Male C57BL/6J mice (6–8 weeks old) were obtained from Beijing Vital River Laboratory Animal Technology (Beijing, China). Vldlr knockout (Vldlr⁻/⁻) mice were generated via CRISPR/Cas9-mediated deletion of exons 2–11 of the Vldlr 207 transcript. The mice were maintained under a 12-h light/dark cycle in a specific-pathogen-free (SPF) facility at Tianjin Medical University. All procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of Tianjin Medical University (No. TMUaMEC 2024030) and conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. For euthanasia, mice were deeply anesthetized with 1.25% tribromoethanol (Avertin; Nanjing Aibei Biotechnology, Nanjing, China) administered intraperitoneally at a dose of 0.2 mL/10 g body weight, followed by cervical dislocation in accordance with the approved ethical guidelines.

The laser-induced choroidal neovascularization (CNV) model was established in C57BL/6J mice (6–8 weeks old) as previously described [21]. Briefly, mice were anesthetized, pupils were dilated with 1% tropicamide, and four laser spots (532 nm, 100 mW, 0.15 s, 100 µm spot size) were applied around the optic disc using a 532-nm photocoagulation system (OcuLight GL, Iridex, Mountain View, CA, USA). Formation of a cavitation bubble was taken as confirmation of Bruch’s membrane rupture. Eyes with retinal hemorrhage were excluded.

For intravitreal injections, the mice were anesthetized with 2%–3% isoflurane in oxygen at a flow rate of 0.5 L/min, and maintained under anesthesia for the duration of the injection procedure (approximately 5-10 min), and pupils were dilated with 1% tropicamide (Sinqi Pharmaceutical, Shenyang, Liaoning, China). A scleral entry site was created with a 30-gauge needle, and a 34-gauge Hamilton syringe was introduced into the vitreous cavity behind the limbus. Mice were randomly assigned to receive 1 µL recombinant human PF4 (0.3 µg/µL in 2% BSA/PBS; R&D Systems, Minneapolis, MN, USA) [22] or vehicle. Topical 0.5% levofloxacin (Sinqi Pharmaceutical, Shenyang, Liaoning, China) was applied before and after injection. Eyes with injection-related hemorrhage or lens injury were excluded from analysis.

FFA was performed at designated time points after laser using a Micron IV retinal imaging system (Phoenix Research Labs, Pleasanton, CA, USA). Mice were anesthetized and pupils dilated with 1% tropicamide. Fluorescein sodium (10%, 100 µL) was injected intraperitoneally, and fundus images were acquired later. Leakage area for each CNV lesion was measured in ImageJ (version 1.53t, National Institutes of Health, Bethesda, MD, USA) by outlining the hyperfluorescent region surrounding the laser spot as described [21]. The mean leakage area per mouse was calculated for statistical analysis.

For RPE–choroid flat mounts, mice were euthanized and eyes were enucleated and fixed in 4% paraformaldehyde for 1 h. Posterior eyecups were dissected, permeabilized in 1% Triton X‑100, and blocked in 3% BSA/0.3% Triton X‑100 in PBS. Tissues were incubated with anti-CD31 and anti-Iba1 primary antibodies, followed by fluorophore-conjugated secondary antibodies, and flat-mounted. Z-stack images of each laser lesion were acquired using a confocal microscope (LSM 800; Carl Zeiss, Oberkochen, Germany). CD31-positive neovascular volume and Iba1-positive microglia/macrophages volume were quantified using Imaris software (version 9.0.1, Bitplane AG, Zurich, Switzerland), and the mean value per mouse was used for analysis. The entire procedure was performed essentially as previously described [23]. Sources of antibodies are listed in Table 1.

For retinal cryosections, eyes were fixed in 4% paraformaldehyde, cryoprotected in 30% sucrose, embedded in Optical Coherence Tomography (OCT) (Sakura Finetek, Torrance, CA, USA), and sectioned at 20 µm. Sections containing laser lesions were permeabilized, blocked in 3% BSA containing 0.3% Triton X-100, and incubated overnight at 4 °C with primary antibodies against CD31, Iba1, vascular endothelial growth factor A (VEGFA), and PF4, followed by the application of fluorophore-conjugated secondary antibodies and DAPI. Sections were imaged by confocal microscopy. Sources of antibodies are listed in Table 1.

Apoptosis was assessed in retinal sections using a One-step TUNEL (terminal deoxynucleotidyl transferase dUTP nick end labeling) Apoptosis Assay Kit (Cat# C1086, Beyotime Biotechnology, Shanghai, China) according to the manufacturer’s instructions. Briefly, cryosections were equilibrated, permeabilized, and incubated with TUNEL reaction mixture, followed by DAPI counterstaining. Images were acquired using a confocal microscope (LSM 800; Carl Zeiss, Oberkochen, Germany).

Human retinal microvascular endothelial cells (HRMECs; Procell Life Science & Technology, Wuhan, China) were cultured in endothelial cell medium (ECM; ScienCell Research Laboratories, Carlsbad, CA, USA) supplemented with 5% fetal bovine serum (FBS), 1% endothelial cell growth supplement (ECGS), and 1% penicillin/streptomycin. Cells were maintained at 37 °C in a humidified 5% CO₂ incubator, and passages 3–6 were used for the experiments. The HRMECs were validated by short tandem repeat (STR) profiling and tested negative for mycoplasma contamination by the supplier.

HRMEC viability in response to PF4 and VEGF was assessed using a CCK‑8 assay (NCM Biotech, Suzhou, China). Cells were treated with PF4 (0.25, 1, or 4 µg/mL) for 24 h in the presence or absence of VEGF (50 ng/mL). CCK‑8 reagent was added for 60 min at 37 °C, and absorbance at 450 nm was measured. Data were normalized to vehicle-treated or VEGF (Sino Biological, Beijing, China) + vehicle controls, as indicated, to identify a non‑toxic, effective PF4 concentration for subsequent functional assays.

HRMEC proliferation was evaluated using a 5-ethynyl-2^′-deoxyuridine (EdU) incorporation assay (BeyoClick™ EdU Cell Proliferation Kit; Cat# C0071S, Beyotime Biotechnology, Shanghai, China). Based on CCK‑8 assays showing that 1 µg/mL PF4 was the minimum concentration that significantly inhibited VEGF-induced HRMEC proliferation without reducing basal viability, this dose was used in subsequent in vitro assays. HRMECs were pretreated with PF4 (1 µg/mL) for 1 h and then stimulated with VEGF (50 ng/mL) for 24 h before EdU labeling and detection.

Cell migration was assessed using a scratch wound assay. HRMEC monolayers in 12-well plates were scratched with a 200-µL pipette tip and washed to remove debris. Cells were pretreated with PF4 (1 µg/mL) for 1 h and then stimulated with VEGF (50 ng/mL). Images were acquired at 0 and 12 h, and wound closure was quantified in ImageJ.

For tube formation, HRMECs were seeded onto growth factor-reduced Matrigel in 96-well plates. Cells were pretreated with PF4 (1 µg/mL) for 1 h and then exposed to VEGF (50 ng/mL). After 6 h, capillary-like networks were imaged, and total tube length and branch points were quantified using the Angiogenesis Analyzer plugin in ImageJ.

PF4 levels in retina–choroid complexes were determined using a commercial ELISA kit (Cat# ML037257, Shanghai Enzyme-linked Biotechnology, Shanghai, China). Tissues were homogenized in PBS and centrifuged, and supernatants were processed according to the manufacturer’s protocol. Absorbance at 450 nm was recorded, and PF4 concentrations were calculated from a standard curve and normalized to total protein.

Protein was extracted from HRMECs or retina–choroid complexes using RIPA buffer with protease and phosphatase inhibitors. Protein concentrations were determined by BCA assay. Equal amounts of protein were separated by SDS-PAGE and transferred to PVDF membranes, blocked with 5% non-fat milk, and incubated overnight at 4 °C with primary antibodies against phospho-ERK (Cat# 9101S), total ERK (Cat# 9102S), phospho-AKT (Cat# 2965T), total AKT (Cat# 4691T), phospho-STAT3 (Cat# 4113T), total STAT3 (Cat# 4904T), VEGFR2 (Cat# 9698), (all 1:1000; Cell Signaling Technology, Danvers, MA, USA), $\beta$-actin (Cat# 4967L, 1:5000; Cell Signaling Technology, Danvers, MA, USA), and VEGFA (1:1000; Cat# 66828-1-Ig, Proteintech, Wuhan, China). After incubation with HRP-conjugated secondary antibodies, signals were visualized by enhanced chemiluminescence and quantified in ImageJ. All procedures were performed essentially as previously reported [24].

Total RNA from retina–choroid complexes was isolated using a commercial RNA extraction kit (Cat# AC0101, Sparkjade, Jinan, Shandong, China) and reverse-transcribed using Hifair III 1st Strand cDNA Synthesis SuperMix (Cat# 11141ES10, Yeasen, Shanghai, China). qPCR was performed with Hieff® qPCR SYBR Green Master Mix (Low Rox Plus) (Cat# 11202ES60, Yeasen, Shanghai, China) on a LightCycler 480 II system (Roche Diagnostics, Basel, Switzerland). The relative mRNA expression levels of target genes were normalized to the internal control Gapdh and quantified using the 2^{-Δ⁢Δ⁢Ct} method. The primer sequences are listed in Table 2.

For transcriptomic analysis, retina–choroid complexes were collected from uninjured controls, laser-injured mice receiving vehicle, and laser-injured mice receiving PF4 (n = 3 per group) at day 7 after laser. Tissues were snap-frozen in liquid nitrogen and stored at –80 °C. RNA extraction, library preparation, sequencing, and bioinformatic analysis were performed by Majorbio (Shanghai, China).

Spectral‑domain OCT (Micron IV, Phoenix Research Labs) was used to assess retinal structure at designated time points after intravitreal injection. Mice were anesthetized and pupils dilated. Circular volume scans centered on the optic disc were obtained, and retinal thickness was measured on ImageJ by masked graders.

Full-field ERGs were recorded using a Celeris D430 system (Diagnosys, Lowell, MA, USA) as described [25]. Mice were dark-adapted overnight, anesthetized under dim red light, and pupils dilated with tropicamide. Corneal electrodes with integrated stimulators were placed on lubricated corneas, and scotopic responses were elicited at 0.01, 0.1, and 1 cd$\cdot$s/m². a- and b-wave amplitudes were analyzed.

To evaluate the potential retinal toxicity of PF4, C57BL/6J mice received a single intravitreal injection of PF4 (0.3 µg in 1 µL) or vehicle (PBS) on day 0. At day 7 post-injection, the mice were subjected to a series of safety assessments. First, fundus imaging and OCT were performed to examine the gross morphology and retinal thickness. Subsequently, retinal function was assessed using ERG as described in the previous section. Following functional testing, the mice were euthanized, and the eyes were harvested for histological analysis, including hematoxylin and eosin (H&E) staining to evaluate structural integrity and a TUNEL assay to detect cell apoptosis.

Data were analyzed using GraphPad Prism version 10 (GraphPad Software, San Diego, CA, USA). All results are expressed as mean $\pm$ standard error of the mean (SEM). For comparisons between two groups, an unpaired two-tailed Student’s t-test was performed. For comparisons among multiple groups, one‑way analysis of variance (ANOVA) followed by Tukey’s post hoc test was used. A p value of less than 0.05 was considered statistically significant.

To define PF4 expression in mouse CNV, we measured PF4 expression in a laser-induced CNV mouse model (Fig. 1A–C). Quantitative real-time PCR (qRT-PCR) showed that Pf4 mRNA levels were not significantly changed at day 7 after laser injury in posterior segment tissues compared with Normal control (i.e., untreated healthy mice) (Fig. 1A). Consistently, ELISA detected no significant difference in PF4 protein levels in posterior eye lysates at day 7 post‑laser (Fig. 1B). Immunofluorescence further revealed only weak PF4 signals in both Normal and laser-injured eyes, without an apparent increase within laser lesions (Fig. 1C).

Fig. 1.

PF4 expression after laser application. (A) Pf4 mRNA levels in the laser lesions of adult mice (6–8 weeks old) at the indicated time points after laser injury, measured by quantitative real-time PCR (qRT-PCR). n = 7 eyes (from 4 mice) per group. (B) PF4 protein levels, measured by ELISA. n = 4 eyes (from 3 mice) per group. (C) Representative immunofluorescence images of PF4 (green) and CD31 (red) in cross-sections of the retinas from Normal control and laser-induced CNV mice at day 7 post-injury; nuclei are counterstained with DAPI (blue). Scale bar = 20 µm. Data are shown as mean $\pm$ SEM. ns p $\ge$ 0.05. Two-tailed Student’s t-test was used for comparison between two groups (A and B). PF4, platelet factor 4; ELISA, enzyme-linked immunosorbent assay; CNV, choroidal neovascularization; DAPI, 4’,6-diamidino-2-phenylindole; SEM, standard error of the mean.

To investigate the role of PF4 in pathological neovascularization, we initially utilized a laser-induced mouse model of CNV. Six-week-old mice underwent laser induction, followed by intravitreal injection of PF4 protein (Fig. 2A). We first assessed the hallmark feature of CNV, vascular leakage, by FFA on day 3 post-laser. FFA images demonstrated that dye leakage was markedly reduced in PF4-treated mice at days 3 and 7 after laser injury (Fig. 2B). Quantitative analysis further confirmed this robust inhibitory effect, showing a significant reduction in leakage area in the PF4-treated group at both time points (Fig. 2C). We then evaluated the extent of neovascularization by performing CD31 immunofluorescence staining on RPE–choroid flat mounts at day 7 post-laser. Representative confocal images revealed that CNV lesions were noticeably smaller in PF4-treated mice than in Vehicle (i.e., mice injected with the carrier solution without PF4) (Fig. 2D). Quantification of CNV volume corroborated these findings, demonstrating a significant decrease in lesion volume in the PF4-treated group (Fig. 2E). To validate these results histologically, we performed H&E staining on retinal cross-sections. H&E images showed a clear reduction of CNV lesions in PF4-treated eyes (Fig. 2F).

Fig. 2.

PF4 administration attenuates vascular leakage and choroidal neovascularization in a laser-induced mouse model of CNV. (A) Schematic of the laser injury, intravitreal injection, and analysis schedule. (B) Representative longitudinal paired FFA images showing vascular leakage in the same eyes at days 3 and 7 after laser injury in vehicle- and PF4-treated mice. (C) Quantification of leakage area at days 3 and 7 is shown in (B). n = 30 lesions from 5 mice per group. (D) Representative confocal images of RPE–choroid flat mounts at day 7 after laser injury from vehicle- and PF4-treated mice, stained for CD31 (green) to visualize CNV lesions. Scale bar = 50 µm. (E) Quantification of CNV volume shown in (D), assessed as CD31-positive fluorescence volume. n = 32 lesions from 5 mice per group. (F) Representative retinal sections stained with H&E from Normal, vehicle-treated, and PF4-treated mice. Scale bar = 50 µm. Data are shown as mean $\pm$ SEM. **p 0.01. Two-tailed Student’s t-test was used for comparison between two groups. FFA, fundus fluorescein angiography; RPE, retinal pigment epithelium; H&E, hematoxylin and eosin. Fig. 2A was created using Adobe Illustrator 2025.

To further evaluate the protective role of PF4, we employed the Vldlr⁻/⁻ mouse model, which recapitulates key features of human neovascular retinal diseases [26]. Vldlr⁻/⁻ mice develop pathological retinal neovascularization (PRNV) from postnatal day 11 (P11), accompanied by progressive retinal ischemia, breakdown of the outer blood–retinal barrier, and functional decline. This pathology stems from impaired RPE lipid transport, resulting in HIF-driven upregulation of proangiogenic factors such as VEGF, mirroring the pathogenesis of proliferative diabetic retinopathy and nAMD [27].

We intravitreally injected PF4 or vehicle into Vldlr⁻/⁻ mice at P11 and analyzed retinal angiogenesis at P18. CD31 immunostaining of retinal flat mounts revealed marked reduction in neovascular lesions in PF4-treated eyes compared with Vehicle, with higher-magnification and three-dimensional (3D)-reconstructed images further demonstrating attenuated aberrant vascular growth (Fig. 3A). Quantitative analysis confirmed that PF4 significantly decreased both the number and area of neovascular lesions (Fig. 3B,C). Similarly, in adult Vldlr⁻/⁻ mice, PF4 injection reduced vascular leakage, as evaluated by FFA at baseline (i.e., the pre-injection time point, day 0) and 7 days post-injection (Fig. 3D). Quantification showed a significant improvement in vascular leakage following PF4 treatment (Fig. 3E).

Fig. 3.

PF4 administration attenuates vascular leakage retinal neovascularization in Vldlr⁻/⁻ mice. (A) Representative confocal images of retinal flat mounts from wild-type (WT), vehicle-treated Vldlr⁻/⁻, and PF4-treated Vldlr⁻/⁻ mice following CD31 (red) immunostaining. Whole-mount views, magnified regions (zoomed in), images from the RPE side, and corresponding three-dimensional (3D) reconstructions illustrating abnormal vascular growth are shown. Scale bar = 500 µm (whole mount), 167 µm (zoomed in), 50 µm (RPE side), and 50 µm (3D). (B,C) Quantifications of the lesion number and lesion area are shown in (A). n = 6 eyes (from 3 mice) per group. (D) Adult Vldlr⁻/⁻ mice received intravitreal injections of the indicated treatments (Vehicle or PF4) and were analyzed on day 7. Representative FFA images show vascular leakage at baseline (day 0, pre-injection) and at day 7 post-injection in Vehicle-treated Vldlr⁻/⁻​ and PF4-treated Vldlr⁻/⁻​ mice, compared with age-matched healthy WT controls. (The two WT images are biological replicates from independent mice, demonstrating the consistent absence of vascular leakage in healthy retinas.) (E) Quantification of leakage area post-treatment is shown in (D). n = 6 eyes (from 3 mice) per group. Data are shown as mean $\pm$ SEM. *p 0.05, **p 0.01. Two-tailed Student’s t-test was used for comparison between two groups. Vldlr⁻/⁻, very low‑density lipoprotein receptor knockout.

Because inflammation contributes to CNV, we examined whether PF4 modulates laser-induced inflammatory responses. Compared with vehicle, PF4 significantly reduced mRNA levels of Il1b, Il6, Tnf, and Nfkb1 in posterior segment retinal tissues at day 7 post-laser injury (Fig. 4A). We next assessed microglia/macrophages recruitment at lesion sites. Immunostaining showed fewer Iba1⁺ microglia/macrophages with reduced aggregation in PF4-treated eyes (Fig. 4B), and quantitative analysis confirmed reduced Iba1⁺ microglia/macrophages volume within lesions (Fig. 4C).

Fig. 4.

PF4 suppresses intraocular inflammation induced by laser injury. (A) Relative mRNA expression of Il1b, Il6, Tnf, and Nfkb1 in the posterior segment 7 days after laser photocoagulation in vehicle- and PF4-treated eyes. (B) Representative confocal images of laser lesions from vehicle- and PF4-treated retinas immunostained for Iba1 (red) and CD31 (green), with nuclei counterstained with DAPI (blue). Scale bar = 50 µm. (C) Quantification of Iba1⁺ microglia/macrophages volume within the laser lesions shown in (B). n = 32 lesions (from 5 eyes) per group. (D) Heatmap of differentially expressed genes in retina–choroid complexes from Normal, vehicle-treated, and PF4-treated mice. (E) Volcano plot of differentially expressed genes in PF4-treated versus vehicle-treated groups. (F) GO enrichment analysis of differentially expressed genes between PF4- and vehicle-treated groups. (G) KEGG pathway enrichment analysis of differentially expressed genes between PF4- and vehicle-treated groups. n = 3 eyes (from 3 mice) per group for RNA sequencing. Data are shown as mean $\pm$ SEM. *p 0.05, **p 0.01. Two-tailed Student’s t-test was used for comparison between two groups. Il1b, interleukin-1​ beta; Il6, interleukin-6; Tnf, tumor necrosis factor; Nfkb1, nuclear factor kappa-B​ subunit 1​; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes.

To comprehensively characterize PF4-mediated transcriptomic changes, we performed bulk RNA sequencing of retina–choroid complexes from Normal, vehicle-treated, and PF4-treated mice. Hierarchical clustering of differentially expressed genes (DEGs) across the three groups revealed clearly segregated expression profiles (Fig. 4D). Vehicle-treated samples displayed a pronounced proinflammatory signature, whereas PF4-treated samples formed a distinct cluster whose expression pattern partially shifted toward that of Normal controls, suggesting that PF4 counteracts injury-induced inflammatory reprogramming. Differential expression analysis between the Vehicle and PF4 groups showed that PF4 induced widespread transcriptional remodeling (Fig. 4E). PF4 significantly downregulated a broad set of inflammation-related genes, including chemokines, cytokines, adhesion molecules, and genes associated with leukocyte trafficking and microglia/macrophages activation, while genes involved in tissue homeostasis were relatively preserved. We interrogated the biological processes associated with PF4-regulated genes using GO enrichment analysis. PF4-downregulated DEGs were predominantly enriched in terms related to innate immune responses, cytokine production, regulation of inflammatory responses, and responses to external stimuli (Fig. 4F), indicating broad suppression of inflammatory activation programs. We then examined pathway changes using KEGG enrichment analysis, which revealed that PF4 strongly inhibited signaling cascades implicated in retinal inflammation and leukocyte recruitment, including the NF-$\kappa$B and JAK–STAT signaling pathways, chemokine signaling, cell adhesion molecule pathways, leukocyte transendothelial migration, and immune-related pathways (Fig. 4G).

Given that inflammatory mediators can induce VEGFA expression [28, 29, 30, 31, 32, 33], we tested whether PF4 affects VEGFA levels in laser-induced CNV. Compared with the Normal control group, laser photocoagulation markedly increased VEGFA protein expression in the RPE–choroid complex, whereas PF4 treatment significantly attenuated this upregulation (Fig. 5A,B). Consistent with the western blot results, immunofluorescence staining of sections through the laser lesions revealed strong immunoreactivity of VEGFA around CD31-positive neovessels in the vehicle group, which was markedly reduced in PF4-treated eyes (Fig. 5C). These findings suggest that PF4 may inhibit CNV formation at least in part by suppressing inflammation-driven upregulation of VEGFA.

Fig. 5.

PF4 attenuates VEGFA expression in laser‑induced CNV lesions. (A) Representative immunoblot of VEGFA expression in Normal (no laser), laser‑injured + vehicle, and laser‑injured + PF4–treated eyes. $\beta$‑Actin served as a loading control. (B) Quantification of VEGFA protein levels normalized to $\beta$-actin and expressed as fold change relative to the sham group. n = 4 eyes (from 4 mice) per group. (C) Representative confocal images of posterior segment from laser-injured eyes treated with vehicle or PF4, showing VEGFA (green), CD31-positive vascular endothelium (red), and nuclei (DAPI, blue), scale bar = 20 µm. Data are shown as mean $\pm$ SEM. *p 0.05, **p 0.01. One-way ANOVA was used for the comparison of multiple groups. ANOVA, analysis of variance.

In addition to suppressing inflammation-driven upregulation of VEGFA in vivo, we sought to investigate whether PF4 could counteract VEGF-induced angiogenic responses in endothelial cells in vitro. We initially investigated the impact of this factor on key angiogenic processes in vitro using HRMECs—a type of vascular endothelial cell that highly expresses VEGFR2 and is highly sensitive to VEGF stimulation [34, 35, 36, 37].

The results showed that the number of proliferating HRMECs was markedly reduced in the PF4-pretreated group compared with the group stimulated with VEGF alone (Fig. 6A). Quantitative analysis further revealed a significant decrease in the percentage of EdU-positive cells (Fig. 6B). We next assessed cell migration capacity using a scratch wound-healing assay, which revealed that PF4 pretreatment significantly suppressed VEGF-stimulated migration of HRMECs (Fig. 6C). The scratch closure rate was significantly lower in the PF4-treated group than in the control group (Fig. 6D). Furthermore, we examined the effect of PF4 on the ability of HRMECs to form capillary-like tubular structures, a critical step in the angiogenesis process. The results demonstrated that PF4 pretreatment disrupted VEGF-driven tube formation in HRMECs (Fig. 6E). The number of branch points was reduced in the PF4-treated group compared with the VEGF control group (Fig. 6F).

Fig. 6.

PF4 inhibits VEGF-induced proliferation, migration, and tube formation of human retinal microvascular endothelial cells. (A) Representative EdU staining of HRMECs after PF4 pretreatment and VEGF stimulation, stained for EdU (green) and DAPI (blue). Scale bar = 50 µm. (B) Quantification of EdU-positive cells as a percentage of total cells in (A). n = 5 per group. (C) Representative images of HRMEC migration in a scratch-wound assay at 0 and 12 hours after PF4 pretreatment and VEGF stimulation. Scale bar = 50 µm. (D) Quantification of gap closure in (C). n = 5 per group. (E) Representative images of tube formation by HRMECs following PF4 pretreatment and VEGF stimulation. Scale bar = 100 µm. (F) Quantification of branch points in (E). n = 5 per group. Data are shown as mean $\pm$ SEM. *p 0.05, **p 0.01, ns p $\ge$ 0.05. One-way ANOVA was used for the comparison of multiple groups. VEGF, vascular endothelial growth factor; HRMECs, human retinal microvascular endothelial cells.

To elucidate the molecular basis underlying these anti-angiogenic effects of PF4, we next examined whether PF4 modulates VEGF/VEGFR2 signaling and its key downstream pathways [38, 39]. In the laser-induced CNV mouse model, PF4 administration markedly reduced VEGFR2 expression, as well as the phosphorylation of ERK and STAT3 in choroidal tissue (Fig. 7A,B), indicating that PF4 attenuates VEGF/VEGFR2-related signal transduction in vivo. Consistent with these findings, PF4 significantly inhibited VEGF-induced phosphorylation of AKT, ERK, and STAT3 in cultured HRMECs (Fig. 7C,D), further supporting its direct suppression of VEGF/VEGFR2-driven pro-angiogenic signaling.

Fig. 7.

PF4 attenuates VEGF/VEGFR2-driven ERK, AKT, and STAT3 signaling in vivo and in vitro.(A) Representative immunoblots of VEGFR2, phosphorylated (p-ERK) and total (t-ERK) ERK, and phosphorylated (p-STAT3) and total (t-STAT3) STAT3 in RPE–choroid–sclera complex tissue from sham (no laser), laser-injured + vehicle, and laser-injured + PF4–treated mice. $\beta$-Actin served as a loading control. (B) Densitometric analysis of VEGFR2/$\beta$-actin, p-ERK/t-ERK, and p-STAT3/t-STAT3 expression in (A), shown as fold change relative to control. n = 4 eyes (from 4 mice) per group. (C) Representative immunoblots of phosphorylated and total AKT (p-AKT, t-AKT), ERK, and STAT3 in HRMECs stimulated with VEGF in the presence or absence of PF4. (D) Quantification of p-AKT/t-AKT, p-ERK/t-ERK, and p-STAT3/t-STAT3 in (C), expressed as fold change relative to unstimulated cells. n = 3 per group. Data are shown as mean $\pm$ SEM. *p 0.05, **p 0.01, ns p $>$ 0.05. One-way ANOVA was used for the comparison of multiple groups. VEGFR2, vascular endothelial growth factor receptor 2; ERK, extracellular signal-regulated kinase; AKT, protein kinase B; STAT3, signal transducer and activator of transcription 3.

We next evaluated the retinal safety of intraocular PF4 administration in mice at day 7 post-injection. OCT imaging showed that retinal morphology and thickness were preserved following intraocular injection of either vehicle or PF4 (Fig. 8A,B). Consistently, ERG revealed no significant reduction in a-wave amplitudes (originating primarily from photoreceptor rods and cones) or b-wave amplitudes (reflecting inner retinal activity, mainly Müller and bipolar cells) in PF4-treated eyes compared with vehicle-treated controls (Fig. 8C,D). Histological examination further confirmed the structural integrity of the retina (Fig. 8E), and TUNEL staining demonstrated an absence of appreciable cell death (Fig. 8F). Taken together, these data indicate that a single intraocular administration of PF4 (0.3 µg/eye) was well tolerated and exhibited a favorable retinal safety profile in mice.

Fig. 8.

Evaluation of retinal toxicity following the intravitreal administration of PF4 in mice. (A) Representative fundus imaging and OCT imaging of normal mice and mice treated with vehicle or PF4. The circular bounding box in the OCT images indicates the standardized region scanned and used for layer thickness measurement. (B) Thickness of total retinal layers. n = 6 eyes (from 3 mice) per group. (C) Representative electroretinogram imaging of normal mice and mice treated with vehicle or PF4, where “a” and “B” denote the a-wave and b-wave amplitudes, respectively. The y-axis represents response amplitude in microvolts (µV). (D) Electroretinogram and quantification of normal mice and mice treated with vehicle or PF4. n = 6 eyes (from 3 mice) per group. (E) H&E staining of eyeball sections from normal mice and mice treated with vehicle or PF4. Scale bar = 100 µm. (F) TUNEL assay of retinal sections from normal mice and mice treated with vehicle or PF4. Scale bar = 50 µm. Data are shown as mean $\pm$ SEM. ns p $\ge$ 0.05. One-way ANOVA was used for the comparison of multiple groups. OCT, Optical Coherence Tomography; TUNEL, terminal deoxynucleotidyl transferase dUTP nick end labeling.