Size-shrinking nanoparticles with high drug-protein payload for efficient, non-invasive treatment of corneal neovascularization

Xiaoling Xie; Zhenyang Xiang; Ruhui Yang; Xuanlu Wu; Jinyu Guo; Ruolin Fang; Xingjie Zan; Wenjuan Huang; Yuhua Tong

doi:10.1186/s12951-026-04775-7

. 2026 Jul 10;24:928. doi: 10.1186/s12951-026-04775-7

Size-shrinking nanoparticles with high drug-protein payload for efficient, non-invasive treatment of corneal neovascularization

Xiaoling Xie ^2,³, Zhenyang Xiang ⁴, Ruhui Yang ^2,³, Xuanlu Wu ⁵, Jinyu Guo ^2,³, Ruolin Fang ⁶, Xingjie Zan ^2,^✉, Wenjuan Huang ^4,^✉, Yuhua Tong ^1,^✉

PMCID: PMC13640248 PMID: 42432700

Abstract

Corneal neovascularization (CNV) is one of the leading causes of corneal blindness, affecting millions of people worldwide. Anti-vascular endothelial growth factor agents, such as Bevacizumab (Beva), offer high specificity and low side effects. However, their limited ability to penetrate the corneal barrier necessitates invasive administration, significantly restricting their clinical application. Herein, we engineered (Beva&C₂G₂R₉)@Zn nanoparticles formed by co-assembling Beva, C₂G₂R₉ peptide and Zn²⁺, which decrease in size over time, as an efficient strategy for noninvasive Beva delivery across the corneal barrier to treat CNV. By combining various technologies (DLS, TEM, XPS, FTIR, and computer simulation), we discovered that the coordination between Beva and Zn²⁺ drives the nanoparticle formation, while the C₂G₂R₉ peptide facilitates its size evolution. Compared to size-stable nanoparticles of Beva@Zn and (Beva&R₉)@Zn, (Beva&C₂G₂R₉)@Zn nanoparticles exhibit rapid cellular internalization, efficient lysosomal escape, and effective corneal barrier penetration, leading to efficiently inhibit HUVEC cell migration and tube formation. Importantly, in a rat alkali-burned CNV model, (Beva&C₂G₂R₉)@Zn nanoparticles exhibited superior efficacy in inhibiting corneal neovascularization compared to size-stable nanoparticles, with the lowest inflammation index. The results of this study highlight the importance of controlling the size of nanoparticles to enable non-invasive delivery of macromolecular drugs across the corneal biological barrier, offering new insights for the design of future nanoparticle-based drug delivery systems.

Graphical Abstract

graphic file with name 12951_2026_4775_Figb_HTML.webp

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12951-026-04775-7.

Keywords: Protein delivery, Corneal neovascularization, Non-invasive delivery, Biobarrier, Size effect

Introduction

Corneal neovascularization (CNV), marked by the abnormal ingrowth of blood vessels into the cornea, often arises from inflammation, infection, or hypoxia, and can lead to significant visual impairment, pain, and scarring [1]. This condition affects millions of individuals globally, particularly those with high contact lens use, widespread dry eye disease, or a history of frequent corneal surgeries [2, 3]. Protein-based therapeutics, such as anti-vascular endothelial growth factor (anti-VEGF) agents like Bevacizumab (Beva) and ranibizumab, are commonly used as first-line treatments due to their high specificity, efficacy, and relatively low side effects [4]. However, the effectiveness of these therapies is significantly hindered by ocular barriers, including the corneal barrier, tear film, drainage system, and conjunctival barrier, which limit drug bioavailability [5]. For protein drugs like anti-VEGF, their large molecular weight and hydrophilicity result in almost negligible ocular permeability, making topical eye drop therapies less effective. Currently, the clinical application of anti-VEGF agents for CNV predominantly involves invasive administration routes, which not only increase the risk of infection but also compromise patient compliance [6]. Consequently, the development of non-invasive drug delivery systems represents a promising avenue for transforming current therapeutic approaches [7].

Nanomedicine eye drops are revolutionizing the treatment of CNV by enhancing drug delivery efficiency and enabling precise, non-invasive targeting [8, 9]. The key to improving the therapeutic efficacy of these formulations lies in prolonging drug retention on the ocular surface and enhancing corneal permeability [10]. The physicochemical properties of nanomedicines, including particle size, charge, shape, surface modification, and hydrophobicity, are critical factors that govern both ocular surface retention and effective corneal penetration. Among these factors, particle size is particularly crucial [11, 12]. Optimizing nanoparticle size to balance retention time and permeability has been a central focus of research [13]. Ultra-small nanoparticles, while capable of deeper tissue penetration, are limited in their ocular surface retention due to the rapid turnover of tear fluid [14]. Furthermore, the tear drainage system quickly clears these ultra-small particles, reducing drug bioavailability. In contrast, larger particles are more resistant to tear fluid renewal, resulting in longer retention times on the ocular surface [15]. However, their larger size hinders effective corneal penetration, thereby limiting therapeutic outcomes [16]. Achieving this balance requires addressing the trade-off between the extended retention of larger particles and the rapid penetration of smaller ones, which remains a significant challenge in the design of effective drug delivery systems [17]. Previous studies have primarily focused on chemically modifying nanomedicines with optimizable particle sizes to extend ocular surface retention time of nanoparticles or enhance corneal tissue barrier permeability, in order to improve the efficacy of nanomedicines in treating CNV [18, 19]. However, such chemical modifications require intricate design and complex synthesis processes, and the potential biological toxicity of the formulations used during these modifications poses significant challenges for clinical translation. We propose that developing nanomedicines with a size that gradually decreases over time could effectively integrate the advantages of large particle long-term retention and small particle tissue penetration, thereby achieving better therapeutic outcomes for CNV.

Previously, we reported a strategy to form Beva@Zn assemblies via a mixing–ultrasonication process of Beva and zinc ions (Zn²⁺) under mild conditions (aqueous environment, pH 8, room temperature), which not only efficiently encapsulates Beva but also preserves its biological activity and protects the encapsulated Beva from proteolytic degradation [20]. Here, we developed a one-step co-assembly strategy based on Beva, the peptide C₂G₂R₉ (Cys-Cys-Gly-Gly-Arg9), and Zn²⁺ to construct (Beva&C₂G₂R₉)@Zn (Scheme 1a). By leveraging the ability of the thiol group in Cys and the guanidino group in Arg to competitively bind Zn²⁺, Zn²⁺ are gradually depleted from the (Beva&C₂G₂R₉)@Zn particles, inducing a progressive size reduction (from an initial 237.0 ± 2.0 nm to 37.8 ± 1.1 nm after 12 h) (Scheme 1b). The advantage of this design lies in the large-size effect of (Beva&C₂G₂R₉)@Zn during the initial phase, which enhances ocular surface retention, followed by the small-size phase, which improves corneal permeability (Scheme 1c). Our results demonstrate that, compared with particles lacking size-evolution properties, such as Beva@Zn and (Beva&R₉@Zn), (Beva&C₂G₂R₉)@Zn exhibits superior therapeutic efficacy in treating CNV. To the best of our knowledge, this is the first size-shrinking nanoparticle system to achieve nearly complete inhibition of CNV via non-invasive eye drop administration in the rat alkali-burn model.

Scheme 1 — Schematic illustration on (a) the fabrication of (Beva&C₂G₂R₉)@Zn nanoparticles, (b) the size-shrinking of (Beva&C₂G₂R₉)@Zn nanoparticles, and (c) the non-invasive delivery of Beva across corneal barriers for CNV treatment by (Beva&C₂G₂R₉)@Zn nanoparticles

Materials and experimental methods

Materials

The C₂G₂R₉ peptide (> 98% purity), FITC-R₉, and FITC-C₂G₂R₉ were purchased from Peptide Institute Inc. (Nanjing, China). Bevacizumab (Beva) was obtained from Roche Pharmaceuticals. Zinc nitrate hexahydrate (Zn(NO₃)₂·6 H₂O, ≥ 99%), polyvinylpyrrolidone (PVP, molecular weight: 58 kDa), 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), sodium hydroxide (NaOH), bovine serum albumin (BSA), fluorescein isothiocyanate (FITC), 10% formaldehyde, acetic acid, and ethanol were purchased from Aladdin (China). Anti-CD31 antibody (ab119339), anti-VEGF-A antibody (ab1316), goat anti-mouse IgG H&L (Alexa Fluor 594), FITC-labeled goat anti-human IgG (H + L), DAPI, and Bradford protein assay kits were obtained from Beyotime (China). Matrigel was purchased from Corning Inc. (USA). Proteinase K was sourced from GlpBio (USA). Bevacizumab ELISA kits were purchased from Ruifan Biotechnology Co., Ltd. (Shanghai, China). Lysosome staining reagent LysoTracker Red DND-99 was obtained from Invitrogen (Thermo Fisher Scientific, USA). Anti-ZO1 tight junction protein rabbit polyclonal antibody was purchased from Servicebio (China). FITC-labeled Beva was synthesized according to a previous report [21]. 3D human corneal epithelial model was purchased from Biocell Biotechnology Co. LTd.

Preparation and optimization of preparation conditions for Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn

Beva (1 mg/mL) was dissolved in 50 mM HEPES sodium buffer (pH 8.0) at 4 °C. Subsequently, different volumes of Zn(NO₃)₂ solution (0.1 M) were added dropwise to the mixture, achieving Beva: Zn²⁺ mass ratios of (92:1, 92:2, 92:3, 92:4, 92:5, and 92:6). The addition of Zn(NO₃)₂ was carried out under sonication until the fixed Zn(NO₃)₂ solution was added, resulting in the solution changing from clear to milky white. The mixture was further sonicated for 30 min, and the resulting Beva@Zn nanoparticles were collected by centrifugation. The supernatant was discarded, and the nanoparticles were re-dispersed in deionized water for characterization. The encapsulation efficiency percentage (EE%) of Beva, particle size zeta potential, and polydispersity index (PDI) were optimized by varying the Beva: Zn²⁺ mass ratios. The optimal Beva: Zn²⁺ mass ratio obtained was used for the synthesis of (Beva&R₉)@Zn and (Beva&C₂G₂R₉)@Zn.

The synthesis of (Beva&R₉)@Zn and (Beva&C₂G₂R₉)@Zn followed a similar procedure to Beva@Zn, with the key difference being that Beva was first pre-mixed with R₉ or C₂G₂R₉ at different mass ratios (1:3, 2:3, 3:3, 4:3, 5:3). The EE% of Beva, particle size zeta potential, and polydispersity index (PDI) were optimized by varying the Beva: R₉ or C₂G₂R₉ mass ratios.

Characterization of Beva-NPs

Size and zeta potential

The particle size, PDI, and zeta-potential of the nanoparticles were measured using a Zeta-sizer instrument (Malvern, ZEN3600, UK). Specifically, 100 µL of the nanoparticle sample was diluted with 1 mL of deionized water, and the particle size and PDI were measured at 25 °C. The measurements were repeated three times. Subsequently, the same sample was transferred to a dedicated electrophoretic cell for zeta-potential determination. All instrument status indicators during the measurements were labeled as “good,” indicating stable particle dispersion and reliable data, which are suitable for further data analysis.

Encapsulation efficiency percentage (EE%) of Beva

FITC-labeled peptides (FITC-R₉ and FITC-C₂G₂R₉) were used to prepare (Beva&FITC-R₉)@Zn and (Beva&FITC-C₂G₂R₉)@Zn nanoparticles. The EE% of Beva and the peptides (R₉, C₂G₂R₉) was calculated using the following formula: EE% = [(m₁ - m₂) / m₁] × 100%, where m₁ represents the initial mass of the nanoparticles added into the reaction system, and m₂ is the free mass of the unencapsulated material in the supernatant after thorough centrifugation. The quantification method for m₂ varies depending on the materials: for Beva, the concentration in the supernatant was determined using by Ultraviolet-Visible(UV–Vis) absorbance measurement at 280 nm and calculated via a standard curve; for R₉ and C₂G₂R₉, the content was quantified using the fluorescence standard curve established for their FITC-labeled counterparts (FITC-R₉ and FITC-C₂G₂R₉). The standard curves were shown in Figure S1a for Beva, Figure S1b for R₉, and Figure S1c for C₂G₂R₉. The Ultraviolet-Visible (UV-vis) spectra were recorded using UV-vis spectrometer (Agilent, CARY 5000, USA), and the absorbance at 280 nm was used for standard curve and quantitative analysis for Beva. And the fluorescence spectra were recorded using Fluorescence Spectrometer (PerkinElmer, LS 55, USA), and the absorbance at 490 nm was used for standard curve and quantitative analysis for FITC-R₉ and FITC-C₂G₂R₉. In addition, other excipients (Zn²⁺, HEPES/PVPON mixture, R₉ and C₂G₂R₉) in the formulation showed no obvious absorption at 280 nm under UV-vis and thus caused no significant interference with the determination, meeting the requirements for relative quantification (Figure S1d). It should be noted that UV–Vis absorbance at 280 nm is a non-specific total protein quantification method, which cannot specifically distinguish Beva from other protein/peptide components in the formulation. Therefore, the quantification results herein provide a relative estimate of Beva encapsulation, rather than an absolute specific quantification of intact Beva.

Loading capacity percentage of (Beva, and R₉ and/or C₂G₂R₉) in Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn

FITC-labeled peptides (FITC-R₉ and FITC-C₂G₂R₉) were used to prepare (Beva&FITC-R₉)@Zn and (Beva&FITC-C₂G₂R₉)@Zn nanoparticles. A certain amount of Beva@Zn, (Beva&FITC-R₉)@Zn, and (Beva&FITC-C₂G₂R₉)@Zn nanoparticles were weighed and dispersed in 200 µL of water. An equal volume of 10 mM EDTA was added to chelate Zn²⁺, thereby disassembling the Beva@Zn, (Beva&R₉)@Zn, or (Beva&C₂G₂R₉)@Zn nanoparticles. Once the solution became completely transparent, the quantities of Beva, R₉, and/or C₂G₂R₉ in the solution were determined according to the calibration curves (FigureS1). The loading capacity percentage (LC%) of Beva, R₉, or C₂G₂R₉ in Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn nanoparticles was calculated using the following formula: LC% = (m₁ / m₀) × 100%, where m₁ represents the mass of Beva, R₉, or C₂G₂R₉, m₀ represents the mass of weight nanoparticles.

Zinc content in Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn

The zinc content in the nanoparticles was determined using Inductively Coupled Plasma Optical Emission Spectroscopy (ICP-OES) (Agilent, ICP-OES 7850, USA) inductively coupled plasma optical emission spectrometer. The sample preparation procedure is as follows: 200 µL of (Beva&C₂G₂R₉)@Zn nanoparticles was mixed with an equal volume of 10 mM EDTA solution. The chelating effect of EDTA on zinc ions was used to dissociate the nanoparticle structure. The mixture was then diluted to a final volume of 5 mL with deionized water and subjected to ICP-OES analysis.

Transmission Electron Microscopy (TEM)

To observe the morphology of the nanoparticles, samples of (Beva&C₂G₂R₉)@Zn prepared at different time points were processed for TEM analysis. First, the samples were diluted approximately 100 times with deionized water and sonicated. A small amount of the sample was then deposited onto a clean copper grid coated with carbon film. Negative staining was performed using phosphotungstic acid solution. After drying, the samples were observed using the TEM (FEI, Talos F200S, USA) equipped with a Thermo Scientific Heratherm OMS 180 sample holder, and images were captured.

Fourier Transform Infrared Spectroscopy (FTIR)

The freeze-dried white powder of (Beva&C₂G₂R₉)@Zn was thoroughly ground with potassium bromide (KBr) in a mortar and then pressed into a transparent pellet using a pellet press. The sample was analyzed using the FTIR Spectrometer (Bruker, TENSOR II, German ) within a wavenumber range of 400–4000 cm⁻¹.

X-ray Photoelectron Spectroscopy (XPS)

The dispersion of Beva and (Beva&C₂G₂R₉)@Zn nanoparticles was deposited onto clean silicon wafers and dried in a Thermo Scientific Heratherm OMS 180 oven at 60 °C. XPS analysis was performed using an ESCALAB 250 Xi spectrometer with a 400 μm spot size and a 1.000 eV step size for data acquisition.

In vitro release of Beva from Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn

The Beva release behavior from Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn was investigated as previous report [20]. Briefly, 1 mL Beva@Zn, (Beva&R₉)@Zn, or (Beva&C₂G₂R₉)@Zn was placed into anti-protein adsorption dialysis tube (molecular weight cutoff: 1000 kDa) in mother solution, 50 mL Bis-Tris buffer (10 mM, pH 7.4) at 37 °C and shaking with 220 rpm. At designated time points, 1 mL samples were sucked from the mother solution and replaced with an equal volume of fresh buffer. The concentration of released Beva was analyzed based the absorbance at 280 nm under UV–Vis spectroscopy and standard curve (FigureS1a). The cumulative Beva release was calculated according to previous report [20]. Since UV–Vis absorbance at 280 nm cannot specifically distinguish Beva from other peptide or protein components in the formulation, the release values presented herein represent a relative estimate of Beva release, rather than an absolutely specific quantification.

Bioactivity of beva

The bioactivity of Beva was evaluated by comparing the activity of free Beva and the Beva released from nanoparticles (Beva@Zn, (Beva&R₉)@Zn, (Beva&C₂G₂R₉)@Zn) upon EDTA dissociation. This comparison was performed to assess the effect of nanoparticle encapsulation on enzyme stability. The activity assays were carried out strictly according to the manufacturer’s instructions for the ELISA kit.

Simulation

In the molecular dynamics (MD) simulation, the Beva dimer complex was first placed in a 23 × 23 × 23 nm³ cubic water box for hydration. The system’s initial configuration, topology files, and CHARMM-36 m force field parameters were constructed using the CHARMM-GUI platform [22, 23]. For the zinc-containing system, in addition to the protein-peptide dimer, the system included approximately 421,541 water molecules, 629 Na⁺ ions, 66 Zn²⁺ ions, and 442 Cl⁻ ions to simulate the buffer environment (25 mM NaCl + 0.75 mM ZnCl₂), resulting in a total of approximately 1,288,000 particles. Periodic boundary conditions were applied in a cubic box, with van der Waals interactions truncated at 1.0–1.2 nm and long-range electrostatic interactions calculated using the Particle Mesh Ewald (PME) method.

After energy minimization via the steepest descent method, the system was equilibrated in two stages: 0.4 ns NVT and 20 ns NPT. During the equilibration process, the restraint force was gradually reduced: from 1000 to 400 kJ·mol⁻¹·nm⁻² in the NVT phase, and further to 40 kJ·mol⁻¹·nm⁻² in the NPT phase, after which the restraint was completely removed. Finally, under the NPT ensemble, a 200 ns all-atom MD simulation was performed using GROMACS 2021 software at 298 K with a 2 fs integration step, generating the trajectory of the dimer complex with different peptides.

In vitro experiments

Cell culture

Human corneal epithelial cells (HCEC) and human umbilical vein endothelial cells (HUVEC) were used to assess the in vitro biocompatibility of the nanoparticles. HCECs and HUVECs were cultured in MEM medium supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin at 37 °C in a humidified incubator with 5% CO₂.

In vitro cytotoxicity

The cytotoxicity of the nanoparticles was evaluated using the CCK-8 assay. HCECs and HUVECs were seeded in 96-well plates at a density of 5 × 10³ cells per well and allowed to adhere for 24 h. Cells were then exposed to a series of concentrations (0, 25, 50, 100, 200 µg Beva/mL) of Beva-NP and cultured for an additional 24 h. After removing the drug-containing medium and washing the cells with PBS, 100 µL of culture medium containing 10% CCK-8 was added to each well. After 2 h of incubation, absorbance was measured at 450 nm using a Thermo Varioskan LUX microplate reader to assess cell viability.

Cellular uptake study

HCECs were seeded at a density of 5 × 10⁴ cells per well on sterile glass coverslips in 24-well plates and cultured for 24 h. After cells had adhered, they were washed three times with PBS and incubated with fresh medium containing either free Beva, (Beva&R₉)@Zn, or (Beva&C₂G₂R₉)@Zn (all groups at a Beva concentration of 100 µg/mL) for 24 h. After incubation, the medium was discarded, and the cells were washed with PBS, followed by fixation with 10% paraformaldehyde. Cell nuclei were stained with DAPI (5 µg/mL). Imaging was performed using a high-resolution laser confocal scanning microscope (HCLSM) (Nikon A1, Japan). FITC-labeled Beva was directly observed.

Lysosomal escape experiment

To observe the intracellular localization of different nanoparticles, HCEC cells were treated with Beva@Zn, (FITC-Beva&R₉)@Zn, or (FITC-Beva&C₂G₂R₉)@Zn (all at a Beva concentration of 100 µg/mL) for 2 h and 6 h, with an untreated medium group serving as a control. After treatment, the cells were labeled with a lysosome probe for 30 min, followed by PBS washing, fixation with 4% paraformaldehyde, and nuclear staining with DAPI. Fluorescent images of the samples were captured using a HCLSM (Nikon A1, Japan), and subsequent colocalization analysis was performed using ImageJ software.

3D Corneal cell model permeation study

To evaluate the corneal permeability of different formulations, the purchased 3D human corneal epithelial model was cultured in Transwell chambers. In the upper chamber, 300 µL of BioGrowth medium containing each FITC-Beva formulation (Free Beva, Beva@Zn, (Beva&R₉)@Zn, (Beva&C₂G₂R₉)@Zn at a concentration of 200 µg/mL) was added, and the lower chamber was filled with 700 µL of DMEM. After 24 h of incubation, the model was washed with PBS, fixed with 4% paraformaldehyde, and stained with DAPI for nuclear visualization. The 3D cell layers adhered to the filter membrane were sealed with an anti-quenching mounting medium, and images were obtained using a HCLSM (Nikon A1, Japan).

Migration inhibition assay

HUVECs were seeded at a density of 4 × 10⁵ cells per well in 12-well plates and cultured overnight until confluence. A sterile 100 µL pipette tip was used to create a scratch in the cell monolayer, and the wells were gently washed three times with PBS to remove detached cells. The medium was then replaced with Beva-NP-containing medium, and cells treated with DMEM served as a control. Images were captured at 0 and 24 h post-scratch using a Carl Zeiss Axio Vert.A1 inverted microscope, and the migration rate width was quantified using ImageJ software.

Tube formation inhibition assay

To evaluate the effect of the drug on angiogenesis, the tube formation ability of HUVECs was assessed on Matrigel-coated plates. Briefly, 150 µL of Matrigel was pre-coated in 48-well plates and allowed to solidify at 37 °C for 1 h. HUVECs (5 × 10⁴ cells) were resuspended in serum-free DMEM and seeded into each well, followed by treatment with different formulations. After 6 h of incubation at 37 °C in a 5% CO₂ incubator, images of tube formation were captured using a Zeiss Axio Vert A1 microscope, and the tubular structures were quantified using ImageJ software.

In vivo experiments

Animals and ethical considerations

Male Sprague-Dawley (SD) rats, aged 6–8 weeks and weighing between 80 and 220 g, were selected for the experiments. The right eye of each rat was designated as the experimental eye for model establishment, while the left eye served as the control. Prior to the experiment, all animals underwent slit-lamp examination to ensure ocular health. Healthy adult New Zealand white rabbits (4 months old, weighing 2.0–3.0 kg) were used for the corneal diffusion experiment.

For the corneal retention experiment, temporary anesthesia was induced using 4% isoflurane, administered for 1–3 min. Rats used for the corneal neovascularization (CNV) model were anesthetized via intraperitoneal injection of pentobarbital (40 mg/kg). Euthanasia of rats was performed by intraperitoneal injection of pentobarbital (40 mg/kg) followed by cervical dislocation. For rabbit euthanasia, after temporary anesthesia, an overdose of pentobarbital (150 mg/kg) was administered intravenously, followed by cervical dislocation.

All animal procedures were approved by the Animal Experimentation Center of Wenzhou Institute, University of Chinese Academy of Sciences (protocol No. WIUCAS22101301).

Evaluation of retention in the cornea

Free Beva, Beva@Zn, (FITC-Beva&R₉)@Zn, and (FITC-Beva&C₂G₂R₉)@Zn particles prepared with FITC-Beva were used as the control groups (Free Beva) and experimental, respectively, to assess the retention time in the cornea of rats. Approximately 20 µL of the drug solution (with concentration FITC-Beva 2 mg/mL) was instilled into the rat’s eye, and the retention time was monitored at 0, 0.5, 1, 2, and 3 h using a multimodal animal imaging system (PerkinElmer IVIS Lumina XRMS Series III, USA). Prior to each imaging session, animals were temporarily anesthetized with 4% isoflurane. Naïve untreated rat eyes served as blank controls to subtract tissue autofluorescence and instrumental background. A fixed circular ROI covering the entire cornea was applied consistently across all images. No autofluorescence quencher was used; all imaging parameters were kept constant.

Percent retention was calculated as:

Where Inline graphic is the fluorescence intensity of the test sample at each time point, is the tissue autofluorescence and instrumental background signal from untreated blank rat eyes, and is the initial fluorescence intensity at 0 h post-administration.

Establishment of alkali burn CNV model

Following inhalation anesthesia with isoflurane, a combination of tropicamide eye drops and 0.5% proparacaine hydrochloride eye drops was applied to the right eye of the rats for pupil dilation and surface anesthesia. Subsequently, 4 µL of a 1 M sodium hydroxide solution was applied to a 3 mm diameter round filter paper, which was fully saturated for 1.5 min. The excess solution in the conjunctival sac was absorbed by the filter paper, and the alkali-saturated filter paper was then placed on the center of the rat’s cornea for 40 s. After quickly removing the filter paper, 30 mL of pre-chilled (4 °C) saline was used to continuously irrigate the conjunctival sac to terminate the burn reaction.

Treatment and medication

Following successful establishment of the alkali-induced CNV model, a total of 75 SD rats were randomly divided into five groups (n = 15 per group): NS control group (0.9% NaCl), free Beva group, (Beva&R₉)@Zn group, Beva@Zn group, and (Beva&C₂G₂R₉)@Zn group. Treatment was initiated immediately after model establishment, with the animals receiving eye drops three times a day for 7 consecutive days, with a dosage of 20 µL per application (with concentration of Beva 2 mg/mL). Rats were euthanized randomly on days 3, 7, and 14 post-treatment, and the eyeball tissues were collected for corneal neovascularization area measurement, histological analysis, and immunohistochemical staining evaluation.

CNV evaluation

The total area of CNV (S) was calculated by dividing the cornea into four quadrants and summing the area (A) of each quadrant. The area of a single quadrant was calculated using the formula:

Where C is the clock hour affected by the quadrant, determined by calibrating the corneal clock hours using ImageJ software; L is the longest distance from the corneal limbus to the leading edge of the neovascularization; r is the radius of the cornea. All parameters were quantified from slit-lamp images(16x magnification) using ImageJ.

Inflammatory index evaluation

The inflammatory index of the anterior segment of the eye was assessed on days 3, 7, and 14 post-alkali burn according to the grading criteria for inflammation indices [24]. The final inflammatory index value was calculated as the sum of the scores for conjunctival congestion, central corneal edema, and peripheral corneal edema, divided by 9. Statistical analysis was performed using appropriate software.

Ex vivo experiments

New Zealand white rabbit corneal diffusion experiment

Healthy adult New Zealand white rabbits (4 months old, weighing 2.0–3.0 kg) were euthanized, and the corneas were immediately excised. The corneas were mounted in a Franz diffusion cell with the epithelial side facing the donor chamber. The receptor chamber was filled with PBS at 37 °C and continuously stirred. In the donor chamber, 200 µL of Beva-containing formulation (free Beva group, (Beva&R₉)@Zn group, Beva@Zn group, and (Beva&C₂G₂R₉)@Zn group, with Beva concentration: 5 mg/mL) was added, and samples were taken from the receptor chamber at designated time points, with equal volumes of fresh medium added into receptor chamber. Beva concentration was measured by UV spectrophotometry, and the cumulative permeation amount was calculated.

Histological sectioning of rat corneal tissues

On the 7th and 14th days post-modeling, rats from each group were randomly euthanized. The eyeballs were immediately harvested and immersed in pre-cooled fixative (40% formaldehyde: acetic acid: 80% ethanol = 1:1:8, v/v/v) and fixed overnight at 4 °C. The lenses were then removed, and the eyeball tissues were subjected to gradient ethanol dehydration and transparency treatment, followed by paraffin embedding. The embedded blocks were sectioned along the sagittal plane at a thickness of 5 μm. After deparaffinization and rehydration, the sections were stained with hematoxylin and eosin (H&E), and images were captured using a Nikon Ni-U optical microscope (Japan).

Immunofluorescence staining

Immunofluorescence staining was performed to investigate the expression of VEGF-A and CD31. For VEGF-A and CD31 expression studies, rats were euthanized on the 7th and 14th days post-administration, and tissue samples were collected. Tissue sections were permeabilized with 2% Triton X-100 and blocked with 5% goat serum. They were then incubated overnight at 4 °C with anti-VEGF-A antibody (1:500) and anti-CD31 antibody (1:500). Subsequently, the sections were incubated for 2 h at room temperature with Alexa Fluor 594-conjugated goat anti-mouse secondary antibody (1:500). After immunostaining, all samples were analyzed using a laser scanning confocal microscope.

Biocompatibility study

Slit-lamp photography

Ten rats were randomly assigned to two groups. After 30 days of three times daily administration of either 0.9% NaCl or (Beva&C₂G₂R₉)@Zn (20 µL per application, Beva concentration: 2.0 mg/mL), slit-lamp images of the anterior segments of the eyes were taken under both yellow and cobalt blue light at the same magnification (16×) to observe the ocular conditions in both the control and experimental groups.

Hematoxylin and Eosin (H&E) staining

Ten SD rats were randomly divided into two groups and administered either 0.9% NaCl (control group) or (Beva&C₂G₂R₉)@Zn nanoparticle formulation three times daily for 30 days. After the treatment period, the rats were randomly euthanized, and their eyeballs were immediately harvested and fixed in pre-cooled fixative (40% formaldehyde: acetic acid: 80% ethanol = 1:1:8, v/v/v) at 4 °C overnight. The lenses were removed, and the eye tissues underwent gradient ethanol dehydration and transparency treatment, followed by paraffin embedding. The embedded blocks were sectioned at a thickness of 5 μm along the sagittal plane. The sections were deparaffinized, rehydrated, and stained with hematoxylin and eosin (H&E), followed by image capture using a Nikon Ni-U optical microscope (Japan).

Corneal endothelial tight junction ZO-1 staining

After 30 days of continuous treatment, rats were euthanized, and their eyeballs were harvested within 30 min. The eyes were immediately immersed in pre-cooled fixative (40% formaldehyde: acetic acid: 80% ethanol = 1:1:8, v/v/v) and fixed overnight at 4 °C. The corneal tissue was then separated under a dissection microscope. Immunofluorescence staining was performed as follows: the corneal tissue was permeabilized with 1% Triton X-100 and blocked with 5% goat serum, followed by overnight incubation at 4 °C with Anti-ZO1 tight junction protein rabbit polyclonal antibody (1:500 dilution). After PBS washing, the samples were incubated at room temperature for 2 h with Alexa Fluor 594-conjugated goat anti-rabbit secondary antibody (1:500 dilution). Finally, the samples were counterstained with DAPI to visualize the nuclei. The stained samples were observed, and images were captured using a laser scanning confocal microscope.

Data analysis

All data are presented as mean ± SEM from at least three independent biological replicates. The normality of data distribution was confirmed using the Shapiro-Wilk test prior to statistical analysis.For comparisons among multiple groups, one-way analysis of variance (ANOVA) was used, followed by Tukey’s post hoc test for pairwise comparisons or Dunnett’s post hoc test for comparisons against a control group.For multiple time-point comparisons, two-way ANOVA was applied, with Bonferroni correction for multiple comparisons to control type I error.Statistical significance was defined as p < 0.05. The exact number of biological replicates and animals (n) for each experiment is clearly stated in the corresponding figure legends. All statistical analyses were performed using SPSS 22.0 software, and graphs were constructed using Origin 2021 software.

Results and discussion

Preparation and optimization of Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn Nanoparticles

As shown in Fig. 1a, the solutions of Beva, Zn(NO₃)₂, C₂G₂R₉+Beva, and R₉+Beva were all clear and transparent. However, when Beva coexists with Zn²⁺, the Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn solutions appeared as milky, translucent mixtures, exhibiting a distinct Tyndall effect under red laser light. This phenomenon confirms that the coordination assembly between Beva and Zn²⁺ drives the formation of Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn nanoparticles [20].

To optimize the mass ratio of Beva to Zn²⁺, the concentration of Beva was fixed at 1 mg/mL, and the size, polydispersity index (PDI), zeta potential, and EE% of Beva@Zn nanoparticles were measured at different Beva to Zn²⁺ mass ratios (92:1, 92:2, 92:3, 92:4, 92:5, and 92:6). As listed in Table S1, with increasing Zn²⁺ mass ratio, the size and EE of Beva@Zn nanoparticles increased, while PDI and zeta potential showed no significant changes. These results suggest that a higher Zn²⁺ mass ratio facilitates the coordination interaction between Beva and Zn²⁺. Excessively large nanoparticle size could hinder corneal tissue barrier penetration, while a low EE% would lead to waste of the expensive Beva, increasing the cost significantly. After considering both the particle size and EE% of Beva@Zn, a Beva to Zn²⁺ mass ratio of 92:5 was selected as the optimal condition for preparing Beva@Zn nanoparticles. At this ratio, the Beva@Zn particle size was 265.9 ± 2.0 nm, PDI was 0.19 ± 0.03, zeta potential was + 18.8 ± 0.4 mV, and the EE% for Beva was 95.0 ± 0.4%. This ratio was also applied for subsequent preparations of (Beva&R₉)@Zn and (Beva&C₂G₂R₉)@Zn.

Further optimization was conducted by screening the mass ratio of peptides (R₉ or C₂G₂R₉) to Beva (1:3, 2:3, 3:3, 4:3, and 5:3). The particle size, PDI, zeta potential, and Beva EE% of (Beva&R₉)@Zn and (Beva&C₂G₂R₉)@Zn are presented in Table S2 and Table S3, respectively. Using the same optimization criteria, a mass ratio of 1:1 for R₉ or C₂G₂R₉ to Beva was suggested to be the optimal condition for preparing (Beva&R₉)@Zn and (Beva&C₂G₂R₉)@Zn nanoparticles. Under this condition, the (Beva&R₉)@Zn particle size was 247.7 ± 2.7 nm, PDI was 0.19 ± 0.08, zeta potential was + 18.8 ± 0.4 mV, and the EE% for Beva was 94.0 ± 0.4%; the (Beva&C₂G₂R₉)@Zn particle size was 237 ± 2.0 nm, PDI was 0.19 ± 0.07, zeta potential was + 17.3 ± 0.5 mV, and the EE% for Beva was 94.0 ± 0.1%.

The drug loading capacity percentage (LC%) or the mass percentage, is critical for enhancing therapeutic efficacy, minimizing side effects of encapsulating materials, improving patient compliance by optimizing the amount of active drug delivered with minimal carrier volume [25, 26]. Component analysis was performed by combining the UV–Vis absorbance measurement at 280 nm for quantification of Beva, UV–Vis absorbance at 490 nm for quantification of FITC-labeled peptides, and ICP-OES for the analysis of Zn²⁺ content. As shown in Fig. 1b, the mass percentages of Beva and Zn in Beva@Zn were 94.6%±0.1%, and 5.4%±0.1%, respectively. In (Beva&R₉)@Zn (Fig. 1c), the mass percentages of Beva, R₉, and Zn were 88.5%±0.1%, 6.3%±0.1%, and 5.2%±0.1%, respectively. In (Beva&C₂G₂R₉)@Zn (Fig. 1d), the mass percentages of Beva, C₂G₂R₉, and Zn were 89.5%±0.1%, 5.3%±0.1%, and 5.2%±0.1%, respectively.

The particle size monitoring over time showed that Beva@Zn (Fig. 1e) and (Beva&R₉)@Zn (Fig. 1f) remained stable (around 240–250 nm) during storage, with no significant changes observed, reflecting the good stability of Beva@Zn and (Beva&R₉)@Zn. In contrast, the particle size of (Beva&C₂G₂R₉)@Zn (Fig. 1g) exhibited a pronounced time-dependent decrease, shrinking from an initial 237 ± 2.0 nm to 122.3 ± 2.2 nm at 3 h and further to 37.8 ± 1.1 nm at 12 h. This size reduction trend was further confirmed by TEM images (Fig. 1h). Additionally, after 12 h incubation, zeta potential measurements indicated that the surface charges of Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn were + 18.8 ± 0.4 mV, + 18.3 ± 0.3 mV, and + 17.3 ± 0.5 mV, respectively suggesting no significant change in zeta potential (Figure S2a). Similarly, the PDI of Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn didn’t show obvious change as well (Figure S2b). The stability of zeta potential and PDI throughout the size evolution process can be attributed to two interrelated mechanisms: the stable surface exposure of positively charged groups maintains a stable zeta potential of nanoparticles; while the stability of PDI arises from interparticle electrostatic repulsion consistent with DLVO theory, which effectively prevents secondary aggregation and maintains a relatively stable size distribution.

The release profiles under simulated physiological conditions (pH = 7.4) are shown in Fig. 1i. In the first 6 h, Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn all exhibited rapid Beva release, with release amounts of 10.6 ± 1.2%, 30.6 ± 1.2%, and 87.1 ± 3.2%, respectively. After approximately 12 h, the release of Beva from Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn slowed down. Notably, (Beva&C₂G₂R₉)@Zn reached equilibrium in Beva release around 12 h, with a release amount of 96.9 ± 0.3%. Beva@Zn and (Beva&R₉)@Zn both reached equilibrium in Beva release after 48 h, with release amounts of 36.9 ± 3.1% and 56.9 ± 5.1%, respectively. At the same time point, (Beva&C₂G₂R₉)@Zn released significantly more Beva than Beva@Zn and (Beva&R₉)@Zn. Specifically, at equilibrium, the final release amount of (Beva&C₂G₂R₉)@Zn was 2.6 and 1.7 times higher than that of Beva@Zn and (Beva&R₉)@Zn, respectively. This may be attributed to the unique, sustained size reduction process of (Beva&C₂G₂R₉)@Zn, which likely facilitated the further release of the encapsulated drug.

The retention of Beva’s biological activity after its incorporation into nanoparticles is crucial for determining its therapeutic efficacy. To assess this, we measured the biological activity of Beva in the nanoparticles. Beva was released from the nanoparticles and its activity was determined using a Beva ELISA kit. The results showed that the Beva released from Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn maintained over 98% of its activity (Figure S3). This indicates that the synthesis method does not compromise the original biological activity of Beva, ensuring the preservation of its therapeutic potential.

Interactions in (Beva&C₂G₂R₉)@Zn

Further characterization revealed the interactions in (Beva&C₂G₂R₉)@Zn. XPS analysis (Fig. 2a) showed distinct Zn characteristic peaks at binding energies of 1044.5 eV (Zn 2p₁/₂) and 1021.7 eV (Zn 2p₃/₂) in (Beva&C₂G₂R₉)@Zn, which were not present in free Beva. The O 1s spectrum (Fig. 2b) exhibited the Zn-O bond in addition to the C-O and C = O peaks. Similarly, the N 1s spectrum (Fig. 2c) revealed the presence of the Zn-NH₂ bond alongside the C-N and C = N peaks. These results confirm that Zn²⁺ coordinates with the O and N atoms in Beva or the peptide, driving the self-assembly of the nanoparticles [27]. Moreover, the O-H/N-H stretching vibration peak in the FTIR spectra shifted from 3275 cm⁻¹ in (Beva&C₂G₂R₉)@Zn to 3282 cm⁻¹ in Beva, indicating a weakening of the original intermolecular hydrogen bond network (Fig. 2d). This phenomenon likely results from the competitive disruption of these hydrogen bonds by Zn²⁺ coordination with the residual groups in Beva, further supporting that coordination is a key driving force for nanoparticle formation [28–30].

Molecular simulation results provided additional kinetic evidence. By steeply sectioning the amino sequence of C₂G₂R₉ and R₉, the root mean square fluctuation (RMSF) of the C₂G₂R₉ peptide in (Beva&C₂G₂R₉)@Zn and R₉ peptide in (Beva&R₉)@Zn was calculated. RMSF of C₂G₂R₉ peptide in (Beva&C₂G₂R₉)@Zn ranged from 3.5 to 7.0 Å (Fig. 2e), which was higher than that of the R₉ peptide in (Beva&R₉)@Zn (4.0 to 5.0 Å), indicating greater conformational flexibility in (Beva&C₂G₂R₉)@Zn. This may be the underlying reason for the dynamic changes in particle size. Additionally, the number of hydrogen bonds formed in the (Beva&C₂G₂R₉)@Zn system was significantly greater than that in (Beva&R₉)@Zn (Fig. 2f), with the stronger hydrogen bond network further stabilizing the nanoparticle structure.

The schematic representation of the Beva protein is shown in Figs. 2g-i. Figure 2g-ii illustrates the binding sites between Beva and R₉ in (R₉&Beva)@Zn. Analysis revealed that hydrogen bonds between the two are primarily distributed in three regions: polar residues on the heavy chain of Beva, including E300 and Y306 in Fig. 2g-ii-B, and N321, Q317, T256, E386, and E388 in Fig. 2g-ii-C, which interact with the positively charged R₉, forming hydrogen bonds. Figure 2g-iii shows the binding mode of Beva and C₂G₂R₉ in (C₂G₂R₉&Beva)@Zn, where the hydrogen bonding interactions extend across four regions. Polar residues on the light chain, including S24, S26, and D28 in Fig. 2g-iii-B, and Q29 in Fig. 2i-C, form hydrogen bonds with the positively charged 9-arginine region of C₂G₂R₉. Additionally, multiple polar residues on the heavy chain, including E386, E388, and E394 in Fig. 2g-iii-D, and D255, T262, E264, E386, E388, Q392, and N395 in Fig. 2g-iii-E, also participate in hydrogen bond formation with C₂G₂R₉.

Cytotoxicity assay

Nanoparticles should exhibit good biocompatibility to enable subsequent cellular studies. The cytotoxicity of Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn in HCECs was found to be low. Even when the concentration Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn reached 200 µg/mL, cell viability remained above 90% (Figure S4a). Similar cytotoxicity results were observed in HUVEC cells (Figure S4b).

Cellular uptake and lysosome escape

To monitor whether the nanoparticles (NPs) can enhance cellular uptake of Beva, FITC-labeled Beva was used in the preparation of the nanoparticles for this experiment. Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn were co-incubated with HCECs for 24 h, and intracellular Beva was observed directly by HCLSM. As shown in Fig. 3a, (Beva&C₂G₂R₉)@Zn exhibited the strongest fluorescence intensity, followed by (Beva&R₉)@Zn, while Beva and Beva@Zn groups showed minimal green fluorescence from FITC-labeled Beva. Figure 3b demonstrates that the fluorescence intensity in the (Beva&C₂G₂R₉)@Zn group was 9.4 and 1.5 times higher than that in Beva@Zn and (Beva&R₉)@Zn groups, respectively. This indicates that (Beva&C₂G₂R₉)@Zn significantly enhances Beva uptake by HCEC cells.

Fig. 3 — Various in vitro assays to assess the biofunctions in cellular levels of Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn nanoparticles. (a) The HCLSM images of HCECs incubated with Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn for 24 h. (b) The statistical fluorescence intensity of intracellular FITC-Beva uptake by in HCECs. (c) The HCLSM images of HCECs incubated with Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn for 2 and 6 h, after staining the lysosomes. (d) Pearson correlation coefficient of lysosome and Beva, in groups of (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn. (e) The migration rate percentage of monolayer HUVECs after treating with Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn nanoparticles for 24 h. (f) The tube formation percentage of HUVECs seeded on matrix gel after treating with Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn nanoparticles for 6 h. The control was the sample treated without any Beva drug. The blue color indicated for the cell nucleus dyed with DAPI, the green for Beva labeled with FITC, and the red for the lysosome dyed with TRIC red. The shared scale bar in (a) and (c): 20 μm. All data are presented as mean ± SEM from three independent biological replicates. The symbols indicated the significance, with * for comparing with control group (no drug), *: P < 0.05; **: P < 0.01; ***: P < 0.001; ****: P < 0.0001, with ^ comparing between drug groups, ^: P < 0.05; ^^: P < 0.01; ^^^: P < 0.001; ^^^^: P < 0.0001

If protein drugs are trapped in the acidic environment of lysosomes, they are prone to inactivation and unable to exert their therapeutic effects. The ability of the protein to escape from lysosomes is crucial for its biological function [31, 32]. FITC-labeled Beva was used in this experiment to observe the lysosomal escape behavior of Beva after nanoparticle endocytosis. As shown in Fig. 3c, the Beva and Beva@Zn groups exhibited only red fluorescence from the lysosome probe inside the cells, with little to no green fluorescence from FITC-labeled Beva. In contrast, the (Beva&R₉)@Zn and (Beva&C₂G₂R₉)@Zn groups showed significant yellow fluorescence at 2 h (resulting from the overlap of red lysosomal probe and green FITC-Beva), with reduced yellow fluorescence observed at 6 h. Pearson correlation coefficient analysis (Fig. 3d) revealed that the co-localization coefficients at 6 h for (Beva&R₉)@Zn and (Beva&C₂G₂R₉)@Zn were 0.21 ± 0.01 and 0.18 ± 0.01, respectively, significantly lower than their co-localization coefficients at 2 h (0.34 ± 0.05 for (Beva&R₉)@Zn and 0.40 ± 0.07 for (Beva&C₂G₂R₉)@Zn). These results indicate that Beva in both (Beva&R₉)@Zn and (Beva&C₂G₂R₉)@Zn groups facilitated the Beva escape from lysosomes. The lysosomal escape capacity of (Beva&C₂G₂R₉)@Zn was slightly superior to that of (Beva&R₉)@Zn (Fig. 3d). The observed lysosomal escape phenotype is likely attributed to the proton sponge effect of guanidinium groups in the acidic lysosomal environment [33]. The efficient lysosomal escape property may help avoid the inactivation of Beva in the acidic lysosomal environment, which is conducive to maintaining the biological activity of the antibody, and may provide a potential mechanistic basis for its application in ocular surface disease treatment.

In vitro anti-migration and anti-tube formation effects on HUVECs

Endothelial cell migration is a critical step in angiogenesis. The migration inhibitory effect on HUVECs was assessed using a scratch assay. Microscopic images of the scratch at 0 and 24 h are shown in Figure S5, and the percentage of scratch closure is presented in Fig. 3e. The results from Figure S5 and Fig. 3e collectively demonstrate that, compared to the control group (without Beva), the drug treatment groups (free Beva, Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn) significantly reduced the migration ability of HUVECs. This indicates that all drug groups inhibit the in vitro migration of endothelial cells. Among the drug groups, the order of inhibitory potency on endothelial cell migration was as follows: (Beva&C₂G₂R₉)@Zn > (Beva&R₉)@Zn > Beva@Zn > free Beva.

The ability of endothelial cells to form tubular structures is another key factor in evaluating the anti-angiogenic effect of a drug in vitro [34]. After 6 h of treatment, HUVECs were imaged (Figure S6). In the control group, HUVECs formed distinct lumen structures. The different drug treatment groups (free Beva, Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn) exhibited various cellular morphologies. In the Beva group, some cells showed lumen structures, and several cells exhibited budding or budding tendencies. The Beva@Zn group showed similar results, but fewer lumen structures were observed. In contrast, in the (Beva&R₉)@Zn and (Beva&C₂G₂R₉)@Zn groups, nearly all HUVECs were in a tube formation-inhibited state, with no apparent lumen structures observed. (Beva&R₉)@Zn and (Beva&C₂G₂R₉)@Zn exhibited significant inhibitory effects on tube formation in vitro. The statistical analysis of tube formation also supported these findings. As shown in Fig. 3f, the drug treatment groups significantly inhibited HUVEC tube formation, with (Beva&C₂G₂R₉)@Zn showing the strongest inhibitory effect among all the drug groups. Notably, (Beva&C₂G₂R₉)@Zn exhibited a significantly better inhibitory effect on HUVEC tube formation compared to (Beva&R₉)@Zn.

Collectively, results from scratch and tube formation assays demonstrate that nanocarrier formulations—particularly (Beva&C₂G₂R₉)@Zn—exhibit significantly enhanced anti-angiogenic efficacy compared to free Beva. Furthermore, based on the intracellular synthesis–secretion pathway of VEGF-A [35] and the demonstrated capacity of nanoparticles to deliver the mAb inside the cell for targeting intracellular components [36], we propose a mechanistic hypothesis that nanocarrier-mediated intracellular delivery of Beva may additionally enable premature intracellular neutralization of nascent VEGF-A, thereby preventing its secretion into the extracellular space and attenuating angiogenic signaling. However, this hypothesis requires further direct experimental validation, and its contribution to the overall anti-angiogenic efficacy remains to be clarified.

Penetration across ocular barriers and retention on ocular surface

Drug delivery across the corneal barrier is crucial for the treatment of corneal neovascularization, particularly for protein-based drugs [37]. The cornea, as a transparent ocular tissue, functions as a robust barrier, with tight junctions between the epithelial and endothelial layers that restrict the penetration of macromolecules [38]. Protein drugs, due to their large molecular size and complex structure, have difficulty penetrating this barrier [39]. Therefore, it is essential to thoroughly evaluate the ability of these nanoparticles (Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn) to cross the corneal barrier. An in vitro reconstructed 3D corneal model was first utilized, as it closely mimics the stratified structure and physiological metabolic functions of the native corneal epithelium, making it a suitable substitute model for evaluating drug penetration. FITC-labeled Beva was used to observe the distribution of Beva within the 3D corneal model. As shown in Fig. 4a, no fluorescence was observed in the free Beva sample, indicating that free Beva has no ability to penetrate the 3D corneal model. The Beva@Zn nanoparticles predominantly penetrated the superficial epithelial layer. In contrast, (Beva&R₉)@Zn and (Beva&C₂G₂R₉)@Zn groups displayed widespread green fluorescence throughout the entire epithelial cell model, indicating significantly enhanced penetration. Fluorescence intensity analysis (Fig. 4b) showed that the amount of Beva penetration in (Beva&R₉)@Zn and (Beva&C₂G₂R₉)@Zn groups was 11.2 and 14.6 times higher than that of free Beva, respectively, and 2.7 and 3.7 times higher than that of Beva@Zn. Clearly, both (Beva&R₉)@Zn and (Beva&C₂G₂R₉)@Zn exhibited excellent penetration capabilities. Compared to (Beva&R₉)@Zn, (Beva&C₂G₂R₉)@Zn showed even better performance in the 3D corneal model, with a statistically significant difference.

Fig. 4 — Penetration across ocular barriers and retention on ocular surface. (a) The HCLSM images of 3D multilayer corneal epithelial, and (b) the statistic fluorescence intensity, after incubating with Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn for 24 h. (c) The schematic illustration of the setup of Franz diffusion cell. The upper part serves as the donor cell for placing the drug solution, while the fixed fresh rabbit cornea is positioned in the center, and the receptor cell at the bottom collects the penetrated drug through the cornea. (d) The cumulative percentage of Beva penetrated across the cornea of different groups: Beva, Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn. (e) The fluorescent images of the rat corneas, and (f) Statistical analysis on the percentage of residual fluorescence intensity on the cornea at different time points after instilling free Beva, Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn, by the multimodal animal imaging system. The blue color indicated for the cell nucleus dyed with DAPI, the green for Beva labeled with FITC (n = 5 per group). The shared scale bar in (a) is 50 μm. The volume of eyedrop is 20 µL containing 40 µg Beva in (e). All data are presented as mean ± SEM from three independent biological replicates (n = 5 per group). The symbols indicated the significance, with * for comparing with control group (no drug), *: P < 0.05; **: P < 0.01; ***: P < 0.001; ****: P < 0.0001, with ^ comparing between drug groups, ^: P < 0.05; ^^: P < 0.01; ^^^: P < 0.001; ^^^^: P < 0.0001

To further investigate the penetration ability of these nanoparticles through a full-thickness cornea, fresh New Zealand rabbit corneas were mounted in a Franz diffusion cell, as shown in Fig. 4c. The upper part of the diffusion cell serves as the donor cell for placing the drug solution, while the fixed fresh rabbit cornea is positioned in the center, and the receptor cell at the bottom collects the drug that penetrates the cornea. By measuring the amount of Beva in the receptor cell over time and calculating the percentage of Beva that passed through, the ability of free Beva, Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn to penetrate biological tissue with an intact corneal structure was evaluated. As shown in Fig. 4d, the Beva penetration for each group was time-dependent. Compared to the free Beva group, the Beva penetration in the Beva@Zn group did not show a significant difference, while the Beva penetration in (Beva&R₉)@Zn and (Beva&C₂G₂R₉)@Zn groups significantly increased. Notably, (Beva&C₂G₂R₉)@Zn exhibited higher diffusion than (Beva&R₉)@Zn. After 6 h, the permeability of (Beva&C₂G₂R₉)@Zn was 18.9 ± 0.70%, which was 11.1, 6.3, and 1.6 times greater than that of the free Beva, Beva@Zn, and (Beva&R₉)@Zn groups, respectively. This improvement is likely due to the smaller nanoparticle size of (Beva&C₂G₂R₉)@Zn, which facilitates better penetration through the corneal tissue, further demonstrating its superior performance in penetrating corneal biological tissues.

The retention time of drug formulations at the anterior cornea is a key parameter for the bioavailability of drugs delivered via eye drops [40]. A longer retention time increases the likelihood of drug penetration through the cornea, improving bioavailability and therapeutic outcomes [41, 42]. Fluorescently labeled Beva was used to evaluate the retention time of Beva-NPs in the anterior cornea of rats via a multimodal animal imaging system. Figure 4e shows fluorescence images of the rat corneas at different time points after instilling free Beva, Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn. Figure 4f provides a statistical analysis of residual fluorescence intensity on corneal, offering a clearer understanding of Beva retention on the cornea. Compared to free Beva, all nanoparticle formulations (Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn) exhibited prolonged retention times. For instance, at 0.5 h, the average retention for the free Beva group was only about 23.6% of the initial dose, while the retention rates for all Beva-NPs groups were all above 70%. At 3 h, fluorescence from the free Beva group was almost undetectable on the cornea, whereas the retention rates for Beva-NPs groups remained above 30%. The difference in corneal retention between Beva-NPs and free Beva may be attributed to the positively charged surface potential of the Beva-NPs, which enhances interaction with the negatively charged ocular mucin. Notably, at all tested time points, (Beva&R₉)@Zn and (Beva&C₂G₂R₉)@Zn demonstrated superior retention compared to Beva@Zn, likely due to the introduction of peptides (R₉ or C₂G₂R₉), which further enhanced the interaction of (Beva&R₉)@Zn and (Beva&C₂G₂R₉)@Zn with ocular mucin. In comparison to (Beva&R₉)@Zn, the retention ability of (Beva&C₂G₂R₉)@Zn in the cornea was slightly reduced, though no significant difference was observed. This may be attributed to the rapid reduction in nanoparticle size in the (Beva&C₂G₂R₉)@Zn group, as smaller particles are more easily washed away by tear fluid, leading to a decrease in retention time. The above results indicate that (Beva&C₂G₂R₉)@Zn not only exhibited enhanced permeability but also demonstrated superior retention in the anterior corneal tissues. This combination allowed Beva to penetrate more deeply into the cornea, thereby leading to improved therapeutic outcomes.

Evaluation of therapeutic efficacy in the rat alkali burn CNV model

A rat alkali burn model was established to further assess their in vivo therapeutic efficacy. Following eye drop administration, the corneal morphology was dynamically observed at different time points using a slit-lamp (Fig. 5a). The results showed that by day 3 post-modeling, all experimental groups exhibited varying degrees of corneal opacity and edema. By day 7, the edema in all groups had generally decreased, with the (Beva&C₂G₂R₉)@Zn group showing the most significant improvement. Additionally, neovascularization extended from the corneal limbus to the central area in the Control, Free Beva, Beva@Zn, and (Beva&R₉)@Zn groups, while no noticeable neovascularization was observed in the (Beva&C₂G₂R₉)@Zn group. By day 14, except for the (Beva&C₂G₂R₉)@Zn group, corneal edema had not fully resolved in the other groups, and neovascularization had worsened compared to day 7, especially in the Control group, where a large number of new blood vessels extended into the central cornea. In contrast, the (Beva&C₂G₂R₉)@Zn group exhibited near complete resolution of corneal edema, transparency restoration, and no apparent neovascularization.

Quantitative analysis of the inflammation index revealed that, on day 3, the inflammation score for the (Beva&C₂G₂R₉)@Zn group was marginally lower than that of the other groups (Fig. 5b). By day 7, the reduction in inflammation was most pronounced in the (Beva&C₂G₂R₉)@Zn group, which exhibited a statistically significant decrease compared to the Control group. This trend was sustained, with the inflammation index remaining at the lowest level in the (Beva&C₂G₂R₉)@Zn group on day 14. Over the entire study period (days 3–14) (Fig. 5b), the inflammation index of the (Beva&C₂G₂R₉)@Zn group was significantly different from those of the Control, Free Beva, Beva@Zn, and (Beva&R₉)@Zn groups. In contrast, no significant differences were observed between the Control, Free Beva, Beva@Zn, and (Beva&R₉)@Zn groups.

With respect to neovascularization area (Fig. 5c), on day 7, the (Beva&C₂G₂R₉)@Zn group displayed minimal neovascularization, with a significantly smaller neovascularization area compared to the other groups. No significant differences were noted between the Control, Free Beva, Beva@Zn, and (Beva&R₉)@Zn groups. On day 14, while all treatment groups exhibited inhibitory effects on neovascularization (with inhibition rates of 44.7 ± 1.1% for Free Beva, 40.5 ± 2.0% for Beva@Zn, and 59.3 ± 3.2% for (Beva&R₉)@Zn), the (Beva&C₂G₂R₉)@Zn group demonstrated the most significant inhibition, achieving an inhibition rate of 99.9 ± 0.1%. Further statistical analysis revealed that no significant difference in inhibitory effects was observed between Free Beva and Beva@Zn. The significant differences were observed between Beva@Zn and (Beva&R₉)@Zn, as well as between (Beva&R₉)@Zn and (Beva&C₂G₂R₉)@Zn.

H&E staining results (Fig. 5d) showed that by day 7, all experimental groups exhibited varying degrees of edema. Additionally, in the Control, Free Beva, and Beva@Zn groups, small amounts of neovascularization formed by endothelial cells were observed, with red blood cells visible within the lumen (indicated by black arrows), suggesting active angiogenesis. By day 14, the Control, Free Beva, Beva@Zn, and (Beva&R₉)@Zn groups still exhibited abundant new blood vessels with clear lumen structures. In contrast, the (Beva&C₂G₂R₉)@Zn group showed no distinct lumen formation. These histological observations were consistent with the slit-lamp findings, further confirming the significant anti-neovascularization effect of (Beva&C₂G₂R₉)@Zn.

Inhibitory effect on the expression of VEGF and CD31

VEGF plays a pivotal role in neovascularization by promoting endothelial cell proliferation and migration, enhancing vascular permeability, and facilitating the formation and maturation of new blood vessels [35]. Additionally, VEGF acts synergistically with other growth factors to further support the process of angiogenesis [43]. Molecular-level validation was performed through VEGF immunofluorescence staining, with the indication by white arrows (Fig. 6a). VEGF signal was observed in all groups, but much weaker in (Beva&C₂G₂R₉)@Zn group. The fluorescence intensity was statistically analyzed and reported in Fig. 6b. In the (Beva&C₂G₂R₉)@Zn group remained the lowest at both day 7 and day 14 (Fig. 6b), indicating the most complete inhibition of VEGF expression. CD31 is a key marker of endothelial cells, primarily expressed on the surface of endothelial cells, particularly in blood vessel endothelium [44]. VEGF promotes the formation of new blood vessels by stimulating endothelial cell proliferation and migration, while also upregulating the expression of CD31 [45, 46]. This upregulation enhances the intercellular connections between endothelial cells, thereby promoting the stability and maturation of newly formed blood vessels. As a marker of neovascularization, CD31 helps maintain the integrity of vascular structures and works synergistically with VEGF in the process of angiogenesis. CD31 staining experiment was carried and the results were shown in Figs. 6c. Similar with the expression of VEGF, the fluorescent signal of CD31 in the (Beva&C₂G₂R₉)@Zn group was weakest, comparing with other groups, which could be confirmed by the statistical analysis in fluorescence intensity (Fig. 6d). Both the results of expression of VEGF and CD31 suggested that no noticeable new blood vessel lumen structures were present throughout the study period. The statistical analysis of the expressions of VEGF (Fig. 6b) and CD31 (Fig. 6d) showed the similar results, which were well consistent with the results in neovascularization area (Fig. 5c), no significant difference between Free Beva and Beva@Zn, significant differences between Beva@Zn and (Beva&R₉)@Zn, as well as between (Beva&R₉)@Zn and (Beva&C₂G₂R₉)@Zn.

Fig. 6 — (a) HCLSM images of immunofluorescence staining VEGF, (b) the statistic analysis of the expression level of VEGF, (c) HCLSM images of immunofluorescence staining CD 31, and (d) the statistical analysis of the expression level of CD31, at day 7 and 14. White arrows indicate VEGF (a) and CD31 (d) staining. The shared scale bar in (a) and (d) are 50 μm. SD rats with CNV model were administered eye drops three times daily, with 20 µL each time for 7 consecutive days (n = 15 per group, n = 5 per group per time point). All drug groups containing Beva 40 µg per time. The blue color indicated for the cell nucleus dyed with DAPI, the red color in (a) and (d) indicated the expression of VEGF stained with anti-VEGF and CD31 stained with anti-CD31 antibody, respectively. All data are presented as mean ± SEM from three independent biological replicates (n = 5 per group per time point). The symbols indicated the significance, with * for comparing with control group (no drug), *: P < 0.05; **: P < 0.01; ***: P < 0.001; ****: P < 0.0001, with ^ comparing between drug groups, ^: P < 0.05; ^^: P < 0.01; ^^^: P < 0.001; ^^^^: P < 0.0001

Notably, all Beva-NPs groups exhibited better inhibition of neovascularization compared to free Beva across all evaluation parameters (corneal morphology, inflammation index, neovascularization area, H&E staining, VEGF-A and CD31 immunofluorescence staining). Among the Beva-NPs groups, the inhibition of neovascularization was strongest in the (Beva&C₂G₂R₉)@Zn group, followed by (Beva&R₉)@Zn, and Beva@Zn. (Beva&C₂G₂R₉)@Zn demonstrated the most optimal therapeutic effect in the alkali-burned corneal model, not only rapidly alleviating edema but also sustaining and efficiently inhibiting neovascularization. Its excellent performance can be attributed to the unique dynamic evolution characteristics of the nanoparticles, their superior epithelial penetration ability, and the synergistic effect between Beva drug and tissue retention, leading to the highest bioavailability and therapeutic efficacy.

Biocompatibility

Considering the excellent therapeutic efficacy of (Beva&C₂G₂R₉)@Zn, its long-term biocompatibility was thoroughly evaluated. A 30-day continuous ocular administration of (Beva&C₂G₂R₉)@Zn to healthy SD rats revealed no significant ocular adverse reactions (Fig. 7a) in the anterior segment. Specifically, there were no signs of inflammation in the anterior segment, the cornea remained transparent, the iris texture was clear, the pupil responded normally to light, the anterior chamber depth was normal, and no lens opacity was observed. Sodium fluorescein staining confirmed the integrity of the corneal epithelium (Fig. 7a, blue colored panel). ZO-1 (Zonula Occludens-1) is a key component of tight junctions in corneal epithelial cells, primarily responsible for maintaining intercellular tight junctions and barrier function, thus protecting the cornea from external harmful substances [47]. The status of ZO-1 is frequently used as a potential biomarker for corneal diseases, serving as a tool for disease assessment and monitoring therapeutic efficacy. ZO-1 immunofluorescence staining further demonstrated that the corneal endothelial cells maintained a regular hexagonal mosaic pattern, with tight junction proteins well preserved (Fig. 7b), in both control and (Beva&C₂G₂R₉)@Zn groups. Additionally, H&E staining of the cornea and retina showed no structural differences compared to the control group (Fig. 7c). Histological analysis of major organs (heart, liver, spleen, and kidneys) revealed no significant deposition of immune complexes, infiltration of inflammatory cells, or tissue structural damage (Fig. 7d). These results indicated that (Beva&C₂G₂R₉)@Zn nanoparticles exhibit good biocompatibility and safety in vivo, highlighting their potential for further applications.

Fig. 7 — Biocompatibility evaluation of (Beva&C₂G₂R₉)@Zn. Healthy SD rats were administered eye drops three times daily, with 20 µL Beva&C₂G₂R₉@Zn (Beva concentration: 2.0 mg/mL), for 30 consecutive days (n = 5 per group). (a) Slit-lamp images of the ocular anterior segment: without sodium fluorescein (uncolored) and with sodium fluorescein stained (blue colored); (b) ZO-1 immunofluorescence staining of the cornea; (c) H&E staining of corneal (left) and retinal (right) tissue sections; (d) H&E staining images of major organs. The control group consisted of healthy rats at day 0, while (Beva&C₂G₂R₉)@Zn represents healthy rats treated with eye drops for 30 days. Scale bars: 500 μm (a), 20 μm (b), 100 μm (c, d)

Discussion

To highlight the advantages of this strategy, Table 1 summarizes the reported strategies for delivering Beva in the treatment of CNV, including the animal models used, drug administration regimens, Beva dosage, and therapeutic outcomes. Compared to strategies involving the co-assembly of His₆ or its derived peptides with Beva, the (Beva&C₂G₂R₉)@Zn nanoparticles demonstrate a significantly lower drug dosage and improved therapeutic efficacy. When compared to Beva-based protein nanoparticle strategies, this approach also shows a reduced drug dose and more favorable treatment outcomes. Although invasive subconjunctival injection-based strategies, such as thermosensitive hydrogel technology, or semi-invasive methods like microneedle-based drug delivery, have been reported to provide better suppression of CNV compared to most other results, (Beva&C₂G₂R₉)@Zn nanoparticles still exhibit lower drug doses and better therapeutic efficacy. Notably, under the rat alkali burn-induced CNV model, the (Beva&C₂G₂R₉)@Zn nanoparticles achieved nearly 100% inhibition of CNV area at day 14 post-treatment with a relatively low Beva dose, which is among the highest inhibitory efficacies reported for non-invasive anti-VEGF delivery systems for CNV treatment to date. It should be noted that direct cross-study comparisons are limited by inherent differences in animal models, dosing regimens, treatment duration, and CNV quantification methods across published reports. This outcome underscores the promising translational potential of (Beva&C₂G₂R₉)@Zn nanoparticles for the treatment of CNV.

Table 1.

Comparing this work with the other reports from the views: drug carrier, animal model, administration way and Beva dosage, and therapeutic outcomes

Reference number	Drug carrier	Animal model	Administration way and Beva dosage	Therapeutic outcomes: CNV area percentage
[48]	HmA@Beva nanoparticles Co-assembled by His₆ peptide and Beva	Rat alkali burn-induced CNV	Eye drops, three times a day, 100 µg per dose, for 7 consecutive days	Day 7: approximately 22%; Day 14: approximately 30%.
[21]	YHmA@Beva nanoparticles Co-assembled by YHis₆ peptide and Beva	Rat alkali burn-induced CNV	Eye drops, three times a day, 50 µg per dose, for 7 consecutive days	Day 7: approximately 5%; Day 14: approximately 1%.
[49]	CHmA@Beva nanoparticles Co-assembled by (CHR)₃ peptide and Beva	Rat alkali burn-induced CNV	Eye drops, three times a day, 50 µg per dose, for 7 consecutive days	Day 7: approximately 8%; Day 14: approximately 20%.
[19]	Human serum albumin and Beva co-precipitated nanoparticle induced by ethanol desolvation	Rat alkali burn-induced CNV	Eye drops, three times a day, 40 µg per dose, for 7 consecutive days	Day 7: approximately 9%; no further evaluations after day 7.
[20]	Beva-G₄ nanoparticles Co-assembled by 4th generation polyamindoamine (G₄) and Beva	Rat alkali burn-induced CNV	Eye drops, three times a day, 100 µg per dose, for 7 consecutive days	Day 7: approximately 50%; Day 14: approximately 15%.
[50]	Poly(ethylene glycol)-poly(ɛ-caprolactone)-poly(ethylene glycol) thermosensitive hydrogel loaded with Beva	Rabbit surgical suture-induced CNV	Invasive subconjunctival injection, a single dose of 2.5 mg, administered once	Day 15: approximately 7%; Day 19: approximately 11%.
[18]	Polylactic acid-hyaluronic acid microneedles loaded with Beva	Rabbit surgical suture-induced CNV	Semi-invasive microneedle, every 3 days, 0.25 mg per dose, for 21 consecutive days.	Day 14: approximately 3%; Day 21: approximately 4%.
This work	(Beva&C₂G₂R₉)@Zn nanoparticles Co-assembled by C₂G₂R₉ peptide and Beva	Rat alkali burn-induced CNV	Eye drops, three times a day, 40 µg per dose, for 7 consecutive days	Day 7: approximately 0.6%; Day 14: approximately 0.1%.

Open in a new tab

The exceptional anti-CNV effects of (Beva&C₂G₂R₉)@Zn nanoparticles can likely be attributed to their size-reducing property over time. This feature allows the nanoparticles to be rapidly internalized by cells and efficiently penetrate the corneal barrier (Fig. 3). Importantly, the gradual reduction in size does not negatively impact their retention on the ocular surface (Fig. 4), which significantly increases their chances of penetrating the cornea. Once internalized, (Beva&C₂G₂R₉)@Zn nanoparticles disassemble within mature lysosomes, where the cationic charge of the R₉ peptide in the C₂G₂R₉ sequence can potentially disrupt the lysosomal membrane, allowing Beva to escape from the lysosome (Fig. 3). These characteristics collectively endow (Beva&C₂G₂R₉)@Zn nanoparticles with excellent anti-CNV capabilities.

In terms of clinical translational potential, (Beva&C₂G₂R₉)@Zn nanoparticles present clear advantages. The nanoparticles exhibit a high encapsulation efficiency for Beva, with a drug loading capacity of 89.5%, an encapsulation efficiency of 94.0%, and Beva retaining over 98% of its bioactivity. These properties not only help reduce the cost of the expensive Beva drug but also facilitate achieving the therapeutic drug concentrations required for effective treatment. Furthermore, (Beva&C₂G₂R₉)@Zn nanoparticles almost completely suppressed CNV, achieving near 100% inhibition. In addition, they demonstrate rapid anti-inflammatory effects, as evidenced by a significantly lower inflammation score after 7 days, indicating their potential to suppress ocular inflammation. These profiles will benefit the clinical translation of (Beva&C₂G₂R₉)@Zn nanoparticles.

There are several limitations of this study that should be explicitly acknowledged. First, although our results confirmed that C₂G₂R₉ peptide alone cannot form nanoparticles in the absence of Zn²⁺, the potential independent biological effects of free C₂G₂R₉ peptide—including cellular uptake, corneal penetration, and in vivo therapeutic efficacy—were not investigated in this study. Therefore, we cannot completely rule out the possibility that free C₂G₂R₉ peptide may contribute to the observed biological effects. This represents a limitation of the present work. Second, clinical translation still requires further systematic safety evaluation. Although this study assessed the ocular safety of (Beva&C₂G₂R₉)@Zn nanoparticles after 30 days of continuous eye drop administration, including evaluations of the anterior and posterior segments of the eye (retina) as well as major organs, these assessments remain relatively simplistic and do not fully meet the rigorous safety standards required for clinical translation. Given that the eye is a relatively closed and highly sensitive organ, the long-term safety of Zn²⁺ and the C₂G₂R₉ peptide requires further investigation. Specifically, although the Zn content in the (Beva&C₂G₂R₉)@Zn nanoparticles administrated during CNV treatment (0.06 µg/daily) is far lower than the recommended daily Zn intake in a healthy diet (2–11 mg for children, and 8–12 mg for adults), Zn²⁺ may accumulate slowly with long-term repeated ocular administration, potentially leading to neurotoxicity. Additionally, the C₂G₂R₉ peptide may induce potential immune responses after long-term administration. Third, the quantification of Beva in encapsulation and release assays was performed using UV–Vis absorbance at 280 nm, which is a non-specific total protein detection method and cannot specifically differentiate intact Beva from other peptide or protein components. Accordingly, all related values represent relative estimates rather than absolutely specific quantification, which constitutes a methodological limitation of this study. These are critical concerns that must be addressed prior to the clinical application of (Beva&C₂G₂R₉)@Zn nanoparticles.

Conclusion

In summary, we have developed a (Beva&C₂G₂R₉)@Zn nanoparticle system with a size-reducing property over time for Beva delivery across the corneal barrier to treat CNV. The encapsulation efficiency of Beva in (Beva&C₂G₂R₉)@Zn nanoparticles is 94.0%, with a drug loading capacity of 89.5%, and the retained bioactivity of Beva is 98.5%. The size of (Beva&C₂G₂R₉)@Zn nanoparticle shrunk from an initial 237 ± 2.0 nm to 122.3 ± 2.2 nm at 3 h and further to 37.8 ± 1.1 nm at 12 h, while the Beva@Zn and (Beva&R₉)@Zn nanoparticles were stable around 240–250 nm. Compared to size-stable nanoparticles of Beva@Zn and (Beva&R₉)@Zn, (Beva&C₂G₂R₉)@Zn nanoparticles are rapidly internalized by cells, and the encapsulated Beva efficiently escapes from lysosomes, significantly inhibiting HUVEC migration and tube formation. Moreover, (Beva&C₂G₂R₉)@Zn nanoparticles can quickly penetrate the corneal barrier and retain at the ocular surface for prolonged periods. In the rat alkaline burn corneal neovascularization model, (Beva&C₂G₂R₉)@Zn nanoparticles demonstrated far superior CNV suppression compared to Beva@Zn and (Beva&R₉)@Zn nanoparticles, with nearly 100% inhibition and the lowest inflammation scores. This study presents a novel design strategy for the development of non-invasive corneal drug delivery systems for protein-based therapeutics in the future.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 1^{(661.8KB, docx)}

Author contributions

X. Xie: Designed the study, conducted most experiments, analyzed the data, and wrote the original manuscript. Z. Xiang: Contributed to the study design, and performed critical data analysis. Ruhui Yang: Performed partial experimental work. R. Fang: Helped the experimental design. X. Zan: Assisted with data collection and analysis, reviewed and revised the manuscript. L. Wu: Contributed to manuscript revision. W. Huang: Led the conceptualization of the project, and acquired the funding. Y. Tong: Supervised the research, provided key intellectual input, and acquired the funding. All authors have read and approved the final manuscript for submission.

Funding

This work was financially supported by National Natural Science Foundation of China (82401294), Zhejiang Provincial Medical and Health Science and Technology Project (2025KY426, 2025KY436), Zhejiang Provincial Science and Technology Program of Traditional Chinese Medicine (2024ZL1212), Zhejiang Provincial Natural Science Foundation (ZCLMS25H1201).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Contributor Information

Xingjie Zan, Email: zanxj@ucas.ac.cn.

Wenjuan Huang, Email: huangwj5388@enzemed.com.

Yuhua Tong, Email: yuhuatong@wmu.edu.cn.

References

1.Nicholas MP, Mysore N. Corneal neovascularization. Exp Eye Res. 2021;202:108363–9. [DOI] [PubMed] [Google Scholar]
2.Lee P, Wang CC, Adamis AP. Ocular neovascularization: an epidemiologic review. Surv Ophthalmol. 1998;43(3):245–69. [DOI] [PubMed] [Google Scholar]
3.Roshandel D, et al. Current and emerging therapies for corneal neovascularization. Ocul Surf. 2018;16(4):398–414. [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Stevenson W, et al. Corneal neovascularization and the utility of topical VEGF inhibition: ranibizumab (Lucentis) vs bevacizumab (Avastin). Ocul Surf. 2012;10(2):67–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Kaur S, et al. Molecular nature of ocular surface barrier function, diseases that affect it, and its relevance for ocular drug delivery. Ocul Surf. 2023;30:3–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Gaballa SA, et al. Corticosteroids in ophthalmology: drug delivery innovations, pharmacology, clinical applications, and future perspectives. Drug Deliv Transl Res. 2021;11(3):866–93. [DOI] [PubMed] [Google Scholar]
7.Lin X, et al. Nanomedicine-Based Ophthalmic Drug Delivery Systems for the Treatment of Ocular Diseases. Int J Nanomed. 2025;20:9221–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Cheng Y, et al. Biomimetic Metal-Organic Antioxidant Eye Drops for Corneal Neovascularization: Visible-Light-Triggered CO Release for Microenvironment Regulation. Nano Lett. 2025;25(30):11703–12. [DOI] [PubMed] [Google Scholar]
9.Wu H, et al. A SU6668 pure nanoparticle-based eyedrops: toward its high drug Accumulation and Long-time treatment for corneal neovascularization. J Nanobiotechnol. 2024;22(1):290. [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Ahmed S, Amin MM, Sayed S. Ocular Drug Delivery: a Comprehensive Review. AAPS PharmSciTech. 2023;24(2):66. [DOI] [PubMed] [Google Scholar]
11.Öztürk K, Kaplan M, Çalış S. Effects of nanoparticle size, shape, and zeta potential on drug delivery. Int J Pharm. 2024;666:124799. [DOI] [PubMed] [Google Scholar]
12.Di J, et al. Size, shape, charge and stealthy surface: Carrier properties affect the drug circulation time in vivo. Asian J Pharm Sci. 2021;16(4):444–58. [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Han H, et al. Polymer- and lipid-based nanocarriers for ocular drug delivery: Current status and future perspectives. Adv Drug Deliv Rev. 2023;196:114770. [DOI] [PubMed] [Google Scholar]
14.Xu J, et al. Applications and challenges of ultra-small particle size nanoparticles in tumor therapy. J Control Release. 2023;353:699–712. [DOI] [PubMed] [Google Scholar]
15.Beirampour N et al. Formulation of Polymeric Nanoparticles Loading Baricitinib as a Topical Approach in Ocular Application. Pharmaceutics 2024;16(8):1092. [DOI] [PMC free article] [PubMed]
16.Mandal A, et al. Polymeric micelles for ocular drug delivery: From structural frameworks to recent preclinical studies. J Control Release. 2017;248:96–116. [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Lu Z, et al. Activated aggregation strategies to construct size-increasing nanoparticles for cancer therapy. Wiley Interdiscip Rev Nanomed Nanobiotechnol. 2023;15(2):e1848. [DOI] [PubMed] [Google Scholar]
18.Zhang Z, et al. Highly Efficient Intrastromal Dual-Drug Delivery by a Dissolving Bilayer Microneedle for Synergistic Therapy against Corneal Neovascularization. ACS Nano. 2025;19(42):37065–81. [DOI] [PubMed] [Google Scholar]
19.Luis de Redin I, et al. In vivo effect of bevacizumab-loaded albumin nanoparticles in the treatment of corneal neovascularization. Exp Eye Res. 2019;185:107697–708. [DOI] [PubMed] [Google Scholar]
20.Xu H, et al. Engineering Matrix-Free Drug Protein Nanoparticles with Promising Penetration through Biobarriers for Treating Corneal Neovascularization. ACS Nano. 2024;18(11):8209–28. [DOI] [PubMed] [Google Scholar]
21.Tang B, et al. Decorating hexahistidine-metal assemblies with tyrosine enhances the ability of proteins to pass through corneal biobarriers. Acta Biomater. 2022;153:231–42. [DOI] [PubMed] [Google Scholar]
22.Lee J, et al. CHARMM-GUI Input Generator for NAMD, GROMACS, AMBER, OpenMM, and CHARMM/OpenMM Simulations Using the CHARMM36 Additive Force Field. J Chem Theory Comput. 2016;12(1):405–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Huang J, et al. CHARMM36m: an improved force field for folded and intrinsically disordered proteins. Nat Methods. 2017;14(1):71–3. [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Laria C, Alio JL, Ruiz-Moreno JM. Combined non-steroidal therapy in experimental corneal injury. Ophthalmic Res. 1997;29(3):145–53. [DOI] [PubMed] [Google Scholar]
25.Zheng Y, et al. Remote drug loading into liposomes via click reaction. Mater Horiz. 2022;9(7):1969–77. [DOI] [PubMed] [Google Scholar]
26.Yang C, et al. A facile doxorubicin-dichloroacetate conjugate nanomedicine with high drug loading for safe drug delivery. Int J Nanomed. 2018;13:1281–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Kaur A, et al. Characterization of Cu(2+) and Zn(2+) binding sites in SUMO1 and its impact on protein stability. Int J Biol Macromol. 2020;151:204–11. [DOI] [PubMed] [Google Scholar]
28.Zou Z, et al. Zn(2+) -Coordination-Driven RNA Assembly with Retained Integrity and Biological Functions. Angew Chem Int Ed Engl. 2021;60(42):22970–6. [DOI] [PubMed] [Google Scholar]
29.Harding MM. The architecture of metal coordination groups in proteins. Acta Crystallogr D Biol Crystallogr. 2004;60(Pt 5):849–59. [DOI] [PubMed] [Google Scholar]
30.Ramirez-Bello V et al. Zinc and Copper Ions Induce Aggregation of Human β-Crystallins. Molecules 2022;27(9):2970. [DOI] [PMC free article] [PubMed]
31.Pei D, Buyanova M. Overcoming Endosomal Entrapment in Drug Delivery. Bioconjug Chem. 2019;30(2):273–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Levy JMM, Towers CG, Thorburn A. Targeting autophagy in cancer. Nat Rev Cancer. 2017;17(9):528–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Koren E, Torchilin VP. Cell-penetrating peptides: breaking through to the other side. Trends Mol Med. 2012;18(7):385–93. [DOI] [PubMed] [Google Scholar]
34.Vailhé B, et al. The formation of tubular structures by endothelial cells is under the control of fibrinolysis and mechanical factors. Angiogenesis. 1998;2(4):331–44. [DOI] [PubMed] [Google Scholar]
35.Ferrara N, Gerber HP, LeCouter J. The biology of VEGF and its receptors. Nat Med. 2003;9(6):669–76. [DOI] [PubMed] [Google Scholar]
36.Singh SR, et al. Intravenous transferrin, RGD peptide and dual-targeted nanoparticles enhance anti-VEGF intraceptor gene delivery to laser-induced CNV. Gene Ther. 2009;16(5):645–59. [DOI] [PMC free article] [PubMed] [Google Scholar]
37.Dastjerdi MH, et al. Corneal penetration of topical and subconjunctival bevacizumab. Invest Ophthalmol Vis Sci. 2011;52(12):8718–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
38.Noske W, et al. Tight junctions of the human corneal endothelium: morphological and electrophysiological features. Ger J Ophthalmol. 1994;3(4–5):253–7. [PubMed] [Google Scholar]
39.Mandal A, et al. Ocular delivery of proteins and peptides: Challenges and novel formulation approaches. Adv Drug Deliv Rev. 2018;126:67–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
40.Silva B, et al. Colloidal nanosystems with mucoadhesive properties designed for ocular topical delivery. Int J Pharm. 2021;606:120873. [DOI] [PubMed] [Google Scholar]
41.Chu YC, et al. Functional Peptide-Loaded Gelatin Nanoparticles as Eyedrops for Cornea Neovascularization Treatment. Int J Nanomed. 2023;18:1413–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Zhang W, et al. Dual-ROS-scavenging and dual-lingering nanozyme-based eye drops alleviate dry eye disease. J Nanobiotechnol. 2024;22(1):229. [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Amadio M, Govoni S, Pascale A. Targeting VEGF in eye neovascularization: What’s new? A comprehensive review on current therapies and oligonucleotide-based interventions under development. Pharmacol Res. 2016;103:253–69. [DOI] [PubMed] [Google Scholar]
44.Goncharov NV et al. Markers of Endothelial Cells in Normal and Pathological Conditions. Biochem (Mosc) Suppl Ser A Membr Cell Biol, 2020. 14(3): pp. 167–183. [DOI] [PMC free article] [PubMed]
45.Kim H, et al. CD31 + cells represent highly angiogenic and vasculogenic cells in bone marrow: novel role of nonendothelial CD31 + cells in neovascularization and their therapeutic effects on ischemic vascular disease. Circ Res. 2010;107(5):602–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
46.Nourse MB, et al. VEGF induces differentiation of functional endothelium from human embryonic stem cells: implications for tissue engineering. Arterioscler Thromb Vasc Biol. 2010;30(1):80–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
47.Ryeom SW, Paul D, Goodenough DA. Truncation mutants of the tight junction protein ZO-1 disrupt corneal epithelial cell morphology. Mol Biol Cell. 2000;11(5):1687–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Xu H, et al. Deliver protein across bio-barriers via hexa-histidine metal assemblies for therapy: a case in corneal neovascularization model. Mater Today Bio. 2021;12:100143. [DOI] [PMC free article] [PubMed] [Google Scholar]
49.Mao B, et al. A promising strategy for ocular noninvasive protein delivery: The case in treating corneal neovascularization. Acta Biomater. 2025;196:307–20. [DOI] [PubMed] [Google Scholar]
50.Xu X, et al. Prevention of corneal neovascularization by subconjunctival injection of avastin^® loaded thermosensitive hydrogels in rabbit model. Int J Pharm. 2018;552(1–2):164–70. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material 1^{(661.8KB, docx)}

Data Availability Statement

No datasets were generated or analysed during the current study.

[CR1] 1.Nicholas MP, Mysore N. Corneal neovascularization. Exp Eye Res. 2021;202:108363–9. [DOI] [PubMed] [Google Scholar]

[CR2] 2.Lee P, Wang CC, Adamis AP. Ocular neovascularization: an epidemiologic review. Surv Ophthalmol. 1998;43(3):245–69. [DOI] [PubMed] [Google Scholar]

[CR3] 3.Roshandel D, et al. Current and emerging therapies for corneal neovascularization. Ocul Surf. 2018;16(4):398–414. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR4] 4.Stevenson W, et al. Corneal neovascularization and the utility of topical VEGF inhibition: ranibizumab (Lucentis) vs bevacizumab (Avastin). Ocul Surf. 2012;10(2):67–83. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR5] 5.Kaur S, et al. Molecular nature of ocular surface barrier function, diseases that affect it, and its relevance for ocular drug delivery. Ocul Surf. 2023;30:3–13. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR6] 6.Gaballa SA, et al. Corticosteroids in ophthalmology: drug delivery innovations, pharmacology, clinical applications, and future perspectives. Drug Deliv Transl Res. 2021;11(3):866–93. [DOI] [PubMed] [Google Scholar]

[CR7] 7.Lin X, et al. Nanomedicine-Based Ophthalmic Drug Delivery Systems for the Treatment of Ocular Diseases. Int J Nanomed. 2025;20:9221–49. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR8] 8.Cheng Y, et al. Biomimetic Metal-Organic Antioxidant Eye Drops for Corneal Neovascularization: Visible-Light-Triggered CO Release for Microenvironment Regulation. Nano Lett. 2025;25(30):11703–12. [DOI] [PubMed] [Google Scholar]

[CR9] 9.Wu H, et al. A SU6668 pure nanoparticle-based eyedrops: toward its high drug Accumulation and Long-time treatment for corneal neovascularization. J Nanobiotechnol. 2024;22(1):290. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR10] 10.Ahmed S, Amin MM, Sayed S. Ocular Drug Delivery: a Comprehensive Review. AAPS PharmSciTech. 2023;24(2):66. [DOI] [PubMed] [Google Scholar]

[CR11] 11.Öztürk K, Kaplan M, Çalış S. Effects of nanoparticle size, shape, and zeta potential on drug delivery. Int J Pharm. 2024;666:124799. [DOI] [PubMed] [Google Scholar]

[CR12] 12.Di J, et al. Size, shape, charge and stealthy surface: Carrier properties affect the drug circulation time in vivo. Asian J Pharm Sci. 2021;16(4):444–58. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR13] 13.Han H, et al. Polymer- and lipid-based nanocarriers for ocular drug delivery: Current status and future perspectives. Adv Drug Deliv Rev. 2023;196:114770. [DOI] [PubMed] [Google Scholar]

[CR14] 14.Xu J, et al. Applications and challenges of ultra-small particle size nanoparticles in tumor therapy. J Control Release. 2023;353:699–712. [DOI] [PubMed] [Google Scholar]

[CR15] 15.Beirampour N et al. Formulation of Polymeric Nanoparticles Loading Baricitinib as a Topical Approach in Ocular Application. Pharmaceutics 2024;16(8):1092. [DOI] [PMC free article] [PubMed]

[CR16] 16.Mandal A, et al. Polymeric micelles for ocular drug delivery: From structural frameworks to recent preclinical studies. J Control Release. 2017;248:96–116. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR17] 17.Lu Z, et al. Activated aggregation strategies to construct size-increasing nanoparticles for cancer therapy. Wiley Interdiscip Rev Nanomed Nanobiotechnol. 2023;15(2):e1848. [DOI] [PubMed] [Google Scholar]

[CR18] 18.Zhang Z, et al. Highly Efficient Intrastromal Dual-Drug Delivery by a Dissolving Bilayer Microneedle for Synergistic Therapy against Corneal Neovascularization. ACS Nano. 2025;19(42):37065–81. [DOI] [PubMed] [Google Scholar]

[CR19] 19.Luis de Redin I, et al. In vivo effect of bevacizumab-loaded albumin nanoparticles in the treatment of corneal neovascularization. Exp Eye Res. 2019;185:107697–708. [DOI] [PubMed] [Google Scholar]

[CR20] 20.Xu H, et al. Engineering Matrix-Free Drug Protein Nanoparticles with Promising Penetration through Biobarriers for Treating Corneal Neovascularization. ACS Nano. 2024;18(11):8209–28. [DOI] [PubMed] [Google Scholar]

[CR21] 21.Tang B, et al. Decorating hexahistidine-metal assemblies with tyrosine enhances the ability of proteins to pass through corneal biobarriers. Acta Biomater. 2022;153:231–42. [DOI] [PubMed] [Google Scholar]

[CR22] 22.Lee J, et al. CHARMM-GUI Input Generator for NAMD, GROMACS, AMBER, OpenMM, and CHARMM/OpenMM Simulations Using the CHARMM36 Additive Force Field. J Chem Theory Comput. 2016;12(1):405–13. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR23] 23.Huang J, et al. CHARMM36m: an improved force field for folded and intrinsically disordered proteins. Nat Methods. 2017;14(1):71–3. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR24] 24.Laria C, Alio JL, Ruiz-Moreno JM. Combined non-steroidal therapy in experimental corneal injury. Ophthalmic Res. 1997;29(3):145–53. [DOI] [PubMed] [Google Scholar]

[CR25] 25.Zheng Y, et al. Remote drug loading into liposomes via click reaction. Mater Horiz. 2022;9(7):1969–77. [DOI] [PubMed] [Google Scholar]

[CR26] 26.Yang C, et al. A facile doxorubicin-dichloroacetate conjugate nanomedicine with high drug loading for safe drug delivery. Int J Nanomed. 2018;13:1281–93. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR27] 27.Kaur A, et al. Characterization of Cu(2+) and Zn(2+) binding sites in SUMO1 and its impact on protein stability. Int J Biol Macromol. 2020;151:204–11. [DOI] [PubMed] [Google Scholar]

[CR28] 28.Zou Z, et al. Zn(2+) -Coordination-Driven RNA Assembly with Retained Integrity and Biological Functions. Angew Chem Int Ed Engl. 2021;60(42):22970–6. [DOI] [PubMed] [Google Scholar]

[CR29] 29.Harding MM. The architecture of metal coordination groups in proteins. Acta Crystallogr D Biol Crystallogr. 2004;60(Pt 5):849–59. [DOI] [PubMed] [Google Scholar]

[CR30] 30.Ramirez-Bello V et al. Zinc and Copper Ions Induce Aggregation of Human β-Crystallins. Molecules 2022;27(9):2970. [DOI] [PMC free article] [PubMed]

[CR31] 31.Pei D, Buyanova M. Overcoming Endosomal Entrapment in Drug Delivery. Bioconjug Chem. 2019;30(2):273–83. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR32] 32.Levy JMM, Towers CG, Thorburn A. Targeting autophagy in cancer. Nat Rev Cancer. 2017;17(9):528–42. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR33] 33.Koren E, Torchilin VP. Cell-penetrating peptides: breaking through to the other side. Trends Mol Med. 2012;18(7):385–93. [DOI] [PubMed] [Google Scholar]

[CR34] 34.Vailhé B, et al. The formation of tubular structures by endothelial cells is under the control of fibrinolysis and mechanical factors. Angiogenesis. 1998;2(4):331–44. [DOI] [PubMed] [Google Scholar]

[CR35] 35.Ferrara N, Gerber HP, LeCouter J. The biology of VEGF and its receptors. Nat Med. 2003;9(6):669–76. [DOI] [PubMed] [Google Scholar]

[CR36] 36.Singh SR, et al. Intravenous transferrin, RGD peptide and dual-targeted nanoparticles enhance anti-VEGF intraceptor gene delivery to laser-induced CNV. Gene Ther. 2009;16(5):645–59. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR37] 37.Dastjerdi MH, et al. Corneal penetration of topical and subconjunctival bevacizumab. Invest Ophthalmol Vis Sci. 2011;52(12):8718–23. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR38] 38.Noske W, et al. Tight junctions of the human corneal endothelium: morphological and electrophysiological features. Ger J Ophthalmol. 1994;3(4–5):253–7. [PubMed] [Google Scholar]

[CR39] 39.Mandal A, et al. Ocular delivery of proteins and peptides: Challenges and novel formulation approaches. Adv Drug Deliv Rev. 2018;126:67–95. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR40] 40.Silva B, et al. Colloidal nanosystems with mucoadhesive properties designed for ocular topical delivery. Int J Pharm. 2021;606:120873. [DOI] [PubMed] [Google Scholar]

[CR41] 41.Chu YC, et al. Functional Peptide-Loaded Gelatin Nanoparticles as Eyedrops for Cornea Neovascularization Treatment. Int J Nanomed. 2023;18:1413–31. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR42] 42.Zhang W, et al. Dual-ROS-scavenging and dual-lingering nanozyme-based eye drops alleviate dry eye disease. J Nanobiotechnol. 2024;22(1):229. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR43] 43.Amadio M, Govoni S, Pascale A. Targeting VEGF in eye neovascularization: What’s new? A comprehensive review on current therapies and oligonucleotide-based interventions under development. Pharmacol Res. 2016;103:253–69. [DOI] [PubMed] [Google Scholar]

[CR44] 44.Goncharov NV et al. Markers of Endothelial Cells in Normal and Pathological Conditions. Biochem (Mosc) Suppl Ser A Membr Cell Biol, 2020. 14(3): pp. 167–183. [DOI] [PMC free article] [PubMed]

[CR45] 45.Kim H, et al. CD31 + cells represent highly angiogenic and vasculogenic cells in bone marrow: novel role of nonendothelial CD31 + cells in neovascularization and their therapeutic effects on ischemic vascular disease. Circ Res. 2010;107(5):602–14. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR46] 46.Nourse MB, et al. VEGF induces differentiation of functional endothelium from human embryonic stem cells: implications for tissue engineering. Arterioscler Thromb Vasc Biol. 2010;30(1):80–9. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR47] 47.Ryeom SW, Paul D, Goodenough DA. Truncation mutants of the tight junction protein ZO-1 disrupt corneal epithelial cell morphology. Mol Biol Cell. 2000;11(5):1687–96. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR48] 48.Xu H, et al. Deliver protein across bio-barriers via hexa-histidine metal assemblies for therapy: a case in corneal neovascularization model. Mater Today Bio. 2021;12:100143. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR49] 49.Mao B, et al. A promising strategy for ocular noninvasive protein delivery: The case in treating corneal neovascularization. Acta Biomater. 2025;196:307–20. [DOI] [PubMed] [Google Scholar]

[CR50] 50.Xu X, et al. Prevention of corneal neovascularization by subconjunctival injection of avastin^® loaded thermosensitive hydrogels in rabbit model. Int J Pharm. 2018;552(1–2):164–70. [DOI] [PubMed] [Google Scholar]

PERMALINK

Size-shrinking nanoparticles with high drug-protein payload for efficient, non-invasive treatment of corneal neovascularization

Xiaoling Xie

Zhenyang Xiang

Ruhui Yang

Xuanlu Wu

Jinyu Guo

Ruolin Fang

Xingjie Zan

Wenjuan Huang

Yuhua Tong

Abstract

Graphical Abstract

Supplementary Information

Introduction

Scheme 1.

Materials and experimental methods

Materials

Preparation and optimization of preparation conditions for Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn

Characterization of Beva-NPs

Size and zeta potential

Encapsulation efficiency percentage (EE%) of Beva

Loading capacity percentage of (Beva, and R₉ and/or C₂G₂R₉) in Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn

Zinc content in Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn

Transmission Electron Microscopy (TEM)

Fourier Transform Infrared Spectroscopy (FTIR)

X-ray Photoelectron Spectroscopy (XPS)

In vitro release of Beva from Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn

Bioactivity of beva

Simulation

In vitro experiments

Cell culture

In vitro cytotoxicity

Cellular uptake study

Lysosomal escape experiment

3D Corneal cell model permeation study

Migration inhibition assay

Tube formation inhibition assay

In vivo experiments

Animals and ethical considerations

Evaluation of retention in the cornea

Establishment of alkali burn CNV model

Treatment and medication

CNV evaluation

Inflammatory index evaluation

Ex vivo experiments

New Zealand white rabbit corneal diffusion experiment

Histological sectioning of rat corneal tissues

Immunofluorescence staining

Biocompatibility study

Slit-lamp photography

Hematoxylin and Eosin (H&E) staining

Corneal endothelial tight junction ZO-1 staining

Data analysis

Results and discussion

Preparation and optimization of Beva@Zn, (Beva&R₉)@Zn, and (Beva&C₂G₂R₉)@Zn Nanoparticles

Fig. 1.

Interactions in (Beva&C₂G₂R₉)@Zn

Fig. 2.

Cytotoxicity assay

Cellular uptake and lysosome escape

Fig. 3.

In vitro anti-migration and anti-tube formation effects on HUVECs

Penetration across ocular barriers and retention on ocular surface

Fig. 4.

Evaluation of therapeutic efficacy in the rat alkali burn CNV model

Fig. 5.

Inhibitory effect on the expression of VEGF and CD31

Fig. 6.

Biocompatibility

Fig. 7.

Discussion

Table 1.

Conclusion

Supplementary Information

Author contributions

Funding

Data availability

Declarations

Competing interests