We used adult 6–8 week-old male C57BL/6J mice (purchased from the Shanghai Laboratory Animal Center of the Chinese Academy of Sciences), weighing 20–30 g for laser-induced CNV studies. Animals were housed under controlled lighting conditions (12/12 h light/dark cycle). All experiments were performed in accordance with the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research. All experiments were approved and monitored by the Institutional Animal Care and Use Committee (IACUC) at the medical academy of Shanghai Jiao Tong University.

Seven days after laser treatments, fluorescein angiography was performed following a previously described protocol (Elizabeth Rakoczy et al., 2006; Shah et al., 2015). In brief, mice were fully anesthetized, and the eyes were dilated, and a drop of lubricant gel was placed on each eye. Intraperitoneal injection of 1 ml of a 2% fluorescein sodium solution (Alcon, Fort Worth, TX, United States) was used to get a clear bright-field image. Fundus images were captured using the platform of the Micron IV fundoscopy system (Phoenix Research Laboratories, Pleasanton, CA, United States). The images were processed using Image J software and quantitatively evaluated by two blind groups of observers following four categories as described previously.

Laser-induced CNV lesions were induced following a standard protocol (Xin et al., 2020). In brief, animals were first anesthetized by intraperitoneal injection of avertin (2,2,2-tribromoethanol) (0.2 ml/10 g) (Nan Jing AIBI Bio-Technology Co., Ltd, Nanjing, China). Their pupils were dilated with 0.5% tropicamide (Santen Pharmaceutical, Osaka, Japan). A drop of levofloxacin gel (Alcon) was placed on the eye to avoid any eye dehydration and infection. Four laser photocoagulations (532 nm, 350 mW, 100 ms) were implemented in the fundus of each eye surrounding the optic nerve and focused on the Bruch’s membrane using a Micron IV retinal image-guided laser system (Phoenix Research Laboratories). The morphologic endpoint of the laser injury was the appearance of an air bubble, which is a sign of BM disruption. Intravitreal injections were performed following earlier protocols (Mitra et al., 2017). In brief, immediately after laser injury, a 30-gauge needle was used to make a precise hole on the sclera of each eye. A 35-gauge needle attached to a 10 μl nanofilm syringe (World Precision Instruments, Sarasota, FL, United States) was then gently inserted through the puncture hole into the vitreous cavity delivering 2 μl of SC (Selleckchem, Cat. No. S5452) at the concentration of 2.5 μM or PBS solution. The injection process was visualized using an operating microscope (Carl Zeiss Surgical, Incorporated, Thornwood, NY, United States).

We measured CNV volume as previously described (Mao et al., 2017; Wei et al., 2018; LeBlanc et al., 2019). In brief, 7 days after laser photocoagulation, we sacrificed the mice and enucleated the eyeballs. After fixing the eyeballs in 4% paraformaldehyde (Electron Microscopy Sciences [EMS], Hatfield, PA) for 1 h, we removed the corneas, the lenses, the vitreous, and the neurosensory retinas. The choroid/RPE/scleral tissue was isolated with four radial incisions and washed three times in PBS. Then we incubated the processed choroid/RPE/scleral complexes with blocking buffer (PBS containing 1% BSA and 0.5% Triton X-100) for 4 h at room temperature. After washing three times in PBS, the RPE complexes (RPE/choroid/sclera) were stained with 10 μg/ml Alexa Fluor^®488 conjugated isolectin GS-IB4 (Invitrogen, Cat. No. 121411) in a volume of 0.5 ml/retina overnight at 4°C and flat-mounted with antifade mounting medium (VECTASHIELD; Vector Laboratories, CA) to preserve the fluorescence. We observed CNV with a Zeiss LSM 710 confocal laser scanning microscope (Carl Zeiss) at a magnification of 100×. An average of 3–4 CNV areas per eye was analyzed. Each CNV spot was considered an independent data point. For each group, at least 25 spots were analyzed from eight mice. All quantifications were conducted under blinded conditions.

Hematoxylin and eosin (H&E) staining was performed according to a previously described procedure (Wei et al., 2018; Xin et al., 2020). Mice were killed, and the eyes were enucleated 3°days (for safety test) or 7°days (for CNV volume analysis) after laser injury and fixed in 4% paraformaldehyde at 4^°C for 24°h, conventionally dehydrated, and embedded in paraffin wax. The fixed tissues were serially sectioned at 3 μm, stained with HE. The tissues were observed and photographed under a light microscope. The TUNEL assay (In situ cell death detection kit-POD; Roche Diagnostics GmbH, Mannheim, Germany) was performed according to the manufacturer’s instructions 3 days after laser injury and was observed under a fluorescence microscope at 200× magnification.

The HRMECs were purchased from the Cell Bank of the Chinese Academy of Sciences and cultured in endothelial cell medium (ECM) containing 5% fetal bovine serum (FBS, Sigma, and Cat. No. F2442), 1%penicillin/streptomycin and 2% endothelial cell growth supplement (ECGS) at 37°C in a humidified incubator with 5% CO₂. HRMECs cells were cultured in serum-free medium (SFM) for 24 h before treatments.

Cell viability was analyzed using a Cell Counting Kit (CCK)-8 (Dojindo Molecular Technologies). HRMECs were seeded in 96-well plates at a density of 10,000 cells/well and incubated at 37°C in a 5% CO₂ environment for 24 h. The culture media from each well was then removed and replaced with 100 μl of a serum-free medium for another 24 h to starve the cells. Then, serum-free medium containing different concentrations (1, 2.5, 7.5, 10, and 20 μM) of SC was added into each well, and the plates were continuously incubated for different times (24 , 48 h). A 10 μl volume of CCK8 solution was gently added per well and incubated at 37°C for 4 h. The absorbance was measured at 492 nm using a microplate reader. Untreated cells served as negative controls, and only media without cells served as background measurements. The cell viability was calculated following the manufacturer’s protocol. All experiments were performed in triplicate.

HRMECs were used for the tube formation assay, as described previously (Mao et al., 2017; Mitra et al., 2017). In brief, Corning Matrigel matrix (Matrigel, Becton Dickinson, NJ, United States) was placed in a 96-well plate at 37°C for 45 min to allow the basement membrane to form a gel. Then 5 × 10⁴ HRMECs were transferred to each well with or without treatment with varying concentrations of SC and VEGF (10 ng/ml). 2-Methoxyestradiol (2-Me, a metabolite of estradiol-17beta) (Pal et al., 2019) at the concentration of 10 nM was used as a positive control. The plate was incubated for 6 h at 37°C with 5% CO₂. Capillary-like tube structures formed by HRMECs on the Matrigel were photographed with a Leica CTR6000 fluorescence microscope (Wetzlar, Germany). Tube formation was quantified by counting the average number of branches per field. The experiments were repeated at least three times.

HRMECs were seeded into 24-well tissue culture plates and cultured for 24 h until approximately 70–80% confluence. After replacing the medium with serum-free ECM for 24 h to starve the cells, we used a sterile 1,000 μl pipette tip to make a straight line on the monolayer across the center of the well in a single direction. Then Similarly, a second straight line was scratched perpendicular to the first line to create a cross-shaped cellular gap in each well. Each well was washed twice with 1 ml of cold PBS to remove the debris and any detached cells. before changed the medium with serum-free medium (control group), 2.5 μM SC or 10 nM 2-Me (positive control) for another 24 h. All cells were incubated at 37°C in a humidified atmosphere containing 5% CO₂ for 24 h before capturing digital images of the cell gaps using a Leica CTR6000 fluorescence microscope (Wetzlar, Germany).

The concentration of VEGF in RPE-choroid complexes was performed 3°days after laser burn with a mouse ELISA kit (R&D Systems, United States) according to the manufacturer’s instruction.

Western blotting was performed following the previous protocol (Mao et al., 2017). HRMECs or isolated retina/choroid complexes were lysed in modified RIPA buffer (Thermo, Rockford, IL, United States) supplemented with protease inhibitor cocktail (Thermo, Rockford, IL, United States) on ice. Proteins obtained from the cell lysates were separated on 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to nitrocellulose (NC) membranes (Millipore, Billerica, MA). Membranes were blocked in Tris-buffer saline Tween 20 buffer containing 5% nonfat milk for 1 h at 37°C, and immunoblotted with the following primary antibodies: rabbit anti-VEGF-A (ab46154, Abcam; 1:1,000), anti-phospho-extracellular signal-regulated kinases (ERK1/2) (Cell Signaling, Danvers, MA, United States), anti-ERK1/2 (Cell Signaling), anti–phospho-protein kinase B (AKT) (Cell Signaling), anti-AKT (Cell Signaling), anti–Phospho-p38 MAPK (Cell Signaling), anti-p38 MAPK (Cell Signaling), and β-actin(Cell Signaling) at 4°C overnight. Membranes were then washed with TBST and incubated with the corresponding secondary antibodies at 37°C for 2 h. The membranes were washed with TBST three times to remove the unlinked secondary antibodies and the chemiluminescence visualized following the manufacturer’s instructions using an enhanced chemiluminescence (ECL) reagent (Pierce, Rockford, IL, United States) with a ChemiDoc MP Imager (BIO-RAD). Each indicated band was quantified and normalized to the corresponding loading control using Image J software.

The experimental data were presented as the mean ± SEM in figures. GraphPad Prism 8.0 software (La Jolla, CA, United States) was used for the figures and data analysis and Photoshop CS5 software for figures organization. T-tests were used for the comparisons between two groups, while multiple group comparisons were analyzed using one-way and two-way ANOVA as appropriate. A p value of less than 0.05 was considered to be significant.

CCK8 assay was used to evaluate the cytotoxicity of SC toward HRMECs. As shown in Figure 1, low concentrations of SC (≤2.5 μM) produced no inhibition on the growth of HRMECs from day 1 to day 2. However, SC (>2.5 μM) began to affect the survival of HRMECs cells in a concentration- and time-dependent manner. Therefore, we chose lower concentrations of SC (≤2.5 μM) for further research.

To investigate the effect of SC on the process of angiogenesis, we performed a VEGF-induced capillary tube formation assay using HRMECs. As can be seen in Figure 2, VEGF treatment led to robust tube formation. 2-Me, an angiogenic inhibitor of HRMECs proliferation and angiogenesis, was used as a positive control in this test and showed a strong inhibitory effect on tube formation. However, SC treatment significantly decreased the number of tubular-like structures in a concentration-dependent manner. Compared to the VEGF-treatment group, the number of capillary-like structures was reduced by 9.76%, 32.02% and 64.45% on the treatment on 0.5, 1, and 2.5 μM SC group, respectively.

As the migration of abnormal endothelial cells plays a crucial role in angiogenic process, we further elevated the effect of SC on HRMECs migration using the scratch-wound assay. As shown in Figure 3, SC treatment reduced 49.75% of wound closure rate compared to the untreated group, which is consistent with the tube formation assay. Our results are in line with the earlier study where SC exhibited a significantly anti-angiogenic effect in various kinds of tumors (Gaziano et al., 2016; Galadari et al., 2017).

To evaluate the inhibitory effect of SC on CNV formation in mice, we first examined the fundus fluorescein angiography (FFA) immediately after laser surgery to confirm the success of the laser burns. Injuries accompanied by choroidal hemorrhage during photocoagulation were excluded. Only the successful laser injuries were included. Then we performed a single, intravitreal injection of SC (2.5 μM, 2 μl/eye) or the equal amount of PBS immediately after photocoagulation in mice. The proliferation of new blood vessels from laser-induced injuries was considered to reach maximum damage at 7 days after laser photocoagulation (Mitra et al., 2017), so we captured the FFA images (Figure 4A) and quantitatively compared the FFA grade score of CNV lesions in each group (Figure 4B). The average FFA grade score of SC treatment group was significantly smaller than that of PBS-treated mice on Day 7 post laser injury. These data demonstrated that intravitreal injection of SC inhibited laser induced CNV formation in mice.

To reconfirm the CNV volume of different groups, we performed RPE/choroidal flat mounts with Alexa Fluor-488-conjugated Griffonia simplicifolia isolectin-IB4 (GS-IB4, Thermo Fisher Scientific, Cat. No. 121411) staining to identify neovascularization and measure CNV area after laser injury (Figure 4C). Twelve (PBS-treated mice) and 22 (SC-treated mice) laser spots out of the respective 50 laser spots were evaluated for endothelial cell marker staining (Figure 4E). The images of each antibody-stained CNV lesion were quantitatively analyzed using Zeiss AxioVision software, which outlined the fluorescent and antibody-stained blood vessels as determined by earlier literature (Mitra et al., 2017). The extent of the isolectin B4-labeled area was much smaller in the SC-treated group than in the PBS-treated group (p < 0.05).

In the HE stainig assay of morphologic cross sections (Figure 4D), CNV in the PBS group was significantly longer and higher than those in the SC group height: PBS: 142.69 ± 15.32 (μm) vs. SC: 109.12 ± 15.48 (μm), p < 0.05 (Figure 4F); (length: PBS: 703 ± 43.1 (μm) vs. SC: 589.59 ± 34.8 (μm), p < 0.05 (Figure 4G), n = 20 laser spots/group.

Seven days after intravitreal injection of SC (2.5 μM, 2 μl/eye), the amount of VEGF protein in the retina/choroid complex was evaluated using Western Blotting (Figure 5A). It was revealed that laser injury increased VEGF protein expression in the retina/choroid complex. However, pretreatment of SC could significantly reduce the expression of VEGF in the retina/choroid complex of laser induced CNV mice (p < 0.05) (Figure 5B).

We next evaluated the VEGF protein in the retina/choroid complex obtained from the eye up using a mouse ELISA kit. As shown in Figure 5C, the amount of VEGF in SC-treated group was significantly lower than that in PBS group (p < 0.05).

Before performing in vivo experiment, we first examined the safety of intravitreal injection of SC (2.5 μM, 2 μl/eye) in CNV eyes at 3 days after photocoagulation using HE staining and TUNEL assay. As shown in Figures 6A,B, no retina abnormalities and inflammatory cells were observed in SC-treated eye and PBS-treated eyes. The TUNEL assay (Figures 6C,D) revealed that there was no difference in TUNEL positive cells between SC group and PBS group.

AKT, p38-MAPK, and ERK1/2 are crucial factors triggered by VEGF to induce the angiogenesis process, including survival, migration and proliferation of endothelial cells (Aiello et al., 1995; Roh et al., 2016; Cheng et al., 2017; Zhang and Huang, 2019). Thus we investigated the effect of SC on VEGF-induced phosphorylation of AKT (p-AKT), p38-MAPK (p-p38-MAPK), and ERK1/2 (p-ERK1/2) using HRMECs (Figure 7A). As shown in Figures 7B–D, VEGF (10 ng/ml, 24, and 48 h exposure) significantly increase the phosphorylation of AKT, p38-MAPK, and ERK1/2. However, the pretreatment of SC (2.5 μM) significantly reduced the VEGF-mediated phosphorylation of AKT, ERK1/2 (p-ERK1/2) and p38-MAPK at all time points.