Regulation of Ocular Angiogenesis by Notch Signaling: Implications in Neovascular Age-Related Macular Degeneration

Abstract

Purpose.: Wet age-related macular degeneration (AMD), which accounts for most AMD-related vision loss, is characterized by choroidal neovascularization (CNV). The underlying mechanism of CNV is poorly understood, but evidence indicates pathologic recruitment of normal angiogenic signaling pathways such as the VEGF pathway. Recent evidence suggests that the VEGF pathway regulates angiogenesis in concert with Notch signaling. Here, the authors examined the role of Notch signaling in CNV in the backdrop of Notch signaling–mediated regulation of retinal angiogenesis.

Methods.: Choroid sclera complexes, after laser-induced CNV, were examined for changes in CNV lesion volume and in proangiogenic and antiangiogenic gene expression after perturbation in Notch signaling. Retinal vessels and angiogenic gene expression in retinal endothelial cells were analyzed in postnatal rats after perturbations in Notch signaling. Notch signaling was activated and inhibited by intravitreal or systemic injection of Jagged1 peptide and gamma secretase inhibitor DAPT, respectively.

Results.: The authors demonstrated that activation of the canonical Notch pathway reduced the volume of CNV lesions as it attenuated the development of postnatal retinal vasculature. In contrast, inhibition of the Notch pathway exacerbated CNV lesions as it led to the development of hyperdense retinal vasculature. The authors also identified genes associated with proangiogenesis (Vegfr2, Ccr3, and Pdgfb) and antiangiogenesis (Vegfr1 and Unc5b) as targets of Notch signaling–mediated vascular homeostasis, the disruption of which might underlie CNV.

Conclusions.: This study suggests that Notch signaling is a key regulator of CNV and thus a molecular target for therapeutic intervention in wet AMD.

Age-related macular degeneration (AMD) is the most common cause of progressive vision loss in patients older than 60 in developed countries. ^1,2 The loss of vision is due to the degeneration of photoreceptors in the macula, the center of visual acuity in retina, thus disabling patients from reading, driving, and recognizing faces. The clinical hallmark of AMD is extracellular deposits called drusen, located between the retinal pigment epithelium (RPE) and the Bruch's membrane. The appearance of drusen with subsequent atrophy of RPE and gradual reduction in central vision characterizes early AMD, also known as dry or nonexudative AMD. ³ A subset of patients with dry AMD progress to late AMD, also known as wet or exudative AMD, in which new vessels sprout from choriocapillaries. This process is known as choroidal neovascularization (CNV), and the resultant fenestrated vessels eventually invade the subretinal space, causing photoreceptor degeneration and rapid vision loss. ^3,4 Wet AMD, in contrast to its dry counterpart, affects only 10% of AMD patients but accounts for most AMD-related vision loss. ^3,5  

The underlying mechanism of wet AMD is not well understood, but evidence has emerged that pathologic recruitment of signaling pathways involved in normal angiogenesis plays a critical role. Studies with a variety of approaches implicate the VEGF pathway, which plays a central role in embryonic vasculogenesis and postnatal angiogenesis, to CNV. ^{6 –8} For example, Vegf, which is constitutively expressed in RPE and is thought to participate in the trophic maintenance of choriocapillaries, ⁹ is overexpressed in the RPE of autopsied eyes and surgically excised CNV membranes of AMD patients. ^3,9,10 Subsequent studies in animals, in which overexpression of Vegf in the RPE led to CNV, ^11,12 and humans, in whom anti-Vegf approaches led to a decrease in growth, leakage, and vision loss, provide strong evidence of the VEGF pathway's involvement in CNV. ^13,14  

Recent studies suggest that the VEGF pathway regulates angiogenesis in concert with Notch signaling. ^15,16 The Notch pathway is an evolutionarily conserved signaling mechanism, recruited during embryonic development and in adults, to regulate multiple processes such as cell commitment, proliferation, and survival. ¹⁷ It is essential for vascular development and postnatal angiogenesis in different tissues, including the retina. ^{15,18 –22} Recent studies implicate the Notch pathway in pathologic angiogenesis associated with tumor growth ^{23 –25} and ischemic stroke. ²⁶ Here, we have examined the role Notch signaling plays in laser-induced CNV along with its involvement as a regulator of normal angiogenesis during postnatal retinal vasculature development in rats. We demonstrate that components of canonical Notch pathways, particularly those that have been implicated in angiogenesis, are expressed in neonatal retinal endothelial cells (ECs) and adult choroid-sclera complex. We demonstrate that Notch signaling, similar to its role in normal retinal angiogenesis, is associated with CNV and that the activation of the canonical Notch pathway by Jagged 1 peptides (Jag1) significantly reduces the volume of CNV lesions just as it compromises development of the postnatal retinal vasculature. In contrast, inhibition of this pathway by the γ-secretase inhibitor DAPT exacerbates CNV lesions just as it causes hyperdense vasculature in the postnatal retina. In addition, we show that, consistent with its effects on CNV and retinal angiogenesis, the perturbation of Notch signaling is associated with corresponding changes in the expression of genes encoding proangiogenic and antiangiogenic factors in choroidal and retinal ECs. Thus, our study proposes Notch signaling as a key regulator of CNV and, thus, a molecular target for therapeutic intervention in wet AMD. 

Methods

Animals

All the experiments were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Institutional Animal Care and Use Committee at the University of Nebraska Medical Center. Animals were housed and bred in the Department of Comparative Medicine at University of Nebraska Medical Center. 

Postnatal Retinal Angiogenesis

To examine the influence of Notch signaling on postnatal retinal angiogenesis, rat pups received subcutaneous injections of DAPT (100 μg/g body weight; Calbiochem, Temecula, CA) or Jag1 ²⁷ (25 μg/g body weight) at postnatal day (P) 3 and P4 and were killed on P5. Controls included sham-injected pups and pups injected with Jag1^scrambled peptides. Given that there was no significant difference between the two control groups, sham-injected controls were used in all subsequent experiments. Eyes were enucleated, and retinas were carefully removed and fixed in 4% paraformaldehyde in PBS at 4°C overnight. After permeabilization with 0.5% Triton in 1% bovine serum albumin for 2 hours, retinas were incubated in FITC-conjugated fluorescent lectin (Isolectin B4; Vector Laboratories, Burlingame, CA) at 4°C overnight. After three washes in PBS, retinas were flat mounted in mounting medium (Vectashield; Vector Laboratories). 

Laser Photocoagulation Model of CNV

Four to six photocoagulation spots were made (50-μm spot size, 150 ms exposure, 200 mW power) using a laser photocoagulator (Ophthalas 432 EyeLite; Alcon, Austin, TX) in the area surrounding the optic nerve in each eye of 6- to 8-week-old Long Evans rats. Jag1 (30 μg/eye) or DAPT (200 μg/eye) was injected intravitreally 30 minutes, 48 hours, and 96 hours after laser treatment, followed by CNV volume measurements at day 7 and day 21. Fluorescein angiography (FA) was carried out on some rats 35 days after laser treatment. Control included animals sham injected or injected with Jag1^scrambled peptides. Given that there was no significant difference between the two control groups, sham-injected controls were used in all subsequent experiments. An additional injection of Jag1/DAPT was given at day 14 before analyses of CNV at day 21. To examine gene expression in the choroid-sclera complex in response to laser photocoagulation, 20 to 25 laser spots per eye were made. 

Isolation of Retinal ECs and E18 Neurospheres Formation

Retinal ECs from P5 rats were isolated by affinity-binding enrichment, as previously described. ²⁸ Briefly, retinal cell dissociates, obtained after the digestion of retina with 0.1% collagenase (Invitrogen, Carlsbad, CA) and 0.01% DNase (Sigma, St. Louis, MO) at 37° for 45 minutes, were incubated with mouse anti-rat CD31 antibody–coated beads (Dynabeads; Dynal Biotech, New Hyde Park, NY), at 4°C for 90 minutes. ECs were released in Dulbecco modified Eagle medium (DMEM) and 10% fetal bovine serum for RNA extraction. For neurosphere formation, E18 embryos were harvested from timed pregnant Sprague-Dawley rats (SASCO) in Hanks' balanced salt solution, the eyes were enucleated, and retinas were dissected and dissociated into single cells. Retinal dissociates were cultured in retinal culture medium (DMEM-F12, 1× N2 supplement, 2 mM l-glutamine, 100 U/mL penicillin, 100 μg/mL streptomycin) containing 20 ng/mL epidermal growth factor for 5 days to generate clonal neurospheres. 

Fluorescein Angiography

FA was performed using a fundus camera (FF450^plus; Carl Zeiss, Jena, Germany). Animals were anesthetized (ketamine 44 mg/kg body weight, xylazine 4 mg/kg body weight, administered intraperitoneally), and 1% tropicamide, 1% cyclopentolate, and 2.5% phenylephrine were topically applied to dilate the pupil. % Sodium fluorescein 25% (100 μL/animal or 0.1μg/kg body weight) was injected intraperitoneally. Fundus was photographed during early (2 minutes), middle (5 minutes), and late (10 minutes) phases of FA. 

Measurement of CNV Volume

CNV volume in whole mount preparations of the choroid-sclera-RPE complex was measured as previously described. ²⁹ Briefly, after removal of the cornea, lens, and retina from enucleated and paraformaldehyde-fixed (4% solution in PBS at 4°C for 1 hour) eyes, the resultant eye cups were washed in ice-cold ICC buffer (0.5% bovine serum albumin, 0.2% Tween 20,and 0.05% sodium azide in PBS), followed by incubation in fluorescent dyes at 4°C, with gentle rotations, for 4 hours. The fluorescent dye consisted of a 1:1000 dilution of a 10-mg solution of 4,6-diamidino-2-phenylindole (DAPI), a 1:100 dilution of a 1 g/L solution of isolectin IB4 conjugated with AlexaFluor 568, and a 1:100 dilution of a 0.2 U/L solution of phalloidin conjugated with AlexaFluor 488 (Invitrogen-Molecular Probes, Eugene, OR) prepared in ICC buffer. The eyecups were washed in cold ICC buffer, radial cuts were made toward the optic nerve head, and choroid-sclera-RPE complexes were flat mounted in mounting medium (Fluoromount G; (Southern Biotech, Birmingham, AL). Multiplane z-series of images from multiple lesions from each experimental group were collected with a Zeiss confocal microscope at 40× and 1024 × 1024-pixel resolution using the sequential scan mode to prevent bleed-through. The z-series, collected as TIFF images, were used to build animated 3D reconstructions of the CNV complex using image analysis software (Velocity Visualization; Improvision Inc., Lexington, MA). The navigation palette was used to turn off the green (phalloidin, RPE) and blue (DAPI, nuclei) channels. CNV complexes were identified with the red channel (isolectin, endothelial cells), and their volumes in cubic millimeters was quantified using velocity classification modules (Improvision Inc.). 

Quantitative PCR

Quantitative PCR (Q-PCR) was performed as previously described. ^30,31 Briefly, total RNA from retinal ECs and choroid sclera complexes was isolated (Mini RNeasy Kit [Qiagen, Valencia, CA] and Tri Reagent [Molecular Research Center Inc., Cincinnati, OH]) according to the manufacturers' instructions. cDNA was prepared from 2 to 3 μg total RNA using a reverse transcriptase kit (SuperScript III; Invitrogen) according to the manufacturer's instructions. Transcripts were amplified and their levels quantified using gene-specific primers (Table 1) and a PCR kit (Quantifast SYBR Green; Qiagen) on a real-time PCR machine (RotorGene 6000; Corbett Robotics, San Francisco, CA). Measurements were performed in triplicate; an RT-negative blank of each sample and a no-template blank served as negative controls. Amplification curves and gene expression were normalized to the housekeeping gene GAPDH, used as an internal standard. 

Table 1.

 

List of Primers

Gene Name	Accession No.	Primer Sequence (5′–3′)	Tm	Size (bp)
Hes1	NM024360	FP: 5′-GCTTTCCTCATCCCCAATG-3′	56	224
		RP: 5′-CGTATTTAGTGTCCGTCAGAAGAG-3′
Hey1	XM_342216	FP: 5′-CGCAGACGAGAATGGAAACTTG-3′	57.1	307
		RP: 5′-AAACCCCAAACTCCGATAGTCC-3′
Hey2	XM_342216	FP: 5′-GGCAAGAAAGAAAAGGAGAGGG-3′	56	225
		RP: 5′-ATAAAGTCCGTGGCAAGGGC-3′
Vegfr1	NM_019306	FP: 5′-TGAGCCAGGAGGACAAAAAGC-3′	56	305
		RP: 5′-AGTGACTGTGATGTTGGGAGACG-3′
Vegfr2	NM_010612	FP: 5′-CCTGGCGATTTTCTCCATCC-3′	55	164
		RP: 5′-CATTCAGTCACCAATACCCTTTCC-3′
Pdgfb	XM_343293	FP: 5′-CCAATGCCAACTTCCTGGTG-3′	60	338
		RP: 5′-AAACTTTCGGTGCTTCCCTTTG-3′
Unc5b	NM_022207	FP: 5′-CACAGGCTTGCGAATACGAGAG-3′	58.6	318
		RP: 5′-ATGGTGAGCAGGAAGTTAGTGTCC-3′
GAPDH	NM_017008.3	F: 5′-ACAGTCCATGCCATCACTGCC-3′	60	266
		R: 5′-GCCTGCTTCACCACCTTCTTG-3′
Notch1	NM008714	F: 5′-TCTGGACAAGATTGATGGCTACG-3′	56	329
		R: 5′-CGTTGACACAAGGGTTGGACTC-3′
Notch2	NM_010928	F: 5′-GGAGGTGAATGAATGCCAGAGC-3′	58	300
		R: 5′-CAGGTGTAGGAGTCAATCCCATCC-3′
Dll1	NM_032063	F: 5′-TTTCTGTTAGCATCATTGGGGC-3′	57	266
		R-5′-CCTTTTTCTGTCGGGAACCTCC-3′
Dll4	XM_230472	F: 5′-CGGACATTATGAGTGCCAACCAG-3′	57	199
		R: 5′-ATGACGACAGCCATTGTGGG-3′
Jagged1	XM_230472	F: 5′-GTGTCCTCAAGGAGTATCAGTCC-3′	56	256
		R: 5′-GTGTTCTGTTTCAGTGTCTGCCAC-3′
Ccr3	NM_053958	F: 5′-AAATGGCATCCAACGAAGAGG-3′	55.5	276
		R: 5′-GGACAGTGAAGAGAAAGAGCAGGTC-3′
Beta actin	XM_037235	F: 5′-GTGGGGCGCCCCAGGCACCA-3′	50	548
		R: 5′-CTCCTTAATGTCACGCACGATTTC-3′

Statistical Analysis

Values were expressed as mean ± SEM. Data were analyzed using the Student's t-test to determine the significance of the differences between treatment and control groups. 

Results

Canonical Notch Pathway Components in Retinal ECs and Choroid-Sclera Complexes

To understand the involvement of Notch signaling in normal and pathologic angiogenesis, we first examined the expression of components of the canonical Notch pathway in retinal ECs and choroid-sclera complexes, which are rich in choroidal vasculature (Fig. 1A). Transcripts corresponding to receptors (Notch1 and Notch2) and ligands (Dll1, Dll4, and Jagged1) were detected in both, but their levels were different, suggesting their differential use during retinal and choroidal angiogenesis. For example, the levels of Notch1 transcripts in choroid-sclera complexes were lower than those of Notch2 in retinal ECs, suggesting that Notch2 might be the dominant receptor in the former. Similarly, levels of Dll4 and Jagged1 transcripts were higher than Dll1 in both retinal ECs and choroid-sclera complexes, suggesting their relative physiological relevance compared with Dll1 in ECs. Both retinal and choroid-sclera complexes expressed transcripts corresponding to Hairy/Enhancer of Split homologue (HES) family member Hes1, and Hairy-related (HER) family members Hey1 and Hey2, known transducers of canonical Notch signaling during angiogenesis. ^32,33 Together, these observations suggested that the canonical Notch pathway, as observed in retinal angiogenesis, may regulate choroidal angiogenesis. 

Figure 1.

 

Expression of components and Jag1 peptide-mediated perturbation of Notch signaling. RT-PCR analysis of P5 retinal endothelial cells (lane 1) and adult choroid sclera complex (lane 2) revealed the expression of transcripts corresponding to components of the canonical Notch pathway (A). Jag1 peptides, corresponding to a sequence in the DSL domain of Jagged1 (B), caused a dose-dependent increase in the number of neurospheres generated by E18 retinal progenitors in a neurosphere culture assay (C), accompanied by a dose-dependent increase in the Hes1 transcript levels (D). Jag1 caused a significant increase in transcript levels corresponding to Hes1 (E, H), Hey1 (F, I), and Hey2 (G) in P5 endothelial cells (E–G) and in the choroid-sclera complex (H–J) compared with controls.

Next, we carried out experimental perturbations of the canonical Notch pathway to examine its involvement in retinal angiogenesis and laser-induced CNV. To activate Notch signaling, we used 17aa peptide (Jag1) with a sequence corresponding to a region in the DSL domain of Jag1 (Fig. 1B). Although Jag1 has been successfully used to activate Notch signaling in normal ¹⁹ and pathologic angiogenesis, ²⁵ the recent report that Jag1 may play an inhibitory role during angiogenesis ³⁴ prompted us to first determine whether Jag1 activates the canonical Notch pathway. First, we examined the effects of Jag1 on retinal progenitors, which require Notch signaling for their maintenance. ³⁵ We observed a dose-dependent increase in the number of neurospheres when retinal progenitors were cultured in increasing concentrations of Jag1 (Fig. 1C). That the effects on the number of neurospheres involved the canonical Notch pathway was demonstrated by a dose-dependent increase in the levels of Hes1 transcripts (Fig. 1D). Next, we ascertained whether Jag1 engaged the canonical Notch pathway in P5 retinal ECs (Figs. 1E–G) and the choroid sclera complexes from eyes that had undergone laser photocoagulation (Figs. 1H–J). In both cases, we observed an increase in the levels of transcripts corresponding to Hes1 and Hey1 in the presence of Jag1 compared with controls. Although there was a significant increase in the levels of Hey2 transcripts in response to Jag1 peptide in ECs, there were no such increases in choroid-sclera complexes. Taken together, these results demonstrated that Jag1 activated the canonical Notch pathway in general and in ECs and choroid-sclera complexes in particular. 

Notch Signaling during Normal Angiogenesis

In rats, retinal angiogenesis begins during the late prenatal or early postnatal (P1) stage, as the gradient of vascular development spreads from the optic stalk to the retinal margin until P10. ³⁶ We perturbed Notch signaling during active developmental angiogenesis by systemic injection of Jag1 or DAPT at P3 and P4, and the density of the retinal vascular bed was examined at P5 (Fig. 2A). The distal (Figs. 2B–D), middle (Figs. 2E–G), and proximal (Figs. 2H–J) regions of the vascular bed were each examined for changes in vascular density (Fig. 2K), branch points (Fig. 2L), and vessel diameter (Fig. 2M). Most indices of vascular development changed in response to Jag1 and DAPT treatments in the distal and middle vascular bed, whereas those in the proximal vascular bed remained unaltered. In the distal vascular bed, there was a significant decrease in vascular density (2.5-fold; P < 0.01) and branch points (1.7-fold, P < 0.05) in pups injected with Jag1 compared with controls. Similarly, in the middle vascular bed, there was a lesser but significant decrease in vascular density (1.4-fold; P < 0.05) and in the numbers of branch points (1.47-fold, P < 0.01) in Jag1-treated pups compared with controls. There was no significant difference in blood vessel diameter between Jag1-treated and control groups in both the distal and the middle vascular bed. In contrast, both in the distal and the middle vascular beds, there was a significant increase in vascular density (2.33-fold, P < 0.001 [distal] and 1.66-fold, P < 0.001 [middle]) and number of branch points (3.0-fold, P < 0.001 [distal] and 1.41-fold, P < 0.001 [middle]) in DAPT-treated pups compared with controls. Interestingly, the vessel diameter that remained unaffected in Jag1-treated pups increased significantly in both the distal (1.48-fold, P < 0.05) and the middle (1.67-fold, P < 0.05) vascular bed on DAPT treatment. In the proximal vascular bed, there were no significant differences in any of the indices of vascular development between Jag1- or DAPT-treated and control groups. Next, we examined the number and morphologic features of tip cells, particularly the number of the filopodia in response to changes in Notch signaling (Figs. 3A–C). The tip of the sprouting front of the distal retinal vascular bed is composed of highly migratory tip cells with exploratory filopodia that lay the groundwork for vessel formation. ³⁷ We observed that in Jag1-treated pups, there was an approximately 2.0-fold decrease in the numbers of tip cells (2.05-fold, P < 0.001) and filopodia (2.08-fold; P < 0.0001) compared with controls (Figs. 3D, 3E). In contrast, there was an approximately 2.0-fold increase in the numbers of tip cells (2.04-fold, P < 0.001) and filopodia (2.62-fold, P < 0.0001) in DAPT-treated pups compared with controls (Figs. 3D, 3E). These observations suggested that activation of the canonical Notch pathway by Jag1 reduced retinal angiogenesis and that its attenuation by DAPT promoted the migratory phenotype of tip cells and vascular bed density. 

Figure 2.

 

Perturbations of Notch signaling affect retinal vasculature. Pups were injected subcutaneously at P3 and P4 with Jag1/DAPT, and morphometric analysis of retinal vasculature was examined on P5 (A). Isolectin B4 staining of retinal vessels revealed that animals treated with Jag1 and DAPT displayed decreases and increases, respectively, in the retinal vascular density (K), branch point numbers (L), in the distal (C) and middle (F) vasculature compared with controls (B, E), whereas no significant changes were observed in these indices in the proximal vasculature (H–J). Although there were no significant changes in vascular diameter (M) in the distal, middle, and proximal vasculature in Jag1-treated animals, those treated with DAPT had significant increases in vessel diameter in the distal (D), and middle (G) vasculature compared with controls (B, E). Scale bar, 50 μm.

Figure 3.

 

Perturbations of Notch signaling affect the morphology and number of tip cells. There were significant decreases and increases in the numbers of tip cells (A–C, white arrows; D) and filopodia (A–C, red arrows; E), in animals treated with Jag1 and DAPT, respectively, compared with controls.

Next, to gain insight into the mechanism underlying the influence of Notch signaling on angiogenesis, we examined the expression of genes that have emerged as positive (Figs. 4A, 4B) and negative (Figs. 4C, 4D) regulators of angiogenesis in ECs isolated from the treatment groups. Among proangiogenic factors, we examined the expression of transcripts corresponding to Vegfr2, the key mediator of Vegf signaling, ^19,38,39 and Pdgfb, a mediator and marker of tip cell differentiation. ^19,40,41 Among antiangiogenic factors, expressions of transcripts corresponding to Vegfr1, a decoy receptor of Vegf, ^42,43 and Unc5b, an inhibitor of filopodia extension of tip cells, ^44,45 were determined. In the Jag1-treated group, in which indices of retinal vascular development were attenuated, there was a significant decrease in levels of Vegfr2 (1.8-fold, P < 0.05) and Pdgfb (2.1-fold, P < 0.006) transcripts (Figs. 4A, 4B) and a significant increase in those of Vegfr1 (1.6-fold, P < 0.02) and Unc5b (6.0-fold, P < 0.001) compared with controls (Figs. 4C, 4D). In contrast, in the DAPT-treated group, which displayed hyperdense retinal vasculature, there were significant increases in transcript levels of Vegfr2 (2.1-fold, P < 0.0001) and Pdgfb (1.6-fold, P < 0.003; Figs. 4A, 4B) and significant decreases in those of Vegfr1 (2.0-fold, P < 0.02) and Unc5b (4.8-fold, P < 0.001) compared with controls (Figs. 4C, 4D). Transcript levels corresponding to transducer of Notch signaling Hes1, Hey1, and Hey2 (with the exception of Hey1) increased and decreased significantly in Jag1- and DAPT-treated groups (Figs. 4E–G), suggesting engagement of the canonical Notch pathway. Together, these results suggested that Notch signaling positively and negatively influenced antiangiogenic and proangiogenic genes, and the loss of Notch signaling led to uncontrolled angiogenesis, resulting in hyperdense vasculature. 

Figure 4.

 

Perturbations of Notch signaling affects the expression of proangiogenic and antiangiogenic factor transcripts during retinal angiogenesis. Q-PCR analysis of transcripts expressed in P5 EC cells revealed that levels of those corresponding to genes positively associated with angiogenesis, Vegfr2 (A) and Pdgfb (B) decreased and increased when cells were cultured in the presence of Jag1 and DAPT, respectively. In contrast, transcripts corresponding to genes negatively associated with angiogenesis, Vegfr1 (C), and Unc5b (D) increased and decreased when cells were cultured in the presence of Jag1 and DAPT, respectively. These changes and those observed in Figure 2 were accompanied by increased and decreased levels of transcripts corresponding to Hes1 (E) and Hey2 (G) in response to Jag1 and DAPT treatment, respectively, as ascertained by Q-PCR analysis. Although DAPT treatment significantly decreased Hey1 transcript levels (F), they remained unchanged in response to Jag1 treatment compared with controls.

Notch Signaling and CNV

To test the hypothesis that the loss or reduction of Notch signaling leads to uncontrolled ocular angiogenesis in pathologic conditions, we examined the status and effects of the perturbation of Notch signaling on the progression of laser-induced CNV in rats. CNV was detected using FA (Figs. 5A, 5B) and by staining sections of choroid-sclera complexes through the laser spots with isolectin B4 to identify endothelial cells, phalloidin to detect filamentous actin and assess RPE integrity, and DAPI to detect nuclei (Figs. 5C, 5D). The sprouting blood vessels were clearly visible in cross-sections of the laser spot areas, accompanied by a loss of RPE integrity. Temporal examination of CNV revealed that the CNV volume increased over time; the volume was 3.0-fold greater at day 33 than at day 7 after laser photocoagulation (Figs. 5E–I). To determine whether Notch signaling was involved in CNV, we examined the levels of Hes and Hey classes of transcripts in the choroid sclera complexes after laser photocoagulation. We observed a significant decrease in Hes1, Hey1, and Hey2 transcript levels in the laser-treated group compared with controls (Figs. 5J–L). Next, to examine the influence of Notch signaling on CNV volume, rats were injected with Jag1/DAPT intravitreally on days 1, 2, and 4 after laser photocoagulation. CNV volume was measured on day 7 (Fig. 6A). We observed a significant decrease (1.77-fold, P < 0.01) in CNV volume in rats treated with Jag1 compared with controls (Figs. 6B, 6C, 6E). In contrast, laser-treated rats that received intravitreal injections of DAPT displayed approximately 3.0-fold increases (2.72-fold, P < 0.0001) in CNV volume compared with controls (Figs. 6B, 6D, 6E). To determine whether the effects of Notch signaling on CNV volume is sustained beyond 7 days, CNV volume was measured at day 21, after additional intravitreal injection of Jag1 or DAPT at day 14. We observed an approximately 4.0-fold decrease (4.3-fold, P < 0.01) in CNV volume in Jag1-treated animals compared with controls at day 21 (Figs. 6F, 6G, 6I). In contrast, the extent of the increase in CNV volume remained the same at day 7 and day 21 (2.7-fold [P < 0.0001] vs. 2.8-fold [P < 0.000]) in response to DAPT treatment compared with controls (Figs. 6F, 6H, 6I). Next, we determined whether Notch signaling could be perturbed systemically by DAPT to influence CNV. Applying Jag1 peptide to activate Notch signaling systemically was cost prohibitive and, therefore, was not used. We observed that two consecutive subcutaneous injections of DAPT (200 μg/g body weight), after laser treatment, increased the CNV volume by approximately 4.0-fold (3.8-fold, P < 0.0001) compared with controls (Figs. 7A–C). Together, these observations suggested that Notch signaling negatively influenced pathologic ocular angiogenesis and that the attenuation of Notch signaling exacerbated the CNV lesion. 

Figure 5.

 

Laser photocoagulation-induced CNV lesion. Fluorescein angiography revealed the presence of leaky CNV lesions (white arrows) in a laser-treated eye (B) compared with an untreated eye (A). A section of the eye through the CNV lesion revealed isolectin B4–stained vasculature projecting through the disorganized RPE (phalloidin-positive) into the subretinal space (C, D, red arrows). The size of the laser-induced CNV lesion increased with time (E–I). Q-PCR analysis of transcripts in choroid-sclera complexes of eyes with 20 laser spots revealed that Hes1, Hey1, and Hey2 transcript levels were significantly decreased in lesioned eyes compared with controls (J–L). Scale bar, 100 μm.

Figure 6.

 

Perturbations of Notch signaling influence laser-induced CNV lesion volume. Adult rats were subjected to laser photocoagulation followed by intravitreal injections of Jag1/DAPT (blue arrows) and morphometric analyses (inverted black arrows) (A). There was significant decreases and increases in the volume of the CNV lesions in animals treated with Jag1 (C, G) and DAPT (D, H), respectively, compared with controls (B, F) at either 7 days (B–E) or 21 days (F–I) after treatment. Scale bar, 100 μm.

Figure 7.

 

Systemic perturbations of Notch signaling influence laser-induced CNV lesion. Adult rats were subjected to laser photocoagulation, followed by two consecutive subcutaneous injections of DAPT to attenuate Notch signaling. There was a significant increase in the volume of the CNV lesions in animals treated with DAPT (B, C), compared with controls (A, C). Scale bar, 100 μm.

Next, to understand the underlying mechanism of Notch signaling influence on CNV, we examined the expression of proangiogenic and antiangiogenic genes in choroid-sclera complexes of rats subjected to laser photocoagulation at approximately 20 spots/eye, followed by intravitreal injection of Jag1/DAPT, as described. In Jag1-treated groups, in which CNV lesion volume was reduced, no significant decrease in transcript levels of proangiogenic factors was observed with the exception of Pdgfb (1.56-fold, P < 0.02) (Figs. 8A–C). However, there was a significant increase in transcript levels of the antiangiogenic factors Vegfr1 (1.6-fold, P < 0.01) and Unc5b (2.8-fold, P < 0.02) in this group compared with controls (Figs. 8D, 8E). In contrast, in the DAPT-treated group that displayed exacerbation of CNV lesion, there were significant increases in the transcript levels of proangiogenic factors Vegfr2 (2.3-fold, P < 0.02), Ccr3 (2.3-fold, P < 0.004), and Pdgfb (1.5-fold, P < 0.006) compared with control (Figs. 8A–C). There were no significant differences in transcript levels of antiangiogenic factors Vegfr1 and Unc5b between the treated and controls groups (Figs. 8D, 8E). Next, to establish a correlation between the changes in expression levels of proangiogenic and antiangiogenic factors and the canonical Notch pathway, we examined the levels of Hes1, Hey1, and Hey2 transcripts. With the exception of Hey2, transcripts levels of all these transducers of the canonical Notch signaling were significantly increased in the Jag1-treated group. In contrast, with the exception of Hey1, transcripts levels of these transducers were significantly decreased in the DAPT-treated group compared with controls (Figs. 8F–H). Together, these results suggested that Notch signaling favorably and adversely affected the antiangiogenic and proangiogenic genes, respectively, and that the loss of Notch signaling (DAPT-treated group) led to uncontrolled CNV, increasing the size of the lesion. 

Figure 8.

 

Perturbations of Notch signaling affects the expression of proangiogenic and antiangiogenic factor transcripts in laser-induced CNV. Q-PCR analysis of transcripts in the choroid-sclera complex of eyes with approximately 20 CNV lesions revealed that levels of those corresponding to genes positively associated with angiogenesis—Vegfr2 (A), Ccr3 (B), and Pdgfb (C)—decreased and increased in eyes that received intraocular injections of Jag1 and DAPT, respectively. In contrast, transcripts corresponding to genes negatively associated with angiogenesis, Vegfr1 (D) and Unc5b (E) increased and decreased in eyes that received intraocular injections of Jag1 and DAPT, respectively. These changes were accompanied by increases and decreases in the levels of transcripts corresponding to Hes1 (F), Hey1 (G), and Hey2 (H) in response to Jag1 and DAPT treatment, respectively, as ascertained by Q-PCR analysis. Although DAPT treatment significantly decreased Hes1 and Hey2 transcript levels, the levels of Hey1 transcripts remained unchanged (G). The effects of Jag1-treatment on Hey2 transcripts was nonsignificant (H) compared with controls.

Discussion

The canonical Notch pathway is an essential regulator of vascular development. Targeted disruption of components of the canonical Notch pathway, such as Notch1, ^15,46,47 Jagged1, ⁴⁸ Dll4, ^47,49,50 Csl, ⁴⁷ Hey1, and Hey2, ^32,33 confers embryonic lethality accompanied by severe vascular defects. During vasculogenesis, when endothelial cells differentiate and join to form vessels, Notch signaling confers arterial fate on these cells. Therefore, mice deficient in Dll4 ⁵⁰ and HEY1 and HEY2, ^32,33 do not express arterial markers, and venous markers expand into the arterial domain. During angiogenesis, when new vessels sprout from existing ones, Notch signaling negatively regulates the specification of endothelial cell fate into tip cells. ¹⁵ Therefore, heterozygous deletion of Dll4 or pharmacologic inhibition of γ-secretase leads to an increase in the number of tip cells, resulting in a hyperdense retinal vasculature. ^{20 –22} Evidence has emerged suggesting that Notch signaling–mediated regulation of tip cell specification is recruited during pathologic angiogenesis, for example in sprouting angiogenesis of tumor blood vessels. ^21,51  

In this study, we have examined the involvement of Notch signaling in laser-induced CNV, an example of pathologic angiogenesis, in the backdrop of its established role in physiological angiogenesis in the developing retina. We have demonstrated that though the activation of the Notch pathway significantly decreased CNV lesions, pharmacologic inhibition of the pathway exacerbated them, suggesting that unregulated angiogenesis in reduced Notch signaling conditions, exemplified by the development of a hyperdense vascular network in the postnatal retina, may also be involved in the development of new vessels during CNV. The attenuation of Notch signaling may alter cell-cell contributions to angiogenic homeostasis. During retinal angiogenesis, Notch signaling maintains homeostasis by offering a counterbalance to proangiogenic pathways, such as the one mediated by VEGF, by regulating the specification of ECs into stalk and tip cells. ^20,22 Consistent with this notion, we observed that when Notch signaling was reduced during retinal angiogenesis, the number of tip cells with filopodia increased, laying the foundation for a hyperdense vascular network. This imbalance, suggested by excessive specification of endothelial cells into tip cells when Notch signaling was attenuated, was also reflected in an accompanying increase in the expression of genes encoding Vegfr2, the dominant receptor mediating the proangiogenic influence of Vegf ^40,41 and Pdgfb, a tip cell maker. ²⁰ In contrast, the hyperdense vascular phenotype in the face of reduced Notch signaling was associated with a decrease in the expression of genes encoding Vegfr1, a Vegf decoy receptor, whose targeted disruption led to lethality caused by vascular overgrowth, ^42,43 and Unc5b netrin receptor, which inhibited sprouting angiogenesis in the developing retina. ^44,45  

A similar imbalance in angiogenic homeostasis in choroidal ECs, caused by attenuation in Notch signaling, may underlie CNV (Fig. 9). Under normal conditions, Notch signaling contributes to homoeostasis by striking a balance between the expression of proangiogenic and antiangiogenic factors in choroidal ECs, either directly or indirectly. This may involve the inhibition of proangiogenic factors Vegfr2 and Ccr3 through HEY-mediated repression of Vegfr2 ⁵² and Ccr3 promoters (Ahmad I, Parameswaran S, unpublished observations, October 2009), and, therefore, their disinhibition in response to reduced Notch signaling may tip the balance toward uncontrolled angiogenesis. This notion is supported by a number of observations. First, it has been observed that the development of laser-induced CNV is associated with an increased expression of Vegfr2 ⁵³ and that its inhibition reduces CNV. ⁵⁴ Second, Ccr3 is expressed in the choroidal endothelia of patients with wet AMD, and its neutralization by antibodies or its absence in Ccr3 knockout mice significantly reduces the generation of abnormal blood vessels in animal models of laser-induced CNV. ⁵⁵ In contrast, Notch signaling positively regulates the expression of antiangiogenic factors Vegfr1 and Unc5b, thus stabilizing angiogenic homeostasis. As a result, the attenuation of Notch signaling in pathologic conditions removes the inhibitory influence of Vegfr1 and Unc5b on angiogenic activities, leading to exacerbation of the CNV lesion. This notion is supported by several observations. One is that Vegfr1 has been demonstrated to inhibit CNV by negatively regulating Vegfr2 function through Src homology domain 2-containing (SH2-containing) tyrosine phosphatase 1 (SHP-1). ⁵⁶ Another is that Unc5b is expressed in choroidal ECs, and systemic administration of Netrin 4 inhibited CNV formation. ⁵⁷ Therefore, it is possible that during the onset of CNV, regardless of pathogenesis, a decrease in Notch signaling, which counterbalances proangiogenic influences, perturbs angiogenic homoeostasis, thus contributing to CNV. If this model holds, then the Notch pathway is a molecular target that, when targeted alone or in conjunction with targeting the Vegf pathway or those involved in immune activation, ⁵⁵ may offer a therapeutic approach to wet AMD, which is the cause of the preponderance of vision loss in this disease. 

Figure 9.

 

Involvement of Notch signaling in neovascular AMD. Both in retinal and choroidal angiogenesis, Notch signaling promoted homeostasis by striking a balance in the expression of proangiogenic factors VEGFR2 and CCR3 (in choroidal vessels) and antiangiogenic factors VEGFR1 and UNC5B. The disease process, similar to laser-induced damage, upsets this balance by which a decrease in Notch signaling upregulates the expression of VEGFR2 and CCR3, leading to tip cell specification. This process is further exacerbated by a Notch-dependent decrease in the expression of VEGFR1 and UNC5B. Forced accentuation of Notch signaling mitigates some of these changes, thereby reducing the size of the CNV lesion.

Footnotes

 Supported by the Lincy Foundation, the Pearson Foundation, the Nebraska Department of Health and Human Services, and Research to Prevent Blindness.

Footnotes

 Disclosure: I. Ahmad, None; S. Balasubramanian, None; C.B. Del Debbio, None; S. Parameswaran, None; A.R. Katz, None; C. Toris, None; R.N. Fariss, None

The authors thank Mercedes Campos for laser-induced photocoagulation training, Anathbandhu Chaudhuri for contributing to CNV and retinal angiogenesis experiments, Parmender Mehta for help with microscopy, Thomas Gridley for critical review of the manuscript, and Timothy J. Smith for technical assistance.