Inhibitory effects of ursolic acid on oxygen-induced mouse retinal neovascularization via intravitreal injection

Animal models and treatment

A total of 75 healthy seven-day-old (P7) specific pathogen-free-grade C57BL/6J mice with a body mass of 4.59 ± 0.57 g were used for this research (regardless of sex). The study animals were purchased from the Laboratory Animal Centre of Shanxi Medical University, China. All animal experiments were approved by the Animal Ethics Committee of the Shanxi Aier Eye Hospital.

The specific treatment methods are follows: For OIR animal modeling, 60 healthy clean-grade C57BL/6J mice on P7 were placed into feeder oxygen tanks with female mice and connected to 100% medical moist pure oxygen with oxygen flow at approximately 1 l/min [7]. The oxygen concentration was regulated and monitored with an oxygen controller at 75% ± 3% and room temperature (23 °C ± 2 °C). Lighting was normal fluorescent light, with a day-night cycle of 12 h:12 h. The mice were observed every second day. Between observation days, the oxygen tanks were opened promptly to perform necessary maintenance operations but no medicine injections were administered to the mice during this period. Additionally, the female mice were replaced along with pad materials, feeding and water. The mice were housed under high-oxygen partial pressure for five days and removed from the oxygen tank on P12.

In the normal control group, 15 newborn mice were kept in a 23 °C ± 2 °C tank with normal air (without oxygen). The mice in all groups were kept in a tank with 23 °C ± 2 °C normal air for five days until P17.

The successfully modeled OIR mice were randomly divided into four groups: OIR group, phosphate-buffered saline (PBS) group, UA group and Lucentis group (positive control group; n=15, each). On P12–P16, 3 μl PBS buffer (PBS group), 3 μl UA (2 g/l; solid powder, purity >90%; Sigm-Aldrich, St. Louis, MO; UA group) and 3 μl Lucentis injection (Novartis, Basel, Switzerland; Lucentis group) were injected into the vitreous cavity of the mice [8,9]. The normal group and the OIR group were not treated.

Sample collection

Sample size (n=75) was determined based on our previous studies to ensure sufficient statistical power. A minimum of 15 mice per group was selected, allowing for adequate biologic replicates while accommodating potential variability in experimental outcomes. On P17 the mice were euthanized via anesthesia using 1.0% sodium pentobarbital (80 mg/kg, 100,203, Xi'an Guoan Biotechnology Co., Ltd., China). One eye was removed from each mouse for retinal flat-mount preparation and Griffonia simplicifolia type I lectin, isolectin B4 (GS-IB4) fluorescence staining. The RNV area and central retina were analyzed using Image-Pro Plus (v.5.1, Media Cybernetics, Rockville, MD) software. The contralateral eyeballs were removed, and retinal tissue was extracted under sterile conditions and immediately stored in liquid nitrogen for western blot and real-time PCR analyses to test for VEGF, MMP-2, MMP-9 and COX-2.

Culture and identification of human retinal capillary endothelial cells

For human retinal capillary endothelial cell (HRCEC) culture, one to two donor eyeballs were obtained per session in a sterile operating room following corneal transplantation. The donor eyeballs were derived from the post-surgical remains of human corneal transplants, ensuring compliance with ethical guidelines and sterility during handling. The eyeballs were carefully incised along the equator under sterile conditions. Using a dissection microscope, the vitreous vessels and membranes of the retina were carefully stripped using micro-instruments, ensuring that the integrity of the retinal cells was preserved. This procedure specifically targeted the removal of vitreous components rather than the retinal nerve fiber layer. The dissected tissue was enzymatically digested using 0.25% trypsin-EDTA at 37 °C for 15 min with intermittent agitation to dissociate the cells. After digestion, the suspension was mechanically fragmented, homogenized and filtered through a 70-μm mesh to remove tissue debris. The enzymatic reaction was terminated by adding 10% fetal bovine serum. The cell suspension was centrifuged at 300 ×g for five min to collect the cells, which were then resuspended in a complete human endothelial cell culture medium (10% fetal bovine serum, 100 µ/ml penicillin, 50 µ/ml gentamicin; #M200500, Gibco, Grand Island, NY). Cells were seeded into culture flasks and incubated in a humidified 5% CO₂ incubator at 37 °C. The fluid was changed according to the cell growth rate approximately once every one to two days. During culture, retinal vascular endothelial cells were morphologically assessed by flattening, and any cobblestone morphology was noted. To prevent interference from red blood cells in the observation and analysis of endothelial cells, as well as to avoid the effects of glial cells on the growth, proliferation and differentiation of endothelial cells, these contaminating cells were carefully removed to ensure accurate experimental results. Miscellaneous cell locations were marked microscopically, and miscellaneous cells, such as red blood cells and glial cells, were removed using cell scraping. Primary cells in the culture vial were subcultured if the adherent area was >80% [10].

For the identification of HRCECs, their growth characteristics and morphological characteristics were observed using an inverted phase-contrast microscope. Cells were assessed for cobblestone morphology, indicating endothelial cell integrity and purity. Cells in good growth condition were used to prepare a suspension at a concentration of approximately 4.5 × 10⁴ cells/ml. Sterile coverslips were placed in a 24-well plate, inoculated with 1 mL of suspension per well and cultured in an incubator. After the cells were evenly distributed on the slides, they were washed twice with PBS, followed by fixation with 4% paraformaldehyde for 30 min. They were subsequently rinsed three times with PBS. Factor VIII-related antigen-antibody was added to confirm endothelial cell identity, and the cells were incubated overnight at 4 °C. After washing, a secondary antibody conjugated with a fluorescent marker was applied, followed by additional PBS rinses. Cells were observed under a fluorescence microscope for antigen expression. Images were captured to document the characteristic features of HRCECs, validating their endothelial nature.

Measurement of retinal pathological neovascularization, nonperfusion area and morphological observation of astrocytes

Mice were taken at the corresponding time points and euthanized via anesthesia. The eyes were carefully removed, fixed in 4% paraformaldehyde (40 min or four hours on ice) and slowly washed for 20 min four times in a PBS liquid shaker (Eppendorf, Hamburg, Germany). A corneal bayonet puncture was performed in the anterior chamber and the anterior eye segment was removed by cutting the eye equator at the angular scleral rim. Then, the retina was carefully dissected and the cohesive vitreous tissue was removed as cleanly as possible. The retina was stained with Isolectin GS-IB4 (1:50; Sigma-Aldrich), glial fibrillary acidic protein (1:50; Sigma-Aldrich) and 4’,6-diamidino-2-phenylindole (1:500; Sigma-Aldrich) at 4 °C overnight, with 25 μl per eyeball. After cutting the radial four-flap of the eye cup, it was neatly tiled on a slide and sealed using an antifluorescence attenuation sealing tablet. The morphology and distribution of RNV and the ischemic nonperfusion area were observed under a fluorescence microscope (×50, 500-μm scale; Leica Microsystems, Wetzlar, Germany) and the relative area of RNV and the ischemic avascular area (proportion of total retinal area) were measured using Image-Pro Plus software (Media Cybernetics). The subtle morphology and distribution of RNV (IB4-positive) and astrocytes (glial fibrillary acidic protein-positive) were observed under a confocal microscope (×400, 100-μm scale; Zeiss, Oberkochen, Germany).

Human retinal capillary endothelial cell proliferation measured with methyl thiazolyl tetrazolium assay

Cells that were well cultured were harvested and the supernatant was removed. Subsequently, 1.0 mL high-glucose culture medium (human endothelial cell culture medium [Gibco, catalog #11965–092] supplemented with 30 mmol/l glucose) and 1.0 ml of normal glucose culture medium (human endothelial cell culture medium [Gibco, catalog #11885–084] supplemented with 5 mmol/l glucose) were added. The cells were then seeded into 96-well plates at a density of approximately 5 × 10³ cells per well, with a total volume of 200 μl per well. This experiment comprised six groups, each containing five replicate wells.

Cell experiment groups: (1) treatment normal glucose (TN) group—normal glucose (glucose concentration, 5.5 mmol/l); and (2) the treatment high-glucose (TH) group—high glucose (glucose concentration, 25 mmol/l) [11]. T1 group—high glucose + 5 μmol/l UA; T2 group—high glucose + 10 μmol/l UA; T3 group—high glucose + 20 μmol/l UA; and T4 group—high glucose + 40 μmol/l UA. The marginal wells of the 96-well plate were only filled with culture medium as a blank control for trimming and evaporation and the culture medium was discarded when cells were adherent. The UA stock solution was then prepared and diluted with a high-glucose medium to achieve the desired concentration, which was subsequently added according to the experimental grouping. In the TN group, 200 μl normal glucose culture medium was added to each well; in the TH group, 200 μl high-glucose culture medium was added to each well. For the T1–T4 groups, diluted UA solution was added to 96-well plates at 200 μl/well according to a UA concentration of 5, 10, 20 and 40 mol/l, respectively. The cells were incubated for 24, 48 and 72 h, respectively, and then 20 μL of a methyl thiazolyl tetrazolium powder solution was added to each well and cultured for 4 h. After discarding the culture medium, the cells were washed three times with PBS. Then, 20 μl of a dimethyl sulfoxide solution was added to each well and shaken for 10 min until its blue–purple crystal was dissolved. The 96-well plate was placed inside the plate reader (BioTek ELx800, BioTek Instruments, Inc., Winooski, VT) with a selected wavelength of 490 nm and the results were calculated according to the following formula: cell survival rate = average optical density (OD) of the experiment groups (T1–T4) / average OD of the control groups (TN, TH) × 100%; cell inhibition rate = (1 − OD in the experiment groups / OD in the control groups) × 100%. All of the above steps were repeated three times to calculate the final results.

Protein expression of vascular endothelial growth factor, matrix metalloproteinase-2, matrix metalloproteinase-9 and cyclo-oxygenase-2 in the retinal tissue samples detected using western blot analysis

The total protein was extracted from each group of tissue samples using a lysis buffer (Sigma-Aldrich) and the protein supernatant was quantified for protein concentration. The level of protein was quantitated using a bicinchoninic acid assay kit (Sigma-Aldrich). A 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis gel and sample protein were prepared in each well (the total protein in each well was equal based on the protein concentration). Electrophoresis was performed using an electrophoresis instrument and transmembrane transfer was performed using gel transfer apparatus. Subsequently, sodium dodecyl sulfate polyacrylamide gel electrophoresis was adopted to separate the protein lysates and moved to polyvinylidene fluoride membranes (Invitrogen, Carlsbad, CA). The protein was blocked in 5% nonfat milk powder for one to two hours and washed three to four times with PBS (Sigma-Aldrich). Samples were incubated with 1‰ VEGF (Abcam, Cambridge, MA), MMP-9 (Abcam) and COX-2 antibodies (Abcam) and monoclonal mouse anti-β-actin antibody (Cell Signaling Technology, Danvers, MA) as an internal control at 4 °C overnight. The samples were then washed three times with PBS before adding the corresponding secondary antibodies (Abcam), incubated for one hour at room temperature and washed three to four times with PBS. After membrane washing, the sample silicon film was exposed using an electrochemiluminescence kit (Sigma-Aldrich). Gray band values were analyzed using ImageJ software (National Institutes of Health, Bethesda, MD). The experiments were repeated three times.

Messenger ribonucleic acid expression of vascular endothelial growth factor, matrix metalloproteinase-2, matrix metalloproteinase-9 and cyclo-oxygenase-2 in retinal tissue determined using real-time polymerase chain reaction

Total ribonucleic acid (RNA) from each tissue was extracted using TRIzol Reagent (Takara, Tokyo, Japan) and tested for purity and concentration using an ultraviolet spectrophotometer. Reverse transcription reactions were performed using a reverse transcription kit; PCR was performed using the fluorescence quantitative PCR kit (DRR036A Takara PCR Amplification Kit); all PCRs were performed using the indicated PCR instrument. Glyceraldehyde triphosphate dehydrogenase was used as the internal reference to calculate the relative quantification values (2^-△△Ct). The specific primer sequences for VEGF, MMP-2, MMP-9, COX-2 and glyceraldehyde 3-phosphate dehydrogenase (Shanghai Sangon Biotechnology Company, Shanghai, China) are detailed in Table l.

Statistical methods

Statistical analysis was processed using SPSS (v.18.0) software (IBM, Armonk, NY) and measurement data were expressed as mean ± standard deviation. The data of each sample were normally distributed (analyzed using the Kolmogorov–Smirnov method), and homogeneity of variance testing confirmed the homogeneity of variance (using the Levene test). The relative protein expression ratio (gray band value) and mRNA expression ratio (relative quantification value) between the experiment and control groups were compared using a one-way ANOVA, and the least significant difference Student t test was used for multiple comparisons between groups; p<0.05 indicated a statistically significant difference.

Hypoxic-phase injection of ursolic acid significantly reduced retinal neovascularization cluster formation and accelerated physiologic vascular reconstruction in mice with oxygen-induced retinopathy

According to immunofluorescence staining of the retinal flat mounts of the P17 mice, when UA was injected at P12–P16 (hypoxia) in the C57BL/6J mice the retinal veins in the OIR and PBS groups were relatively beaded, with a large central avascular perfusion area and a large number of highly fluorescence-stained neovascular groups were visible on the periphery when compared with the normal group. There was no statistical difference between the OIR group and the PBS group. Compared with the OIR group, the retinal neovascular area decreased by 75.4% and 76.5% in the UA and Lucentis groups, respectively (t=8.79, 7.28; all p<0.01), whereas the nonperfusion area decreased by 73.3% and 74.1% in the UA and Lucentis groups, respectively (t=8.17, 7.44; all p<0.01). However, there was no statistical difference between the UA group and the Lucentis group (Figure 1).

Hypoxic-phase injection of ursolic acid morphologically reduced the formation of retinal neovascularization clusters in mice with oxygen-induced retinopathy and alleviated hypoxia-induced damage to retinal astrocytes

According to the confocal microscopy results of the retinal flat mounts of P17 mice in each group, the degree of neovascularization in the OIR and PBS groups was significantly increased, and the shape was polyneoplastic and massive when compared with the normal group. Neovascularization not only proliferated in the same plane but also protruded into the vitreous cavity, forming the mass accumulation of neovascularization. Astrocytes were primarily (or even entirely) heteromorphic and cell body arrangement was more disordered as the astrocytes had lost their normal stellate and dendritic shape. In the UA and Lucentis groups, the proliferation of neovascularization was not significant and lacked significant polytumors and large mass accumulations compared with the OIR group. Neovascularization was isolated or proliferated only within the same plane and rarely protruded into the vitreous cavity. The morphology and cell arrangement of astrocytes closely resembled normal stellate and dendritic morphology, with few heteromorphic parts (Figure 2).

Expression of vascular endothelial growth factor, matrix metalloproteinase-2, matrix metalloproteinase-9 and cyclo-oxygenase-2 proteins in the retinal tissue samples of each group

VEGF protein expression was significantly higher in the OIR and PBS groups than in the UA, Lucentis and normal groups, with significant statistical differences (p<0.05); however, there was no significant difference between the OIR and PBS groups (p>0.05). VEGF protein expression in the UA and Lucentis groups was slightly higher than in the normal group; the difference was statistically significant (p<0.05). However, there was no significant difference between the UA and Lucentis groups (p>0.05; Figure 3A). Protein expression of MMP-2, MMP-9 and COX-2 was significantly higher in the OIR, PBS and Lucentis groups than in the UA and normal groups; the difference was statistically significant (p<0.05). However, there was no significant difference between the OIR, PBS and Lucentis groups (p>0.05). Protein expression of MMP-2, MMP-9 and COX-2 was slightly higher in the UA group than in the normal group; the difference was statistically significant (p<0.05; Figure 3B–D).

Vascular endothelial growth factor, matrix metalloproteinase-2, matrix metalloproteinase-9 and cyclo-oxygenase-2 messenger ribonucleic acid levels in the retinal tissue samples of each group

VEGF mRNA levels were significantly higher in the OIR and PBS groups than in the UA, Lucentis and normal groups (p<0.05); however, there was no significant difference between the OIR group and the PBS group (p>0.05). VEGF mRNA levels were slightly higher in the UA and Lucentis groups than in the normal group; the difference was statistically significant (p<0.05). However, there was no significant difference between the UA and Lucentis groups (p>0.05; Figure 4A).

MMP-2, MMP-9 and COX-2 mRNA levels were significantly higher in the OIR, PBS and Lucentis groups than in the UA group and the normal group; the difference was statistically significant (p<0.05). However, there was no significant difference among the OIR, PBS and Lucentis groups (p>0.05). MMP-2, MMP-9 and COX-2 mRNA levels were slightly higher in the UA group than in the normal group; the difference was statistically significant (p<0.05; Figure 4B–D).

Effect of different concentrations of ursolic acid on human retinal capillary endothelial cell proliferation in a high-glucose environment

The HRCECs were seeded in 96-well plates. During the same culture period, the number of HRCECs in the normal glucose group was small compared with the high-glucose group (p<0.05). After 24, 48 and 72 h of HRCEC culture, pairwise comparisons between groups with high-glucose + different concentrations of UA groups (5, 10, 20, 40 μmol/l) showed that the inhibition rate of the HRCECs increased with an increased concentration and reflected a statistically significant difference (p<0.05). After intervention of the same concentration of UA, the inhibition rate of the HRCECs was increased along with the drug intervention time; the difference was statistically significant (p<0. 05; Table 2 and Table 3).

Effect of different concentrations of ursolic acid on the protein expression of vascular endothelial growth factor, matrix metalloproteinase-2, matrix metalloproteinase-9 and cyclo-oxygenase-2 in human retinal capillary endothelial cells in a high-glucose environment

The protein expression of VEGF, MMP-2, MMP-9 and COX-2 in different HRCEC groups was determined by protein immunoblot (western blot) as follows:

(1) After HRCEC cultures were performed simultaneously, the protein expression of VEGF, MMP-2, MMP-9 and COX-2 in HRCECs was higher in the high-glucose group than in the normal glucose group; analysis results of the gray band protein value were significantly different (p<0.05; Figure 5A).

(2) After the same culture time, pairwise comparisons between the high-glucose and different-concentration (5, 10, 20, 40 μmol/l) UA groups showed that the expression of VEGF, MMP-2, MMP-9 and COX-2 in HRCECs decreased with increased concentration of UA; the differences were statistically significant (p<0.05). With the extension of UA intervention time, there was a gradual decrease in the expression levels of VEGF, MMP-2, MMP-9 and COX-2. Specifically, the levels observed were lower in the 72-h group than in the 48-h group, and the 48-h group exhibited lower levels compared with the 24-h group; the differences were statistically significant (p<0.05; Figure 5B–D).

Effect of different concentrations of ursolic acid on the messenger ribonucleic acid levels of vascular endothelial growth factor, matrix metalloproteinase-2, matrix metalloproteinase-9 and cyclo-oxygenase-2 in human retinal capillary endothelial cells in a high-glucose environment

Real-time PCR was conducted to detect the mRNA levels of VEGF, MMP-2, MMP-9 and COX-2 in the HRCECs of different groups. Following simultaneous HRCEC culture, the mRNA levels of VEGF, MMP-2, MMP-9 and COX-2 were higher in the high-glucose group than in the normal glucose group; the differences were statistically significant (p<0.05). With the same culture time, pairwise comparisons between groups with high-glucose + different concentrations (5, 10, 20, 40 μmol/L) of UA showed that the mRNA levels of VEGF, MMP-2, MMP-9 and COX-2 in HRCECs decreased with an increased concentration of UA; the differences were statistically significant (p<0.05). mRNA levels of HRCECs, MMP-2, MMP-9 and COX-2 decreased with prolonged drug intervention; the differences were statistically significant (p<0.05; Figure 6).