Figure 1. Suppression of choroidal neovascularization (CNV) by propranolol hydrochloride. Representative images of vehicle-treated (A) and propranolol-treated (B) choroidal-scleral flat mounts stained with intercellular adhesion molecule 2. After quantitative analysis of vehicle- and propranolol-treated eyes (area of CNV is measured in mean pixel intensity), propranolol was shown to reduce CNV by 50% (C). * P < .01, 20 vehicle- vs 13 propranolol-treated eyes. Whiskers represent mean standard error.
Figure 2. Quantitative polymerase chain reaction assessment of choroidal endothelial cells (ChECs) and retinal pigment epithelial (RPE) cells. Messenger RNA (mRNA) expression of vascular endothelial growth factor (Vegf) (A), β1-adrenoreceptor (B), β2-adrenoreceptor (C), and β3-adrenoreceptor (D) is measured in arbitrary units. Data are normalized to expression of the ribosomal protein L13A housekeeping gene. * P < .001. † P < .01. Whiskers represent mean standard error.
Figure 3. Stimulation of vascular endothelial growth factor (VEGF) by norepinephrine bitartrate compared with vehicle. Retinal pigment epithelial (RPE) cells (6 eyes) (A) and choroidal endothelial cells (ChECs) (7 eyes) (C) were treated for 2 hours with vehicle or 10μM norepinephrine bitartrate, increasing Vegf messenger RNA (mRNA) production 4-fold. Secretion of VEGF in RPE (B) and ChEC media (D) after 24 hours of treatment (6 eyes for both groups) increased 1.6-fold and 1.2-fold, respectively. * P < .01. † P < .05. Whiskers represent mean standard error.
Figure 4. Prevention of norepinephrine bitartrate–stimulated vascular endothelial growth factor (Vegf) messenger RNA (mRNA) expression by propranolol hydrochloride compared with vehicle. Retinal pigment epithelial (RPE) cells (A) and choroidal endothelial cells (ChECs) (B) were preincubated with 1μM propranolol hydrochloride for 30 minutes followed by incubation with 10μM norepinephrine bitartrate for 2 hours (5 eyes for both groups). * P < .001.
Figure 5. Effect of β2-adrenoreceptors on vascular endothelial growth factor (Vegf) messenger RNA (mRNA) expression compared with vehicle. Retinal pigment epithelial (RPE) cells (A) and choroidal endothelial cells (ChECs) (B) were preincubated with 1μM β1-adrenoreceptor antagonist or 100nM β2- and β3-adrenoreceptor antagonists (4 eyes for both groups). Retinal pigment epithelial cells (3 eyes) (C) and ChECs (5 eyes) (D) were incubated with 100nM β1- and β2-adrenoreceptor agonists or 1μM β3-adrenoreceptor agonists. * P < .01. † P < .05. ‡ P < .001. NE indicates norepinephrine.
Lavine JA, Sang Y, Wang S, Ip MS, Sheibani N. Attenuation of Choroidal Neovascularization by β2-Adrenoreceptor Antagonism. JAMA Ophthalmol. 2013;131(3):376-382. doi:10.1001/jamaophthalmol.2013.1476
Author Affiliations: Departments of Ophthalmology and Visual Science (Drs Lavine, Sang, Wang, Ip, and Sheibani) and Pharmacology (Dr Sheibani), University of Wisconsin School of Medicine and Public Health, Madison.
Objectives To determine whether β-adrenergic blockade inhibits choroidal neovascularization (CNV) in a mouse model of laser-induced CNV and to investigate the mechanism by which β-adrenoreceptor antagonism blunts CNV.
Design Mice were subjected to laser burns, inducing CNV, and were treated with daily intraperitoneal injections of propranolol hydrochloride. Neovascularization was measured on choroidal-scleral flat mounts using intercellular adhesion molecule 2 immunofluorescence staining. The effect of β-adrenoreceptor signaling on expression of vascular endothelial growth factor (VEGF) was investigated using primary mouse choroidal endothelial cells (ChECs) and retinal pigment epithelial (RPE) cells. These cells were incubated with β-adrenoreceptor agonists and/or antagonists and assayed for Vegf messenger RNA and protein levels.
Setting University of Wisconsin School of Medicine and Public Health.
Participants Wild-type 6-week-old female C57BL/6j mice.
Main Outcome Measures Inhibition of CNV after propranolol treatment and Vegf messenger RNA and protein expression after treatment with β-adrenoreceptor agonists and antagonists.
Results Propranolol-treated mice demonstrated a 50% reduction in laser-induced CNV. Treatment with norepinephrine bitartrate stimulated Vegf messenger RNA expression and protein secretion in ChECs and RPE cells. This effect was blocked by β2-adrenoreceptor antagonism and mimicked by β2-adrenoreceptor agonists.
Conclusions Attenuation of CNV is achieved by β-adrenergic blockade. The β2-adrenoreceptors regulate VEGF expression in ChECs and RPE cells.
Clinical Relevance Antagonists of β-adrenoreceptors are safe and well tolerated in patients with glaucoma and cardiovascular disease. Thus, blockade of β-adrenoreceptors may provide a new avenue to inhibit VEGF expression in CNV.
Exudative age-related macular degeneration (AMD) is one of the leading causes of blindness worldwide.1 Exudative AMD develops when vascular tissue from the choriocapillaris invades through a break in the Bruch membrane into the subretinal pigment epithelium and/or subretinal space; this process is termed choroidal neovascularization (CNV). Leakage of blood, serous fluid, and lipid due to CNV causes central vision loss and ultimately leads to an irreversible fibrovascular scar.1
Histopathological studies of CNV membranes have provided valuable insight into the pathogenesis of exudative AMD. Several studies identified choroidal endothelial cells (ChECs), retinal pigment epithelial (RPE) cells, fibroblasts, myofibroblasts, photoreceptors, glial cells, and macrophages within surgically excised CNV membranes.2- 6 Follow-up studies focusing on growth factor expression demonstrated interleukin 1β,7 tumor necrosis factor,7 tissue factor,8 monocyte chemotactic protein,8 transforming growth factor β,9,10 acidic and basic fibroblast growth factors,9,10 platelet-derived growth factor,10 and vascular endothelial growth factor (VEGF)6- 8,10- 13 within these CNV membranes. Expression of VEGF is localized to RPE, choroidal endothelium, fibroblasts, and macrophages.6- 8,10- 13 Because VEGF is a secreted protein that stimulates vascular permeability and angiogenesis,14,15 we hypothesized that VEFG plays a central role in CNV.
Experimental models demonstrate a causative role for VEGF in CNV. In the laser-induced model, VEGF is upregulated in the RPE, choroidal endothelium, fibroblasts, and macrophages.16 In this same model, inhibition of VEGF signaling prevents CNV.17 These studies demonstrate that VEGF is correlated with and necessary for CNV. Overexpression of VEGF in photoreceptors18 and in the RPE19 produces retinal and intrachoroidal neovascularization, respectively, that did not cross the Bruch membrane. However, adenoviral administration into the subretinal space, which damages the Bruch membrane and stimulates overexpression of VEGF in the RPE, causes CNV.20 These studies demonstrate that a break in the Bruch membrane and elevated VEGF levels are sufficient for CNV. These studies led to the ANCHOR (Anti-VEGF Antibody for the Treatment of Predominantly Classic Choroidal Neovascularization in AMD)21 and MARINA (Minimally Classic/Occult Trial of the Anti-VEGF Antibody Ranibizumab in the Treatment of Neovascular AMD)22 trials, which established anti-VEGF therapy as the criterion standard for treatment of exudative AMD.
Despite the significant breakthrough of anti-VEGF therapy, patients must undergo a large treatment burden, consisting of monthly intravitreal injections. Recently, it was discovered serendipitously that the β-adrenoreceptor blocker propranolol hydrochloride reduces periocular hemangiomas.23,24 The putative mechanism for this phenotype is the inhibitory action of β-adrenergic blockers on VEGF production in multiple nonocular cell types,25- 27 including endothelial cells.28- 31 In the retina, propranolol or β2-adrenoreceptor blockade inhibits neovascularization in the oxygen-induced retinopathy (OIR) model via reduced VEGF production.32,33 These studies suggest that β-adrenoreceptor blockade could be an alternative or adjunctive anti-VEGF treatment.
To test this hypothesis, we determined whether propranolol could suppress CNV in the laser-induced model. Because RPE and choroidal endothelium have been shown consistently to be sources of VEGF production in CNV, we established a model of norepinephrine bitartrate–induced VEGF production in primary mouse RPE cells and ChECs. We finally investigated which β-adrenoreceptor drives norepinephrine-stimulated VEGF expression.
Norepinephrine bitartrate (10mM; catalogue No. A7257, Sigma-Aldrich Corp) was dissolved in 0.5M hydrochloride. We dissolved the following β-adrenoreceptor agonists and antagonists in water: propranolol hydrochloride (1mM; catalogue No. P0884; Sigma-Aldrich Corp), β1-adrenoreceptor antagonist CGP 20712 dihydrochloride (1mM; catalogue No. 1024; Tocris Bioscience, a division of R&D Systems), β2-adrenoreceptor antagonist ICI118 551 (100μM; catalogue No. 0812; Tocris Bioscience), β3-adrenoreceptor antagonist SR 59230A hydrochloride (100μM; catalogue No. 1511; Tocris Bioscience), β1-adrenoreceptor agonist xamoterol hemifumarate (100μM; catalogue No. 0950; Tocris Bioscience), β2-adrenoreceptor agonist formoterol hemifumarate (100μM; catalogue No. 1448; Tocris Bioscience), and β3-adrenoreceptor agonist BRL 37344 (1mM; catalogue No. 0948; Tocris Bioscience).
All research using mouse models of CNV was performed in accordance with the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research and approved by the Institutional Animal Use and Care Committee of the University of Wisconsin School of Medicine and Public Health. Wild-type 6-week-old female C57BL/6j mice were housed on a 12-hour light-dark cycle and provided with food and water ad libitum. Laser-induced CNV experiments were performed as previously described.34 Briefly, mice were anesthetized and treated with 3 focal laser burns in each eye. Propranolol hydrochloride was dissolved in citrate buffer and delivered daily via intraperitoneal injection at 20 mg/kg. After 14 days, the mice were killed and CNV was measured on choroidal-scleral flat mounts using intercellular adhesion molecule 2 (BD BioSciences) immunofluorescence staining. Images were analyzed using public domain open-source software (http://rsbweb.nih.gov/ij/docs/concepts.html).
Sheep anti-rat magnetic beads (Dynabeads; Invitrogen) were washed 3 times with serum-free basal medium (Dulbecco Modified Eagle Medium [DMEM]; Invitrogen) and then incubated with platelet endothelial cell adhesion molecule 1 antibody (MEC13.3; BD Pharmingen) overnight at 4°C (10-μL beads in 1mL DMEM). After incubation, beads were washed 3 times with DMEM containing 10% fetal bovine serum (FBS), resuspended in the same medium, and stored at 4°C.
Eyes from 1 litter (6 to 10 pups) of 4-week-old immortomice (mice from Charles River Laboratories were crossed to C57BL/6j for several generations) were enucleated, and all connective tissue and muscle were removed from the sclera. In cold DMEM, the anterior eye was removed, followed by the lens, vitreous, retina, and optic nerve, leaving only a tissue composed of RPE, choroid, and sclera. At this point, ChECs were isolated from the RPE-choroid-sclera complex identically to retinal endothelial cells as previously described, using anti–platelet endothelial cell adhesion molecule 1 antibody-coated magnetic beads. The ChECs were analyzed by means of fluorescent-activated cell sorting to document purity and grown identically to retinal endothelial cells.35
For RPE cells, 2 sets of tweezers were used to remove the RPE sheet, which was then digested in 5 mL of type I collagenase (1 mg/mL in serum-free DMEM; Worthington Biochemical Company) for 45 minutes at 37°C. After digestion, DMEM with 10% FBS was added, and cells were centrifuged at 400 g for 10 minutes. The cellular digest was filtered through a double layer of sterile 40-μm nylon mesh (Fisher Scientific), centrifuged, and washed twice with DMEM containing 10% FBS. The RPE cells were plated into a single well of a 24-well plate precoated with a 2-μg/mL solution of human fibronectin (BD Biosciences). To document purity, RPE cells were analyzed by means of fluorescent-activated cell sorting as previously described35 using antibestrophin (MAB5466; Millipore) and anti-RPE65 (MAB5428; Millipore) antibodies. The RPE cells were grown in low-glucose DMEM containing 10% FBS, 2mM L-glutamine, 100 μg/mL of streptomycin and 100 U/mL of penicillin per milliliter, and murine-recombinant interferon-γ at 44 U/mL.
The ChECs and RPE cells were maintained at 33°C with 5% carbon dioxide in 1% gelatin-coated 60-mm dishes. Cells were not allowed to grow beyond 20 passages. Before the experiments, cells were serum starved overnight in serum-free medium. Serum-free medium was identical to the growth medium in the previous paragraphs except it lacked 10% FBS, and the ChEC medium also lacked endothelial cell growth supplement.
For messenger RNA (mRNA) analysis, cells were preincubated for 30 minutes and/or incubated for 2 hours in 24-well plates. Cells were then washed with phosphate-buffered saline (strength, 1×), lysed in buffer (RLT plus; Qiagen), and frozen at −20°C. Messenger RNA was extracted using a commercially available kit (RNeasy Plus Mini Kit; Qiagen). The complementary DNA was synthesized using a reverse transcription–polymerase chain reaction kit (Sprint RT Complete-Double PrePrimed; Clontech). We measured VEGF mRNA levels using quantitative polymerase chain reaction analysis (Eppendorf) and normalized to the housekeeping gene for ribosomal protein L13A (RpL13A) by generating a delta threshold cycle value. Fold values were generated by normalizing to the vehicle control. Vehicle control samples were used to assay for baseline levels of VEGF and β-adrenoreceptors. The following primers were used: VEGFA-164 (forward, 5′-ATCTTCAAGCCGTCCTGTGTGC-3′; reverse, 5′-CAAGGCTCACAGTGATTTT-CTGG-3′), β1-adrenoreceptor (forward, 5′-GCCAGGAAGTGTTTCCCTTTGCTT-3′; reverse, 5′-AAGGTGAACCGTGCTCCACACATA-3′), β2-adrenoreceptor (forward, 5′-GGTC-ATCACAGCCATTGCCAAGTT-3′; reverse, 5′-AAGTCCAGAACTCGCACCAGAAGT-3′), β3-adrenoreceptor (forward, 5′-ACCGCTCAACAGGTTTGATGGCTA-3′; reverse, 5′-TGTTGCATCCATAGCCGTTGCTTG-3′), and RpL13A (forward, 5′-TCTCAAGGTTGTT-CGGCTGAA-3′; reverse, 5′-GCCAGACGCCCCAGGTA-3′).
For VEGF-secreted protein analysis, cells were incubated for 24 hours in 12-well plates. The VEGF levels were measured as previously described35 and normalized to total protein from cell lysates using a protein assay kit (BCA; Pierce). Fold values were generated by normalizing to the vehicle control.
For CNV and gene expression comparisons between cell lines, we performed the unpaired t test. For cell culture, each biological sample size was generated by an experiment on a unique passage day. Thus, we performed the paired (2-tailed) t test to compare 2 groups. For multiple comparisons, repeated-measures analysis of variance was performed using the Dunnett multiple comparison posttest against the vehicle control.
To determine whether β-adrenergic signaling plays a role in AMD, we tested propranolol in the in vivo laser-induced CNV model. Briefly, mice were anesthetized and treated with focal laser burns, which rupture the Bruch membrane and induce CNV. Mice were treated with propranolol or citrate vehicle control for 14 days and underwent assessment for CNV. We used intraperitoneal drug delivery because prior studies demonstrated positive results with this method.32,33 We found that propranolol reduced the average CNV area by 50% (Figure 1).
We next sought to investigate the mechanism by which propranolol inhibits CNV. As discussed earlier, the choroidal endothelium and RPE have been consistently identified as key sources of VEGF production in histopathological studies of AMD. We thus began our in vitro analysis by assessing the levels of VEGF and β-adrenoreceptors in primary mouse ChECs and RPE cells. We found that ChECs and RPE cells express equal mRNA levels of Vegf and the β2-adrenoreceptors (Figure 2). However, ChECs displayed significantly less β1- and β3-adrenoreceptor mRNA compared with RPE cells (Figure 2B and D).
Because primary ChECs and RPE cells express VEGF and all 3 β-adrenoreceptors, these cells are a good in vitro model to investigate the role of β-adrenergic signaling in VEGF production. To test this hypothesis, we incubated ChECs and RPE cells with norepinephrine, a naturally produced neurotransmitter and hormone that stimulates α- and β-adrenoreceptors. We found that norepinephrine increased Vegf mRNA production 4-fold in ChECs and RPE cells (Figure 3A and C). In conditioned medium from these cells, norepinephrine incubation augmented VEGF secretion by 1.6- and 1.2-fold in RPE cells and ChECs, respectively (Figure 3B and D).
We next sought to inhibit the demonstrated effect with propranolol to replicate our CNV model in vitro and confirm that this effect is driven by β-adrenoreceptors. Propranolol completely inhibited norepinephrine-stimulated Vegf mRNA production in ChECs and RPE cells (Figure 4). Furthermore, propranolol prevented Vegf mRNA production only in the presence of norepinephrine and had no effect on baseline Vegf mRNA levels.
Because propranolol is a nonselective β-adrenoreceptor blocker, we next investigated which β-adrenoreceptor drives VEGF production. Before norepinephrine stimulation, we preincubated ChECs and RPE cells with selective β-adrenoreceptor antagonists CGP 20712 (β1-adrenoreceptor), ICI118 551 (β2-adrenoreceptor), and SR 59230A (β3-adrenoreceptor). We found that only the β2-adrenoreceptor antagonist could prevent norepinephrine-stimulated Vegf mRNA production (Figure 5A and B). To confirm this result, we stimulated ChECs and RPE cells with selective β-adrenoreceptor agonists xamoterol (β1-adrenoreceptor), formoterol (β2-adrenoreceptor), and BRL 37344 (β3-adrenoreceptor). Only the β2-selective agonist formoterol significantly increased Vegf mRNA production (Figure 5C and D). Treatment with the β3-adrenoreceptor agonist also trended toward a small positive effect on Vegf mRNA production (Figure 5C and D).
Herein we demonstrated that β-adrenoreceptor antagonism was effective to diminish CNV in the laser-induced mouse model. Furthermore, we showed that norepinephrine stimulates VEGF mRNA production and secretion via the β2-adrenoreceptor in primary ChECs and RPE cells. These results suggest that β2-adrenoreceptor blockers could be effective anti-VEGF therapeutics in CNV-driven diseases, such as exudative AMD, traumatic choroidal rupture, myopic CNV, and ocular histoplasmosis.
The use of β-adrenergic antagonism to suppress VEGF production and neovascularization has been shown previously in rodent models of retinopathy of prematurity. In the rat, topical timolol maleate (a β-adrenoreceptor antagonist with highest affinity for the β2-adrenoreceptor36) diminishes retinal neovascularization during OIR.37 More recently, a group from the University of Pisa showed that propranolol32 and a selective β2-adrenoreceptor blocker33 can attenuate retinal neovascularization during OIR in mice by decreasing VEGF production. However, another group failed to demonstrate that propranolol could prevent retinal neovascularization in OIR mice.38 Although Ristori et al32 hypothesized that the β3-adrenoreceptor stimulated VEGF expression because the β3-adrenoreceptor was upregulated in OIR mice, Martini et al33 conclusively demonstrated that the β2-adrenoreceptor modulates this effect via receptor-specific antagonism. However, Dal Monte et al39 showed that long-term administration of isoproterenol (a β1- and β2-adrenoreceptor agonist) attenuates retinal neovascularization in OIR mice, a result that is counterintuitive compared with Martini et al.33 Attenuation occurs by agonist-induced desensitization of the β2-adrenoreceptor. Our results agree with the findings of Martini et al33 and Dal Monte et al,39 that the β2-adrenoreceptor is key in regulating VEGF expression. However, we see a trend toward increased Vegf mRNA production with β3-adrenoreceptor agonism (Figure 5), similar to the hypothesis of Ristori et al.32 The β3-adrenoreceptor might play a minor biological role in VEGF production, or we might be seeing off-target effects because the pharmacological modulators of β2- and β3-adrenoreceptors have some overlap with each other. In summary, our data agree with prior reports, which demonstrate a role for β2-adrenoreceptor–driven VEGF expression in OIR mice, and we extend these results to the laser-induced CNV model.
Several published reports examine models of reduced ocular adrenergic signaling, but the results are conflicting. Chronic loss of adrenergic signaling in rats via surgical sympathectomy40 or propranolol treatment41 increases choroidal and retinal vascularity. Contrary to our findings, these results suggest that β-adrenergic signaling is antiangiogenic. On the other hand, several reports agree with our findings, showing that β-adrenergic signaling is proangiogenic. First, β3-adrenoreceptor agonism stimulates elongation, migration, and proliferation of human retinal cells42 and ChECs.43 In addition, surgical sympathectomy in rats decreases VEGF levels,44 whereas isoproterenol treatment in human ChECs increases VEGF expression.45 These studies demonstrate that β-adrenoreceptor signaling is proangiogenic and drives VEGF expression, in agreement with our results.
In addition, surgical sympathectomy in rats,46Dbh -null mice (which cannot synthesize norepinephrine),47 and mice null for β1-adrenoreceptors48 display degenerate capillaries, pericyte loss, and increased basement membrane thickness, which are all hallmarks of nonproliferative diabetic retinopathy. In agreement, β1-adrenergic agonism reduces human retinal endothelial cell apoptosis.49,50 These results suggest that loss of β1-adrenergic signaling can mimic early diabetic retinopathy. Our results identify the β2-adrenoreceptor as a driver of VEGF expression; specific therapeutic antagonism of the β2-adrenoreceptor should avoid the deleterious effects of β1-adrenergic blockade.
Despite the wealth of knowledge about VEGF and its expression, the specific driver of VEGF expression in exudative AMD remains unknown. In proliferative diabetic retinopathy and central retinal vessel occlusions, VEGF expression is driven by hypoxia.51 However, the choroid provides more blood flow than is necessary, and the role of hypoxia in AMD is unclear. Alternative hypotheses for drivers of VEGF expression in AMD include oxidative stress,52 insulinlike growth factor 1,53 inflammatory cytokines,7 transforming growth factor β,54 age,55 and contact with choroidal endothelium.55 Our results suggest that adrenergic signaling could also contribute to VEGF expression in CNV.
A potential negative consequence of anti-VEGF therapy is the inhibition of RPE-derived VEGF for normal visual function. Complete loss of RPE-derived VEGF expression causes absence of choriocapillaris development and poor vision.56 Loss of RPE-derived soluble VEGF leads to choriocapillaris atrophy, RPE and Bruch membrane abnormalities, and photoreceptor death.57 Importantly, β-adrenergic antagonism does not suppress baseline VEGF expression in RPE cells (Figure 4), suggesting that therapeutic β2-adrenergic blockade would not have these complications.
Our studies have several important limitations and considerations for clinical translation. First, the laser-induced CNV model more closely approximates posttraumatic CNV after choroidal rupture and postinflammatory CNV in the context of diseases such as histoplasmosis than it does exudative AMD. Second, our studies used systemic, intraperitoneal drug delivery. Prior studies show that systemic propranolol therapy affects retinal phenotypes, including the electroretinogram58 and neovascularization,32,33 demonstrating that systemic propranolol penetrates to the posterior eye. In addition, pharmacokinetic studies find that highly lipophilic β-adrenoreceptor blockers, such as propranolol, are most highly concentrated in the RPE and choroid after intravitreal or subconjunctival injection.59,60 Therefore, local propranolol could be highly effective for CNV owing to its lipophilic nature. These possibilities are subjects of future investigations in our laboratory.
Finally, our studies used primary RPE cells isolated from mouse retina that were successfully proliferated and readily passaged in culture without a significant effect on expression of RPE cell markers, including bestrophin and RPE65. We recognize that primary RPE cells may lose some of their in vivo quiescent characteristics on proliferation and passage in culture; thus, we have limited the extent of these cells' passage in culture. We normally do not see any significant changes in expression of RPE cell-specific markers in these cells at least up to the 25th passage. These potential limitations could be further addressed using primary human fetal RPE cells.
In summary, we have identified β2-adrenergic signaling as a potential driver of RPE and ChEC VEGF expression promoting CNV. Future studies will extend these results to human models and expand the mechanism beyond the β2-adrenoreceptor.
Correspondence: Nader Sheibani, PhD, Department of Ophthalmology and Visual Science, University of Wisconsin School of Medicine and Public Health, 600 Highland Ave, K6/456 CSC, Madison, WI 53792 (email@example.com).
Submitted for Publication: July 17, 2012; final revision received September 12, 2012; accepted September 13, 2012.
Published Online: January 3, 2013. doi:10.1001/jamaophthalmol.2013.1476
Author Contributions: Dr Sheibani had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was supported by grants EY016995, EY021357 (Dr Sheibani), and P30-EY016665 from the National Institutes of Health; an unrestricted departmental award from Research to Prevent Blindness; the Medical Scientist Training Program (Dr Lavine); research award 1-10-BS-160 from the American Diabetes Association (Dr Sheibani); and the Retina Research Foundation.
Additional Contributions: Elizabeth A. Scheef, MS, contributed expertise and help with cell culture studies. SunYoung Park, MS, assisted with the VEGF enzyme-linked immunosorbent assays.