β-LGND2, an ERβ Selective Agonist, Inhibits Pathologic Retinal Neovascularization

Abstract

Purpose.: The goal of our study was to evaluate the in vitro and in vivo anti-angiogenic effects of ERβ selective agonist, β-LGND2, using human retinal microvascular endothelial cell (HRMVEC) cultures and a mouse model for oxygen-induced retinopathy (OIR).

Methods.: The selectivity of β-LGND2 was determined using binding and transactivation assays. The effects of β-LGND2 on pathologic neovascularization were evaluated in OIR mice by histology and retinal mounts stained with isolectin B4 to quantify aberrant angiogenesis. Gene expression and protein levels were evaluated using Q-PCR, angiogenesis protein array, and Western blotting. A cell death detection ELISA kit was used to evaluate HRMVECs following hypoxic and hyperoxic conditions. In vitro angiogenesis was evaluated by growth factor-induced proliferation, tube formation, and cell migration assays.

Results.: β-LGND2-treated OIR mice had a reduced number of neovascular tufts compared to vehicle-treated animals and a significant amount of normal blood vessel maturation similar to normoxia controls. β-LGND2 inhibited in vitro hypoxia- or hyperoxia-induced cell death and the formation of endothelial tubular structures in an ERβ-specific mechanism. However, β-LGND2 did not inhibit significantly growth factor-induced HRMVEC proliferation and migration. Gene and protein studies revealed that OIR mice treated with β-LGND2 had lower levels of pro-angiogenic factors, like VEGF and HIF1α.

Conclusions.: β-LGND2 inhibited in vitro and in vivo pathologic neovascularization in the retina in an ERβ-specific mechanism. These results show that β-LGND2, a non-steroidal ERβ selective agonist, could be a useful therapeutic for ocular diseases involving aberrant angiogenesis, like ROP, wet-AMD, and diabetic retinopathy.

Introduction

Ocular diseases, like wet age-related macular degeneration (AMD), diabetic retinopathy (DR), retinopathy of prematurity (ROP), and retinal vein occlusion, are characterized by pathologic neovascularization and vascular leakage. ¹ These diseases arise due to loss of retinal vasculature, which subsequently results in retinal ischemia. To compensate for a loss of oxygen in the retina, new blood vessels are formed aberrantly. Multiple signaling pathways are altered in the diseased retina that ultimately lead to activation of angiogenic factors (VEGF, TGFβ, FGF, IGF1, and HIF, and so forth) as well as anti-angiogenic factors (pigment epithelium derived factor [PEDF], thrombospondins, endostatin, somatostatin, and so forth). A balance in the levels of pro- and anti-angiogenic proteins is critical in the formation and maintenance of normal retinal vasculature. In pathologic states associated with angiogenic factors, the balance is tipped in favor of abnormal blood vessel formation. ^1–5 Various therapeutic strategies, such as the use of antioxidants, anti-VEGF therapy, photodynamic therapy, laser photocoagulation, and corticosteroids, have been evaluated for the treatment of pathologic neovascularization. ^6–9 Although these therapeutic strategies have helped reduce symptoms, they do not cure the disease completely and can be associated with undesirable side effects. 

The oxygen-induced retinopathy (OIR) model in rodents has been very useful in understanding the mechanisms of pathologic neovascularization in the retina. ^10–12 During OIR, exposure to hyperoxia results in apoptosis of retinal endothelial cells and vascular obliteration. Vaso-obliteration results in a physiologic hypoxia and activation of pathologic angiogenesis. ^13,14 Previous studies have shown that estrogens can inhibit pathologic neovascularization in the mouse OIR model. ^15,16 Treatment with 17-β estradiol (E₂) during the hyperoxia phase inhibited retinal endothelial cell apoptosis and reduced the vaso-obliteration. Further, E₂ inhibited extraretinal neovascularization, and the expression of hypoxia induced pro-angiogenic factors VEGF and HIF1α. ¹⁵ These data indicate that an estrogen could be used as a treatment for the pathologic neovascularization in this model. 

The actions of estrogens are mediated through two nuclear receptors: estrogen receptor (ER) α and β. It is unknown whether the estrogen-induced protection from pathologic neovascularization is mediated through its interaction with ERα or ERβ receptors. Previously, we showed that ERβ, and not ERα, protects retinal pigmented epithelial cells (RPEs) from apoptosis induced by oxidative stress. ¹⁷ In our study, we extended our work with an ERβ selective agonist, β-LGND2, to explore its ability to reduce pathologic neovascularization in the mouse OIR model. Our results show that this ERβ selective agonist inhibits hyperoxia- and hypoxia-induced apoptosis in retinal vascular endothelial cells, VEGF/FGF-induced tube formation, and pathologic neovascularization in the mouse OIR model. 

Materials and Methods

Reagents

Phenol red-free Dulbecco's modified Eagle's medium/Ham's F12 (DMEM/F12), PBS containing Ca⁺⁺ and Mg⁺⁺, 1 M Tris pH 8.0, and 0.5 M EDTA were obtained from Mediatech (Manassas, VA). Tetrahydrochrysene (THC) and ICI-182,780 (ICI) were obtained from Tocris (Ellisville, MO). Charcoal/dextran treated fetal bovine serum (csFBS) was obtained from Atlanta Biologicals (Lawrenceville, GA). Trypsin-0.25% EDTA and 4% to 20% gradient SDS gels were obtained from Invitrogen Corporation (Carlsbad, CA). VEGF and bFGF were obtained from R & D Systems (Minneapolis, MN). Glutaraldehyde, paraformaldehyde, osmium tetroxide (OsO₄), propylene oxide, and Epon plastic were purchased from Electron Microscopy Sciences (Hatfield, PA). Oxygen cylinders (80% and 1%) were purchased from NexAir (Memphis, TN). 

Cell Culture

Human retinal microvascular endothelial cells (HRMVECs) were obtained from Cell Systems Corporation (CSC, Kirkland, WA). Cells were cultured at 37°C in a 5% CO₂ incubator in CSC complete medium and passaged using CSC passage reagent (CSC). HEK-293, human umbilical vein endothelial cells (HUVECs), and ARPE-19 cells were obtained from American Tissue Culture Collection (ATCC, Manassas, VA) and cultured in DMEM (HEK-293) or DMEM-F12 supplemented with 10% fetal bovine serum (FBS). 

Radioligand Binding Assay

Radioligand binding assays for β-LGND2 were performed as described previously. ¹⁸ Competitive receptor binding assay was performed using [³H]E₂ and bacterial lysates expressing rat ERα or ERβ ligand binding domain (LBD). The equilibrium dissociation constant (K _d) was determined first by incubating cell lysates containing LBD protein with increasing concentrations of [³H]E₂ in buffer A (10 mM Tris, pH 7.4, 1.5 mM disodium EDTA, 0.25 M sucrose, 10 mM sodium molybdate, 1 mM PMSF), with and without a high concentration of unlabeled E₂ at 4°C for 18 hours. After the incubation period, plates were harvested with GF/B filters using the Unifilter-96 Harvester (PerkinElmer Inc., Waltham, MA). Filters were washed three times with ice cold buffer B (50 mM Tris, pH 7.2) and then dried at room temperature (RT). Microscint-O cocktail was added to the filter plates and radioactive counts were determined using a TopCount NXT Microplate Scintillation Counter (PerkinElmer). The K _d for ER (ERα 0.71 nM, ERβ 1.13 nM) was determined by computer fitting the data with Sigma Plot (Systat Software, San Jose, CA) and nonlinear regression analysis with the ligand binding one site saturation curve. For competitive binding, β-LGND2 was dissolved in dimethyl sulfoxide (DMSO) and diluted to the appropriate concentration in assay buffer A. Increasing concentrations of β-LGND2 (range 10⁻¹¹–10⁻⁵ M) were incubated with [³H]E₂ (4 nM) and ER LBD, and the assay was performed as detailed above. The concentration of compound that reduced the specific binding of [³H]E₂ by 50% (IC₅₀) was determined by computer-fitting the data with Sigma Plot using nonlinear regression with the four parameter logistic curve. The equilibrium binding constant (K _i) of β-LGND2 was calculated by: $K_i=K_d\times I{C}_{50}/(K_d+L)$, where K_d is the equilibrium dissociation constant of [³H]E₂, and L is the concentration of [³H]E₂. 

Transactivation Assay

Transactivation assays were performed as described previously. ^18,19 Rat ERα and ERβ sequences were amplified from rat ovarian cDNA and cloned into a pCR3.1 (Invitrogen) plasmid vector backbone. For the assay, HEK-293 cells (1.25 × 10⁵ cells/well) were plated into 24-well plates in phenol red-free DMEM containing 5% csFBS. Cells were transfected with ERE-LUC (firefly luciferase), constitutively expressing CMV-LUC (renilla luciferase), and ERα or ERβ using Lipofectamine (Invitrogen) as per the manufacturer's instructions. Then, 24 hours after transfection, cells were treated with increasing concentration of β-LGND2 (range 10⁻¹⁰–10⁻⁵ M) in DMSO, E₂ (range 10⁻¹³–10⁻⁷ M) in ethanol, or a combination of β-LGND2 and E₂. Then, 48 hours after drug treatment, Dual-Luciferase Reporter Assay (Promega, Madison, WI) was performed using a Victor³ V multi-label plate reader (PerkinElmer) to measure firefly and renilla luminescence. The concentration of β-LGND2 that stimulated the activity of the receptor by 50% (EC₅₀) of the maximal response (E_max) was determined by computer-fitting the data with SigmaPlot and nonlinear regression with the four parameter logistic curve. The concentration of drug that inhibits 50% of the E_max (IC₅₀) induced by 1 nM E₂ also was determined by the same method with SigmaPlot. 

Mouse Model of Oxygen-Induced Retinopathy

All experimental and animal care procedures were in compliance 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 of the University of Tennessee Health Science Center. 

Timed pregnant C57BL/6 mice (obtained from Harlan Sprague Dawley Inc., Indianapolis, IN) were housed in 12:12 light:dark cycle with food and water ad libitum. The day pups were born was labeled as post-natal day 0 (PN0). On PN7, the pups (n = 6–7) along with their mothers were exposed to hyperoxia (75% ± 2%, O₂) until PN12. On PN12, the pups were taken out of the hyperoxia chamber and then were housed in room air. Vehicle (DMSO:PEG-300, 15: 85) or drug (1 mg/kg/day) was administered subcutaneously immediately after the pups were taken out of the chamber (PN12) and dosing was continued until PN17. The pups were euthanized and eyes/retinas were collected for whole mounts, histology, or protein expression studies. 

Retinal Whole Mounts

Eyes were enucleated and a puncture was made on the cornea. The eyeball was incubated in ice cold 4% paraformaldehyde in 0.1 M PBS for 2 to 3 minutes. Retinas were dissected out as a cup and 4 radial cuts were made. The retina was transferred onto a glass slide and incubated in ice cold 4% paraformaldehyde in 0.1 M PBS for 15 minutes. Each retina was washed three times with PBS and incubated with isolectin GS-B₄ from Griffonia simplicifolia , Alexa Fluor 488 conjugate (1:100; Invitrogen) in fresh PBS containing 0.5 mM CaCl₂ for 24 hours. Retinas were washed with PBS and mounted using Vectashield (Vector Laboratories, Burlingame, CA). Images were obtained using a Zeiss Axiophot fluorescent microscope and a Zeiss LSM 510 confocal microscope (Carl Zeiss MicroImaging, LLC, Oberkochen, Germany). The images were processed uniformly using LSM Image Browser and Adobe Photoshop CS5. Total retinal area and avascular zone were measured using the Adobe Photoshop CS5 Extended version (Adobe Systems Incorporated, San Jose, CA). Avascular zone was quantified as described by Aguilar et al. ¹⁴  

Retinal Histology

Mice were sacrificed on PN17, eyes were enucleated, and a puncture was made on the cornea. Anterior segments were removed, and incubated in a mixture of 2% glutaraldehyde and 2% paraformaldehyde in 0.1 M phosphate buffer for 24 hours at 4°C. Later, the eyes were post-fixed with 1% OsO₄ in PBS for 1 hour at RT. The eyes were washed with PBS and dehydrated with serial incubations of 50% to 100% ethanol for 20 minutes. The eyes were infiltrated with 1:1 ethanol:propylene oxide and then with 100% propylene oxide for 20 minutes. Later, the eyes were infiltrated with 1:1 propylene oxide:Plastic (Epon-812) for 1 hour and incubated with 100% Plastic for 12 hours. After the incubation period, the eyes were embedded in molds and cured for 24 hours at 60°C. Sections (5 μm thick) were obtained 200 μm on either side of the optic nerve. The slides were scanned using an Aperio ScanScope Digital Slide Scanner (Aperio Technologies, Inc., Vista, CA). The extent of neovascularization was quantified as described by Aguilar et al. ¹⁴ Neovascular nuclei were counted at the vitreous zone using Aperio's software Spectrum from at least 15 to 20 sections/eye for morphometric analysis. Enumeration was done by three separate observers and the data were averaged. The data represent counts from 6 to 10 animals from 3 separate experiments (±SEM, ** = P < 0.01, relative to vehicle-treated group). 

Quantitative Real-Time PCR (qPCR)

Retinas were obtained from mice at different ages for gene expression analysis. Retinas (3–4) were pooled for each sample and flash frozen on dry ice. qPCR was performed as detailed previously. ¹⁸ Samples were amplified using mouse ABI probes for genes (VEGF, HIF1a, HIF2a, and GAPDH). Quantification was performed by ddCT method using GAPDH as loading control. 

Western Blotting

Cell lysates were obtained from 80% confluent cultures in a 15 cm plate. The lysates were centrifuged at 3000g for 5 minutes and the supernatants were resolved on a 4% to 20% SDS-PAGE gradient gel. Proteins were transferred to nitrocellulose membranes and immunoblotting was done with antibodies to mouse anti-ERα (Cell Signaling Technology, Cambridge, MA), rabbit anti-ERβ, and mouse anti-β–actin (Millipore, Billerica, MA). For OIR whole retinal lysates, the eyes were enucleated, and the retina was removed carefully from the eye cup. Retinas (4–6) were pooled for each sample and flash frozen on dry ice. Frozen retinas were suspended in 250 μL of lysis buffer (Cell Signaling Technology) containing protease and phosphate inhibitor cocktail, and homogenized using a hand-held homogenizer. The lysates were incubated on ice for 30 minutes, and protein concentration was determined by the Bradford method. For angiogenesis array, 250 μg of protein were loaded and blotted as per the manufacturer's instructions (RayBiotech, Norcross, GA). For Western blotting, 40 to 50 μg of retinal lysate were loaded and resolved on 4% to 20% SDS-PAGE gradient gel. Proteins were transferred to nitrocellulose membranes and immunoblotting was done with antibodies to mouse anti-HIF1α, rabbit anti-VEGF, mouse anti-GFAP, rabbit anti-CD11b (Abcam, Cambridge, MA). Immunodetection was done with secondary antibody conjugated to HRP and enhanced chemiluminescence kit (Amersham, Piscataway, NJ). Densitometry was performed using Adobe Photoshop CS5. 

Cell Death Assay

HRMVECs (1 × 10⁵ cells/well) were cultured in 6 well plates in CSC medium containing 10% csFBS and incubated at 37°C for 24 hours. After the incubation period, the media was replaced with complete medium containing vehicle or 10 nM β-LGND2 or E₂ and incubated overnight. Later, the cells were exposed to hyperoxia (80% O₂) or hypoxia (1% O₂) for 24 hours and cell death was determined using Cell Death Detection ELISA Kit (Roche Diagnostics, Mannheim, Germany). Data are presented as an average of three separate experiments performed in triplicate (±SEM, ** = P < 0.01, relative to hyperoxia only cells). 

HRMVEC Proliferation Assay

HRMVECs (3000 cells/well) were cultured in a 96-well plates in complete medium and incubated overnight at 37°C in a 5% CO₂ incubator. After incubation, the media was replaced with serum-free media and incubated for 2 hours. Cells then were treated with 20 ng/mL VEGF + FGF and vehicle or different dilutions of β-LGND2, and incubated at 37°C for 72 hours. Cell proliferation was measured using BrdU Cell Proliferation Assay Kit as per the manufacturer's instructions (Calbiochem, Gibbstown, NJ). The data are expressed as an average of three separate experiments performed in triplicate (±SEM, ** = P < 0.01, relative to VEGF/FGF-only treated cells). 

HRMVEC Migration Assay

Oris Cell Migration Assay – Collagen I coated kit from Platypus Technologies (Madison, WI) was used to evaluate the VEGF-induced migration of HRMVECs. The assay was performed according to the manufacturer's instructions. Briefly, 100 μL of cell suspension (40 × 10³ cells/well) were loaded in each well in CSC media (growth factor free). The cells were allowed to adhere for 6 hours at 37°C. After cells were attached, the stopper was removed, the cells were washed, and 100 μL of fresh media containing 50 ng/mL VEGF with vehicle or drug were added to each well and incubated for 16 hours at 37°C. Following incubation, the wells were washed with fresh media, and calcein AM (1 μM in fresh media) was added to each well and incubated for 30 minutes at 37°C. Images were obtained using a Zeiss Axiophot fluorescent microscope. Oris Detection Mask was applied to the bottom of the plate and fluorescence intensity (Excitation/Emission = 488 nm/520 nm) from each well was measured using a BioTek Synergy 4 plate reader (BioTek, Winooski, VT). 

Tube Formation Assay

BD Matrigel matrix, growth factor reduced, phenol-red free (BD Biosciences, San Jose, CA), with 20 ng/mL VEGF + FGF or 10% FCS was coated (200 μL) onto a 24-well plate and incubated at 37°C for 1 hour. HRMVECs (5 × 10⁴ cells/well) in serum-free medium were added to each well, along with vehicle or 1 μM β-LGND2 and incubated for 16 hours. After the incubation period, photomicrographs were taken using a Zeiss Axiophot microscope. Tubes were counted from three random fields in each well. The counts represent three separate experiments performed in duplicate (±SEM, ** = P < 0.01, relative to vehicle-treated group). 

Results

Binding Affinity and Transactivation Potential of β-LGND2

Previously reported data show that β-LGND2 binds to both ER subtypes, but has significantly higher affinity and selectivity for ERβ than ERα. ¹⁹ Briefly, β-LGND2 binds ERβ with a similar affinity (K _i = 2.11 nM) as E₂ (K _i = 2.33 nM), but binds ERα with significantly lower affinity compared to E₂ (β-LGND2 K _i = 39.74 nM, E₂ K _i = 0.36 nM). RBA, calculated as the ratio of the K _i values for β-LGND2 to E₂, indicated that β-LGND2 binds preferentially to ERβ (∼112-fold). In vitro transactivation assays in HEK-293 cells showed that β-LGND2 activated the transcription of ERβ with ∼30-fold higher potency than that observed with ERα. Also, β-LGND2 did not antagonize E₂-mediated transactivation of either ERα or ERβ (>10,000 nM). β-LGND2 induced near maximal transactivation for ERα or ERβ (ERα E _max = 90.3%, ERβ E _max = 83.2%) suggesting that β-LGND2 is an ERβ selective full agonist. 

Effect of β-LGND2 on Aberrant Neovascularization in Mice OIR Model

Mice that are reared under normoxic conditions (room air, 21% O₂) develop normal vasculature throughout the retina by PN17 (Fig. 1A). Smith et al. showed that exposure of pups from PN7–PN12 to hyperoxia (75% O₂), followed by exposure to normoxia, results in physiologic hypoxia in the retina and death of retinal endothelial cells. ¹³ This physiologic hypoxia also activates the angiogenic mechanisms resulting in aberrant neovascularization characterized by extraretinal tufts that are apparent by PN17. ¹³ Interestingly, the aberrant neovascularization is observed only in pups that are stressed by environmental oxygen tension and they either have significant weight loss or very little weight gain compared to their cage mates (Louis Smith, personal communication). We found the same pattern in our studies, and hence we used only pups that weighed 4.0 ± 0.3 g on PN7 and ∼ 6.5 ± 0.5 g on PN17 in our study. As expected, exposure of the pups to hyperoxia (PN7–PN12) followed by normoxia (PN12–PN17) resulted in aberrant neovascularization and extraretinal tufts in vehicle-treated groups (Fig. 1A). Administration of β-LGND2 during the physiologic hypoxic phase resulted in significant reduction in the formation of extraretinal tufts (Figs. 1A, 10× magnification, 1C). The retinal vessels in the peripheral zone of the drug-treated mice appeared more mature with good anastomosing networks (Fig. 1A, 10× magnification). Interestingly, though the neovascular tufts were inhibited, the area of avascular zone (AV zone) relative to total retinal area was not significantly different between control and β-LGND2-treated retinas (Fig. 1B). Histologic evaluation of the OIR retina revealed endothelial cells/neovascular tufts in the extraretinal zone (vitreous layer, Figs. 1C, 1D). The extraretinal zone in the hyperoxia-exposed β-LGND2-treated animals had very few to almost no neovascular tufts, similar to that seen in the normoxia retina (Fig. 1D). Morphometric counting showed that the number of extraretinal endothelial cells in the β-LGND2-treated OIR retinas was reduced by 85%, relative to vehicle-treated OIR retina (Fig. 1C). No significant differences in neovascular tuft formation or response to treatment were observed between male or female pups used in our study. These results suggested that β-LGND2 inhibits pathologic neovascularization and facilitates normal vasculature formation in the mouse OIR model. 

Figure 1. 

 

β-LGND2 (1 mg/kg/day) inhibits aberrant neovascularization in the OIR model. (A) Isolectin B4-stained images show significant reduction in the neovascular tufts and also normal vasculature in β-LGND2-treated retina, relative to vehicle-treated OIR retina. 2.5× magnification images were obtained from a Zeiss Axiophot fluorescent microscope and the high magnification images were obtained from Zeiss LSM510 confocal microscope. (B) Ratio of total retinal area/avascular zone measures using Adobe Photoshop CS5 (extended version) show no significant difference between vehicle- and drug-treated groups (n = 8–10 retinas from 3 separate experiments). (C) Morphometric counting of the neovascular nuclei show significant reduction in β-LGND2-treated OIR retinas relative to vehicle. Neovascular nuclei were counted at the vitreous zone using Aperio's software Spectrum from at least 15 to 20 sections/eye for morphometric analysis. The cell count data represent counts from 6 to 10 animals from 3 separate experiments (±SEM, **P < 0.01, relative to vehicle-treated group). (D) Toluidine blue stained representative images obtained from Aperio ScanScope Digital Slide Scanner at 20× magnification. Arrows: indicate the neovascular nuclei in the extra retinal zone.

ERβ Protein Expression

Previous studies have shown that ER is expressed in the retinal layers and RPE. ^17,20,21 In our study, we checked for the expression of ERα and ERβ protein in retinal microvascular cells of human origin (HRMVECs). Our results showed that HRMVECs expressed both types of ER and that the level of ERβ was higher relative to that of ERα (Figs. 2A, 2B). Our results also verified earlier studies that showed ER expression in endothelial cells of umbilical origin (HUVECs), which were used as controls along with MCF-7 cells. ^22,23  

Figure 2. 

 

(A) Expression of estrogen receptor (α and β) in HRMVECs. Lane 1: ARPE-19. Lane 2: HRMVEC. Lane 3: MCF7. Lane 4: HUVEC. (B) Densitometric analysis of Western blots. The graph shows protein levels of ERα and ERβ in HRMVEC, MCF-7, and HUVEC cell lines, relative to ARPE-19 cells (±SEM).

Effect of β-LGND2 on Hyperoxic and Hypoxic Stress-Induced Cell Death

HRMVECs were exposed to hyperoxia (80% O₂) or hypoxia (1% O₂) for 24 hours, and cell death was evaluated by measuring the amount of DNA fragmentation. After exposure to hyperoxia or hypoxia, only 20% of the cells survived, relative to cells grown under normoxia (Figs. 3A, 3B). Extensive dose-response studies with ERβ agonists showed that 1 μM concentration produced maximal effect ¹⁸ and, therefore, we chose this concentration for all our in vitro studies. Pretreatment with β-LGND2 (1 μM) inhibited the hyperoxia- or hypoxia-induced cell death up to ∼80% of normoxia control (Figs. 3A, 3B). To evaluate whether this is receptor-mediated protection, we pretreated the cells with ICI (ERα and -β antagonist) and THC (ERβ specific antagonist). Pretreatment with ICI or THC (1 μM) prevented the protection of HRMVECs from hyperoxia-induced cell death, indicating a receptor-specific protection (Fig. 3B). There was no significant effect of ICI or THC alone on hyperoxia-induced cell death. 

Figure 3. 

 

β-LGND2 protects HRMVECs from hypoxia (A)- and hyperoxia (B)-induced cell death. The above data is an average of three separate experiments performed in triplicate (±SEM, **P < 0.01, relative to hypoxia or hyperoxia only cells).

Effect of β-LGND2 on Growth Factor-Induced Proliferation, Migration, and In Vitro Angiogenesis

Treatment with VEGF and FGF increased the proliferation of HRMVECs by 20–25%, whereas cotreatment with β-LGND2 did not inhibit the proliferation of HRMVECs at concentrations <10 μM (Fig. 4A). In addition, VEGF (50 ng/mL) treatment resulted in migration of HRMVECs to the central zone, while treatment with β-LGND2 did not inhibit the migration of HRMVECs at concentrations <10 μM (Fig. 4B), suggesting that β-LGND2 is not a strong inhibitor of VEGF/FGF-induced proliferation or VEGF-induced migration. 

Figure 4. 

 

(A) β-LGND2 did not inhibit VEGF- and FGF-induced HRMVEC proliferation. The above data is an average of three separate experiments performed in triplicate (±SEM, **P < 0.01, relative to VEGF/FGF-only treated cells). (B) β-LGND2 did not inhibit VEGF-induced HRMVEC migration. Fluorescence microscopy image after calcein AM staining and relative fluorescence determined using plate reader (inset). The above data is an average of two separate experiments performed in triplicate (±SEM, **P < 0.01, relative to 50 ng/mL VEGF-only treated cells).

The ability of β-LGND2 to inhibit in vitro angiogenesis was evaluated by tube formation assay. HRMVECs form tubular structures when incubated with serum or the growth factors VEGF and FGF (Fig. 5A). There was no significant difference in the amount of tube formation between serum, and VEGF and FGF conditions. Therefore, the effect of β-LGND2 on tube formation was evaluated under VEGF and FGF growth conditions. Our results showed that pre-incubation of HRMVECs with β-LGND2 (1 μM) inhibited tube formation by 40% (Figs. 5A, 5B). 

Figure 5. 

 

β-LGND2 inhibits VEGF- and FGF-induced tube formation of HRMVECs. (A) Images showing the tubular structures at 10× magnification. (B) Number of tubes was counted in 3 fields/well. The counts represent three separate experiments performed in duplicate (±SEM, **P < 0.01, relative to vehicle-treated group).

Effect of β-LGND2 on HIF1α and VEGF Expression in the Mouse OIR Model

To examine the mechanism(s) of inhibition of aberrant neovascularization in β-LGND2-treated OIR retinas, we evaluated the differences in the protein expression between PN12 OIR vehicle-treated retinas, and also PN15 OIR vehicle- and β-LGND2-treated retinas. VEGF protein was not detected in the angiogenesis protein array in either vehicle- or β-LGND2-treated OIR retinas, possibly due to low sensitivity of this antibody on the array. Since VEGF is a key player in angiogenesis, we further investigated gene expression changes at the mRNA level, and we ran Western blots to evaluate VEGF protein levels. An increase in the levels of growth factors bFGF and VEGF was seen between PN12 and PN15 in vehicle-treated OIR retinas (Figs. 6A–6C). Interestingly, in the β-LGND2-treated PN15 retina, the protein levels of HIF1α and VEGF were reduced significantly compared to vehicle-treated PN15 retina (Figs. 6B, 6C). No significant difference in the bFGF protein levels was seen in β-LGND2-treated retinas, compared to vehicle group. To evaluate whether the decrease in protein levels was due to decreased transcription (mRNA levels), we performed qPCR for VEGF, HIF1α, and HIF2α. The gene expression of VEGF was down regulated 3.6-fold following β-LGND2 treatment, while no changes in the gene expression of HIF1α or HIF2α were seen (Fig. 6D). These results suggested that the differences in HIF1α protein levels following β-LGND2 treatment likely are due to post-translational regulation. 

Figure 6. 

 

β-LGND2 inhibits HIF1α protein, and VEGF protein and gene expression in the OIR model. The experiment was run using 4 to 6 pooled retinas, and the data are the average of 3 separate experiments. (A) Angiogenesis array: vehicle- and β-LGND2-treated whole retinal lysates (250 μg) were loaded onto the array. The expression of 24 proteins was determined using the antibody cocktail provided by the manufacturer. Inset: shows the blots of bFGF and MCP1 proteins from the array, and also the layout of the proteins on the angiogenesis array. (B) Western blotting: 40 to 50 μg of whole retinal lysates were run on 4% to 20% gradient SDS-PAGE, and the blots were probed with antibodies against HIF1α, VEGF, GFAP, CD11b, and β-Actin (loading control). (C) Densitometric analysis of the angiogenesis array and Western blots. *P < 0.05 (PN12 vehicle vs. PN15 vehicle). #P < 0.05 (PN15-vehicle vs. PN15-β-LGND2). (D) QPCR analysis of OIR-PN15 retinas. Data represented here are the fold change relative to the OIR PN15 vehicle-treated retinas (baseline; n = pooled retinas from 3 separate experiments).

The expression of the angiogenic and anti-angiogenic factors in the retina are regulated by multiple cell types, including astrocytes, macrophages, and glial cells. ^24–26 We evaluated the expression of retinal glial markers, GFAP – astrocytes, Müller glia, and CD11b – microglia in the OIR retina. There was an increase in the levels of GFAP, while the level of CD11b was reduced between vehicle-treated PN12 and PN15 retinas (Figs. 6A–6C). Further, comparison between β-LGND2-treated and vehicle-treated retinas at PN15 revealed a decrease in GFAP, and only a slight increase in CD11b and MCP1 (macrophage marker) protein levels (Figs. 6A–6C). 

Discussion

Pathologic neovascularization in the ocular tissues occurs due to ischemia-induced proliferative retinopathies. The rodent OIR is a classic model that mimics ischemia-induced retinopathy, and exhibits hyperoxic and hypoxic stresses that activate pathologic spatiotemporal events involving apoptosis, proliferation, and aberrant angiogenesis in the retina. ^3,13 In our study we showed that β-LGND2, a selective ER-β agonist, inhibits pathologic neovascularization and facilitates normal vasculature in the mouse OIR model. Further, using in vitro models, we showed that this ERβ agonist inhibits hyperoxia- and hypoxia-induced HRMVEC apoptosis in an ER-β dependent manner. Our results also demonstrated that the compound inhibits tube formation, but not the proliferation or migration of HRMVECs. 

Hyperoxic conditions during OIR generate reactive oxygen species (ROS), like nitric oxide, peroxynitrite, and superoxide, which lead to endothelial cell death, vessel loss, and vaso-obliteration. ^13,27–29 Inhibition of ROS production by increased expression of superoxide dismutase or by inhibiting iNOS protects pups against OIR-induced neovascularization. ^30,31 Antioxidant therapy by supplementation with vitamin C or 3-polyunsaturated fatty acids reduces hypoxic stress and the pathologic neovascularization in the rodent OIR models. ^32,33 Studies have shown that estrogens also exhibit antioxidant properties. ^17,18,34,35 Previously, we demonstrated that 17β-estradiol and GTx-822 (a structurally-related selective ERβ agonist) completely inhibit oxidative damage and apoptosis of RPEs in an ERβ receptor-dependent manner. ^17,18 This anti-apoptotic and antioxidant effect was mediated by mitochondrial protection through activation of MAPK/PI-3K pathway and up-regulation of HO-1 and GPx2 genes. ^17,18 In our current study, we showed that β-LGND2 inhibits hyperoxia- and hypoxia-induced cell death. We speculated that ERβ agonist-mediated protection of HRMVECs is related to its ability to prevent oxidative stress, likely through the same mechanism of mitochondrial protection as shown in our earlier studies with 17β-estradiol and an ERβ agonist. 

In the OIR model, exposure of pups to hyperoxia results in endothelial cell death and vaso-obliteration leading to a physiologic hypoxia. ¹³ Similar events occur in other ocular retinopathies characterized with aberrant neovascularization. Angiogenesis and neovascularization are dynamic processes mediated by positive molecular regulators, like VEGF, bFGF, and HIF, and negative regulators, like PEDF, Fas/FasL, endostatin, and thrombospondin-1. ^1,3–5,36 HIFs are known to be the primary molecules that induce aberrant angiogenesis during OIR. HIF proteins further activate the expression of growth factors, like VEGF and bFGF. ^37–39 These growth factors activate angiogenic processes, such as endothelial cell proliferation, migration, and pericyte recruitment. Interestingly, β-LGND2 treatment significantly reduced the retinal protein levels of HIF1α and VEGF, relative to the age-matched vehicle-treated OIR animals (Fig. 6C). We speculated that the down regulation of VEGF (mRNA and protein) was a result of decreased levels of HIF1α protein. Although the exact mechanism for the decreased expression of HIF1α is not known, we speculated that β-LGND2 treatment activated the HIF1α degradation pathway, as there was no decrease in transcription. Prolyl hydroxylase domain proteins (PHD1-3) target the HIFs for degradation when oxygen levels are normal, ^39,40 but during hypoxia PHDs become less active and, thus, increase the levels of HIF to activate further the pro-angiogenic gene expression. No change in the gene expression of PHD 1, 2, or 3 was seen in either vehicle- or β-LGND2-treated OIR retinas (data not shown). We hypothesize that β-LGND2 reduced the hypoxic and oxidative stress in OIR retinas. The reduced stress could result in oxygen tensions being sensed as normal by the PHD proteins, which leads further to HIF protein degradation. β-LGND2 protected retinal microvascular cells from hyperoxia-induced apoptosis in vitro (Fig. 3B). This inhibition of apoptosis should be reflected as a decrease in the apoptosis of the retinal endothelial cells leading to reduced avascular zone in OIR retina (in vivo). Surprisingly, we did not see a decrease in the avascular zone in β-LGND2-treated animals (Fig. 1B). The difference between in vitro and in vivo results could be due to the fact that the drug treatment in animals was started after hyperoxia exposure. 

The expression and localization of the molecular regulators are under the control of multiple cell types. Major cell types that regulate retinal angiogenesis during normal development and OIR are retinal endothelial cells, astrocytes, and microglia/macrophages. ^24–26 In β-LGND2-treated animals, we observed an increase in the protein levels of MCP1 and CD11b, which are markers for macrophages and microglia. Interestingly, the astrocyte marker GFAP protein levels were reduced in β-LGND2-treated retinas. Macrophages and microglia are required for the endothelial cell apoptosis, vessel regression of ocular capillaries, and at the same time express pro-angiogenic factors, ^24,25,41,42 whereas retinal astrocytes are known to normalize the neovascularization during OIR. ^26,43 These data suggest that β-LGND2 could be reducing the inflammatory component of OIR, which can contribute to prevention of aberrant angiogenesis. Further studies to evaluate the spatiotemporal localization of these markers and cell types in this model would help in understanding the causal-effect relationship. 

Overall, our studies suggested that β-LGND2 inhibits the hyperoxia- and hypoxia-induced stress in an ERβ-dependent mechanism. We showed that β-LGND2 reduces hypoxic stress, and modulates the spatiotemporal expression of molecular and cellular regulators of neovascularization in the rodent OIR model. Current standard of care for neovascular diseases of the retina is intravitreal injection with large molecule antibodies directed against VEGF. Several studies have shown that oxidative stress and cell death are the precursors for the proliferation of blood vessels in retinal diseases, like OIR and AMD. ^13,27–31 Our data from previous studies ^17,18 together with the data in the current study indicate that ERβ ligands could help in targeting this initial stage of disease as well as the downstream VEGF-induced angiogenesis. Moreover, β-LGND2 used in our study is a small molecule that potentially can be delivered orally or as ocular drops. 

Footnotes

 Supported by GTx, Inc.

Footnotes

 Disclosure: A. Giddabasappa, GTx (F, I, E, P); J.R. Eswaraka, GTx (F, I, E, P); C.M. Barrett, GTx (F, I, E, P); M.N. Bauler, GTx (F, I, E); Z. Wu, GTx (F, I, E, P); M. Yepuru, GTx (F, I, E, P); D.D. Miller, GTx (F, I, E, P); J.T. Dalton, GTx (F, I, E, P)