Early Vascular and Neuronal Changes in a VEGF Transgenic Mouse Model of Retinal Neovascularization

Abstract

purpose. To investigate early retinal changes in a vascular endothelial growth factor (VEGF) transgenic mouse (tr029VEGF; rhodopsin promoter) with long-term damage that mimics nonproliferative diabetic retinopathy (NPDR) and mild proliferative diabetic retinopathy (PDR).

methods. Rhodopsin and VEGF expression was assessed up to postnatal day (P)28. Vascular and retinal changes were charted at P7 and P28 using sections and wholemounts stained with hematoxylin and eosin or isolectin IB4 Griffonia simplicifolia Samples were examined using light, fluorescence, and confocal microscopy.

results. Rhodopsin was detected at P5 and reached mature levels by P15; VEGF protein expression was transient, peaking at P10 to P15. In wild-type (wt) mice at P7, vessels had formed in the nerve fiber/retinal ganglion cell layer and showed a centroperipheral maturational gradient; some capillaries had formed a second bed on the vitread side of the inner nuclear layer (INL). By P28, the retinal vasculature had three mature capillary beds, the third abutting the sclerad aspect of the INL. In tr029VEGF mice, capillary bed formation was accelerated compared with that in wt, with abnormal vessels extending to the sclerad side of the INL by P7 and abnormally penetrating the photoreceptors by P28. Compared with P7, vascular lesions were more numerous at P28 when capillary dropout was also evident. At both stages, retinal layers were thinned most where abnormal vessel growth was greatest.

conclusions. Concomitant damage to the vasculature and neural retina at early stages in tr029VEGF suggest that both tissues are affected, providing opportunities to examine early cellular events that lead to long-term disease.

Clinically, diabetic retinopathy (DR) has chronic progression, with early nonproliferative diabetic retinopathy (NPDR) being characterized by increased vessel permeability, microaneurysms, and capillary loss. ^{1 2 3 4 5 6} Ensuing retinal ischemia triggers proliferative DR (PDR), with neovascularization leading to fibrovascular changes, retinal traction, detachment, and blindness. ⁷ Loss of cells in the neural retina as well as glial changes occur early in DR and often precede vascular disease. ^{8 9 10 11}  

The usefulness of animal models for DR depends on the degree to which the pathologic features mimic those seen clinically. STZ models suffer the disadvantages of only transient and often late-onset changes with no neural loss, as well as systemic toxicity. ^{12 13 14} The Ins2^Akita mouse is spontaneously hyperglycemic and characterized by mild and late NPDR with some neuronal loss in the inner nuclear layer (INL) and inner plexiform layer (IPL), ¹⁵ whereas the IGF-1 transgenic progresses to severe PDR with retinal detachment and cataract ¹⁶ ; neuronal damage has not been examined. Eye-specific VEGF transgenics have also been generated, and while some show severe and rapid progression to PDR, ^{17 18 19} others display retinopathy that is relatively slow and mild (tr029VEGF). ^{19 20} Neural damage in the VEGF models was either not examined ^{18 21} or limited to a qualitative description of INL and outer nuclear layer (ONL) thinning. ¹⁷  

We have recently charted vascular changes in tr029VEGF from 6 weeks after birth and show progression from NPDR to mild PDR with increased permeability; pericyte and endothelial cell loss; vessel tortuosity leukostasis; and capillary blockage, dropout, and hemorrhage. ²⁰ In this study, we investigated earlier stages, from before the onset of transgene expression (embryonic day [E]18) until 4 weeks after birth, to examine the sequence of abnormal vessel formation and compare initial changes in the vasculature with those in the neural retina. Part of this work has been published in abstract form. ²²  

Materials and Methods

Animals and Anesthesia

Procedures were performed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and the Institutional Animal Ethics Committee. Fourth- and fifth-generation tr029VEGF litters were bred as previously described. ¹⁹ Mice were screened by amplification of tail genomic DNA. Anesthesia was administered with ketamine and xylazine (100 and 10 mg/kg body weight intraperitoneally [IP], respectively; Troy Laboratories Pty. Ltd., Smithfield, NSW, Australia); euthanasia was performed with pentobarbitone (200 mg/kg body weight IP; Jurox, Newcastle, NSW, Australia). Wild-type (wt) littermates served as control subjects. 

Color Fundus Photography and Fluorescein Fundus Angiography

The retinal vasculature was viewed at P28 (n = 3) using color fundus photography and fluorescein fundus angiography to examine microaneurysm formation and vascular leakage, as previously described. ^{19 23}  

RNA Extraction and cDNA Synthesis

Retinal RNA was extracted (Masterpure DNA and RNA purification kit; MC 85200) and quality and concentration checked spectrophotometrically. For cDNA synthesis, 1 μg of RNA was mixed with 500 μM of oligo dT and 40 mM dNTPs, incubated at 70°C for 5 minutes with 200 units of reverse transcriptase (Superscript II; Invitrogen Life Technologies), five times first-strand buffer, 40 units of RNase inhibitor (Invitrogen Life Technologies, Sydney, Australia) and 5 mM dithiothreitol (DTT) were added and incubated (42°C, 50 minutes), followed by heat inactivation (70°C, 15 minutes). One unit of RNaseH (Invitrogen Life Technologies) was added and incubated (30°C, 20 minutes); cDNA was stored at −20°C. 

Real-Time PCR Analysis

PCR reaction components for quantitation were 2 picomoles of primer, nucleic acid stain (SYBR Green), 10× reaction buffer, 2 mM MgCl₂, 0.2 mM dNTPs, 0.5 units polymerase (Platinum Taq; Invitrogen Life Technologies), and 1 μL of cDNA per 20 μL. Reactions were performed (Rotorgene 3000; Corbett Research, Mortlake, NSW, Australia), dissociation profiles were used to check single-product amplification, and random samples were sequenced for verification. 

Protein Analysis

Ocular hVEGF₁₆₅ was quantified in tr029VEGF at P10, P15, P20, and P28 (n = 4 eyes per stage) using ELISA (Quantikine; R&D Systems, Minneapolis, MN). ²⁰  

Tissue Preparation: Sectioned Material

Eyes were marked dorsally, fixed in 4% paraformaldehyde for 2 hours, placed in 70% ethanol overnight, and wax embedded. Orientation was so as to section ventrodorsally and reveal the nasotemporal axis; 1 of every 10 sections (6-μm sections) was collected. 

Series were stained with hematoxylin and eosin, rhodopsin, isolectin IB4 (Griffonia simplicifolia), or glial fibrillary acidic protein (GFAP). Sections were permeabilized (1% with Triton X-100 plus cations in PBS, blocked with 0.3% H₂O₂) and washed in PBS. Primary antibodies were diluted in PBS Triton X/cations: rhodopsin Rho-4D2 (1:1000; the kind gift of David Hicks, INSERM, City University Hospital, Strasbourg, France), isolectin IB4 Griffonia simplicifolia (1:100; Vector Laboratories, Burlingame, CA), GFAP (1:1000; Dako, Botany, NSW Australia) for incubation overnight at 4°C. Slides were washed, incubated with streptavidin-horseradish peroxidase (50% in PBS, LSAB Kit; DakoCytomation) at room temperature for 1 hour, washed, incubated in 3,3′-diaminobenzidine (DAB) substrate, washed, dehydrated, cleared, and mounted. 

Wholemounts

Retinas with a dorsal orientation mark were fixed in 4% paraformaldehyde (PBS, pH 7.4, 20 minutes) and washed in PBS/cations (0.1 mM MnCl₂, 0.1 mM MgCl₂, and 0.1 mM CaCl₂ in PBS) with Triton X-100 (1%). Wholemounts were incubated in biotinylated isolectin IB4 (1:100 dilution, in PBS/cations; Vector Laboratories) at 4°C overnight, washed, incubated in a peroxidase conjugate (Extravidin; 1:150 in PBS; Sigma-Aldrich) for 1 hour at room temperature, washed, coverslipped in PBS, and kept moist at 4°C. 

Tissue Analysis: Sectioned Material

Retinal structure at E18 (n = 3 wt and n = 3 tr029VEGF), before transgene expression, was assessed in hematoxylin and eosin (H&E)–stained sections. Retinal thickness and its component nuclear and plexiform layers were measured in H&E-stained sections at P7 (n = 3) and P28 (n = 3) with commercial software (ImagePro Plus; Media Cybernetics, Silver Spring, MD) To ensure comparability of data, measurements were confined to sections encompassing the optic disc and were taken from central (200–500 μm of the optic disc) and nasal and temporal peripheral (200–500 μm of the limbus) locations. For tr029VEGF, we analyzed central regions according to whether they had abnormal vessels. Wt nasal and temporal peripheral retinas were not significantly different (P > 0.05), and those data were pooled; central and peripheral locations differed significantly (P < 0.05) and were treated separately. 

Blood vessel location within the different retinal layers was charted in sections stained with isolectin IB4 at P7 (n = 3) and P28 (n = 3). Low-power images from the midventral, central, and middorsal retina were captured and used as a template to map blood vessels, which were identified at high power. Cup-shaped maps of individual sections were converted to straight lines. ²⁴  

Wholemounts

The panretinal distribution of the retinal vasculature was examined in isolectin IB4-stained wholemounts from the vitread surface at P7 (n = 7) and P28 (n = 8) with fluorescence or confocal microscopy, and photomontages were generated. For tg029VEGF, we mapped regions of capillary dropout and the location of vascular lesions that were defined as bright fluorescent spots resembling a spectrum from small microaneurysms to larger intraretinal microvascular abnormalities (IRMAs) seen both experimentally and clinically. ^{19 20 25} The numbers of vascular lesions was compared between the dorsonasal, dorsotemporal, ventronasal and ventrotemporal retinal quadrants and between central and peripheral retinal halves. 

The detailed structure of the vascular beds and vascular lesions was analyzed with confocal microscopy. Representative areas were captured in the z-plane in 1.5- to 2.0-μm steps. Color coding was used to distinguish vessels in the upper (red: nerve fiber–retinal ganglion cell layers, NFL/RGCL) middle (green), and lower (blue) capillary beds, which abutted, respectively, the vitread and sclerad margins of the INL. Yellow was used for vessels that had extended into the photoreceptor outer segments (OS). 

Statistics

Data were compared by using the two-sample Student’s t-test assuming unequal variances (significance P < 0.05; Excel; Microsoft, Inc., Redmond, WA). 

Results

Rhodopsin Expression

In wt mice, rhodopsin mRNA expression was detectable at P5 and increased thereafter (Fig. 1A) ; expression did not differ significantly between wt and tr029VEGF (Fig. 1A ; P > 0.05). At P7 and P28, rhodopsin immunostaining was uniformly intense from the center to the periphery (Figs. 1B 1C 1D 1E 1F 1G 1H 1I) . By P28 in both wt and tr029VEGF, an immunopositive ONL and OS could be distinguished centrally and peripherally (Figs. 1F 1G 1H 1I) . 

VEGF Expression

hVEGF₁₆₅ mRNA transcript was detected at all stages tested (Fig. 1J) . In agreement with previous findings, ²³ hVEGF₁₆₅ protein expression was transient—being, respectively, 38.9 ± 3.7 and 36.6 ± 2.3 pg/mg at P10 and P15 but declining to 10.4 ± 2.4 and 15.2 ± 1.1 pg/mg at P20 and P28 (Fig. 1K) . 

Vascular Leakage

In wt mice at P28, the retinal vasculature was mature, having major retinal vessels that radiated from the optic disc and an evenly distributed capillary bed; fluorescein leakage was not observed (Figs. 1L) . In tr029VEGF, mild changes were seen with enlarged major vessels and evidence for vascular lesions and some leakage (Figs. 1M) . 

Retinal Vascular Morphology and Distribution of Blood Vessels

At E18, H&E sections revealed no discernible differences in overall retinal architecture between wt and tr029VEGF (Fig. 2) . At P7, IB4-stained wholemounts of both wt and tr029VEGF revealed a regular alternating pattern of radial retinal arteries and veins and an upper capillary bed that had extended to within 100 to 200 μm of the periphery (Figs. 3A 3B 3M 3N) . In addition, in both wt and tr029VEGF, a maturational centroperipheral gradient was seen within the upper capillary bed. Centrally, the capillaries displayed a regular, open network whereas peripherally, the arrangement was more complex and the advancing front of newly formed vessels could be seen at the far periphery (Figs. 3E 3F 3G 3H) . By P28 in wt, the vasculature was mature (Figs. 3C 3I 3J) . 

Vascular lesions were present at P7 in tr029VEGF but were more numerous and prominent at P28, although their numbers were variable (Figs. 3B 3D 3G 3H 3K 3L 3M 3N 3O 3P) . At P7, the number of vascular lesions was 54 ± 31, did not favor any retinal quadrant (dorsonasal: 14 ± 11; dorsotemporal: 16 ± 20; ventronasal: 10 ± 6; ventrotemporal: 13 ± 15; P > 0.05) but were concentrated more centrally than peripherally (central: 41 ± 20; peripheral: 11 ± 16; P < 0.05). By P28, the number of vascular lesions had increased (Figs. 3D 3K 3L 3O 3P)to 153 ± 128, did not favor any retinal quadrant (dorsonasal: 30 ± 42; dorsotemporal: 42 ± 45; ventronasal: 33 ± 22; ventrotemporal: 47 ± 33; P > 0.05) and showed no significant difference centrally compared with peripherally (central: 72 ± 73; peripheral: 81± 58; P > 0.05). Centrally, some regions showed capillary dropout (Figs. 3O 3P) . 

IB4 staining in retinal wholemounts showed that vessel formation in tr029VEGF was accelerated compared with wt. At P7 in wt, the upper capillary bed displayed a regular arrangement and initial sprouting toward the inner retina. The first growth of vessels to form the middle bed was seen as bulbous sprouts connected to parent capillaries in the upper bed by a thin bridge (Figs. 4A 4B) . Areas with more mature, continuous vessels were also seen (Fig. 4C) . However, some peripheral regions had only an upper capillary bed and had yet to start forming the middle bed (Fig. 4D) . In tr029VEGF, some retinal regions appeared normal (Fig. 4E)but, in contrast to wt, most had begun to form the middle bed. Sprouts forming the middle bed were larger and more bulbous compared with wt (Figs. 4E 4F) . In other regions, abnormal vessels had formed and extended farther toward the outer retina (Figs. 4G 4H) . Some vascular lesions were discrete, and the upper and middle beds appeared similar to normal (Fig. 4G) . In others, vascular lesions were large and tortuous, and both the upper and middle beds showed dropout (Fig. 4H) . Vascular lesions were not observed in the upper capillary bed. 

By P28 in wt, three (upper, middle, and lower) capillary beds had formed and appeared as a regular and nonoverlapping array (Fig. 4I) . A possible sequence of vascular lesion formation is shown in Figures 4J4K4L4M4N4O4P . In tr029VEGF, some regions appeared normal (not shown), whereas others had slight capillary enlargements in the middle and lower beds (Fig. 4J) . Vascular lesions were more common than at P7. Some were continuous between the middle and lower beds (Figs. 4K4F) . Others formed a continuous structure between the lower bed and the OS (Figs. 4K4L)and appeared to enlarge by folding back on themselves (Figs. 4M4N) . Vascular lesions were larger in the lower bed and in the OS compared with those in the inner bed (Figs. 4O4P) . Furthermore, capillary beds appeared normal away from the vascular lesions (Figs. 4J4L)but suffered dropout immediately above and surrounding them (Figs. 4F4G4H4I4J4K4L4M4N4O4P) . Reconstructions through the z-plane (Fig. 5 ; Movies 1–4) revealed that vascular lesions comprised continuous capillaries that were highly tortuous and had a diameter of 27.8 ±1 0.3 μm (range, 10–46 μm). 

H&E sections allowed three-dimensional mapping of capillaries in relation to the retinal layers and confirmed the acceleration of capillary bed formation in tr029VEGF (Figs. 6A 6B 6C 6D 6E 6F 6G 6H) . In tr029VEGF, abnormal vessels also invaded and disrupted the ONL and OS as well as the retinal pigment epithelium (Figs. 6D 6H 6I 6J 6K) . GFAP staining of Müller cell processes showed no difference between wt and tr029VEGF at P7, with immunostaining being restricted to the vicinity of the inner limiting membrane (ILM; Figs. 6L 6M ). By P28, the OLM was also stained in both wt and tr029VEGF although immunoreactivity was greater in tr029VEGF and increased in regions associated with vascular lesions (Figs. 6N 6O) . 

Changes in the thickness of the retinal layers were quantitated from H&E-stained sections. At P7 in tr029VEGF, regions both with and without vascular lesions, were significantly reduced, the effect being greater where lesions were present. In central regions with vascular lesions, the overall thickness of the retina was considerably reduced compared with (wt: 190.2 ± 9.3 μm; tg: 160.8 ± 7.2 μm; P < 0.05) with significant reductions in the NFL, INL, and ONL (P < 0.05; Figs. 7C 7D 7F 7H 7J ). The overall thickness of central retinal regions lacking vascular lesions was also significantly less than in the wt (tg: 169.6 ± 3.3 μm; P < 0.05), a difference attributable to a thinner INL (P < 0.05; Figs. 7A 7B 7F 7J ). The same pattern was seen in the peripheral retina lacking vascular lesions, with overall retinal thickness being less in tr029VEGF than in wt (tg: 166.6 μm ± 3.9; wt: 189.9 ± 3.8 μm; P < 0.01), again with the difference being attributable to a thinner INL and NFL (P < 0.01; Figs. 7D 7F 7J ). 

At P28 in tr029VEGF, there was a similar pattern of greater thinning associated with the vascular lesions, but different retinal layers were affected compared with P7. For central retina with vascular lesions, overall retinal thickness was reduced compared with wt with losses attributable to reductions in the IPL and OS (tg: 126.5 ± 16.6 μm; wt: 180.5 ± 3.1 μm; P < 0.05; Figs. 7M 7O 7S 7T ). Overall thickness of central retinal regions lacking vascular lesions was also reduced compared with wt (tg: 168.1 ± 5.5 μm; P < 0.05) with differences being attributed to reductions in OS (Figs. 7K 7L 7S 7T) . The overall thickness of peripheral retina which lacked retinal lesions did not differ compared with wt (tg: 150.16 μm; wt: 150.5 ± 2.7 μm; P > 0.05; Fig. 7T ). 

Discussion

Early changes within the retinal vasculature and neural retina of tr029VEGF were mild and coincided with early hVEGF expression. Initially, vascular lesions were small and resembled microaneurysms that were visible clinically but then progressed to larger structures with contorted capillaries that resembled IRMAs. Thinning of the neural retina coincided with the formation of vascular lesions, but not all retinal layers were affected equally. Furthermore, although retinal regions with vascular lesions experienced the most damage, regions distant from lesions were also somewhat affected. 

Our results support previous evidence of an early but transient upregulation in hVEGF₁₆₅ protein in tr029VEGF ^{19 23} which, as we showed, coincides with early damage to the vasculature and neural retina. Such damage persists up to 24 weeks, even in the absence of continued protein expression, ²³ suggesting that an initial “spike” in hVEGF₁₆₅ is sufficient to induce long-term changes characteristic of NPDR. The greater vascular and retinal damage seen in advanced PDR in other models is presumably related to sustained and high levels of VEGF expression. ¹⁸  

VEGF models differ from DR because abnormal vessel growth is toward the transgene source within the photoreceptors in the outer retina rather than toward the inner retina and vitreous. ^{17 18 19 20 21} However, although the direction of abnormal vessel growth might be different in our VEGF model compared with human DR, we showed that the process of new vessel formation and associated complications was the same. Furthermore, the damage that we and others have observed within photoreceptors is also seen in human DR, suggesting that these cells are particularly vulnerable. ²⁶ We have also recently shown that changes within the retinal vasculature in tr029VEGF are accompanied by structural damage to the choroid that mimics that in DR, ^{3 27 28 29} suggesting that pathophysiological changes in the vascular beds either side of the photoreceptors are associated with their loss. 

Induction of VEGF in photoreceptors in adult retinas leads to profound and rapid damage, with detachment and obliteration of the retina within a few days ²¹ precluding analysis of initial cellular changes or of how NPDR might progress to PDR. VEGF transgenic models with mild damage offer the advantage of being able to track early cellular changes within the vasculature and neural retina and examine their interrelationships. Nevertheless, such changes occur within a developmental context. 

Although rhodopsin is first expressed in the murine retina at P5, ³⁰ the inner retina has yet to fully mature, and the ONL and OS are not yet distinct. Also, the central-to-peripheral gradients of cell division and synaptogenesis are not complete until P11. ^{31 32} Our immunohistochemical analysis of rhodopsin expression and the thickness of the OS is similar peripherally and centrally and that at P7. Taken together, the data suggest that elevated hVEGF₁₆₅ concentrations are likely to be equal across the retina. Nevertheless, in tr029VEGF at P7, growth of vessels toward the outer retina is accelerated centrally compared with peripherally and in addition, vascular lesions appear first centrally. The pattern matches the central-to-peripheral and inner-to-outer sequence of normal vessel formation observed in mammals ^{33 34 35 36 37 38} and suggests that older vessels are more susceptible than younger ones. Of note, the initial central location of vascular lesions in tr029VEGF is similar to DR in which neovascular changes consistently originate close to the optic disc in 90% of cases. ³⁹  

The bulbous sprouts that extend into the INL at P7 and which are connected to parent capillaries by thin bridges are identical with microaneuryms observed histologically in adult human DR (Fig. 1Bin Ref. ²⁵ ), indicating a similar mechanism of formation in the developing mouse and in diabetes. Furthermore, microaneurysms in wholemounts of both tr029VEGF and humans appear as bright fluorescent spots of similar size (humans, 15–70 μm diameter ²⁵ ; tr029VEGF, 10–46 μm diameter). Confocal 3-D analysis of small lesions in tr029VEGF and of microaneurysms in humans revealed limited tortuosity, ²⁵ whereas large lesions in tr029VEGF were highly tortuous vessels and resembled the IRMAs seen clinically. ^{40 41 42} Our observations indicate a continuum from microaneurysm to IRMA in tr029VEGF and by analogy also in clinical DR. In humans, microaneurysms are often blocked by blood cells. ^{25 43} Presumably, the tortuous nature of the vascular lesions in tr029VEGF also contributes to capillary blockage in addition to the leukostasis that we have observed in this model. ^{20 44 45}  

Although, by far, most studies on DR have focused on retinal vascular changes, early observations in diabetic patients have also revealed neural losses. ^{46 47} More recently, attention has returned to the neural retina with evidence for early functional changes as revealed by abnormal ERG recordings, thinning of the different retinal layers, and neural apoptosis. ^{10 11 48 49 50 51 52 53 54}  

Our findings show that overall retinal thickness was reduced at P7 and further by P28 and support the clinical scenario, although not all layers were affected equally. At P7, reductions occurred within the NFL/RGCL, INL, and ONL suggesting losses within each nuclear layer, but by P28, no significant differences were seen between wt and tr029VEGF, indicating recovery. However, thinning of nuclear layers could reflect the same number of cells but with smaller sizes and therefore increased density or changed extracellular space, a feature that is likely, since edema occurs in DR. ^{1 2}  

By P28, however, the IPL was thinner than in wt animals, suggesting reductions in dendritic tree size and/or synaptogenesis. Indeed, ERG recordings at early stages of DR reveal changes in oscillatory potentials indicative of degraded amacrine cell function. ⁵⁵ However, the most striking loss was within the OS which were considerably shortened. Similarly, in human DR donor tissue, photoreceptors appear to be particularly susceptible. ^{56 57}  

We have yet to determine the mechanisms whereby elevated VEGF leads to early onset of both vascular and neural damage. Nevertheless, we showed that vascular and neural damage coincided chronologically, since we saw changes in both components very soon after transgene expression. Damage was also spatially correlated, since regions with vascular lesions displayed more neural damage than those without. Furthermore, neural retina lacking vascular lesions was also affected, but less so, suggesting long-range effects possibly from diffusion of soluble VEGF. Transgenic models such as tr029VEGF provide an opportunity to examine cellular changes that mimic those seen in DR and to determine the extent to which damage cascades within the vasculature impinge on the neural retina and vice versa. 

 Supported by the Juvenile Diabetes Research Foundation, a National Health and Medical Research Council (NH&MRC) Program Grant (EPR), and a Westpac Banking Corporation grant-in-aid. SAD is an NH&MRC Senior Research Fellow (Grant 254670).

 Submitted for publication March 9, 2006; revised June 20, 2006; accepted August 21, 2006.

 Disclosure: P.E. van Eeden, None; L.B.G. Tee, None; S. Lukehurst, None; C.-M. Lai, None; E.P. Rakoczy, Westpac Banking (F); L. D. Beazley, Westpac Banking (F); S.A. Dunlop, Westpac Banking (F)

 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) Rhodopsin promoter as assessed by quantitative RT-PCR in wt and tr029VEGF. Rhodopsin was barely detectable at P5 and increased to mature levels by P15. (B–E) Photoreceptor-specific opsin expression at P7 in wt and tr029VEGF retina as assessed immunohistochemically. Immature immunopositive photoreceptors, which mostly lacked OS, were seen. In both wt and tr029VEGF, staining intensity was similar centrally compared with peripherally. (F–I) At P28, a similar immunostaining pattern was seen in both wt and tr029VEGF compared with P7, except that immunopositive somata could be distinguished from OS. Staining intensity was similar centrally and peripherally. Immunopositive cells in the presumptive INL at P7 and the INL at P28 (arrows) are ectopic photoreceptors. (J, K) hVEGF₁₆₅ mRNA transcript was present throughout, whereas protein expression decreased with time. (L, M) Fundus fluorescein angiography at P28 showing microaneurysms and vascular leakage in tg (M) but not wt (L) animals. Scale bar, 25 μm.

Figure 2.

 

Sectioned wild-type (A) and transgenic (B) retinas stained with H&E at E18 showing a similar overall architecture before the onset of transgene expression. The hyaloid vasculature can be readily distinguished (arrow). Scale bar, 200 μm.

Figure 3.

 

(A–D) Photomontages of retinal wholemounts stained with IB4 at P7 and P28 in wt and tr029VEGF animals in central (E, G, I, K) and peripheral (F, H, J, L) retina. The centroperipheral maturational gradient at P7 was seen in both wt and tr029VEGF. At P28, the vasculature was mature in wt but showed vascular lesions and dropout in tr029VEGF. (M–P) Representative maps drawn from IB4-stained tr029VEGF wholemounts at P7 (M, N) and P28 (O, P) showing distribution of vascular lesions (dots) and major areas of capillary dropout (hatched). Arteries and veins are represented by lines. Scale bars: (A–D) 1 mm; (E–L) 200 μm; (M–P) 1 mm.

Figure 4.

 

Retinal wholemounts at P7 (A–H) and P28 (I–P) stained with IB4 showing the upper (NFL/RGCL: red), middle (vitread side of the INL: green), and lower (sclerad side of the INL: blue) capillary beds in wt (P7: A–D; P28: I) and tr029VEGF (P7: E–H; P28: J–P) animals. (A–D) In wt animals at P7, the upper capillary bed (red) had formed and had thin bridges (green, arrows) tipped by bulbous sprouts (green, open arrows) which extended toward the INL (A, B). In some instances, (C, arrowhead) capillaries in the middle bed (green) were continuous rather than ending. In the far periphery, the upper capillary bed (red) lacked capillary sprouting toward the INL. (E–H) Micrographs showing proposed sequence of vascular lesion formation at P7. Capillary beds showing initial bulbous structures sprouting toward the INL were numerous compared with wt (E, green, open arrows) and were connected to the upper capillaries (red) by thin bridges (E, green, arrow). Continuous, but swollen, capillaries in the middle capillary bed were also seen (E, F, green, arrowheads). In addition, vascular lesions extended toward the outer retina (G, H, blue). In G, the vascular lesion (blue) was relatively confined, whereas the upper (red) and middle (green) capillary beds appeared normal (*). (H) The vascular lesion (blue) was more extensive, and both the upper (red) and middle (green) capillary beds showed capillary dropout. (I) At P28 in wt animals, the upper (red), middle (green), and lower (blue) vascular beds formed a regular, mostly nonoverlapping array. (J–P) Sequence of micrographs showing presumed progression of vascular lesions at P28 in tr029VEGF, from discrete swellings to large complex structures. (J) Normal-appearing region with the exception of mild vascular lesions appearing as swellings along capillaries in the lower bed (blue, arrowheads). (K) Discrete vascular lesions showing slight round enlargements of capillaries in the inner (green; arrowhead) and lower beds (blue; arrowhead) and with relatively normal-appearing capillaries elsewhere (*). (L) Vascular lesions in the middle bed (green, arrowhead) and a larger lesion that appeared to have arisen by elongation and folding (J, blue; arrowheads). (M–P) Progressively larger lesions in the lower bed (blue, arrowheads) accompanied by further folding and extension into the photoreceptor layer (yellow, arrowheads). Scale bars, 100 μm.

Figure 5.

 

Z-stacks of vascular lesions in Figures 4F(A), 4L(B), 4N(C), and 4P(D). Vascular lesions were not simple enlargements of capillaries but rather were composed of continuous and contorted capillaries that were folded on themselves that resemble microaneurysms and IRMAs seen in clinical DR. Scale bar, 50 μm.

Figure 6.

 

Sectioned retinas stained with isolectin IB4 at P7 and P28 in wt (A, C) and tr029VEGF (B, D) showing the location of vessels in relation to retinal layers. Vessels formed the upper capillary bed in the NFL/RGCL (red arrows) and the middle and lower capillary beds on the vitread (green arrows) and sclerad (blue arrows) sides of the INL. In addition, in tr029VEGF, vessels extended into the OS (yellow arrow). Some vessels in the outer retina in tr029VEGF appeared abnormal and tortuous (B, D; blue arrows). (E–H) Representative reconstructions of three sections from dorsal (top row), central (middle row), and ventral (bottom row) retina depicting blood vessels in the upper (red dots), middle (green dots), and lower (blue dots) capillary beds and vessels extending into the photoreceptors (yellow dots) at P7 (E, F) and P28 (G, H) in wt (E, G) and tr029VEGF (F, H). Lines in rows of red dots indicate the extent of retinal tissue, (i.e., to the limbus). (I–K) H&E-stained sections of vascular lesions showing disruption of the outer retina and RPE. (L–O) Sectioned retinas stained with GFAP at P7 (L, M) and P28 (N, O) in wt (L, N) and tr029VEGF (M, O), showing location of GFAP-immunopositive Müller cell processes. At P7, no differences were detected between wt and tr029VEGF, with staining confined to the ILM. By P28 in wt animals, staining was also present in the OLM but in tr029VEGF, was upregulated and localized immunopositive Müller cell processes, which extended into the photoreceptor layer and were associated with vascular lesions (arrows). (D, V, N, T) Dorsal, ventral, nasal, and temporal, respectively. Scale bars: (A–D, I) 100 μm; (J, K) 20 μm; (L–O) 50 μm.

Figure 7.

 

Sectioned retinas stained with H&E at P7 (A–C) and P28 (K–M) from central retina in wt (A, K) and tr029VEGF (B, C, L, M) animals. (B, L) Representative micrographs showing minimal vascular abnormalities and minimal changes to the neural retina; (C, M) regions with greater vascular damage and concomitantly greater disruption of the neural retina and thinning of the INL. (D–J, N–T) histograms of the overall thickness of the retina and its component layers at P7 (D–J) and P28 (N–T) for retinal regions without (transgenic: no lesion) or with (transgenic: with lesion) vascular lesions. NFL/RGCL is nerve fiber–retinal ganglion cell layer. *Significantly different from normal (P < 0.05). Scale bars: (A–C, K–M) 25 μm.

Supplementary Materials

Movie 1 - 2.39 MB (.mov) - Supplement to Figure 5A. 

Movie 2 - 568 KB (.mov) - Supplement to Figure 5B. 

Movie 3 - 716 KB (.mov) - Supplement to Figure 5C. 

Movie 4 - 704 KB (.mov) - Supplement to Figure 5D. 

The authors thank Marissa Penrose and Michael Archer for technical assistance and the Centre for Microscopy and Microanalysis.