Effects of Topic Delivery of an Inhibitor of Serine Racemase on Laser-Induced Choroidal Vasculopathy

Abstract

Purpose: Intravitreal injection of anti-VEGF antibodies remains the primary therapy for exudative age-related macular degeneration (exAMD), although its efficacy is limited. Previous research has demonstrated that both a loss-of-function mutation of srr and the intravenous injection of a serine racemase inhibitor, L-aspartic acid β-hydroxamate (L-ABH), significantly inhibit laser-induced choroidal neovascularization (CNV) in mice. Given that L-ABH is a small molecule, this study investigated the effects of L-ABH administered via eye drops on CNV, aiming to develop a noninvasive treatment strategy for exAMD.

Methods: CNV models in mice and rhesus macaques were established through laser photocoagulation. Seven monkeys were randomly assigned to receive either saline solution or L-ABH eye drops. Intraperitoneal or intravenous injection of fluorescein characterized CNV in both mice and monkeys. Fluorescein fundus angiography was used to assess leakage, whereas optical coherence tomography measured retinal thickness in the monkeys.

Results: L-ABH eye drops significantly reduced fluorescein leakage in laser-injured mice (P < 0.001 compared to saline). In laser-injured rhesus macaques, the average percent changes in leakage areas treated with L-ABH were 2.5% ± 25.8% (P = 0.004) and 1.5% ± 75.7% (P = 0.023 compared to saline solution) on day 14 and day 28, respectively. However, L-ABH eye drops did not significantly affect the number of grade IV laser spots or retinal thickness, whereas bevacizumab did.

Conclusions: This study demonstrates the potential efficacy of an SRR inhibitor in two animal models of laser-induced CNV.

Translational Relevance: This represents the first investigation into the effects of topical delivery of an SRR inhibitor on CNV.

Introduction

Age-related macular degeneration (AMD) is the primary cause of central visual impairment and blindness in the elderly worldwide.¹ Exudative AMD (exAMD) affects approximately 15% of AMD patients, emerging suddenly and rapidly progressing to blindness if untreated. In contrast, dry-type AMD is a chronic disorder affecting about 85% to 90% of AMD patients, leading gradually to vision deterioration.² As the principal treatment for exAMD, intravitreal injection of anti-VEGF antibodies stabilizes vision and reduces the risk of vision loss in 30% to 40% of patients.^3,4 Currently, intravitreal injections are the sole treatment method for exAMD. However, the limited efficacy of anti-VEGF antibodies under certain conditions and the potential adverse effects from repeated injections warrant the exploration of alternative treatments. 

Mammalian serine racemase (SRR) is a pyridoxal-5′-phosphate–dependent enzyme that converts l-serine (Ser) to D-Ser. D-Ser is a physiological co-agonist of the N-methyl-D-aspartate receptor (NMDAR), alongside glutamate, mediating physiological and pathological responses through NMDAR activation. Depletion of D-Ser significantly reduces NMDAR-mediated neurotransmission and excitotoxicity.^5–7 Given this, modulating SRR activity presents a promising strategy for treating disorders associated with excitotoxicity, a common mechanism of neuronal damage in conditions such as Alzheimer's disease and diabetic retinopathy. In this context, we are pioneering research into the role of SRR in diabetic retinopathy. Studies, including our own, have shown that knockout of srr, the gene encoding SRR, or overexpression of a D-Ser-degrading enzyme, D-amino acid oxidase, significantly attenuates retinal neurovascular abnormalities in diabetic rodents.^8–10 

Brain cultures from SRR-deleted mice show a significant reduction in nitric oxide (NO) production.⁷ NO is pro-angiogenic, influencing angiogenesis partially through the synthesis of VEGF. For instance, optimal levels of NO enhance VEGF production in both tumor cells and vascular smooth muscle cells.^11,12 All three isoforms of nitric oxide synthase (NOS) are implicated in ocular angiogenesis: deficiencies in inducible NOS (iNOS) and neuronal NOS (nNOS) suppress laser-induced CNV), while a lack of endothelial NOS (eNOS) inhibits retinal neovascularization in oxygen-induced ischemic retinopathy.^13,14 We have previously demonstrated that deletion of SRR in the RPE leads to reduced production of iNOS and VEGF under inflammatory conditions.¹⁵ Therefore, a loss-of-function mutation in srr likely diminishes NO production in the retina, inhibiting laser-induced CNV. Consequently, reducing SRR activity with an inhibitor may somewhat decrease NO production. 

SRR inhibitors are classified into four categories: malonates, malonate-based derivatives, dipeptides, and hydroxamic acid derivatives.¹⁶ L-Aspartic acid β-hydroxamate (L-ABH), a competitive SRR-selective inhibitor, has a Ki of 97.5 ± 23.7 µM. Given its small molecular weight of approximately 148 daltons, we explored the effectiveness of L-ABH eye drops in CNV animal models. This investigation aims to develop a noninvasive treatment strategy for exAMD. 

Material and Methods

Material

Avastin (bevacizumab) was provided by Roche Diagnostics GmbH (Basel, Switzerland). L-aspartic acid β-hydroxamate was supplied by Sigma-Aldrich Corp. (St. Louis, MO, USA). Tropicamide/phenylephrine eye drops were manufactured by Santen Pharmaceutical Co. Ltd. (Osaka, Japan). Fluorescite was obtained from Alcon Laboratories, Inc. (Geneva, Switzerland). Ketamine hydrochloride/xylazine was produced by Zhong Mu Bei Kang Pharmaceutical Co. Ltd. (Jiangsu Province, China). Erythromycin eye ointment was sourced from Chen Xin Fu Dou Pharmaceutical Co. Ltd. (Shandong Province, China). Pentobarbital sodium was provided by Shanghai Boyun Biotech Co., Ltd. (Shanghai, China). C57BL/6J mice were sourced from the Shanghai Laboratory Animal Center, Chinese Academy of Science (SLACCAS), Shanghai, China. Rhesus monkeys were housed and cared for at the Research Center of Sichuan PriMed Shines Bio-Tech Co., Ltd. (Ya'an, Sichuan, China). All experiments involving monkeys were conducted by the center's physician-scientists. 

Ethics

All experimental procedures adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. The Wenzhou Medical University Ethics Committee approved the procedures on mice (approval no. wydw2023-0014). Additionally, the procedures on rhesus macaques were approved by the Laboratory Animal Ethics Committee of the Eye Hospital at Wenzhou Medical University and the Institutional Care and Use Committee of Sichuan PriMed Shines Bio-Tech Co., Ltd. (approval no. AW2016). The physician-scientists at Sichuan PriMed Shines Bio-Tech Co., Ltd., conducted the experiments on rhesus macaques. 

Experiments With C57BL/6J Mice

Induction of CNV in C57BL/6J Mice

Both male and female C57BL/6J mice were randomly assigned to either saline solution– or L-ABH-treated groups at approximately 20 g in weight and two months of age. The CNV model was established using our previously documented protocol.^17,18 Four laser spots were created around the optic disc at the 3-, 6-, 9-, and 12-o'clock positions using a Micron IV system (Phoenix Research Laboratories, Pleasanton, CA, USA) equipped with a krypton red laser (0.05-second interval, 0.07-second duration, 240 mW). 

Inclusion and Exclusion of Laser Spots

Because of variability in laser-induced CNV, laser spots were included or excluded based on criteria established in previous studies.^15,19–21 Laser burns that were hemorrhagic and linear in shape were excluded. Burns that were either more than five times larger or one-fifth smaller than the second largest or smallest spot in a group, respectively, were considered outliers and excluded. Only laser burns that produced a bubble at the injury site were selected for leakage quantification. A single eye per mouse underwent laser photocoagulation. To generate sufficient spots for leakage quantification, 12 mice per group underwent this procedure. 

Assessment of Fluorescein Leakage in the Mice With Fundus Fluorescein Angiography (FFA)

Mice were prepared by dilating the pupils with two drops of a tropicamide/phenylephrine hydrochloride solution. They were then anesthetized with an intramuscular injection of a ketamine/xylazine mixture (0.08 mL/10g body weight, 10 mg/kg). After anesthesia, the mice received an intraperitoneal injection of 0.2 mL of 1% Fluorescite. A 2.5% methylcellulose solution was applied to the corneal surface, and the Micro IV objective was positioned against the experimental eye. Images were captured using a filter set for excitation (480 nm) and emission (525 nm) wavelengths, facilitated by imaging software provided by Phoenix Research Laboratories. The images were acquired within three minutes after anesthetization to prevent cataract formation. 

Reverse-Phase HPLC

To assess the effect of L-ABH on SRR activity in the retina and the ocular posterior segment, the retina and RPE/choroid tissues were collected to measure l-serine and D-serine levels using reverse-phase HPLC, established protocol.¹⁷ Choroid/RPE tissues were homogenized in phosphate-buffered saline solution, and 10% trichloroacetic acid was used to precipitate the homogenates, which were then spun in a centrifuge. Water-saturated ether was used to remove trichloroacetic acid. The supernatants were derivatized using a mixture of solution I (30 mg/mL t-BOC-L-cysteine and 30 mg/mL o-phthaldialdehyde in methanol) and solution II (100 mM sodium tetraborate, pH 9.4), at a ratio of 3:7. Separation was achieved using a 3.5-µm Zorbax Eclipse AAA column (150 × 4.6 mm) (Agilent Technologies, Winooski, VT, USA). The amino acids were eluted with a linear gradient starting from buffer A (0.1 M sodium acetate buffer, pH 6; 7% acetonitrile; 3% tetrahydrofuran) and gradually transitioning to 100% buffer B (0.1 M sodium acetate buffer, pH 6; 47% acetonitrile; 3% tetrahydrofuran) over 60 minutes at a flow rate of 0.8 mL/min. D-serine was detected using an ultraviolet detector with excitation at 344 nM and emission at 443 nM. 

Experiments With Rhesus Macaque

Seven male monkeys, weighing between 4.31 and 5.53 kg and aged three to four years, were used. The experimental protocol included a 28-day adaption period, 21 days for laser-induced CNV formation, and 28 days for drug treatment. Day 1 marked the start of drug administration, and laser photocoagulation was performed on day −21 of the experimental timeline. 

 

To minimize the observer bias, data collection was conducted by an examiner blinded to the treatment groups. Each monkey was fed daily with 200 g of balanced food comprising 17% protein, 5% fat, and 63% carbohydrates. They had free access to water throughout the study and were pair-housed each day. Body weight measurements were taken once before laser induction, once before drug administration, once on day 14, and once on day 28 during drug administration. 

Induction of CNV in Rhesus Monkeys

Both eyes of each monkey underwent laser photocoagulation for CNV induction. Physician-scientists performed this procedure and the subsequent drug delivery at the Primed Non-human Primate Research Center (Ya'an, Sichuan, China). Seven rhesus monkeys were used for CNV induction. After pupil dilation with tropicamide/phenylephrine eye drops to more than 6 mm, monkeys were anesthetized with an intramuscular injection of ketamine (0.1 mL/kg body weight, 50 mg/mL). The monkeys' heads were then immobilized in a scaffold for laser photocoagulation using a Vitra2 532 nm laser (Quantel Medical, Clermont-Ferrand, France), applying a 50 µm spot size at 0.1 second duration and 675 mW. Each eye received nine laser spots surrounding the fovea. Of the 14 eyes subjected to laser photocoagulation, 13 exhibited grade IV fluorescein leakage and were selected for subsequent dosing, whereas one eye showing grade I leakage was excluded from drug delivery. The monkeys were randomly assigned to placebo (two monkeys), bevacizumab (two monkeys), and L-ABH eye drops (three monkeys) groups. The methods section below provides further details on CNV assessment and fluorescein leakage grading. 

Intravitreal Injection of Bevacizumab

Monkeys were anesthetized with an intramuscular injection of 0.1 mL/kg body weight of ketamine (50 mg/mL), followed by an intravenous injection of 0.5–1.0 mL/kg propofol to maintain anesthesia. They were positioned supine on an operating table, and the skin around the eyes, eyelids, and eyelashes was sterilized with 5% povidone-iodine. A surgical drape covered the eye undergoing surgery, and a speculum was used to keep the eyelid open. Then 0.5% povidone-iodine was applied to the conjunctival sac for further sterilization. After 90 seconds, the eye was flushed with sterilized saline solution and immobilized with a sterilized cotton swab for injection. Using a 29-gauge needle, bevacizumab was injected 2 to 3 mm posterior to the limbus in the superior temporal quadrant, with the needle penetrating 4 to 6 mm into the eyeball. After injection, the site was pressed with a sterilized cotton swab for one minute, and erythromycin eye ointment was applied to the eyelid. Monkeys were then returned to their cages after regaining consciousness. 

Eye Drop Medication

The monkeys were trained for topical delivery of saline solution without premedication. Eye drop administration was conducted while the monkeys were awake and restrained in a monkey chair. The eyelid was opened using a speculum, and two drops of either placebo or L-ABH eye drops were instilled into the conjunctival sac. The nasolacrimal duct was pressed for one minute after instillation to enhance absorption, and the monkeys were subsequently returned to their cages. 

Fundus Photography (FP)

FP examination was conducted as outlined in the experimental protocols. Monkeys were anesthetized with an intramuscular injection of 0.1 mL/kg of ketamine (50 mg/mL), and two drops of tropicamide/phenylephrine eye drops (Santen Pharmaceutical Co., Ltd, Osaka, Japan) were administered to each eye for pupil dilation. Monkeys were then placed in a dark room until the pupil diameter exceeded 6 mm. FP was performed using a retinal camera (VX-20; Kowa Company, Ltd, Tokyo, Japan). 

Assessment of Retinal Thickness With OCT

The retinal thickness of laser spots was assessed using OCT (Spectralis OCT Plus; Heidelberg Engineering GmbH, Heidelberg, Germany), as described in the experimental protocols. OCT imaging was conducted after FP and FFA. The fast-macular scan protocol was used to examine the eyes, scanning a macular area of 9 × 9 mm to assess retinal thickness. The eyes were evenly illuminated to ensure high-quality images, and the camera was precisely focused on the fovea. The built-in software of the Heidelberg OCT automatically measured retinal thickness by determining the distance between the inner limiting membrane and Bruch's membrane. Occasionally, the inner limiting membrane and Bruch's membrane lines were manually located. The location exhibiting maximal thickness around the burnt spot was selected for measurement. A tracking mode was used to maintain consistency in the analysis. Alterations in retinal thickness before and after dosing were expressed as a percentage: (Retinal thickness after dosing − Retinal thickness before dosing)/(Retinal thickness before dosing − Retinal thickness before laser). 

Assessment of Fluorescein Leakage With FFA

Fluorescein angiography was performed according to the experimental schemes. A rapid series of photographs centered on the fovea was captured every three seconds during the first one minute after fluorescein injection. Additionally, two photographs of both eyes, centered on the fovea, were taken five and 10 minutes after injection. CNV was graded on a scale from I to IV: Grade I indicated no hyperfluorescence; grade II, lesions with hyperfluorescence but no leakage; grade III, lesions with hyperfluorescence and late leakage; and grade IV, lesions with bright hyperfluorescence and late leakage that extended beyond the borders of the burnt area. Lesions were systematically graded starting from the 12 o'clock position clockwise. The grade IV fluorescein leakage spots were designated as the region of interest (ROI).²² The ROI area was quantified by tracing the borders of fluorescence leakage using ImageJ software (NIH, Bethesda, MD, USA). Changes in retinal fluorescein leakage were expressed as (Leakage areas after dosing − Leakage areas before dosing)/(Leakage areas before dosing) × 100%. 

Statistics

Values were reported as mean ± standard deviation. The normality of data was assessed using the Kolmogorov-Smirnov test. For data with a normal distribution, comparisons between two groups were conducted using the unpaired Student's t-test, while differences among multiple groups were analyzed with one-way analysis of variance (ANOVA), using Dunnett's post hoc test for comparisons to a single control group, or Tukey's post hoc test for comparisons among multiple groups. For data with a non-normal distribution, the Mann-Whitney U test was employed for comparing two groups, and the Kruskal-Wallis test for differences among multiple groups (SPSS15.0.1; SPSS Inc., Chicago, IL). When analyzing the number of grade IV spots, Fisher's exact test was used because the total sample sizes were fewer than forty. A P value < 0.05 was considered statistically significant. 

Results

L-ABH Eye Drops Attenuated Fluorescein Leakage in Laser-Injured CNV Mice

Our previous studies indicated that intravenous injection of L-ABH reduces CNV in mice.¹⁷ L-ABH is a small molecule with a molecular weight of approximately 148 daltons. This study investigated whether L-ABH formulated as eye drops affects CNV. A krypton red laser was used to create four injury spots evenly distributed around the optic disc (see Supplementary Fig. S1). After laser photocoagulation, mice were randomly assigned to receive either saline solution or L-ABH treatment. The lesion sizes in mice treated with L-ABH eye drops were 6314.7 ± 406.4 µm², which were comparable to those in the saline group, 7136.4 ± 1143.57 µm² (P = 0.088) (refer to Supplementary Fig. S1). L-ABH was dissolved in saline at a concentration of 5 mg/ml and applied topically twice daily at 8 AM and 4 PM, starting on the day of laser treatment and continuing for seven consecutive days. FFA was performed on the seventh day after laser treatment to assess leakage. Under the current dosing regimen, the topical application of L-ABH significantly reduced leakage. The leakage areas for L-ABH eye drops were significantly reduced with leakage areas at 5855.9 ± 1379.74 µm², significantly less than the values, 18230.1 ± 5842.06 µm² in saline solution (Fig. 1) (P < 0.001). Because we did not successively establish methods to examine L-ABH, reverse-phase HPLC was used to examine the ratios of L/D-Ser rather than detect L-ABH in the retina and RPE/choroid tissues, thus indirectly reflecting the effect of L-ABH on SRR. In the retina, L-ABH eye drops decreased D-Ser at the levels of 0.43 ± 0.022 versus 0.55 ± 0.034 µg/mL in saline solution (P = 0.003) but did not change L-Ser contents at the levels of 2.63 ± 0.539 versus 2.22 ± 0.497 µg/mL in saline solution (P = 0.184) (Fig. 2A). Correspondingly, the ratios of L/D-Ser in L-ABH treatment were 5.39 ± 1.134, significantly higher than 4.06 ± 0.781 in saline solution treatment (P = 0.025) (Fig. 2B). In the RPE/choroid tissues, L-ABH increased L-Ser contents at the values of 1.71± 0.202 versus 0.96 ± 0.126 µg/mL in saline solution (P < 0.0001) and decreased D-Ser at 0.49 ± 0.029 versus 0.55 ± 0.030 µg/mL in saline solution treatment (P = 0.006) (Fig. 2C). Correspondingly, the ratios of L/D-Ser in L-ABH treatment were 3.46 ± 0.355, significantly higher than 1.76 ± 0.296 in saline solution (P < 0.0001) (Fig. 2D). The increased ratios of L/D-Ser in the RPE/choroid tissue mean that D-Ser production was inhibited and L-ABH exerted its effect in the RPE/choroid tissue, suggesting that L-ABH reaches the ocular posterior segment of the mice. 

Figure 1.

 

L-ABH eye drops inhibited fluorescein leakage in laser-injured mice. L-ABH was dissolved in saline solution with a final concentration of 5 mg/mL. C57BL/6J mice at the age of two months were subjected to laser photocoagulation. On the day after the laser, one drop (∼10 µL) of saline solution (Con) or l-aspartic acid β-hydroxamate (L-ABH, 5 mg/mL) was instilled via the conjunctival sac for seven consecutive days, twice daily at 8 AM and 4 PM. (A) Fundus fluorescein angiography was conducted to evaluate fluorescein leakage, and representative images were indicated. (B) The leakage areas were quantified. Student's t-test was used to compare the differences, ***P < 0.001. Twelve mice were used for each group; 21 laser spots were quantified for the group of L-ABH and 26 spots for saline solution, averaged from three experiments. Each laser lesion was considered an independent experimental observation.

Figure 2.

 

L-ABH eye drops increased the ratios of L/D-Ser either in the retina or in the RPE/choroid tissue. C57BL/6J mice at the age of two months were subjected to topic delivery of saline or L-ABH eye drops (5 mg/mL) via the conjunctival sac for seven consecutive days, twice daily at 8 AM and 4 PM, respectively. The contents of L/D-Ser in the retina (A) or in the RPE /choroid tissues (C) were analyzed with rp-HPLC. The ratios of L/D-Ser were calculated for the retina (B) and the RPE/choroid tissues (D). All the differences were analyzed with Student's t-test except that Mann-Whitney U test was used to compare the ratios in D, *P < 0.05, **P < 0.01, and ****P < 0.0001. The retinae or RPE/choroid tissues from three or four eyeballs were pooled as one sample, and seven independent analyses were conducted for saline solution or L-ABH treatment. The results were from a one-time experiment.

L-ABH Eye Drops Attenuate Fluorescein Leakage in Laser-Injured Monkeys

A total of seven rhesus macaques were used for this experiment. Two monkeys received placebo (saline solution), two were treated with bevacizumab, and three received L-ABH eye drops (Table 1). As outlined in the section on experiments with rhesus macaques, the animals were adaptively fed for 28 days, followed by laser photocoagulation. Starting from the twenty-first day after laser treatment, L-ABH eye drops were topically administered for 28 days: for the initial 20 days, 10 mg/mL eye drops were applied twice daily, and for the remaining eight days the concentration was increased to 15 mg/mL, also applied twice daily (Table 1). During the adaption period, laser treatment, CNV formation period, and on days 14 and 28 after dosing, FP, FFA, OCT, and slit lamp examinations were conducted per the experimental protocols. Additionally, body weight (Supplementary Table 1) and intraocular pressure (Supplementary Table 2) were monitored throughout the experiment. Importantly, no keratitis, eye redness, uveal swelling, or iris color changes were observed in the treated monkeys. 

Table 1.

 

Information for Drug Delivery

 

Leakage Area of Grade IV Spots

Nineteen days post-photocoagulation, fluorescein leakage indicative of CNV formation was observed in the monkeys' eyes (Fig. 3A). In the saline-treated eyes, the leakage area increased by 87.0 ± 61.9% and 219.3 ± 158.2% on Days 14 and 28, respectively (Fig. 3B). In contrast, the eyes treated with bevacizumab exhibited significantly reduced leakage areas with average percent changes of −79.6% ± 40.9% on day 14 (P = 0.001 vs. saline solution) and −82.8% ± 34.5% on day 28 (P = 0.004 vs. saline solution) (Fig. 3). In the eyes treated with L-ABH, leakage slightly increased compared to pretreatment levels, but the increases were significantly smaller than those observed in the saline solution treatment group. The average percent changes of leakage for L-ABH were 2.5% ± 25.8% on day 14 (P = 0.004 vs. saline solution) and 1.5% ± 75.7% on day 28 (P = 0.023 vs. saline solution), significantly less than those in saline solution–treated eyes (Fig. 3). Significant differences were observed between the bevacizumab and L-ABH group on Day 14 (P = 0.046), whereas no difference was observed on day 28 (P = 0.455). 

Figure 3.

 

The effect of L-ABH eye drops on fluorescein leakage in the laser-injured monkeys. (A) Laser induction was conducted at day −21 and drug delivery at day 1. Fundus fluorescein angiography was conducted two days before drug delivery (Day−2) and at 14 (Day 14) and 28 days (Day 28) after dosing. Saline solution and bevacizumab were delivered for two monkeys (four eyes), respectively; L-ABH was used for three monkeys (five eyes) because one eye indicated grade I fluorescence leakage before dosing and was excluded for drug delivery. Arrows indicate representative leakage laser spots. (B) The leakage areas from FFA measurements were quantified. Four eyes for saline solution, four for bevacizumab, and five for L-ABH. One-way ANOVA with Tukey's post hoc test was used to compare the differences on day 14 and day 28. *P < 0.05, **P < 0.01, ***P < 0.001. The results were from a one-time experiment.

Amount of Grade IV Spots

In saline solution–treated eyes, grade IV leakage spots increased from baseline (17 vs. 12 at day 14; 20 vs. 12 at day 18, respectively). In bevacizumab-treated eyes, the number of grade IV leakage spots significantly decreased on day 14 (8 vs. 16, P = 0.001) and day 28 (8 vs. 16, P < 0.0001), compared to saline solution (Table 2). In the eyes treated with L-ABH, the number of grade IV leakage spots decreased on day 28 (12 vs. 15) compared to the pretreatment values. However, this reduction did not reach statistical significance compared to the vehicle on day 28 (P = 0.07) (Table 2). No significant differences were observed between the bevacizumab and L-ABH treatments at either timepoint (P = 0.536 on day 14 and P = 0.135 on day 28) (Table 2). 

Table 2.

 

Numbers of Grade IV Leakage Spots

 

Retinal Thickness

Nineteen days after laser treatment, a highly reflective echogenic mass indicative of CNV formation was observed in the monkeys' eyes (Fig. 4A). In the saline solution–treated group, the average percent change in retinal thickness was 8.4% ± 39.8% on day 14 and 26.3% ± 65.4% on day 28 (Fig. 4B). In bevacizumab-treated eyes, the average percent change in retinal thickness was −105% ± 32.2% on day 14 and −114.9% ± 28.8% on day 28 (Fig. 4B). Bevacizumab significantly reduced retinal thickness on both day 14 (P = 0.008) and day 28 (P = 0.036), compared with saline solution (Fig. 4). L-ABH modestly reduced retinal thickness but did not achieve significance compared with saline solution. In L-ABH-treated eyes, the average percent changes in retinal thickness relative to pretreatment were −16.8% ± 48.8% on day 14 (P = 0.654 vs. saline solution) and −8.0% ± 87.9% on day 28 (P = 0.74 vs. saline solution) (Fig. 4). A significant difference was observed between the bevacizumab and L-ABH groups on day 14 (P = 0.025), but no significant difference was found on day 28 (P = 0.095). 

Figure 4.

 

The effect of L-ABH eye drops on retinal thickness in laser-injured monkeys. (A) Choroidal neovascularization was induced in the monkeys, and drugs were administered as mentioned above. OCT was conducted during the adaption period (Day−35), two days before drug delivery (Day−2), and at 14 (Day 14) and 28 days (Day 28) after dosing. Saline solution and Bevacizumab were used for two monkeys (four eyes), respectively. L-ABH was used for three monkeys (five eyes). Arrows indicated representative laser spots. (B) The built-in software of Heidelberg OCT was used to measure the retinal thickness by automatically determining the distance between the inner limiting membrane and Bruch's membrane. The location with the maximal thickness around the burnt spot was chosen for measurement. Four eyes for saline solution, four for bevacizumab, and five for L-ABH. One-way ANOVA with Tukey's post hoc test was used to compare the differences on day 14 and day 28. *P < 0.05, **P < 0.01. The results were from a one-time experiment.

Discussion

Laser-induced CNV in primates, previously studied in rhesus and cynomolgus monkeys,^23–25 has become the preferred model for preclinical evaluation of therapies for wet AMD. For instance, the efficacy of ranibizumab has been demonstrated in a cynomolgus monkey model²⁶; similarly, bevacizumab has been shown to attenuate CNV.²⁷ In contrast, neither ranibizumab nor bevacizumab improves leakage or neovascularization in rodent CNV models^28,29 because of structural differences between rodent and human VEGF. Given the similar physiology and anatomy of human and nonhuman primate eyes, we also evaluated the effect of L-ABH eye drops on laser-induced CNV in rhesus macaques. 

L-ABH eye drops significantly inhibited fluorescein leakage compared to saline solution in both mice and rhesus macaques. In mice, the increased ratios of L/D-Ser in the choroid/RPE tissue suggest that L-ABH reached the ocular posterior segment. However, a similar investigation was not possible in monkeys because they were not subject to euthanasia. We hypothesize that the amount of L-ABH reaching the posterior segment in monkeys was considerably less because of the greater distance from the ocular surface to the posterior segment than in mice. L-ABH exhibited a mild effect on grade IV spots and retinal thickness in rhesus macaques, whereas bevacizumab demonstrated significant improvements. Initially, 10 mg/mL of L-ABH eye drops were administered twice daily for 20 days, which did not achieve the desired effect on retinal thickness. Consequently, the concentration was increased to 15 mg/mL for the remaining eight days. Despite optimizing the dosing scheme towards the end of the treatment period, the impact on retinal thickness remained minimal. Potential factors contributing to this lack of efficacy include limited intraocular penetration and suboptimal dosing. No local or systemic adverse effects were observed throughout treatment in either the mice or the monkeys. 

Diseases of the anterior segment are typically treated with eye drops, whereas delivering medication to the posterior segment via topical application is challenging because of the need to overcome ocular barriers to reach the back of the eye.³⁰ Additionally, the effect of conventional eye drops is transient. Ocular hydrogels have been used to prolong drug release, but current research primarily utilizes intravitreal injections to achieve sustained drug release from the vitreous reservoir.³¹ Other methods like inserts and implants prolong drug release to the posterior segment.³² However, these delivery systems require invasive procedures performed by an ophthalmologist, and repeated injections may lead to various adverse effects. Therefore conventional eye drops remain highly desirable as they provide an easy-to-use option. However, because of the high concentration of active ingredients typically required in eye drops, substantial amounts may be absorbed unwanted, leading to unfavorable systemic bioavailability.^33,34 

If effective, combining anti-VEGF antibodies with specific eye drops could reduce the frequency of intravitreal antibody injections and minimize the systemic effects of active ingredients in eye drops. For instance, eye drops made from pazopanib—a multitarget tyrosine kinase inhibitor that blocks the VEGF receptor and the platelet-derived growth factor pathway—have been shown to prevent the progression of laser-induced CNV in animal models.^35,36 However, in human clinical trials, pazopanib eye drops do not reduce the need for ranibizumab injections, nor do they affect the number of CNVs and lesion characteristics.³⁷ 

This study highlights the potential effects of L-ABH in two animal models of laser-induced CNV, although the efficacy did not meet expectations. Future research may explore appropriate dosing schemes, administration routes, or the combination with anti-VEGF antibodies for preclinical trials in CNV. Given that subclinical CNV in AMD carries a significant risk of developing into exudative AMD—with an annual incidence of around 20%^38,39—it is worthwhile to investigate the effects of L-ABH eye drops in subclinical CNV, especially given their notable inhibitory impact on fluorescein leakage. 

The mechanisms underlying the efficacy of L-ABH eye drops are not fully understood. Our study demonstrated that deletion of srr in the RPE leads to decreased production of iNOS.¹⁵ Furthermore, blocking SRR with L-ABH in the RPE significantly reduces the production of monocyte chemoattractant protein-1 and VEGF under inflammatory stimuli.¹⁷ Studies in brain cultures from SRR-deleted mice also show markedly diminished nitric oxide formation.⁷ These findings support the hypothesis that L-ABH eye drops inhibit the production of proinflammatory factors and VEGF, thereby mitigating the angiogenic effects induced by nitric oxide. 

In the retina, SRR is expressed in neurons,⁴⁰ the glial cells,^41,42 and the RPE.¹⁵ It is challenging to determine the type of cell through which inhibition of SRR led to CNV inhibition in the mice. Thus we did not examine the L/D-Ser contents and NO production in each cell type under the influence of L-ABH eye drops. Astrocyte-mediated neurovascular coupling in the brain involves the role of D-Ser by coactivation of NMDAR and eNOS.⁴³ Thus the possibility that decreased levels of D-Ser in the RPE/choroid tissues due to L-ABH eye drops affect angiogenesis is worth investigating. 

In summary, our study demonstrates the potential effect of an SRR inhibitor delivered topically in two animal models of laser-induced CNV. However, the progression to clinical and translational medicine is still early, requiring further preclinical trials to optimize dosing and administration routes or to explore combinations with an anti-VEGF antibody in nonhuman primates. 

Acknowledgments

Supported by Wenzhou Municipal Scientific and Technology Bureau(#H20210012) and an intramural grant, Integrated Project of State Key Laboratory of School of Optometry and Ophthalmology and Eye Hospital (#J02-20190204), Wenzhou Medical University. 

Disclosure: S. Wang, None; Y. Liu, None; D. Xu, None; K. Pei, None; H. Jiang, None; L. Gong, None; W. Zeng, None; Y. Liu, None; S. Wu, None