 90 minutes. After they were washed, the sections were postfixed with 0.5% OsO₄ for 15 minutes, dehydrated in an ascending series of alcohols, and embedded in Epon according to standards. Ultrathin sections were cut with a microtome (Reichert) and the sections viewed with an electron microscope (EM 902; Carl Zeiss Meditec, Oberkochen, Germany). Control experiments were performed in the same way as described for immunohistochemistry. 

Western-Blot Analysis

To confirm specificity of the endostatin antibody, Western blot analysis was performed. Eight control mice 2 to 3 months old were used. Lens capsule and retina were dissected and lysed in SDS sample buffer for gel analysis. The samples were boiled 5 minutes, and protein content was measured using bicinchoninic acid (BCA) protein assay reagent (Pierce, Rockford, IL). Proteins (5 μg/lane) were loaded and separated by polyacrylamide gel electrophoresis (PAGE), using a 5% SDS-polyacrylamide stacking gel and a 12% SDS-polyacrylamide separating gel. After electrophoresis, the proteins were transferred by semidry blotting (Bio-Rad, Hercules, CA) onto a polyvinyl difluoride membrane (Roche, Mannheim, Germany). The membrane was incubated with PBS, containing 0.1% Tween-20 (PBST; pH 7.2) and 3% BSA for 1 hour. An anti-rabbit endostatin antibody (diluted 1:1000; Chemicon International) was added and allowed to react overnight at 4°C. As a preabsorption control, human recombinant endostatin was added to the primary antibody solution. After a wash with PBST, alkaline phosphatase-conjugated goat anti-rabbit IgG (diluted 1:1000; Promega, Madison, WI) was added for 30 minutes. Visualization of alkaline phosphatase was achieved, using chemiluminescence. CDP-star (Roche) was diluted 1:100 in detection buffer, and the filters were incubated for 5 minutes at room temperature. Chemiluminescence was detected with an imaging workstation (Lumi-Imager; Roche). Exposure times ranged between 1 and 5 minutes. 

To compare the expression of endostatin in 5-day-old ND mice and controls with similar morphology of the vitreoretinal wholemounts, we removed the lenses from the globes and processed the remaining tissues for Western blot analysis as described earlier. Tissues of four eyes (two animals) of each group were pooled. Intensities of hybridization signals were determined using the imaging system software (Lumi-Analyst; Roche). 

Results

Postnatal Days 1 to 5

Control Mice.

In male mice with the wild-type allele of the ND-gene (controls), the hyaloid vessels spread radially from the optic disc to the periphery, forming the typical polygonal network of interconnecting capillaries. ³⁹ In the periphery, the vessels gathered so that the hyaloid vasculature appeared to be surrounded by a vascular ring (Fig. 1a) . 

At P1 a superficial layer of capillaries started to develop in the retina. A dense network of capillaries was present at the optic disc, covering less than one third of the inner retina (Fig. 1b) . Elongated cell processes connected to the capillary tubes spread toward the periphery. In the vicinity of and between the retinal and hyaloid vessels, numerous round or dendriform lectin-labeled cells were seen. The round cells were often located directly at the vessel walls, whereas the dendriform cells extended into the intervascular spaces (Figs. 1a 1b) . Double-labeling with lectin and the marker for macrophages (F4/80) revealed that most of the lectin-positive cells also stained for F4/80. 

Double labeling for lectin and endostatin clearly showed that all hyaloid vessels were endostatin-positive (Fig. 1b) . The sprouting superficial capillaries of the retina, however, were only slightly stained. The cells sprouting from the vessels toward the periphery remained unstained (Fig. 1b) . In midsagittal sections through the entire globe, intense staining for endostatin was seen in the hyaloid vessels and the lens capsule. The peripheral retina was unstained (Fig. 1c) . 

Norrie Mice.

Between P1 and P5, the staining pattern and shape of the hyaloid vessels in ND mice were the same as described for the control mice. This was also true for the superficial capillaries of the retina. In ND mice, there were also numerous lectin and F4/80-labeled cells located adjacent to or between the hyaloid vessels. In Norrie mice as in the control animals, all hyaloid vessels stained intensely for endostatin (Fig. 1d) , whereas the superficial central retinal capillaries were only slightly labeled. As in the control animals, the peripheral retina remained unstained (Fig. 1d) . 

Postnatal Days 6 to 9

Control Mice.

By P6, hyaloid vessels were still present running from the optic disc toward the ringlike vessel in the periphery. The number of interconnecting vessels, especially in the central hyaloid, appeared reduced, however. At this stage of development, the network of superficial capillaries in the retina covered more than two thirds of the inner retina. In the following days, regression of hyaloid vessels continued so that by P8 to P9, most of the interconnecting vessels had disappeared. Some of the remaining vessels appeared considerably narrow. Thinning of this kind was seen in the entire length of these vessels. Lectin- and F4/80-stained cells were still visible along and between capillaries. The hyaloid vessels at this stage of development still revealed positive staining for endostatin, and even degenerating ones remained endostatin immunoreactive (IR; Fig. 2a ). The retinal vessels now covered the entire surface of the retina. At this stage, angiogenesis of the retina had started, so that capillaries were now also seen in the outer retina (Fig. 2b) . None of the retinal vessels was labeled with the endostatin antibody. 

Norrie Mice.

In ND-mice up to P8 and P9, most of the hyaloid vessels persisted (Fig. 2c) . The vessels still gathered in an interconnecting ring at the periphery. Only some of the interconnecting capillaries of the hyaloid vasculature showed signs of regression with thinning of the lumen (Fig. 2c) . Numerous lectin- and F4/80-positive cells were found throughout the wholemount and adjacent to regressing vessels (Fig. 2c) . 

The pattern of the superficial retinal vessels differed considerably from that in the controls. Between days 6 and 9, the capillary network appeared more irregularly arranged and the intervascular spaces enlarged (Fig. 2c) . The hyaloid vessels all stained for endostatin, whereas the retinal vessels remained unstained. Serial sagittal sections through the posterior eye segment revealed that within the outer retina, nearly no lectin-labeled vessels were present. Lectin labeling of vessels was restricted to the innermost retina (Fig. 2d) . 

Older than P9

Control Mice.

In the controls, regression of hyaloid vessels continued up to P14 to P21, when the entire hyaloid vessels had disappeared. Concomitantly, vasculogenesis and angiogenesis of the retina were complete. In the posterior eye segment, staining for endostatin was restricted to the inner limiting membrane. Retinal vessels remained unstained for endostatin. 

Norrie Mice.

In Norrie mice of all age groups, hyaloid vessels still persisted. The pattern of the persisting vessels showed individual differences. In some animals, less than 1 year old, the pattern was similar to that shown for the 8-day-old mice (Figs. 2c 2d) , whereas in others some more interconnecting capillaries had regressed. Most of the straight vessels were present, and in the periphery, numerous interconnections were still visible. All persisting hyaloid vessels stained for endostatin, whereas the vessels in the inner retina remained unstained. There were numerous lectin- and F4/80-positive cells in the vitreous. In ND mice older than 1 year, the pattern of the persisting hyaloid vessels changed compared with that in younger mice. The larger vessels no longer followed a straight course but were irregularly arranged (Fig. 3) . Some of the interconnecting vessels formed loops. At places, connections between endostatin-positive hyaloid vessels and the endostatin-negative lectin-labeled retinal vessels were seen (Fig. 3) . These intervascular connections between the hyaloid vessels represent straight connections of single vessels. Sprouting of more than one vessel from the retina into the vitreous was never observed. 

At the ultrastructural level, the persisting hyaloid vessels in ND mice showed the same morphology as that of young control mice. The capillary endothelium was surrounded by pericytes (Fig. 4a) . Adjacent to the outer surface of the basement membrane, macrophage-like cells (hyalocytes) ensheathing the pericytes were seen. The basement membrane of the persisting hyaloid vessels showed immunogold labeling for endostatin, especially at the outer surface. There was, however, also slight labeling at that part of the basement membrane lining the endothelial cells (Fig. 4b) . This pattern of endostatin staining was the same as that seen in young control animals. 

Western Blot Analysis

The presence of endostatin was confirmed by Western blot analysis. Two bands were found at approximately 35 to 38 kDa in the lens capsule of adult control mice (Fig. 5a) . After addition of human recombinant endostatin as absorption control, both bands were no longer visible. Endostatin can be generated from collagen XVIII by action of various proteases and at different cleavage sites ⁴⁰ . Therefore, more than one band was also found by other investigators, demonstrating expression of endostatin. ^{32 40} In 5-day-old ND and control mice with still similar morphology of the vitreoretinal wholemounts, there were no quantitative differences in protein expression of endostatin in the eye tissues (Fig. 5b) . 

Discussion

Our studies clearly demonstrate that the vitreous membranes in adult mice with knockout of the Norrie gene mainly consist of persistent hyaloid vessels. 

In the nonaffected male controls (ND controls), regression of hyaloid vessels occurred in a similar time course and pattern as has been described in normal mouse eyes. ³⁹ In control eyes, regression started around P6 in the peripheral hyaloid vessels, especially in the interconnecting branches. At P16 to P21, the hyaloid vessels had completely disappeared. Concomitantly in the same time span, the vascularization and capillarization of the retina took place. ^{41 42} In contrast, in ND-mice, degeneration of hyaloid vessels between days P6 and P21 was barely noticeable; only single interconnecting branches in the peripheral hyaloid disappeared. 

The function of the ND gene product is not yet known, and antibodies for demonstration of the localization of this protein are not available. In previous studies ND mRNA has been localized in the retina; however, only adult mice without hyaloid vessels have been investigated. ^{10 41} Trials to localize the mRNA in vitreoretinal wholemounts did not provide sufficient evidence. It is therefore not known whether the ND gene product is normally present in the hyaloid vasculature, or whether knockout of the ND gene influences other factors involved in normal regression of hyaloid vessels. 

In previous studies, several factors have been hypothesized to be responsible for the normal regression of the hyaloid vasculature (e.g., stretching of the vessels due to enlargement of the eye ⁴² or changes in blood flow ⁴³ that might be related to the development of the superficial retinal vasculature). The eyes of ND mice enlarge during the first days of development as in the control mice and the inner retina becomes vascularized. Therefore, purely mechanical factors cannot be causative for the persistence of hyaloid vessels. 

In ND mice capillarization of the retina and the stria vascularis of the inner ear is impaired. ^{12 44} Do such developmental changes also occur in hyaloid vessels and secondarily influence regression of the vasculature? In 1- to 5-day-old ND mice, the morphology and pattern of the hyaloid vasculature was nearly the same as in normal controls. At the ultrastructural level, even in old ND mice the different parts of the hyaloid vasculature showed the typical morphology of healthy young mice. Therefore, primary vascular changes are presumably not responsible for the persistence of the hyaloid vessels in ND mice. 

In mice with knockout of endostatin, hyaloid vessels persist. ³⁰ Could knockout of the ND gene influence endostatin expression? Our investigations on the expression of endostatin in ND mice and controls using immunohistochemical, biochemical, and immunocytochemical methods did not reveal any difference between ND mice and their age-matched controls. Even the persistent hyaloid vessels in older ND mice remained endostatin positive. 

Another factor involved in regression of hyaloid vessels is the phagocytosis of vascular cells. ^{45 46 47 48 49} Could macrophages lack in the vitreous of ND mice? The presence of vitreal macrophages was investigated using lectin and the F4/80 antibody. In ND mice, numerous F4/80-labeled cells were seen in the vitreous and also adjacent to the few regressing capillaries, similar to what has been found in normal young mice. 

The present immunohistochemical studies performed on wholemount preparations confirmed that the persistence of hyaloid vessels in ND mice was accompanied by impaired angiogenesis of the retina. ¹² Although in the first days of postnatal development, the superficial layer of retinal capillaries sprouted to the periphery in a manner similar to that in the controls, remodeling of these vessels and sprouting into the outer retina was sparse. This impaired capillarization was restricted to the retina. The other vascular beds of the eye as the choroidal vasculature or the limbal vessel were normal (Lütjen-Drecoll E, Ohlmann AV, unpublished observation, 2002). Vascularization of the retina and outgrowth of capillaries into the outer retina are impaired by lack of vascular endothelial growth factor (VEGF). ⁵⁰ Hypoxia upregulates VEGF, ^{51 52 53} whereas hyperoxia suppresses VEGF expression. ^{51 53} Physiological hypoxia can be induced by an increased retinal thickness and increased metabolism of retinal cells concomitant with regression of hyaloid vessels. This has been discussed as being the driving force for retinal capillarization. ^{50 53} On the contrary, in retinopathy of the prematurity (ROP), high oxygen levels have been discussed as the most important factors in the pathogenesis of the impaired vascularization of the retina. ⁵² This increase in oxygen levels in ROP inhibits capillarization in the peripheral retina. Secondarily, proliferation of retinal vessels into the vitreous occurs with decreasing oxygen levels and an increase in VEGF. ⁵⁴ In ND mice, as in ROP, angiogenesis is impaired. It is tempting to speculate that in addition to the absence of the ND gene product in the retina, increased oxygen levels due to persistence of hyaloid vessels may contribute to the observed changes in retinal vasculature in ND mice. In contrast, in ND mice, the morphology of the retinal and hyaloid vasculature differed from that in ROP. In ND mice, we found no sprouting of retinal vessels into the vitreous. The restriction of endostatin IR to the hyaloid vessels allowed a clear distinction between hyaloid and retinal vessels. There were only straight connections between vitreal vessels and retinal vasculature in ND mice, ¹² and they were only seen in mice older than ∼1 year. Therefore, we assume that these connections occur secondarily. It is possible that sprouting of retinal vessels does not occur in ND mice because the oxygen levels in the vitreous presumably remain constantly increased. ⁵⁵  

In summary, our findings show that in mice with knockout of the ND gene there is the persistence of hyaloid vessels accompanied by impaired angiogenesis of the retina. Whether the ND gene product influences both processes separately or whether impaired sprouting of retinal vessels into the outer retina is not only due to the lack of the ND gene product in the retina, but in addition to the increased oxygen levels in the vitreous due to persistence of hyaloid vessels has to be clarified. 

 Supported by Interdisziplinäre Zentrum Klinische Forschung (IZKF).

 Submitted for publication November 6, 2003; revised March 9, 2004; accepted March 11, 2004.

 Disclosure: A.V. Ohlmann, None; E. Adamek, None; A. Ohlmann, None; E. Lütjen-Drecoll, 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.

 

Figure 1.

 

(a, b) Controls P1: Vitreoretinal wholemounts double labeled with lectin (green) and antibodies against endostatin (red). (a) All hyaloid vessels stained for endostatin and lectin (yellow). In the periphery the hyaloid vasculature forms the typical polygonal network of interconnecting vessels, which gather in a ring-like vessel (arrow). In mice of this age, there was also slight staining for endostatin in the superficial retinal vessels. (b) In the area of the optic disc the superficial retinal vessels formed a dense network of the superficial capillaries covering one third of the inner retina (arrows). (c, d) Midsagittal sections through the globe of 5-day-old mice stained with antibodies against endostatin. (c) Control mouse: there was intense staining of hyaloid vessels in the vitreous and at the lens capsule as well as of the lens capsule itself (arrowheads). The lens (L) itself and the peripheral retina (R) remained unstained. (d) ND mouse: at P5 there was no difference in staining pattern (arrowheads) between ND mice and control animals.

Figure 2.

 

Control mice P8: (a) vitreoretinal wholemount doublestained for endostatin (red) and lectin (green). Most of the interconnecting hyaloid vessels had disappeared. The single hyaloid vessels (arrowheads) leading from the optic disc to the periphery still stained for endostatin. In the retina there was a dense superficial capillary network labeled for lectin (arrows). This network was present at the entire inner retina. The retinal vessels were not stained for endostatin. (b) Sagittal section through the posterior eye segment with optic disc labeled for lectin. There was intense labeling of vessels in the inner and outer retina (arrows). Remnants of the hyaloid vessels were seen adjacent to the optic disc (arrowhead). ND-mice P8: (c) vitreoretinal wholemount double stained for endostatin (red) and lectin (green). Most of the hyaloid vessels persisted (arrowheads). Only some interconnecting capillaries had disappeared, and some of the straight running vessels showed narrowing of their lumen. All hyaloid vessels were still double stained for lectin and endostatin. The retinal capillaries stained only for lectin (arrows). They formed an irregular sparse superficial network. (d) Midsagittal section through the posterior globe. Lectin-labeled vessels were present in the vitreous (arrowheads) and in the inner retina (arrows). In the remaining retina no labeled vessels were seen.

Figure 3.

 

ND mouse 1 year old: vitreoretinal wholemount double stained for endostatin (red) and lectin (green). The hyaloid vessels were still present and intensely stained for lectin and endostatin. They formed an irregular vascular network. At some places, straight connections between hyaloid (yellow) and retinal vessels (green) formed (arrows).

Figure 4.

 

Electron micrographs of hyaloid vessels of adult ND mice: (a) after normal staining and (b) after immunogold labeling for endostatin. Comparison of (a) and (b) shows that most intense staining for endostatin was present at the outer BM of the vessel. There was, however, also some staining at the inner portion of the BM between endothelial cells and pericytes. E, endothelial cells; BM, basement membrane ensheathing the pericytes (P).

Figure 5.

 

Western blot analysis for endostatin: (a) In the lens capsule of adult control mice a double band was visible at ∼35 kDa. Expression of more than one signal has also been described by other investigators. ^{37 40} When preabsorption controls with endostatin were used, both bands were no longer visible. (b) Western blot analysis of eye tissues without lens. The amount of endostatin was compared in 5-day-old normal (WT) and ND mice. No difference in endostatin expression was observed.

The authors thank Hong Nguyen, Anke Fischer, and Heide Wiederschein for excellent technical assistance and Marco Gösswein who expertly prepared the micrographs.