Levodopa Suppresses Choroidal Neovascularization Through a Tyrosinase-Dependent Dual Mechanism

Abstract

Purpose

Levodopa (L-DOPA), a precursor for melanin and dopamine, has been linked to reduced intravitreal injection burden and delayed onset of neovascular age-related macular degeneration (AMD). Further, L-DOPA and dopamine receptor D2 (DRD2) agonists inhibit laser-induced choroidal neovascularization (CNV). However, the contributions of endogenous versus exogenous L-DOPA signaling, their effects in alternative CNV models, and the contributions of DRD2 versus GPR143, the receptor for L-DOPA, signaling remain unresolved.

Methods

Choroidal sprouting assays (CSA) were performed using wild-type (WT) and tyrosinase-mutant (Tyr^−/−) mice with and without dopamine pathway agonists and antagonists. CNV number and area were measured in pigmented Vldlr^−/− and albino Vldlr^−/−Tyr^−/− mice with and without L-DOPA or the DRD2 agonist quinpirole.

Results

Endogenous L-DOPA deficiency (Tyr^−/−) did not affect CSA sprouting or CNV in Vldlr^−/− mice. Exogenous L-DOPA suppressed angiogenesis ex vivo in both WT and Tyr^−/− choroidal explants in a dose-dependent manner. In vivo, L-DOPA reduced CNV lesion number, lesion area, and macrophage infiltration in albino but not pigmented Vldlr^−/− mice. Dopamine and quinpirole produced modest anti-angiogenic effects, and eticlopride partially reversed L-DOPA inhibition in choroidal explants. Quinpirole suppressed CNV lesion number, lesion area, and macrophage infiltration in pigmented Vldlr^−/− mice.

Conclusions

Our findings show that L-DOPA's anti-angiogenic effects are exogenous, more effective in tyrosinase-mutant mice, and mediated by both the DRD2 and non-DRD2 pathways, potentially GPR143. The saturation of GPR143 signaling in pigmented eyes provides a mechanistic basis for reduced responsiveness, highlighting the importance of pigmentation biology in the development of L-DOPA-based therapeutics.

Age-related macular degeneration (AMD) is one of the leading causes of irreversible central vision loss among individuals over 60 years of age in developed countries.¹ Neovascular AMD (nAMD) is particularly devastating, accounting for the majority of severe visual impairment associated with AMD.² nAMD is driven by pathologic choroidal neovascularization (CNV), which disrupts the retinal architecture through vascular leakage, hemorrhage, and eventual fibrotic scar formation, resulting in rapid and often profound central vision loss. Whereas intravitreal anti-vascular endothelial growth factor (VEGF) agents have transformed the management of nAMD by stabilizing and, in some cases, improving visual outcomes, these therapies remain reactive rather than preventive.³ The only widely recommended strategy for reducing the risk of progression from intermediate to nAMD is oral supplementation with the Age-Related Eye Disease Study 2 (AREDS2) formulation.⁴ Consequently, there remains a critical unmet need for effective preventive approaches that can delay or halt the onset of nAMD, thereby reducing the burden of irreversible vision loss.

Pigmentation has long been associated with differential susceptibility to AMD. Epidemiological studies suggest that individuals with lighter iris pigmentation are at a higher risk of AMD, indicating a protective role for melanin and melanin-related pathways in retinal health.⁵ One molecule central to both pigmentation biology and neuroprotection is levodopa (L-DOPA), the precursor to melanin synthesis as well as dopamine. Several lines of evidence suggest a potential protective link between L-DOPA and AMD. Epidemiologic analyses have reported that individuals receiving exogenous L-DOPA, commonly prescribed for neurological disorders, exhibit a reduced incidence and delayed onset of AMD compared with matched controls.⁶ In support of this study, exogenous L-DOPA treatment can delay the need for anti-VEGF treatment in patients with new-onset nAMD and can reduce anti-VEGF injection burden in a small, prospective cohort study.⁷ Furthermore, L-DOPA exposure has been shown to reduce anti-VEGF injection burden and delay the onset of nAMD in large database studies from the United States⁸ and France.⁹ However, these studies were confounded by a large proportion of patients with Parkinson's disease. Importantly, exogenous L-DOPA supplementation in patients without Parksinon's disease is well tolerated and restores diabetes-induced neuroretinal dysfunction.^10,11 Thus, exogenous L-DOPA could be a preventative treatment for patients at high risk for conversion to nAMD or as an anti-VEGF reducing agent for patients with nAMD undergoing anti-VEGF injections.

However, several important gaps remain in our understanding of the L-DOPA–AMD axis. First, the precise mechanism by which L-DOPA exerts its protective effects is unclear. L-DOPA can signal through the G protein-coupled receptor GPR143, which is expressed in retinal pigment epithelial cells and melanocytes,¹² reducing the expression of VEGF and increasing the expression of pigment epithelial-derived factor (PEDF), both of which are anti-angiogenic changes.¹³ In addition, L-DOPA can be converted to dopamine and signal through the dopamine receptor D2 (DRD2), which decreases VEGF receptor 2 (VEGFR2) signaling.^9,14 Second, current experimental evidence is largely derived from the laser-induced CNV mouse model.¹⁵ Although the laser-induced CNV model is widely used and has provided valuable insights like the importance of VEGF,¹⁶ this model reflects an acute injury response rather than the chronic degenerative processes that characterize human AMD. Third, it remains uncertain how endogenous pigmentation levels, which leads to more melanin and simultaneously more melanocyte- and retinal pigment epithelium (RPE)-derived endogenous L-DOPA production, influence response to L-DOPA or DRD2 agonism.

To address these unresolved questions, we performed a series of ex vivo and in vivo experiments. Using choroidal sprouting assays (CSA) and the Vldlr^−/− model of spontaneous CNV, we examined the role of melanocyte- and RPE-derived endogenous L-DOPA versus exogenous L-DOPA, the influence of tyrosinase function, and the relative contributions of DRD2 and non-DRD2 signaling. These studies allowed us to test whether L-DOPA acts directly to suppress angiogenesis, whether tyrosinase function modulates its effectiveness, and which receptor pathways mediate its protective effects in AMD-relevant contexts.

Methods

Sex as a biological variable was considered in all experiments. Each group included at least 10 animals of each sex. Data were assessed for sex-dependent differences.

Animals

Wild-type C57BL/6J (#000664), Vldlr^−/− (#002529), and C57BL/6J Tyr^c-2J/c-2J (#000058, henceforth referred to as Tyr^−/−) mice were obtained from Jackson Laboratory (Bar Harbor, ME, USA). Wild-type (WT) C57BL/6J, C57BL/6J Tyr^−/−, and Vldlr^−/− lines were subsequently bred and maintained under pathogen-free barrier conditions at the Center for Comparative Medicine, Northwestern University. Vldlr^−/−Tyr^+/+, Vldlr^−/−Tyr^+/−, and Vldlr^−/−Tyr^−/− mice were generated as described in the results. As Vldlr^−/− mice are maintained on a mixed genetic background, all experimental comparisons were made against littermate controls. Genotyping was performed by Transnetyx (Cordova, TN, USA) to verify the absence of the rd8 allele (Crb1⁻). All animal procedures conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Northwestern University Institutional Animal Care and Use Committee. All in vivo experiments were conducted on mice aged 5 to 7 weeks of age.

Choroidal Sprouting Assay

CSAs were conducted using WT and Tyr^−/− male and female mice aged 8 to 12 weeks of age. WT and Tyr^−/− eyes were enucleated and dissected in EGM2-MV medium (CC3202; Lonza, Walkersville, MD, USA) lacking hydrocortisone supplementation. Posterior eyecups containing the sclera, RPE, and choroid were isolated. The peripheral choroid was separated from the central region using a 2 mm dermal hole punch (#33-31; Integra Lifesciences, Princeton, NJ, USA), sectioned into approximately 0.5 × 0.5 mm fragments, and embedded in growth factor-reduced Matrigel (#356231; Corning, Bedford, MA, USA) in 48-well plates on ice. Matrigel was polymerized at 37°C for 10 minutes, after which EGM2-MV medium was added. Media were replaced every 1 to 2 days throughout the assay.

On day 2 post-dissection, dopamine agonists and antagonists were added daily until day 7. Exogenous L-DOPA (D1066; Spectrum Chemicals) was dissolved in PBS daily and tested at concentrations of 0, 0.015, 1.5, and 150 µM in choroidal explants derived from Tyr^−/− mice. Exogenous dopamine (62-31-7; Thermo Scientific Chemicals, USA) was dissolved in PBS daily and tested at 0, 1.5, and 150 µM in WT explants. Quinpirole hydrochloride (Q102; Sigma, USA), a DRD2 agonist, was dissolved in PBS daily and tested at 0, 10 nM, 1.0, and 100 µM in WT explants. To assess the role of DRD2 antagonism, eticlopride hydrochloride (1847; Tocris Bioscience, USA) was dissolved in PBS and administered at 10 µM in combination with 150 µM L-DOPA to determine whether DRD2 blockade prevented the inhibitory effects of L-DOPA on choroidal angiogenesis.

Explant cultures were imaged and quantified between days 5 and 7 using brightfield microscopy. Images were acquired with a Nikon Ti2 widefield microscope (Buffalo Grove, IL, USA) equipped with a 4 × objective and Nikon NIS Elements software. Image analysis was performed using the Nikon Elements General Analysis module, as previously described.¹⁷ Pre-processing included edge detection and segmentation by thresholding. For each explant, the largest angiogenic area was measured, and the central choroidal tissue area was subtracted to calculate the net sprouting area. All experiments were independently repeated at least three times.

Dopamine Agonist Administration In Vivo

L-DOPA and benserazide hydrochloride (B7283-16; Sigma, St. Louis, MO, USA) were dissolved in PBS daily. Mice received four daily intraperitoneal (IP) injections of L-DOPA (30 mg/kg) combined with benserazide hydrochloride (12 mg/kg), or PBS as a control. Benserazide hydrochloride is a decarboxylase inhibitor that decreases peripheral conversion of L-DOPA to dopamine so that more L-DOPA reaches the central nervous system. Quinpirole hydrochloride, a selective DRD2 agonist, was dissolved in PBS. Mice were treated with four daily IP injections of quinpirole (5 mg/kg) or PBS. Animals were euthanized on day 5 for choroidal wholemount immunofluorescence imaging.

Immunofluorescence Imaging of Vldlr^−/− Choroidal Flatmounts

Mice were euthanized, and the eyes were immediately enucleated and fixed in 4% paraformaldehyde (#15713-S; Electron Microscopy Sciences, Hatfield, PA, USA) for 1 hour at room temperature. Choroidal flatmounts were subsequently dissected and incubated overnight at 4°C in blocking buffer consisting of (tris-buffered saline) TBS supplemented with 5% donkey serum (#S30; Sigma-Aldrich, St. Louis, MO), 2.5% bovine serum albumin (A2153; Sigma), and 0.5% Triton X-100 (X100; Sigma). Choroidal flatmounts were incubated in primary antibodies (see the Table) overnight at 4°C, followed by 5 washes in TBS-T (TBS with 0.5% Tween-20; 00777; Amresco, Solon, OH, USA). Secondary antibody incubations (see the Table) were performed overnight at 4°C, after which flatmounts were washed 5 times in TBS-T and mounted onto HistoBond microscope slides (#16004–406; VWR, Batavia, IL, USA) using Immu-Mount (#9990402; Thermo Fisher, Carlsbad, CA, USA). Imaging was conducted on a Nikon W1 Dual CAM Spinning Disk Microscope using Nikon NIS Elements software. Image analysis was performed in a blinded manner, as previously described.¹⁸ Briefly, CNV number and area were quantified in FIJI using the ROI Manager tool. Macrophages were quantified by thresholding the IBA1 channel, excluding the optic nerve, and applying the “analyze particles” function to count cells larger than 20 µm in the area.

Table.

Immunofluorescence Antibodies

Target	Fluorophore	Manufacturer, Product Number, Dilution	Use
Goat anti-mouse-podocalyxin	—	Invitrogen, AF155, 1:250	Primary (IF)
Rabbit anti-mouse IBA1	—	Wako, 019–19,741, 1:500	Primary (IF)
Donkey anti-goat (H + L)	Alexa fluor 488	Invitrogen, A11055, 1:500	Secondary (IF)
Donkey anti-rabbit (H + L)	Alexa fluor 647	Invitrogen, A31573, 1:500	Secondary (IF)
Donkey anti-rat (H + L)	Alexa fluor 555	Invitrogen, A34055, 1:100	Secondary (IF)

Statistical Analysis

CSAs were analyzed using 2-way ANOVA with Tukey's (when comparing between all groups) or Dunnett's (when comparing only to control) multiple comparisons test. For CNV lesion number, CNV total area, and IBA1⁺ macrophage counts, data were assessed for normality using the Shapiro-Wilk test. Normally distributed data with two groups was compared using Student's unpaired t-test. Non-parametric data were analyzed using the Mann-Whitney test. For data with three groups, comparisons were made using 1-way ANOVA with Tukey's multiple comparisons test. A P value < 0.05 was considered statistically significant. All data are presented as mean ± standard error of the mean.

Results

To determine whether the absence of melanocyte- and RPE-derived endogenous L-DOPA affects choroidal angiogenic activity, we performed the CSA comparing C57BL/6J WT and Tyrosinase-mutant (C57BL/6J Tyr^−/−) mice, which are deficient in both melanocyte- and RPE-derived L-DOPA and melanin. Choroidal explants from both genotypes were dissected and cultured under identical ex vivo conditions for 7 days. The choroidal angiogenesis area was measured on days 5, 6, and 7. Representative images at day 5 and day 7 are shown in Figures 1A and 1B. Quantitative analysis of the sprouting area (Fig. 1C) demonstrated equivalent growth between WT and Tyr^−/− explants. These results suggest that the absence of Tyrosinase, and thereby the lack of endogenous L-DOPA production from melanocytes and RPE in choroidal tissue, does not affect ex vivo choroidal angiogenesis.

Figure 1.

Lack of tyrosinase does not affect choroidal angiogenic activity ex vivo. (A, B) Representative images of choroidal explants at day 5 and day 7. Boundaries for day 5 are illustrated in red and day 7 in green. (C) Quantification of the sprouting area revealed no significant difference between WT and Tyr^−/− explants across all time points (2-way ANOVA followed by Sidak's multiple comparisons test, P > 0.05, N = 37–40 explants per group).

To further investigate the influence of endogenous melanocyte- and RPE-derived L-DOPA production in vivo, we utilized the Vldlr^−/− mouse model of spontaneous CNV. The Vldlr^−/− mouse is an alternative to the laser-induced CNV model that spontaneously develops CNV in the subretinal space connecting the retinal and choroidal circulations, mimicking type 3 CNV in patients with nAMD and macular telangiectasia.¹⁹ Very-low-density lipoprotein receptor (VLDLR) deficiency leads to angiogenesis because photoreceptors are starved for energy from both loss of triglyceride uptake and paradoxical downregulation of glucose transporters, leading to stabilization of hypoxia-inducible factor 1 and increased VEGF expression.^19,20 We chose to investigate the Vldlr^−/− mouse because laser-induced CNV studies with L-DOPA have been previously performed,⁹ and the role of pigmentation in the Vldlr^−/− mouse can be investigated, unlike the laser-induced CNV model. Heterozygous Vldlr^−/−Tyr^+/− mice were crossed to generate Vldlr^−/−Tyr^+/+, Vldlr^−/−Tyr^+/−, and Vldlr^−/−Tyr^−/− mice (albino) mice (Fig. 2A). Six-week-old male and female mice from each genotype were euthanized, choroidal flatmounts were dissected, and immunofluorescence imaging was performed. Podocalyxin, a sialomucin expressed on endothelial cells, was used to delineate CNV lesions²¹ and IBA1 was utilized to identify and quantify macrophages.^18,22 Representative flatmount images from Vldlr^−/−Tyr^+/+, Vldlr^−/−Tyr^+/−, and Vldlr^−/−Tyr^−/− mice are shown in Figures 2B to 2D. Quantitative assessment of lesion number and total CNV lesion area revealed comparable outcomes across the three groups (Figs. 2E, 2F). No statistically significant differences were observed in lesion number or area between Vldlr^–/–Tyr^+/+ and Vldlr^–/–Tyr^–/– mice. No sex-specific effects were identified (Supplementary Fig. S1). These results indicate that endogenous melanocyte- and RPE-derived L-DOPA does not affect CNV in Vldlr^−/− mice.

Figure 2.

Endogenous L-DOPA does not influence CNV development in Vldlr^−/−mice. (A) Schematic for breeding strategy of Vldlr^−/−mice. (B–D) Representative choroidal flatmount images of Vldlr^−/− Tyr^+/+ (B), Vldlr^−/− Tyr^+/− (C), and Vldlr^−/− Tyr^−/− (D) stained with Podocalyxin (green, CNV) and IBA1 (magenta, macrophages). (E, F) Quantification of CNV lesion number (E) and total lesion area (F) showed no significant differences between wild-type and albino eyes. One-way ANOVA followed by Tukey's multiple comparisons test, **P < 0.01, N = 24 to 42 mice per group.

Next, we investigated whether exogenous L-DOPA could decrease choroidal angiogenesis in the CSA using both WT and Tyr^−/− mice. On day 2 after dissection, L-DOPA was added to the growth medium daily and images were analyzed for growth on days 5, 6, and 7. Representative day 7 images from control and L-DOPA treated WT and Tyr^−/− mice are shown in Figures 3A and 3B. In WT explants, L-DOPA decreased choroidal angiogenesis by 24.5% (P < 0.05) on day 5, 30.3% (P < 0.01) on day 6, and 35.1% on day 7 (P < 0.001; Fig. 3C). In Tyr^−/− explants, L-DOPA reduced angiogenic growth by 28.7% (P < 0.01) on day 5, 18.6% (P = 0.17) on day 6, and 39.6% (P < 0.0001) on day 7 (see Fig. 3C). No significant differences between WT and Tyr^−/− L-DOPA treated explants were detected. Next, we performed a dose-response curve in Tyr^−/− explants. On day 7, L-DOPA produced a dose-dependent reduction in angiogenic sprouting, suppressing vessel outgrowth by 28.8% at 1.5 µM (P < 0.01) and 66.4% at 150 µM (P < 0.0001; Fig. 3D). These data suggest that exogenous L-DOPA suppresses choroidal angiogenesis ex vivo equally in WT and Tyr^−/− explants.

Figure 3.

Exogenous L-DOPA suppresses choroidal angiogenesis ex vivo. (A, B) Representative day 7 images of control and L-DOPA treated WT (A) and Tyr^−/− (B) explants. (C) L-DOPA reduced sprouting in WT explants and Tyr^−/−explants (2-way ANOVA followed by Tukey's multiple comparisons test, N = 30–32 explants per group). (D) L-DOPA dose–response curve in Tyr^−/− (2-way ANOVA followed by Dunnett's multiple comparisons test, N = 21–39 explants per group). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

To investigate the effects of exogenous L-DOPA in vivo, we treated 5 to 6-week-old male and female Vldlr^−/− mice with daily IP injections of L-DOPA (30 mg/kg) and benserazide hydrochloride (12 mg/kg) compared to PBS control for 4 consecutive days (Fig. 4A). Animals were euthanized on day 5, and the eyes were dissected for choroidal wholemount immunofluorescence imaging. Representative choroidal wholemount from PBS and L-DOPA treated mice are shown in Figures 4B and 4C. Quantitative analysis demonstrated no significant difference in the number of CNV lesions (Fig. 4D), total lesion area (Fig. 4E), nor number of IBA1⁺ cells (Fig. 4F) between PBS and L-DOPA treatment groups. No sex-specific effects were observed (Supplementary Fig. S2). These data suggest, in contrast to the laser-induced CNV model, that exogenous L-DOPA does not influence spontaneous CNV number or area in pigmented Vldlr^−/− mice.

Figure 4.

Exogenous L-DOPA does not affect spontaneous CNV in Vldlr^−/− mice. (A) Schematic of injection strategy. (B, C) Representative images from PBS and L-DOPA–treated mice. (D, E) CNV lesion number and area were unchanged between groups (Mann–Whitney test, N = 29–30 mice per group). (F) IBA1⁺ macrophage counts were also unaffected (unpaired t-test, N = 29–30 mice per group). ns, not significant.

To further assess the role of pigmentation, the same experimental paradigm was applied to albino Vldlr^−/−Tyr^−/− mice (Fig. 5A). Representative images of CNV lesions from PBS-treated and L-DOPA treated eyes are shown in Figures 5B and 5C, respectively. Quantitative analysis demonstrated that L-DOPA treatment reduced CNV lesion number by 33.0% (P < 0.0001; Fig. 5D), decreased total CNV lesion area by 23.4% (P < 0.01; Fig. 5E), and lessened the number of IBA1⁺ cells by 21.4% (P < 0.05; Fig. 5F). L-DOPA was equally effective in male and female mice (Supplementary Fig S3). These data suggest that in vivo, L-DOPA is more effective at reducing the CNV number and area in albino eyes compared with pigmented eyes, likely due to greater levels of endogenous L-DOPA in pigmented eyes.

Figure 5.

L-DOPA reduces CNV in albino Vldlr^−/−Tyr^−/−mice. (A) Schematic of injection strategy. (B, C) Representative images from PBS and L-DOPA-treated eyes. L-DOPA reduced CNV lesion number (D, Mann-Whitney test), CNV lesion area (E, Mann–Whitney test), and IBA1⁺ cells (F, unpaired t-test). N = 26 mice per group. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

The mechanism by which L-DOPA influences CNV has been hypothesized to include direct agonism of GPR143^23,24 and/or conversion to dopamine and agonism of DRD2.⁹ To further investigate these pathways, we first added exogenous dopamine to the CSA in WT mice. Dopamine at 150 µM decreased sprouting growth by 23.5% (P < 0.05) on day 6 and by 19.9% (P < 0.0001) on day 7 (Fig. 6A). These data suggest that dopamine reduces choroidal sprouting but with a lesser effect size compared to L-DOPA (40%–60% reductions) at equal concentrations (see Fig. 3). Next, we added the DRD2 agonist quinpirole hydrochloride to the CSA. On day 7, quinpirole decreased sprouting angiogenesis by 19.9% at 10 nM (P < 0.01), 24.1% at 1 µM (P < 0.0001), and 14.8% at 100 µM (P < 0.01; Fig. 6B). These findings indicate that DRD2 agonism by quinpirole suppresses choroidal angiogenesis with similar effects sizes of approximately 20% to dopamine.

Figure 6.

DRD2 signaling contributes to the anti-angiogenic effects of L-DOPA. CSAs were performed in the presence of increased doses of dopamine (A, 2-way ANOVA with Dunnett's multiple comparisons test, N = 31–32 explants per group) and quinpirole (B, 2-way ANOVA with Dunnett's multiple comparisons test, N = 16–46 explants per group). Co-treatment with the DRD2 antagonist, Eticlopride attenuated the inhibitory effect of L-DOPA (C, 2-way ANOVA followed by Tukey's multiple comparisons test, N = 38–40 explants per group). Colored asterisks indicate significant differences between treatment (matching the color) and vehicle control. The day is indicated by the location of the asterisk either above/below the data on day 6 and/or day 7. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 versus controls, ##P < 0.01 between groups.

Based upon these findings that dopamine and DRD2 block choroidal angiogenesis with a lesser effect size than L-DOPA, we hypothesized that L-DOPA was exerting its anti-angiogenic effects through both GPR143 and DRD2. To test this hypothesis, we assessed whether DRD2 antagonism could counteract the inhibitory effects of L-DOPA on choroidal angiogenesis. Eticlopride hydrochloride (10 µM), a selective DRD2 antagonist, was co-administered with 150 µM L-DOPA in WT choroidal explants. Sprouting activity was evaluated at days 5 to 7. On day 6, L-DOPA + PBS reduced angiogenic sprouting by 33.0% (P < 0.0001) compared with controls (Fig. 6C). Explants treated with L-DOPA + eticlopride decreased sprouting by 17.9% compared with controls (P < 0.05) with no significant difference from L-DOPA + PBS (see Fig. 6C). On day 7, the inhibitory effect of L-DOPA + PBS remained robust at 21.0% compared with controls (P < 0.0001). However, L-DOPA + eticlopride explants displayed significantly greater sprouting (P < 0.01) compared to L-DOPA + PBS, and equal sprouting compared with controls (see Fig. 6C). These results indicate that DRD2 blockade partially attenuates the anti-angiogenic effect of L-DOPA, suggesting that the DRD2 and a second pathway, likely GPR143, regulate the anti-angiogenic effects of L-DOPA.

To further assess the role of DRD2 activation in vivo, pigmented Vldlr^−/− mice received daily IP injections of quinpirole hydrochloride (5 mg/kg) or PBS as a control for 4 consecutive days. Animals were euthanized on day 5, and choroidal wholemount immunofluorescence imaging was performed (Fig. 7A). Representative choroidal flatmounts from PBS- and quinpirole-treated eyes are shown in Figures 7B and 7C, respectively. Quantitative analysis demonstrated that quinpirole treatment significantly reduced the number of CNV lesions by 22.9% (P < 0.01; Fig. 7D) and decreased total CNV lesion area by 23.4% (P < 0.05; Fig. 7E). In addition, quinpirole lessened the number of IBA1⁺ macrophages by 19.1% (P < 0.01) compared with controls (Fig. 7F). A significant effect was observed in male mice with only trends toward a reduction in female mice (Supplementary Fig. S4). These findings suggest that DRD2 agonism can suppress CNV in pigmented Vldlr^−/− mice.

Figure 7.

DRD2 agonism suppresses CNV in pigmented Vldlr^−/− mice. (A) Schematic of injection strategy. (B, C) Representative whole mounts from PBS- and quinpirole-treated eyes. Quinpirole reduced CNV lesion number (D, Mann-Whitney test), CNV lesion area (E, Mann–Whitney test), and IBA1⁺ macrophage counts (F, unpaired t-test). N = 38 to 40 mice per group. *P < 0.05, **P < 0.01.

Discussion

In this study, we investigated the role of L-DOPA in choroidal angiogenesis using complementary ex vivo and in vivo approaches. We found that loss of melanocyte- and RPE-derived endogenous L-DOPA (Tyr^−/− mice) did not alter angiogenic growth in CSA or spontaneous CNV in Vldlr^−/− mice. These results indicate that baseline angiogenesis is not dependent on endogenous melanocyte- and RPE-derived L-DOPA in these models. By contrast, exogenous L-DOPA consistently suppressed angiogenesis ex vivo in a dose-dependent manner, whereas its in vivo effects were more pronounced in albino mice compared with pigmented mice. These results suggest that the L-DOPA pathway may be saturated in pigmented mice. However, DRD2 agonism was effective in vivo in pigmented mice and ex vivo but with reduced effect size. Further, DRD2 antagonism only partially reduced the effects of L-DOPA to inhibit angiogenesis ex vivo. These data suggest that L-DOPA exerts its anti-angiogenic effects by potentially activating both GPR143 and DRD2 (Fig. 8).

Figure 8.

Model of how L-DOPA influences choroidal angiogenesis. Left: In the CSA, L-DOPA inhibits angiogenesis through DRD2 and non-DRD2 pathways. Right: L-DOPA inhibits CNV in albino Vldlr^−/− mice and DRD2 agonism inhibits CNV in pigmented Vldlr^−/− mice.

Because L-DOPA is produced during melanogenesis and is a potential link between pigmentation and AMD risk, pigmentation levels are an important variable to consider in animal models. In the laser-induced CNV model, both L-DOPA and DRD2 agonism are equally effective CNV inhibitors.⁹ In contrast, L-DOPA was only capable of reducing CNV in albino Vldlr^−/− mice, whereas DRD2 agonism was effective in pigmented Vldlr^−/− mice. One possible explanation is that in pigmented eyes, endogenous L-DOPA and melanin metabolites already saturate GPR143 signaling, limiting the impact of additional exogenous supplementation. A potential mechanism for the discrepancy between laser-induced CNV and Vldlr^−/− mice is that the laser is absorbed by pigmented RPE and melanocyte cells, creating thermal injury, which could result in less melanin synthesis and decreased endogenous L-DOPA production at CNV lesions. Because the half-life of L-DOPA is on the order of hours and CNV does not develop until days 5 to 7 after laser injury,²⁵ it is possible that laser injury in pigmented mice reduces L-DOPA levels below the saturation of GPR143 receptors, thus allowing L-DOPA and GPR143 agonism to suppress laser-induced CNV. Based upon these data, it will be important to consider patient pigmentation levels as a potential confounding variable if L-DOPA is translated into a therapeutic for the prevention or treatment of nAMD.

Despite the importance of pigmentation levels on responsiveness to L-DOPA, endogenous melanocyte- and RPE-derived L-DOPA levels did not influence the ex vivo angiogenic sprouting nor CNV in Vldlr^−/− mice. Because Tyr mutations are known to increase intraocular angiogenesis, specifically iris neovascularization,²⁶ this result was surprising. Interestingly, Vldlr^−/−Tyr^+/− mice did demonstrate significantly greater lesion area (see Fig. 2F). A possible explanation for this discrepancy is that the CNV lesions exist between the retina (intraocular) and the RPE (extraocular). Considering that Tyr mutations decrease extraocular angiogenesis, specifically in the cornea,²⁶ it is possible that the Tyr heterozygous state has complex effects at the barrier between the extraocular and intraocular space. A second possible explanation is that Tyr mutations only reduce RPE- and melanocyte-derived L-DOPA without influencing neuronally derived L-DOPA. It is possible that neuronal L-DOPA is key for endogenous angiogenesis. Third, in Vldlr^−/− mice, CNV is driven by photoreceptors that highly express VEGF due to metabolic starvation from both loss of beta-oxidation and glycolysis.²⁰ Endogenously produced L-DOPA agonism of GPR143 on RPE would have minimal effect to reduce VEGF total levels because the photoreceptors are the major source of VEGF production in Vldlr^−/− mice, which is in contrast to laser-induced CNV, where RPE and macrophages are key VEGF sources.^27–30 Furthermore, in the CSA, the angiogenic front is quickly distant from the melanin-producing choroidal tissue, potentially explaining why endogenous L-DOPA is dispensable.

We hypothesized 2 potential mechanisms of action for L-DOPA: agonism of GPR143 to reduce VEGF and increase PEDF levels,¹³ and conversion to dopamine followed by agonism of DRD2 to inhibit VEGFR2 signaling.^9,14 Our ex vivo analyses revealed that dopamine and DRD2 agonism produced only modest inhibition of angiogenesis, whereas L-DOPA elicited a substantially stronger effect. Further, pharmacologic blockade of DRD2 partially reversed the anti-angiogenic activity of L-DOPA, suggesting that both DRD2 and non-DRD2, potentially GPR143, contribute to its inhibitory effects. In confirmation, DRD2 agonism suppressed CNV in pigmented Vldlr^−/− mice where L-DOPA was ineffective; as discussed above, because the photoreceptors are the key VEGF source in Vldlr^−/− mice, inhibition of VEGFR2 phosphorylation is logically more effective. Additionally, L-DOPA was capable of inhibiting CNV in albino Vldlr^−/− mice where no endogenous GPR143 signaling is occurring and conversion to dopamine can agonize DRD2. Collectively, these findings suggest that exogenous L-DOPA exerts anti-angiogenic effects through a dual receptor mechanism involving DRD2 and non-DRD2 pathways, likely GPR143.

Although our study provides novel insights into the anti-angiogenic activity of L-DOPA and its modulation by pigmentation, certain limitations should be acknowledged. First, we did not repeat our experiments in the laser-induced CNV model, which remains the most widely used preclinical model of neovascular AMD. Because studies on DRD2 agonism and L-DOPA on laser-induced CNV are already published by an experienced laboratory,⁹ we focused upon Vldlr^−/− mice and the CSA. Second, in pigmented Vldlr^−/− mice, we did not evaluate the effects of higher doses of L-DOPA. It remains possible that more robust dosing could overcome receptor saturation or pharmacokinetic constraints, thereby revealing anti-angiogenic activity even in the pigmented context. Future studies incorporating both alternative CNV models and dose-escalation strategies could help to fully define the therapeutic potential and limitations of L-DOPA in nAMD. Third, quinpirole reduced CNV in male Vldlr^−/− mice more effectively than female mice. This sex-specific effect warrants further investigation. Fourth, Tyr mutant mice only reduce RPE- and melanocyte-derived L-DOPA and do not reduce neuronally derived L-DOPA, so the influence of neuronal L-DOPA was not investigated thoroughly. Fifth, Tyr mutant mice have reduced melanin and L-DOPA, it is possible that the difference between pigmented and albino mice are related to the properties of melanin rather than L-DOPA. Finally, we suspect that the non-DRD2 pathway is through GPR143, however, GPR143 knockout mice were not investigated and warrant more detailed studies.

Taken together, our results demonstrate that L-DOPA's anti-angiogenic activity is exogenous, more effective in mice without pigmentation, and mediated through dual DRD2 and non-DRD2 pathways, potentially GPR143. The observation that pigmentation potentially saturates GPR143 signaling provides a mechanistic explanation for reduced responsiveness in pigmented eyes and underscores the need to consider pigmentation biology when developing L-DOPA-based therapies.