Suppression of aberrant choroidal neovascularization through activation of the aryl hydrocarbon receptor

Abstract

The aryl hydrocarbon receptor (AhR) is a ligand activated transcription factor, initially discovered for its role in regulating xenobiotic metabolism. There is extensive evidence supporting a multi-faceted role for AhR, modulating physiological pathways important in cell health and disease. Recently we demonstrated that the AhR plays a role in the pathogenesis of age-related macular degeneration (AMD), the leading cause of vision loss in the elderly. We found that loss of AhR exacerbates choroidal neovascular (CNV) lesion formation in a murine model. Herein we tested the therapeutic impact of AhR activation on CNV lesion formation and factors associated with aberrant neovascularization. We screened a panel of synthetic drugs and endogenous AhR ligands, assessed their ability to activate AhR in choroidal endothelial cells, and inhibit angiogenesis in vitro. Drugs with an anti-angiogenic profile were then administered to a murine model of CNV. Two compounds, leflunomide and flutamide, significantly inhibited CNV formation concurrent with positive modifying effects on angiogenesis, inflammation, extracellular matrix remodeling, and fibrosis. These results validate the role of the AhR pathway in regulating CNV pathogenesis, identify mechanisms of AhR-based therapies in the eye, and argue in favor of developing AhR as a drug target for the treatment of neovascular AMD.

1. Introduction

Age-related macular degeneration (AMD) is the leading cause of irreversible vision loss in the industrialized nations (1–3), with limited treatment options. It currently affects over 60 million people Worldwide, and this figure is expected to surge to 196 million people by 2020 (4). In the early stages of AMD, characterized by the accumulation of insoluble, extracellular, lipid-protein rich deposits, called drusen, beneath the retinal pigment epithelium (RPE), patients can be asymptomatic. In contrast, in the later stages known as geographic atrophy and exudative/wet AMD, each characterized by degeneration of the RPE, and formation of immature new vessels originating from the outer choroiocapillary blood supply, respectively, patients experience varying levels of visual impairment, often rapidly progressing to blindness (5–9). Over the years, treatment options for wet AMD have evolved from laser photocoagulation to surgery, radiation and steroids, photodynamic therapy, and finally to administration of anti-vascular endothelial growth factor (VEGF) agents (10). Though anti-VEGF therapy provides vision improvement in 25–40% of patients with wet AMD, there is still a significant population of non-responders (11–13). Thus there is a need to not only identify other signaling pathways involved in neovascular development and stability, but also discover and test new therapeutic approaches for the treatment of this disease (14).

The aryl hydrocarbon receptor (AhR) is a heterodimeric transcription factor, traditionally investigated for its ability to regulate xenobiotic metabolism and the toxicity of 2,3,7,8-tetrachlorodibenzodioxin (TCDD) (15–17). There is accumulating evidence that AhR plays multiple roles mediating a variety of physiological processes, critical to cell homeostasis (18–20). Some of the physiological pathways modified by AhR identified to date include lipid metabolism, apoptosis, angiogenesis/vasculogenesis, and extracellular matrix regulation, to name a few (21–31). These pathways are dysregulated in vascular, autoimmune, neurodegenerative and systemic diseases (32–37). Importantly, dysfunction of these pathways have also been linked with development of retinal diseases including AMD (38–42). We recently showed that AhR activation modifies multiple pathways involved in the pathogenesis of dry and wet AMD, and the loss of AhR in vivo exacerbates the formation and severity of laser-induced neovascular lesions in mice (43, 44). These findings led us to consider AhR as a new viable target to treat angiogenesis and fibrosis, pathways associated with not only development but also progression of neovascularization in AMD patients. To test our hypothesis, we screened a panel of AhR-active pharmaceutical agents and endogenous AhR ligands (45, 46) for their ability to activate the AhR signaling pathway in choroidal endothelial cells, the primary cells vulnerable in choroidal neovascular formation. Furthermore, we assessed their effect in functional angiogenesis assays. Based on the in vitro results, we identified two AhR-active pharmaceuticals, leflunomide and flutamide, with the most promising anti-angiogenic profiles, as candidate drugs to be tested in an animal model of laser-induced neovascularization. We found that daily intraperitoneal treatment of mice with leflunomide and flutamide as monotherapy was able to alleviate the severity of laser-induced lesions in aged wild-type mice, compared to vehicle control. Immunohistochemical analysis of the lesions revealed that AhR activation controlled extracellular matrix deposition and specifically led to a decrease in the accumulation of collagen type IV in the CNV lesions. Furthermore, AhR activation resulted in an increase in expression of microglia and macrophage markers within the lesion as revealed by an increase in recruitment of F4/80⁺ and Iba1⁺ cells to the lesion site. Collectively, these findings validate the role of the AhR pathway in regulating the pathogenesis of CNV lesion formation and support our hypothesis that pharmacologic targeting of AhR may be used as potential therapy for the treatment of neovascular AMD. Finally, our results have identified mechanisms of AhR-based therapies, potentially though regulation of fibrosis and immune-cell recruitment, in the eye.

2. Materials and methods

2.1. Cell lines

RF/6A cells, a spontaneously transformed choroidal endothelial cell line derived from the eyes of a rhesus macaque fetus, was obtained from ATCC (Manassas, VA, USA) and propagated in minimum essential media (MEM) and 10% fetal bovine serum (FBS) as previously described (43, 47). Passages 35–40 were used in this study. All our cell lines are periodically tested for mycoplasma contamination.

2.2. Transcriptional activation assay

The transcriptional activity of AhR was measured using a luciferase-based reporter assay and AhR target gene expression was quantified using qPCR (43, 44). Briefly, 50,000 RF/6A cells/well were seeded in 24-well plates in phenol red-free medium supplemented with 7.5% charcoal-stripped FBS and cultured overnight. Lipofectin (Invitrogen)-mediated transfection was performed the following day, using plasmids encoding AhR-tk – luciferase reporter, CMV-β-galactosidase or pBSII. Following overnight culture, cells were treated with the panel of pharmacological drugs and endogenous ligands at doses listed in Supplementary Table 1 and 2. The cells were lysed 24 hours later for luminescence reading. Luciferase (reporter) and β-galactosidase [chlorophenol red β-D-galactopyranoside (CPRG) as substrate; transfection normalization] activities were measured using a Perkin-Elmer fusion instrument. Concomitantly, cells were treated with these same compounds for RNA isolation and target gene expression studies. All samples were run in triplicate and experiments were performed a minimum of three times (technical and biological replicates=3).

2.3. RNA Isolation and qPCR

Total RNA isolation from cultured cells, RNA quality assessment, cDNA reverse transcription, and qPCR were completed as previously described (43, 44, 47, 48). qPCR was performed using the Bio-Rad CFX96 Realtime PCR Detection System (Bio-Rad). Melt curves for each pair of primers were inspected to confirm a single amplicon. The Ct values were normalized to a housekeeping gene (acidic ribosomal phosphoprotein P0, 36B4). Gene expression fold changes were calculated using the ΔΔCT method. Primer sequences used were selected from Primer Bank, http://pga.mgh.harvard.edu/primerbank and are presented in Supplementary Table 3.

2.4. Scrape wound assay

125,000 RF/6A cells were added to each well in twelve well culture plates. RF/6A cells were pre-treated with AhR ligands, at doses listed in Supplementary Table 1 and 2 for 24 hours. The cell monolayer was scraped in a direction perpendicular to a horizontal line, using a 1000 μl pipette tip, to create a wound. Human recombinant VEGF-165 (Waltham, Massachusetts, USA; 100 ng/ml)-induced cell motility was observed at t = 0 and 36 hours post-scraping. The total number of cells migrating into the wound at t = 36 were counted using ImageJ (developed by Wayne Rasband, National Institute of Health, Bethesda, MD) and normalized to vehicle control. Data were generated from four fields of view/experiment in a total of three biological replicates.

2.5. Tube-formation assay

Tube formation assay was used as a model for angiogenesis. Geltrex™ (Life Technologies, Grand Island, NY, USA) was thawed overnight at 4°C. Using cold pipette tips, 10 μl/well of Geltrex™ was added to a μ-slide angiogenesis plate (ibidi GmBH, Germany). The Geltrex™ solidified into a thin layer after incubation at 37°C for 1 hour. RF/6A endothelial cells were pre-treated with drugs or endogenous ligands for 24 hours (Supplementary Table 1 and 2), trypsinized and then plated onto the Geltrex™-coated wells (12,000 cells/well). Network formation was examined after 3 hours, using an inverted phase contrast microscope, and quantified as total tube length formed by endothelial cells normalized to DMSO control, using ImageJ. Four fields of view/experiment were examined, in a total of three biological replicates.

2.6. Cell viability

RF/6A cells were plated in 96 well plates at a cell density of 10,000 cells/well. Twenty four hours later, cells were pre-treated with AhR ligands for 4 days, (Supplementary Table 1 and 2); fresh media with AhR ligands was added at day 2. Cell viability was measured using CellTiter-Blue® (Promega, Madison, WI), according to the manufacturer’s protocol. Briefly, measurements are based on the ability of living cells to convert a redox dye (resazurin) to a fluorescent end product (resorufin). Nonviable cells do not have the metabolic capacity to carry out the redox conversion and hence, they do not generate a fluorescent signal. The 96-well plate was read at an excitation wavelength of 560 nm and an emission wavelength of 590 nm.

2.7. Animals

11–12 month old male and female mice on the C57BL/6J background (n=39) obtained from Jax mice, and 10–12 month old male and female AhR^−/− mice (n=6) on the C57BL/6J background (43, 44) were maintained at room temperature (25 °C), in a light controlled (12h light/12h dark) environment, and provided standard mouse chow ad libitum. AhR^−/− mice were screened for the confounding retinal degeneration 8 mutation and its absence was confirmed as previously described (44).

2.8. Immunohistochemistry and morphology

Eyes were fixed in 4% paraformaldehyde and cryopreserved as previously described (43, 44, 47). Specimens were cryosectioned from the superior cup through the optic nerve to the inferior cup in 10 μm increments. Cryosections from the superior, central and inferior regions of the eye were probed with antibodies (Supplementary Table 4) specific to collagen IV (COL4), fibronectin (FN1), adhesion G-protein coupled receptor E1, also known as F4/80, and the microglial marker allograft inflammatory factor 1 [alternate name, ionized calcium binding adaptor molecule 1 (Iba1)]. Non-specific immunostaining in sections was blocked with normal serum (Jackson Immunoresearch, West Grove, PA, USA) appropriate to the secondary antibody species. Secondary antibodies were conjugated to AlexaFluor 568 and 488 (Thermo-Fisher Scientific, Grand Island, NY). Control slides containing sequential sections were probed with non-immune serum and buffer without primary antibody. Nuclei were stained with Hoechst 33258 (Thermo-Fisher Scientific, Grand Island, NY). Images were collected on a Nikon C1 or C2si confocal microscope (Nikon Corporation, Tokyo, Japan). and visualized and processed using Adobe Photoshop CS4.

2.9. Mouse model of CNV

Laser photocoagulation was performed in cohorts of aged C57BL/6J (11–12 month old) and AhR^−/− (10–12 month old) mice, as previously described (43, 47). Briefly, four thermal burns were induced in each eye around the optic nerve, using a slit lamp delivery system. To test the efficacy of AhR drugs in mice expressing the AhR in the eye, C57BL/6J mice were divided into three cohorts and treated with vehicle control, n=14 eyes (1% DMSO in saline), leflunomide, n=17 eyes (Sigma-Aldrich, St. Louis, MO; 20 mg/kg/day, i.p.) or flutamide, n=8 eyes (Sigma-Aldrich, St. Louis, MO; 5 mg/kg/day, i.p.). To test the potential off-target effects of AhR agonists, AhR^−/− (n=4 eyes/cohort) were treated with vehicle, lefluonomide, or flutamide at the same concentrations listed above. All mice were pre-treated with the drugs daily for 2 days prior to laser CNV induction and five days/week there after until 3 weeks post-laser treatment, when they were euthanized. Blood was collected from the sub-mandibular vein and eyes were harvested for assessment of laser-induced CNV in posterior pole flat-mounts, or cryopreserved for immunohistochemistry and morphology experiments, as described above. CNV lesion volume, area and size were measured in flat-mounts stained with isolectin GS-IB₄ (Griffonia simplicifolia isolectin type IB₄) Alexa Fluor^® conjugate (Life Technologies, Grand Island, NY) to examine vascularity of the neovascular lesion as previously described (43, 47).

2.10. In vivo imaging

A Micron IV (Phoenix Research Laboratories Inc, Pleasanton, CA) retinal imaging microscope was used for in vivo imaging. Fundus images were obtained from anesthetized C57/BL6J and AhR^−/− mice. The optical coherence tomography (OCT) module of the Micron IV was used to image retinal layers guided from the bright field. Fluorescein angiography was also obtained on anesthetized mice following an intraperitoneal injection with 10% sodium fluorescein (AK-FLUOR, Akorn, Decatur, IL) at a dose of 0.02 ml/5–6 gram body weight.

2.11. Evaluation of mouse CNV lesions

Following laser CNV, ex vivo flatmounts of the posterior pole were stained with isolectin GS-IB₄ Alexa Fluor® conjugate according to manufacturer’s protocol and as previously described (43, 47), to examine vascularity and size of the neovascular lesion. Lesions were visualized by a Nikon C2si confocal microscope. Horizontal optical section or z-stack images of the flatmounts were obtained at 1.50 μm intervals using NIS-elements microscope imaging software. Total area of the CNV lesions per lesion per eye was measured using ImageJ software by setting the scale (79 pixels = 100 μm), and drawing a free shape around the CNV lesions. The total area of CNV lesions per eye was calculated and divided by the number of laser burns to calculate ‘mean CNV lesion area per eye’. The volume of the CNV lesions was also measured using the NIS-Elements software. The threshold intensity was set using the central z-slice, following which volume measurement function was used to calculate the volume across the z-stack. All measurements are reported as μm² for area and μm³ for volume. Cryosections from the lasered eyes containing the mid point of the lesions were stained with F4/80 and Iba1 antibodies (Supplementary Table 4). CNV was demarcated and total number of cells staining positive for F4/80 and Iba1 were counted and plotted. To study extracellular matrix deposition, cryosections were stained with FN1 and COL4 antibodies. CNV lesions were demarcated and fluorescence intensity of FN1 and COL4 was measured by ImageJ. Mean intensity was plotted.

2.12. Cytokine analysis

Blood was collected from mice via the sub-mandibular route at the end of drug treatment in tubes with anti-coagulant. Tubes were spun at 8,000 rpm for 15 minutes at 4°C to pellet the red blood cells. Plasma was carefully removed and stored in a separate tube. Protein quantitation was performed using BCA assay and was used to normalize the amount of protein loaded on to each membrane from Mouse Cytokine Array C3 (RayBiotech, Norcross, GA). Each treatment group (vehicle, leflunomide and flutamide) had three biological replicates. The manufacturer’s protocol was followed for processing the membranes. The images were inverted to neutralize the effect of any differences in background, and intensity of the dots was measured using ImageJ. Since each antibody on the membrane was blotted in duplicate, the mean intensity of the dots was calculated, then averaged across the biological replicates and normalized to vehicle group.

2.13. Statistics

Statistical methods for data analysis included two-tailed Student’s t-test and two-way ANOVA, with Sidak’s multiple comparison test using GraphPad Prism. Values were considered statistically significant at p < 0.05 and are indicated in the figures.

2.14. Study approval

All animal protocols used were approved by the Duke University Institutional Animal Care and Use Committee (IACUC), and experiments were performed in accordance with the guidelines of the ARVO statement for the Use of Animals in Ophthalmic and Vision Research.

3. Results

3.1. AhR activation profile of pharmacological drugs and endogenous receptor ligands in choroidal endothelial cells in vitro

Previously we reported that AhR is present in the normal human choroid, and the signaling pathway can be activated in choroidal endothelial cells in vitro (43, 44). Given the potential for different ligands to produce unique physiological effects based on their target tissue (16, 46), we investigated the efficiency of a panel of AhR-active pharmaceuticals drugs and endogenous AhR ligands (Supplementary Tables 1 and 2), to activate AhR, concurrent with their ability to induce AhR target gene expression in choroidal endothelial cells. We selected this population of cells as our target given they migrate through breaks in bruch’s membrane towards the RPE, leading to CNV development, and as such are vulnerable in wet AMD. We found that all seven pharmacological ligands and three out of four of the known AhR endogenous ligands induced activation of the AhR-tk-luciferase reporter plasmid in choroidal endothelial cells (Figure 1A). 2,3,7,8-tetrachlorodibenzodioxin (TCDD), a known agonist of AhR was used as a positive control and was able to efficiently activate the AhR promoter in choroidal endothelial cells at similar levels to that of the drug panel and endogenous ligands. In contrast, stemreginin1 (SR1), a purine derivative and a potent AhR antagonist, did not induce the activity of the AhR promoter in choroidal endothelial cells. The AhR signaling pathway was further assessed by determining the expression of known receptor target genes in response to the same set of AhR ligands (Figure 1B, C) that were optimized and did not negatively affect cell viability (Supplementary Figure 1). We found significant induction of the expression of two AhR-specific target genes [cytochrome P450, family 1, subfamily A, polypeptide 2 (CYP1A2) and cytochrome P450, family 1, subfamily b, polypeptide 1 (CYP1B1)], important in xenobiotic metabolism, following treatment with the AhR-active pharmaceuticals and endogenous ligands (Figure 1B, C). Collectively, our data support that, with the exception of ITE, an indole endogenous ligand for AhR, all other synthetic and endogenous ligands tested were able to activate the AhR signaling pathway in choroidal endothelial cells in vitro at levels comparable to TCDD and these were observed at non-cytotoxic concentrations.

Figure 1.

Effect of pharmacological and endogenous ligands on AhR activity in choroidal endothelial cells. (A) AhR activity (values are expressed as mean and SEM) in choroidal endothelial cells transfected with the AhR–tk–luciferase reporter; cells were treated with pharmacological, endogenous ligands, or DMSO as control (n = 3): *, p < 0.05 relative to DMSO-treated cells. Relative fold change in expression of CYP1A2 (B) and CYP1B1 (C) in choroidal endothelial cells in response to pharmacological, endogenous ligands, or DMSO treatment (values are expressed as mean and SEM) (n = 3); *, p < 0.05 relative to DMSO treated cells; ns = not significant.

3.2. Leflunomide and flutamide inhibit VEGF-induced choroidal endothelial cell migration and tube formation

Key processes in neovascularization include endothelial cell migration and sprouting (49) of new vessels from the pre-existing vasculature. To study the effect of AhR activation on new vessel growth we examined the effect of the panel of synthetic and endogenous ligands on choroidal endothelial cell migration, tube formation and proliferation, using VEGF as a stimulant (Figure 2). As expected, VEGF was able to induce choroidal endothelial cell migration in comparison to control. Importantly, we found that leflunomide and flutamide were able to significantly inhibit VEGF-induced migration of choroidal endothelial cells into the ‘scraped’ wound (Figure 2A, C). On the other hand, none of the endogenous ligands were able to significantly impact VEGF-induced migration (Figure 2B, D). We also tested the effect of AhR agonists on the ability of endothelial cells to undergo morphogenesis and organize into tube-like structures in a 3-D matrix. In the presence of VEGF, which induces tube formation, we visualized and measured tube-like structures. Total tube length was used as a surrogate of vascular morphogenesis. We found that pre-treatment with leflunomide and flutamide was sufficient to inhibit VEGF-induced tube formation in choroidal endothelial cells (Supplementary Figure 2). Based on these findings we concluded that leflunomide and flutamide demonstrate an anti-angiogenic profile in choroidal endothelial cells and selected them as our drug candidates to move to the next phase; examining the effect of AhR activation in an in vivo model of experimental CNV.

Figure 2.

Effect of AhR activation by pharmacological and endogenous ligands on VEGF-induced migration in choroidal endothelial cells. The effect of pharmacological drugs (A) and endogenous ligands (B) on cell migration was analyzed in a VEGF-induced wound healing assay (n = 3, representative images at t = 36 hours are shown); dotted lines demarcate the boarders of the scrape wound. Ctrl: media only, VEGF: vascular endothelial growth factor (100 ng/ml). Cells migrating following treatment with pharmacological drugs (C) and endogenous ligands (D) into the scraped region were counted using ImageJ (mean and SEM; n = 3; *, p < 0.05 relative to Ctrl; ns = not significant; one way ANOVA, Tukey's multiple comparisons test).

3.3. Leflunomide and flutamide treatment inhibits laser-induced CNV in aged mice

The effects of AhR activation on the development of laser-induced CNV, were evaluated by live imaging using fundus photography, OCT, and fluorescein angiography; and post mortem evaluation of flatmounts of posterior eye cups following isolectin GS-IB₄ staining, to highlight the vasculature. Given the nature of AMD and its occurrence in the elderly, aged 11–12 month old C57BL/6J mice, which have been shown to express AhR, were used (43, 44). The treatment protocol involved intraperitoneal injections of 20 mg/kg/day leflunomide, 5 mg/kg/day flutamide or sterile vehicle, two days prior to laser photocoagulation and every day thereafter for 21 days. The drugs at the above-mentioned doses were found to be non-toxic to the animals as evident by the negligible changes (3–5%) in the weights, pre- and post-treatment (Supplementary Figure 3). On day 19, Micron IV (Phoenix Research Labs, Pleasanton, CA) was employed for in vivo imaging and qualitative analysis of the impact of the AhR ligand on CNV lesions. We performed fluorescein angiography to visualize and examine leakage of dye from the neovascular lesions, which is proportional to lesion size. A drug dependent decrease in lesion size and dye leakage, was measured following treatment, in the leflunomide and flutamide cohorts, which displayed relatively smaller lesions and less dye leakage in comparison to vehicle treated mice (Figure 3A). Fundus visualization of the eyes from each cohort corroborated the CNV size observed by fluorescein angiography, and OCT imaging allowed for identification of clearly defined boundaries, visualized as hyper-reflective regions, in cross sections through the laser-induced CNV lesions in vivo (Figure 3B, C). The displacement of retinal layers overlying the center of the lesion was a common characteristic of lesions from all the groups and was dependent on the lesion size. We confirmed that the AhR ligands administered were indeed able to activate the AhR pathway in the target tissue by examining expression of mouse AhR target genes (Cyp1a1 and Cyp1a2), which were elevated compared to the vehicle treated in the RPE/choroid tissue complex (Supplementary Figure 4). Quantitative evaluation of the CNV lesions was performed by measuring the area, distribution, and volume of isolectin GS-IB₄ stained flatmounts. We discovered that the severity and complexity of the lesions decreased following AhR activation. Specifically, the incidence of three merged CNV lesions was substantially greater in the vehicle cohort in comparison to the leflunomide and flutamide treatment groups (Figure 4A, B). Quantitatively, the mean lesion area of the two treatment groups were significantly lower than that of the vehicle group; 51% lower in the leflunomide and 61% lower in the flutamide group (Figure 4C). We found that two-dimensional quantifications of the lesion area translated into the three-dimensional measurements of the lesion volume. Three-dimensional reconstruction of images spanning the lesion thickness allowed us to measure the volume of the CNV lesion, which exhibited a significant decrease in ligand treated mice (55% decrease in volume in leflunomide cohort and 56% decrease in flutamide cohort; Figure 4D, E). Therefore, leflunomide and flutamide, which were selected on the basis of their ability to inhibit choroidal endothelial migration and tube formation in vitro, while effectively activating the AhR signaling pathway, were found to be successful in inhibiting lesion formation in an animal model of laser-induced choroidal neovascularization.

Figure 3.

In vivo evaluation of the effect of AhR activtion on lesion formation in a laser-induced CNV mouse model. Micron IV based in vivo imaging taken of the posterior pole of mice treated with vehicle, leflunomide and flutamide 19 days post CNV induction. Representative images are shown. (A) Regions of leakage (dotted line) from the CNV lesions are visible in fluorescein angiography images. (B) Fundus images of the posterior eye showing the extent of the CNV lesions (dotted line). (C) OCT images displaying cross-sections of the lesions (corresponding to the green line in the bright-field images). CNV lesions are marked by dotted lines. Retinal layers are labeled in the OCT image of a leflunomide treated eye: IPL: inner plexiform layer, INL: inner nuclear layer, ONL: outer nuclear layer, OS: outer segments.

Figure 4.

Ex vivo evaluation of the effect of AhR activation on lesion formation in a laser-induced CNV mouse model. (A) Representative images of the posterior flatmounts stained with isolectin GS-IB₄ from three treatment groups; vehicle, leflunomide and flutamide. Two examples from each cohort are shown. Dotted circle delineates the CNV lesion; solid circle delineates the optic nerve (ON). (B) Distribution of number of eyes with individual versus merged lesions in reatment groups. (C) Meanlesion area/eye was measured using ImageJ (mean and SEM.; *p < 0.05, two tailed t-test). (D) Mean lesion volume/eye (mean and SEM for each group; *p < 0.05, two tailed t-test). (E) Representative images of 3-D reconstruction of lesions from the treatment groups.

3.4. AhR activation results in decreased collagen deposition in the CNV lesions

We have previously reported that loss of AhR results in increased expression of collagen type IV (COL4) by RPE cells in vitro and within the eye in vivo (43, 44). COL4 has been found to be a component of sub-RPE deposits that accumulate in the eyes of aged AhR^−/− mouse eyes, as well as neovascular lesions in aged AhR^{− −} mice (43, 44). This is significant with respect to AMD pathology because COL4 has been found to be a component of sub-RPE deposits and CNV membranes in human AMD as well (50, 51). Additionally, fibronectin (FN1) is a known component of Bruch’s membrane, sub-RPE deposits and associated with CNV lesions (51–54). In light of these observations, we examined the distribution of these two proteins in the CNV lesions (Figure 5A). We discovered that while AhR ligand treatment did not affect fibronectin staining in the lesion, it significantly lowered the staining intensity of COL4 as compared to the control group treatment (Leflunomide: 32% Flutamide: 32%; Figure 5b). These observations suggest that leflunomide and flutamide may regulate the deposition of extracellular matrix and possibly, fibrosis in the neovascular lesions, supporting a potential mechanism for AhR-based therapies in controlling CNV lesion formation.

Figure 5.

AhR activation decreases collagen type IV (COL4) deposition in CNV lesions (A) Fibronectin (FN1;green) and COL4 (red) immunolocalization in CNV lesions of mice treated with vehicle control, leflunomide, or flutamide. Dotted oval demarcates the lesion area; nuclei are stained blue with Hoechst; representative images are shown; scale bar = 50 μm. (B) COL4 and FN1 staining intensity was quantified in the CNV lesions using ImageJ (mean and SEM; n = 3/group; *p < 0.01; one way ANOVA, Tukey's multiple comparisons test).

3.5. AhR activating ligands modulate inflammation in vivo

AhR has been shown to regulate immune cell localization in CNV lesions (43). With this in mind, we assessed inflammation in CNV lesions from vehicle and drug treated mice by probing retinal/RPE/chorid cross sections with antibodies to Iba1, which labels microglial cells and macrophages, and F4/80, which labels mature macrophages (Figure 6). Whereas leflunomide treatment did not have an effect on the number of Iba1⁺ and F4/80⁺ cells in CNV lesions, flutamide caused a significant upregulation of both Iba1 and F4/80 stained cells in the CNV lesions (Figure 6A, B). To further corroborate the effect of leflunomide and flutamide on inflammation, we measured the relative systemic cytokine levels in plasma collected from our treated mouse cohorts. We observed that out of a panel of 62 cytokines, leflunomide and flutamide treatment caused a significant reduction in the expression of eight and six cytokines, respectively, with a primarily pro-inflammatory profile in systemic circulation (Figure 7). The representative images used for the intensity measurements are also shown (Supplementary Figure 5). In summary, AhR activation significantly reduced the systemic levels of pro-inflammatory cytokines in mice, potentially contributing to the inhibition of lesion formation as a result of laser burns.

Figure 6.

AhR activation promotes microglial infiltration into CNV lesions (A) F4/80 (green) and Iba1 (red) immunolocalization in CNV lesions of mice treated with vehicle control, leflunomide, or flutamide. Dotted oval demarcates the lesion area; nuclei are stained blue with Hoechst; representative images are shown; scale bar = 50 μm. Inset panels are high magnification views of F4/80 and Iba1 immunopositive cells indicated by white arrowheads. (B) F4/80 and Iba1 positive cells were quantified in the CNV lesions using ImageJ (mean and SEM.; n = 3/group; *p < 0.01; ns: not significant, one way ANOVA, Tukey's multiple comparisons test).

Figure 7.

AhR activation regulates circulating cytokine profiles. C3 Mouse cytokine array was used to assess systemic levels of 62 cytokines in the plasma of mice treated with vehicle, leflunomide and flutamide. (A) Dot plots demonstrating select cytokines, which were significantly modulated by drug treatment. Two examples/cohort are shown. (B) Plot of dot intensity for cytokines depicted in panel A (mean and SEM.; n = 3/group; *p < 0.01; ns: not significant, Multiple unpaired t-tests, Holm-Sidak correction for multiple comparisons).

3.6. Treatment with leflunomide and flutamide does not affect lesion severity in aged AhR^−/− mice

We have established that leflunomide and flutamide can act as AhR agonists in vitro in choroidal endothelial cells and also in vivo, as evident by induction of AhR target genes in the mouse RPE/choroid tissue following ligand treatment. To further corroborate specificity of the AhR activating ligands in vivo, we tested the therapeutic potential of leflunomide and flutamide to inhibit CNV lesion formation in aged AhR^−/− mice. This experiment was done to address the concern of off-target effects of the AhR ligands. Following the previous experimental protocol, aged 10–12-month old AhR^−/− mice on the C57BL/6J background, received intraperitoneal injections of 20 mg/kg/day leflunomide, 5 mg/kg/day flutamide or sterile vehicle, 2 days prior to induction of laser burns into the back of the eye, and every day thereafter for 21 days. We imaged the mice on day 19 and obtained fluorescein angiography and OCT images (Supplementary Figure 6A, B). Qualitative evaluation of the images substantiated a lack of therapeutic efficacy in the absence of AhR, with the drug treatment. Upon further analyses and quantification of the lesion area and volume (Supplementary Figure 6C, D), no significant differences were seen, thus confirming that the effects of leflunomide and flutamide observed on CNV lesion formation in the eye are mediated through the AhR signaling pathway rather than off-target effects.

4. Discussion

Historically, the AhR was discovered as the mediator of the expression of xenobiotic metabolizing enzymes and the toxicity of TCDD, as such it was originally dubbed the dioxin receptor (16–18). More recently, its role has been expanded to include regulation of several other molecular pathways including angiogenesis/vasculogenesis, inflammation and extracellular matrix regulation (21–31). Consequently, AhR is now known to be important in the development of diseases including cancer, cardiovascular disease, neurodegenerative and autoimmune disorders to name a few (32–37). Of direct relevance to our study, many of these diseases share common pathogenic pathways with AMD (38, 41, 42). Recently, we confirmed that AhR plays a role in the development of pathologies characteristic of AMD in mice, and demonstrated that in the absence of AhR, there is accumulation of sub-RPE debris, progressive choroidal thinning, and exacerbation of laser-induced CNV lesion formation (43, 44). This led us to evaluate the AhR signaling pathway as a drug target for the treatment of pathological neovascularization associated with wet AMD, and cognizant that ligands may behave differently based on the targeted tissue (16, 46), to investigate mechanisms underlying AhR-based therapies in the eye. With this goal in mind, we screened a panel of known AhR-active compounds to evaluate their ability to activate this receptor in choroidal endothelial cells, cells vulnerable in wet AMD, and determine their impact on angiogenesis, extracellular matrix regulation and inflammation in the eye.

The most common treatment for wet AMD, characterized by the leakage of vessels under the RPE and retina, is anti-VEGF therapy. Though effective in a sub-population of patients, up to 60% of wet AMD patients see partial to no improvement (11–13). This highlights the need to identify new signaling pathways that can be effectively targeted to alleviate neovascular leakage and fibrosis. Herein, we examined the therapeutic potential of targeting the AhR signaling pathway in the eye. We screened a panel of synthetic and endogenous AhR ligands for their ability to activate the receptor in choroidal endothelial cells and prevent angiogenesis. The endogenous ligands used in this study are all derivatives of amino acid metabolism (Supplementary Table 2) and the panel of pharmacological drugs used were derived from a recent report which investigated the effect of eight AhR-active pharmaceuticals including, leflunomide (anti-inflammatory), nimodipine (calcium channel blocker), sulindac (anti-inflammatory), tranilast (anti-allergic), mexiletine, flutamide (anti-androgen) and 4-hydroxytamoxifen on AhR activity in two different breast tumor cell lines where they found the ligands exhibited structure- and cell context-dependent AhR agonist/antagonist activities (46). Of the drugs tested, two candidate ligands that displayed promising anti-angiogenic profiles were leflunomide and flutamide, drugs that have been extensively tested and considered as viable treatments for cancer and autoimmune diseases. Leflunomide, has been investigated for its ability to inhibit tumor angiogenesis in breast tumor cell lines (46), and subcutaneous tumors established from human colon carcinoma cells (55), as well as treatment of rheumatoid arthritis (56). Flutamide, has been shown to have anti-androgen properties and is effective at reducing tumor vessel density (57). It is noteworthy that the magnitude of AhR activation in choroidal endothelial cells by endogenous ligands, used at levels that did not negatively affect cell viability, was considerably lower than that of the synthetic drug panel. This observation may have downstream implications as reflected by the results obtained in functional angiogenesis assays. Whereas leflunomide and flutamide were both successful in inhibiting VEGF-induced cell migration in a 2-dimensional culture and tube formation in a 3-dimensional space, which measure cell motility and reorganization respectively, none of the endogenous ligands were able to induce inhibitory effects in angiogenesis assays.

Endothelial cell migration and tube formation assays allowed us to identify two candidate drugs to be tested in vivo. We utilized the power of a well-established in vivo model of CNV, experimentally induced by laser burns, which result in breaks through bruch’s membrane, allowing the formation of new vessels below the retina (43, 47, 58). Following systemic administration of leflunomide and flutamide, we found that both treatments resulted in an increase in AhR target gene expression in RPE/choroid tissue, confirming the presence of the drug in the target tissue. Importantly, the treatments demonstrated a therapeutic effect as measured by a decrease in lesion area, volume and severity. To identify mechanisms underlying the action of these drugs, we probed the retina/RPE/choroid tissues with markers of extracellular matrix molecules and immune cells. Extracellular matrix remodeling and immune cell infiltration into CNV lesions have consistently been reported to be associated with the pathogenesis of AMD (38, 59–63). We discovered that deposition of collagen type IV, an extracellular matrix molecule associated with human CNV membranes, was downregulated as a result of leflunomide and flutamide treatment, suggesting a reduction in the formation of fibrosis or scar tissue. Leflunomide has previously been shown to prevent renal injury in diabetic rats by inhibition of TGFβ1 mediated extracellular matrix deposition and tubulointerstitial fibrosis, as well as by inhibiting tubular epithelial-myofibroblast transdifferentiation, corroborating the inhibitory effect of leflunomide on fibrosis (64). Whereas flutamide treatment has not been reported to directly regulate extracellular matrix production or deposition, therefore, this is the first report showing the regulation of collagen IV by flutamide.

Inflammation is a major regulator of the different clinical sub-types of AMD, including geographic atrophy and neovascular AMD (38). Inflammation has been proposed to contribute to AMD progression at two levels. Changes in inflammatory markers and immune cells locally within the retinal tissue, and contributions from circulation. Results of studies that have investigated changes in local distribution of immune cells are mixed such that some studies have indicated the presence of resident microglia and recruited macrophages, in the sub-retinal space to be associated with exacerbation of the laser CNV phenotype (43, 65), while other studies have found local immune cells to be associated with amelioration of CNV lesion formation (47, 66, 67). To determine the effect of AhR activation on immune cell distribution in conjunction with CNV severity, we probed retina/RPE/choroid cross sections with antibody markers for macrophages (F4/80) and microglial cells (Iba1). Interestingly, we did not observe a change in the relative numbers of Iba1 and F4/80 positive cells in CNV lesions in the leflunomide treated group, which itself is an anti-inflammatory drug. On the other hand, the flutamide cohort displayed a significant upregulation of F4/80 and Iba1-positive cells in the CNV lesions. Previously, flutamide has been reported to regulate recruitment of microglial cells in an organotypic model of cerebellar slices used in studies of nerve myelation. It was shown that flutamide treatment was able to reverse the remyelating effects of testosterone and caused recruitment of astrocytes and microglia (68). Overall, this is an interesting observation as immune cells such as macrophages are known to differentiate into distinct phenotypes. The M1 macrophages are known for their pro-fibrotic and pro-inflammatory properties whereas, M2 macrophages, are characterized by their immunosuppressive and tissue remodeling properties (38, 69, 70). Given the functionally distinct phenotypes in which macrophages can differentiate into, they may have either pro-inflammatory or immunosuppressive effects within the sub-retinal space. This suggests that further investigation is necessary to identify the effect of AhR ligands as well as laser injury on the differentiation status of macrophages in the ocular space. Our results vary from previous studies that found macrophage depletion diminishes lesion size. This may in part be due to the fact that we examined the distribution of immune cells at only one time point, three weeks post-laser injury, whereas other studies tended to examine lesion size only up to 7 days post-laser (58, 59, 63, 65, 67). Our rationale for choosing a longer time point, was to allow for the formation of a mature lesion with some degree of fibrosis. Thus it is plausible that macrophages accumulating at the later stages of lesion formation may demonstrate an anti-inflammatory phenotype. We also examined the potential systemic contribution of inflammation to CNV development. Screening of the circulating cytokines, revealed an overall downregulation of select cytokines some of which have been shown to be elevated in AMD patients, following AhR activation. For example, IL-17A, which is a pro-inflammatory cytokine, has been shown to elicit a pro-angiogenic effect on human choroidal endothelial cells, and its receptor and IL-17 receptor C, are upregulated in patients with wet AMD (71). CXCL1 was another cytokine modulated as a result of AhR activation. This cytokine has been shown to be secreted by CD11b⁺ cells, which populate at the site of laser injury, and potentially contribute to a pro-inflammatory microenvironment (72). Interestingly we also observed that the levels of leptin and its receptor (leptin-R) were significantly downregulated in leflunomide treated mice. Leptin is also known as the ‘satiety hormone’ (73) . It is secreted by adipose cells that help regulate energy balance by inhibiting hunger. This could have an indirect effect on wet AMD by regulating visceral fat ratio, a proposed risk factor for wet AMD (74). CXCL5, is another pro-inflammatory cytokine, which is produced following stimulation of cells with inflammatory cytokines such as TNF-α and has been implicated in connective tissue remodeling (75), an important molecular pathway in scar formation in wet AMD. Furthermore, its expression has been shown to be elevated in RPE-endothelial co-cultures resulting in upregulation of both pro-angiogenic and pro-inflammatory processes (76). Finally, CXCL2, also known as MIP-2 (macrophage inflammatory protein-2), has been shown to be secreted by microglial cells and acts as a strong pro-angiogenic molecule in glioma pathogenesis (77). Collectively, these results support that AhR activation regulates multiple inflammatory pathways both systemically and within the eye following injury.

The complexity of wet AMD is noteworthy as multiple cell types contribute to its pathology, including choroidal endothelial cells, RPE cells, and immune cells, all of which express AhR (43, 44, 78). It remains to be shown if one cell type is affected more or less by AhR activiation in the posterior eye. In our study, while treatment with AhR-active compounds did not effect the integrity and/or morphology of regions of the murine RPE and choroid not subjected to laser burns, it was able to ameliorate lesion size and composition of the RPE/choroid subjected to laser injury. Previously we have shown that the activity of AhR in ARPE19 cells, a spontaneous human RPE cell line derived from a 19 year old donor, is greater than that of RF/6A, a choroidal endothelial cell line derived from the macaque monkey (43). Though there are limitations to the cell culture model systems used and their fidelity to human cells, the results may indicate that systemic administration of an AhR ligand could potentially activate the receptor in both of these ‘AMD-vulnerable’ cell types and certainly since circulating monocytes may be recruited to a neovascular lesion (66), activation of AhR in monocytes may also be a contributing factor. Growth factors such as VEGF and pigment epithelial derived factor (PEDF) are also known regulators of CNV development and we have previously reported that knockdown of AhR in RPE cells results in an increase in VEGFA and decrease of PEDF expression. Interestingly, AhR knockdown in choroidal endothelial cells did not effect VEGFA expression. Collectively, these data support the potential beneficial impact of AhR activation at the level of multiple ‘AMD-vulnerable’ cells in the treatment of wet AMD and futher suggest that AhR ligands may be used as an adjuvant therapy along with anti-VEGF agents to tackle both vascular leakage and fibrosis.

TCDD and related halogenated aromatic compounds bind with high affinity to the AhR and induce a well characterized set of toxic responses that include wasting syndrome, thymic atrophy and immunotoxicity (79). However, upon identification of other AhR-responsive pathways such as those involved in autoimmune disease, it has also been shown that TCDD can inhibit development of symptoms of multiple sclerosis in mouse models. These findings have led to identification of selective AhR modulators (SAhRMs) that do not exhibit classical ‘dioxin-like’ toxicities, for potential clinical application in the treatment of diseases including multiple sclerosis and breast cancer (80, 81). Similarly, a recent study reported that systemic treatment with TCDD exacerbated laser-induced CNV formation in mice (82). These results collectively support the hypothesis that the consequences of AhR activation are dictated by its downstream effectors and highlight the importance of utilizing the power of SAhRMs to achieve therapeutic benefit (46). Of significance, in our study we addressed the specificity of AhR activation by evaluating the effect of leflunomide and flutamide in AhR^−/− mice following laser-induced CNV induction. We did not see a therapeutic benefit in our end-points of lesion size and volume in the drug-treated groups, suggesting that leflunomide and flutamide reaching the eye, act through the AhR pathway.

6. Conclusions

Previous studies have demonstrated that screening known AhR-active pharmaceuticals (83) can be used to identify individual compounds or SAhRMs that inhibit breast and pancreatic cancer cell invasion (45, 46, 84). Herein we took a similar approach in our study and have shown for the first time, the ability of two AhR-active pharmaceuticals to provide therapeutic benefit on pathologies associated with wet AMD, specifically reducing pathological neovascularization in the sub-retinal space, potentially by regulating fibrosis and inflammation. These results argue in favor of developing AhR as a drug target for the treatment of neovascular AMD.

Highlights.

• The aryl hydrocarbon receptor regulates the severity of aberrant new vessel growth in the back of the eye.

• Lefluonomide and flutamide are two selective modulators for the aryl hydrocarbon receptor capable of ameliorating angiogenesis in vitro.

• Activation of the aryl hydrocarbon receptor suppresses volume and severity of choroidal neovascular lesions.

• Several mechanisms regulated by the aryl hydrocarbon receptor in aberrant neovascularization in the eye include angiogenesis, extracellular matrix turnover, inflammation and fibrosis.