Hyaluronic Acid-PEG-PLGA Nanoparticles Encapsulating Myricetin Attenuate NMDA-Induced Retinal Ganglion Cell Injury Associated with Microglia-Related Neuroinflammatory Responses

Abstract

Background

Retinal ganglion cell (RGC) injury caused by excitotoxic stress is an important component of retinal neurodegeneration. Microglia-mediated inflammatory responses can amplify neuronal damage following injury. Myricetin (MY), a natural flavonoid with anti-inflammatory properties, has potential neuroprotective activity but is limited by poor aqueous delivery characteristics. This study aimed to develop a nanoparticle-based delivery system for MY and evaluate its protective effects against NMDA-induced retinal injury.

Methods

MY was encapsulated into hyaluronic acid–polyethylene glycol–poly (lactic-co-glycolic acid) nanoparticles (MY-NPs). Nanoparticle characteristics, including particle size, drug loading, encapsulation efficiency, morphology, and in vitro release profile, were evaluated. An NMDA-induced retinal injury model was established in mice, followed by intravitreal administration of MY-NPs. Retinal function, RGC survival, apoptosis, inflammatory responses, and microglial activation were assessed using electroretinography, optomotor testing, immunofluorescence, histological analysis, TUNEL staining, RNA sequencing, and quantitative PCR. The anti-inflammatory effects of MY-NPs were further evaluated in activated BV2 microglia and microglia-conditioned medium-treated R28 cells.

Results

MY-NPs showed favorable physicochemical properties, with an average diameter of 67.1 nm, drug loading of approximately 7.4%, and encapsulation efficiency above 80%. The nanoparticles exhibited a biphasic release profile with prolonged myricetin release over 96 h. In NMDA-treated mice, MY-NPs improved retinal function, preserved RGC density, reduced retinal apoptosis, and decreased TNF-α expression. Transcriptomic analysis indicated that NMDA-induced injury was associated with inflammatory pathways, including cytokine–cytokine receptor interaction, JAK–STAT signaling, and TNF signaling, whereas MY-NPs attenuated the expression of inflammatory-related genes. MY-NPs also reduced pro-inflammatory activation markers and inflammatory mediator production in activated BV2 microglia and alleviated apoptosis of R28 cells induced by activated microglial conditioned medium.

Conclusion

HA-PEG-PLGA nanoparticles provide an effective delivery platform for myricetin and MY-NPs are associated with protection against NMDA-induced excitotoxic retinal injury. The observed neuroprotective effects were accompanied by reduced microglial activation and inflammatory responses, suggesting that attenuation of microglia-associated inflammation may contribute to RGC preservation. Further validation in chronic glaucoma-relevant models is required to determine the translational potential of this strategy.

Graphical Abstract

Introduction

Glutamate is the principal excitatory neurotransmitter in the central nervous system. However, under pathological conditions, its excessive accumulation can cause damage or even death of neurons through sustained activation of glutamate receptors—a process known as excitotoxicity.^1,2 Glaucoma and other optic neuropathies represent major causes of irreversible visual impairment worldwide, affecting tens of millions of individuals and imposing a substantial global health burden.³ In glaucoma and various other optic neuropathies, glutamate-induced excitotoxicity is considered a key pathological mechanism contributing to RGC degeneration,⁴ therefore, identifying therapeutic strategies targeting excitotoxic injury and associated neuroinflammatory responses represents an important research priority. The N-methyl-D-aspartate receptor (NMDAR), a major subtype of glutamate receptor, plays a central role in excitotoxicity. Nearly all RGCs express NMDARs, making them particularly sensitive and vulnerable to abnormal elevations of glutamate levels.¹ Therefore, intravitreal injection of NMDA has become a classical, stable, and reproducible animal model for studying the mechanisms of RGC death and for screening neuroprotective agents.⁴

Traditionally, NMDA-induced excitotoxicity has been considered a cell-autonomous process, meaning that the damage occurs entirely within the RGCs themselves. Excessive NMDA (or glutamate) binds to NMDARs on the surface of RGCs, triggering death signaling pathways that converge on the activation of the caspase family, thereby executing apoptosis.⁵ Although the classical cell-autonomous model provides a clear picture of the intracellular death processes in RGCs, it fails to explain why direct NMDAR antagonists have repeatedly failed in clinical trials.⁶ These clinical outcomes strongly suggest that RGC death is not solely determined by overactivation of their own NMDARs; rather, the classical cell-autonomous pathway represents only part of the overall pathological landscape. This has led to the consideration that neighboring non-neuronal cells—particularly glial cells—may play a crucial role in mediating and amplifying excitotoxic damage.^2,7

Microglia are the resident macrophages of the retina and serve as the intrinsic immune cells of neural tissue. They are responsible for immune surveillance, maintaining neuronal homeostasis, and clearing cellular debris.⁸ When the retina is exposed to injury or stress stimuli—such as NMDA-induced excitotoxicity—microglia become rapidly activated. They retract their processes, adopt an amoeboid morphology, and migrate toward the site of injury.⁹ This activation represents one of the earliest events in neurodegeneration and marks the onset of neuroinflammation.¹⁰ It is now widely accepted that NMDA-induced retinal injury in vivo results from a combination of direct neuronal effects and indirect effects mediated by the retinal inflammatory microenvironment—the latter of which may even be more critical.¹¹ This complexity poses challenges for therapy but also provides a rationale for developing broad-spectrum neuroprotective drugs. Compared with targeting a specific receptor subtype, focusing on downstream pathological pathways—such as neuroinflammation and oxidative stress—may represent a more effective and universal therapeutic strategy.

Myricetin (MY), a natural flavonoid, has recently gained attention for its potential therapeutic applications.¹² Studies have demonstrated its anticancer properties, including the ability to induce tumor cell apoptosis, inhibit proliferation and migration, and enhance chemotherapy sensitivity.¹³ Additionally, MY has shown promise in diabetes management by lowering blood glucose levels and mitigating diabetic complications.¹⁴ It has also been reported to protect against non-alcoholic fatty liver disease by regulating gut microbiota, reducing hepatic lipid synthesis, and decreasing inflammation.¹⁵ The MY has recently attracted growing interest in ophthalmology due to its potential therapeutic benefits. Research has found that oral administration of MY can help manage diabetic retinopathy by reducing blood sugar levels and oxidative stress.¹⁶ Moreover, it could effectively lower inflammatory cytokine levels in the aqueous humor.¹⁷ The multiple mechanisms of action of myricetin make it an ideal candidate drug for combating the aforementioned neuroinflammatory pathology in NMDA-induced RGC damage. Its core anti-inflammatory effects are closely related to the inhibition of key inflammatory signaling pathways. Studies have clearly shown that myricetin can significantly suppress the activation of several critical inflammatory cascades, particularly the mitogen-activated protein kinase (MAPK) family (including p38, JNK, and ERK) and the NF-κB pathway.¹⁸ Myricetin effectively downregulates the expression and release of multiple proinflammatory mediators in activated microglia. These include inducible nitric oxide synthase (iNOS), cyclooxygenase-2 (COX-2), and key neurotoxic cytokines such as TNF-α and IL-1β.¹⁸ This action directly weakens the molecular arsenal that drives neuronal death in the NMDA-induced model.

However, the poor water solubility, instability in aqueous environments, and low bioavailability of MY remain significant barriers to its clinical translation in ophthalmology. Therefore, some studies have prepared micellar nanocarriers to carry flavonoids and improve their corneal penetration when applied as eye drops.^19,20 However, the drug delivery system composed of these micelles may not be suitable for clinical treatment. When administered intravitreally, the micelles are easily affected by dilution or environmental changes in the body, and their stability is relatively compromised.²¹ Therefore, we selected an HA-PEG-PLGA nanoparticle platform because it integrates several features advantageous for intravitreal drug delivery. PLGA provides a biodegradable polymeric matrix capable of encapsulating hydrophobic compounds and enabling sustained drug release,²² while PEG can improve colloidal stability and may reduce nonspecific interactions within the vitreous.²³ In addition, HA is biocompatible and negatively charged and can interact with CD44, a receptor expressed in retinal tissues and upregulated in activated microglia, thereby providing a potential basis for enhanced interaction with activated microglia,²⁴ enhancing ocular retention and potential active targeting.²⁵ Thus, rather than relying on a single delivery property, the HA-PEG-PLGA system was chosen to combine drug stabilization, sustained release, favorable physicochemical characteristics for intravitreal delivery, and potential CD44-mediated cellular interaction.

In this study, this nanoparticle-based approach was designed to improve the delivery characteristics of myricetin and evaluate its potential RGC-protective effects. Based on these considerations, we hypothesized that HA-PEG-PLGA nanoparticles could improve the ocular delivery characteristics of myricetin and provide neuroprotective effects against NMDA-induced RGC injury. Specifically, we proposed that MY-NPs may attenuate microglia-associated inflammatory responses, thereby reducing secondary inflammatory damage and contributing to RGC preservation.

Materials and Methods

Preparation of MY-Loaded NPs (MY-NPs)

The research period is from December 2023 to October 2024. HA (20k)-PLGA (LA:GA = 50:50, 10k)-PEG (2k) was supplied by Chongqing Yusi Medical Technology Cable Co., Ltd. Briefly, 2 mg of MY (Yuanye, China, S25978) was dissolved in 0.2 mL of dimethyl sulfoxide (DMSO, MP bio, 219605580). Then the MY solution was added to 10 mg of HA-PLGA-PEG in DMSO/dimethylformamide (DMF) solution (1:1). The mixture was then slowly added to deionized water (3 mL) and stirred for 30 minutes. The obtained NPs suspension was dialyzed in deionized water with a dialysis bag (MWCO 8000–12000) for 12 h to remove excess organic solvent. The physicochemical properties of MY-NPs were characterized as follows. Particle size, polydispersity index (PDI), and zeta potential were measured using dynamic light scattering (DLS). Measurements were performed after appropriate dilution of nanoparticle suspensions in deionized water, and data were presented as mean ± standard deviation from repeated measurements. Nanoparticle size in the nanometer range and a low PDI value were considered indicators of favorable particle uniformity. Surface charge was evaluated by zeta potential measurement to assess colloidal stability. The amount of encapsulated MY was quantified using UV spectrophotometry. Nanoparticle morphology and structural characteristics were examined using transmission electron microscopy (TEM). Samples were placed on copper grids, negatively stained, and visualized under TEM. The MY standards of known concentrations were prepared in the same solvent used for the NP samples, and their absorbance was measured at 380 nm. The calibration equation was y=0.0677x-0.0332, with an R² of 0.9997, where y represents the absorbance at 380 nm and x represents the MY concentration. Encapsulation efficiency was defined as the amount of MY encapsulated in the nanoparticles relative to the total amount of MY initially added, whereas drug loading was defined as the amount of encapsulated MY relative to the total mass of the MY-loaded nanoparticles. In addition, the in vitro drug release properties of the NPs were examined according to previously reported methods.²⁶ The in vitro release behavior of MY-NPs was evaluated using 1 mL of MY-NP suspension (0.8 mg/mL). The NPs were first immersed in PBS (Bioss, C01-01001) solution containing 1% polysorbate 20 and shaken at 150 rpm at 37 °C. At predetermined time points (0, 1, 2, 4, 6, 12, 24, 48, 72, and 96 h), aliquots of the release medium were collected and replaced with an equal volume of fresh release medium. The concentration of released myricetin was quantified, and the cumulative release percentage was calculated. To quantitatively characterize the in vitro release behavior of MY-NPs, the cumulative drug-release data were fitted using nonlinear least-squares regression to zero-order, first-order, Higuchi, Weibull, and Korsmeyer–Peppas models. The Korsmeyer–Peppas model was applied only to the initial release phase corresponding to ≤60% cumulative drug release.

Animals and the Model of NMDA-Induced Retinal Ganglion Cell Damage

The animal experiments commenced in December 2023. The number of animals per group was determined based on statistical power analysis and ethical considerations to minimize animal use. Adult C57BL/6J mice (7 weeks old, male) were purchased from Hunan Slake Jingda Experimental Animal Co., Ltd. They were maintained in a Specific Pathogen Free (SPF) grade environment with a 12-h light/dark cycle in the Laboratory Animal Resource Center of Aier Eye Institute, and the animal ethics approval has been obtained (AEI20230079) and the welfare of these laboratory animals complies with GB/T 42011–2022 General Requirements. Food and water were provided ad libitum. Mice were anesthetized via intraperitoneal injection of 1.5% Zolazepam (Virbac, catalog number 3597133073812) and 0.2% Xylazine (Shengxin, catalog number 070031777) at a dose of 4 μL/g. Under a surgical microscope, a 33-gauge Hamilton syringe was inserted trans-sclerally into the vitreous cavity 1 mm posterior to the limbus, avoiding lens injury. The right eye received a single intravitreal injection of 1μL (40mM) NMDA (Sigma M3262), while the control group was injected with an equal volume of PBS. A single intravitreal injection of MY-NPs (1μL) was performed using a microsyringe through the same hole. Mice were euthanized by cervical dislocation, which was performed after the mouse had been deeply anesthetized. The anesthetized mouse was then manually restrained by the tail base, while the thumb and forefinger applied pressure at the atlanto-occipital joint. A rapid, forceful extension of the head backward, combined with a caudal pull on the tail, resulted in immediate separation of the cervical vertebrae and transection of the spinal cord. The procedure was carried out away from other animals to reduce potential stress. All samples were collected and analyzed three days after model establishment, except for RNA-seq samples, which were collected 24h after model establishment.

Electroretinography (ERG)

ERG (Roland, RETI-Scan 21) examinations were conducted using a dark-adapted (DA) 3.0 protocol to assess retinal function under scotopic conditions. Animals were dark-adapted for 12 h, then anesthetized, and their pupils were dilated using a topical mydriatic. Using a Ganzfeld dome, 3.0 cd·s/m² flashes were delivered, and retinal responses were recorded using an ERG system. The a-wave, which reflects photoreceptor cell responses (mainly rod cell activity), and the b-wave, which represents the activity of inner retinal bipolar cells and Müller cells, were analyzed to assess functional changes. The wave amplitude and implicit time were measured as indicators of retinal health.

Small Animal Vision Function Detection

Visual function in animals was evaluated using a virtual optokinetic system (OptoDrum, STRIA-TECH, Germany) to assess the optomotor response. The animals were placed on an elevated platform at the center of a cylindrical arena surrounded by a rotating digital display of vertical black-and-white stripes. By gradually adjusting the spatial frequency and rotation speed of the stripes, the animal’s visual acuity and contrast sensitivity were measured. Visual acuity was determined by the spatial-frequency threshold, expressed in cycles per degree (cpd), which reflects the animal’s capacity to distinguish fine details. A higher threshold indicates better visual acuity.

Histological and Immunohistochemical Analyses

Histological and immunohistochemical analyses were conducted to assess retinal structure, inflammation, and cell apoptosis. After enucleation, eyeballs were fixed in 4% paraformaldehyde (Sinopharm, 80096628) embedded in paraffin, and sectioned. Hematoxylin and eosin (H&E) staining was used to examine overall retinal morphology and structural integrity. For the evaluation of inflammation, TNF-α immunohistochemistry was performed to evaluate inflammation, using a specific anti-TNF-α antibody (Abcam, ab183218, 1:100). For quantitative analysis of TNF-α staining, images were analyzed using ImageJ software. All images used for comparison were acquired under identical imaging settings. The resulting fluorescence intensity was normalized to the mean value of the control group, which was defined as 1.0. Apoptotic cells were detected with the TUNEL assay (Beyotime, C1086) in combination with DAPI staining. The ZEN confocal microscope software was used to measure the area of the ganglion cell layer (GCL) in each randomly selected field of view. The number of TUNEL⁺ cells in the GCL was manually counted.

RGCs Counting

To quantify RGCs, whole-mount retinal flat-mounts were prepared and stained with immunofluorescence for βIII-tubulin (Abcam, ab78078, 1:200) and RNA-binding protein with multiple splicing (RBPMS, Abcam, ab152101, 1:100). The stained retinas were imaged using fluorescence or confocal microscopy. For RGC quantification, images were taken from four quadrants of each retina at standardized distances from the optic nerve head. The results were reported as RBPMS⁺/βIII-tubulin⁺ (cells/0.1mm²) and compared among experimental groups to evaluate treatment efficacy or disease progression.

Microglial Activation Testing in vivo

To assess microglial activation in the retina, retinal flat mounts were immunostained for IBA1 (Wako 019–19,741, 1:500). The stained retinas were imaged using confocal microscopy, and z-stack images were acquired to visualize two distinct layers of IBA1-positive microglia: Layer 1 is located between the nerve fiber layer and the ganglion cell layer; Layer 2 is located between the inner nuclear layer and the Inner plexiform layer. Microglia were counted by manually identifying IBA1⁺ cells using image analysis software. Microglial morphology was further assessed by measuring cell body diameter and branch length using ImageJ. An increase in cell body diameter together with shortening of branch was interpreted as being consistent with a more activated microglial morphology. These parameters were used as selected morphometric indicators of microglial activation rather than as a comprehensive characterization of microglial morphology.

RNA Sequencing

Retinal tissues were collected, and total RNA was extracted using a commercial RNA isolation kit (Agilent 2100 bioanalyzer, CA, USA). RNA sequencing was carried out by Beijing Novo Biotech Co., Ltd. After library construction and quality assessment, libraries that met quality standards were pooled based on the required data volume for the target effective concentration. Illumina sequencing was then performed to generate 150 bp paired-end reads. Sequencing-by-synthesis was used to obtain sequence information for the target fragments, followed by bioinformatics analysis. Raw data quality control was conducted using fastp to obtain clean reads. Reference genome and gene model annotation files were downloaded from the genome database, and an index of the reference genome was built using HISAT2 v2.0.5. FeatureCounts was used to quantify reads mapped to each gene, and FPKM values were calculated based on gene length. Differential expression analysis was performed using DESeq2, and KEGG enrichment analysis was conducted using clusterProfiler.

qPCR

To quantify the transcription levels of MyD88, TNF-α, MMP3, and CD86, quantitative PCR (qPCR) was performed. qPCR was conducted using specific primers for MyD88, TNF-α, MMP3, and CD86 et al (as shown in Table S1). All qPCR reactions were performed in triplicate. Gene transcription levels were quantified using the comparative Ct method (ΔΔCt), normalized to reference genes (Actin), and expressed as fold changes relative to the control group.

BV2 Microglia Intervention

BV2 microglia (from National Collection of Authenticated Cell Cultures, GNM45) were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) (Thermo, C11995500BT) supplemented with 10% fetal bovine serum (FBS, Thermo, 10270106) and 1% penicillin/streptomycin (Thermo, 15140122) under standard incubator conditions (37°C, 5% CO₂) in an incubator. To induce microglial activation, BV2 cells were exposed to a combination of 100 ng/mL lipopolysaccharide (LPS, Sigma, L2880) and 10 ng/mL interferon-γ (IFN-γ, Peprotech, 3150520) for 8 h. During the intervention, MY-NPs were applied to activated BV2 cells at varying concentrations. To assess microglial activation and inflammatory responses, the expression of CD86 (a marker of activated microglia, Abcam, ab220188, 1:100), IBA1 (a microglia marker), and iNOS (inducible nitric oxide synthase, a pro-inflammatory microglial marker, Abcam, ab178945, 1:100) was detected.

Preparation of BV2 Cell Supernatant and Its Influence on R28

Preparation of cell supernatant: After culturing BV2 cells to 80% confluence, the treatment group was incubated with LPS (100ng/mL) + IFN-y (10ng/mL) for 8h, and the drug treatment group was treated with LPS (100ng/mL) + IFN-y (10ng/mL) + MY-NPs at different concentrations for 8h (0.2 mg/mL, 0.4 mg/mL, 0.8 mg/mL MY-NPs, maintaining a consistent doubling interval), then the cell supernatant was removed and replaced with new DMEM high-glucose medium for 24h. After 24h of culture, the cell supernatant was collected and filtered with a 0.22 μm filter. The cells were concentrated with a 10 KDa ultrafiltration tube at 4°C and 4000 rpm for 40 min. After concentration, the cell concentrate was placed at −80°C for standby use.

Concentrations of IL-6 and TNF-α in cell supernatants were quantified using commercial ELISA kits (NeoBioscience, IL-6: EMC004, TNF-α: EMC102a, China). Nitric oxide (NO) production was assessed by a kit measuring nitrite accumulation using the Griess reaction (Beyotime, S0021S, China).

R28 cells (provided by Glaucoma Institute, Changsha Aier Eye Hospital, the use of the cell lines was approved by the Aier Eye Institute) were treated with cell concentrates from different treatment groups for 8 h (the dilution ratio of the concentrate was the concentration multiple of the original cell supernatant). Caspase 3 Activity Assay Kit (Beyotime, C1115, China) was used for quantification of caspase 3 enzyme activity. Cell apoptosis was also assessed using the Annexin V-FITC apoptosis detection kit (Biyuntian, C1062S). The flow cytometer is BD FACSCelesta. For the flow cytometry data, we utilized FlowJo software for all data analysis and statistical gating. Specifically, when using the Annexin V-FITC antibody, the study focused on quantifying the apoptotic cell populations. The gating strategy involved the following: Q2 (Annexin V-FITC positive, Propidium Iodide positive): This quadrant represents the late apoptotic cells. Q3 (Annexin V-FITC positive, Propidium Iodide negative): This quadrant represents the early apoptotic cells. The percentages of cells in Q2 and Q3 were summed to represent the total apoptotic level for each sample.

Immunofluorescence Staining for Microglial CD44 Expression and Targeting of MY-NPs

BV2 cells were seeded on coverslips in 24-well plates and stimulated with LPS (100 ng/mL) plus IFN-γ (10 ng/mL) for 16 hours. After fixation with 4% PFA, cells were permeabilized, blocked, and incubated with anti-IBA1 (1:100) and anti-CD44 antibodies (1:100 Servicebio GB112054) overnight, followed by fluorescent secondary antibody and DAPI staining before mounting.

To detect the targeting capacity of this system, BV2 cells treated with LPS/IFN-γ (which highly express CD44 receptors after activation²⁷) were used to detect the quantity of Nile red absorbed by the cells after Nile-NPs intervention (For labeling of the NPs, MY was substituted with Nile red due to its stronger fluorescent properties).

Ocular Biodistribution and Retention Assessment of MY

NMDA-treated C57 mice were anesthetized, followed by intravitreal injection of 1 µL myricetin nanoparticles (MY-NPs). Tissue collection was performed at 8 h, 24 h, and 72 h post-injection.

For vitreous humor and retinal tissue: Considering the very small volume of vitreous humor, we euthanized the mice via carbon dioxide overdose and enucleated the eyeballs. The eyeballs were rinsed with PBS, and after removing the peripheral muscle tissues, they were dried with dust-free paper. The eyeballs were then placed in EP tube lids on ice. After dissection and separation of the retinal tissue, the dissected eyeballs were rinsed with 100 μL of saline, and the rinse fluid was collected into new EP tubes and stored on ice. This rinse fluid was defined as diluted intraocular vitreous humor. The dissected retinas were separately transferred into another EP tube for storage.

10 µL of vitreous humor was aliquoted into a 1.5 mL polyethylene centrifuge tube. Then, 10 µL of 50% methanol-water solution was added, followed by 10 µL of internal standard (2000 ng/mL) and 60 µL of 0.1% formic acid-methanol solution. The mixture was vortexed for 30s, centrifuged at 14,000 rpm for 10 min, and the supernatant was collected for analysis.

The retina tissues were placed in 0.2 mL homogenization tubes. A 4:1 methanol-water solution (v/v) was added at a ratio of 4 µL per mg of tissue (1:4, w/v). The homogenate was stored at −80°C for 4 h, followed by lyophilization in a vacuum freeze concentrator (LABCONCO) for 16 h. The dried sample was then reconstituted in 15 µL of methanol-water solution.

Myricetin (MY) was quantified using liquid chromatography–mass spectrometry (Nexera LC-40, 8060RX).

OCT Imaging

The fundus imaging was performed using OCT (Phoenix Micron IV). Following fundus imaging, a 5% sodium fluorescein solution was administered via intraperitoneal injection at a dose of 100 μL per 10 g of body weight. Three minutes after the injection, fundus fluorescence imaging was uniformly performed on all mice.

Statistical Analysis

Data are presented as means ± standard deviations (SDs) and analyzed using one-way ANOVA with Tukey’s post hoc test or unpaired Student’s t test. Statistical significance was set at p < 0.05. Significance levels are indicated as follows: *p < 0.05, **p < 0.01, ***p < 0.001, ns p > 0.05. All analyses were performed using GraphPad Prism 9.0.

Results

Characterization of Myricetin-Loaded Nanoparticles (MY-NPs)

The MY-NPs achieved a drug loading of approximately 7.4% and an encapsulation efficiency of more than 80%. The MY-NPs had an average hydrodynamic diameter of 67.1 nm with a narrow size distribution (PDI <0.1), indicating good nanoparticle uniformity. The zeta potential was approximately −20 mV, suggesting a negatively charged nanoparticle surface. A distinct core-shell structure could be observed from the TEM images (Figures 1A, B and S1). The in vitro release profile of MY-NPs exhibited a biphasic pattern. Approximately 55.84% of the encapsulated MY was released within the first 24 h, followed by a slower release phase, with cumulative release reaching 74.83% at 96 h (Figure 1C). Kinetic modeling showed that the Weibull model provided the best fit to the complete release profile (R² = 0.9784), outperforming the zero-order, first-order, and Higuchi models. The Weibull shape parameter was β = 0.5004, consistent with a progressively decreasing release rate over time. For the initial release phase (≤60% cumulative release), the Korsmeyer–Peppas model yielded an R² of 0.9523. Together, these findings quantitatively support the biphasic and prolonged release behavior of MY-NPs.

Figure 1.

Characterization of MY-NPs and its protective effects against NMDA-induced retinal ganglion cell injury. (A) Electron microscope image of MY-NPs. (B) DLS analysis showing hydrodynamic diameter and particle size distribution of MY-NPs (67.1 nm, PDI <0.1). (C) In vitro sustained release curve of MY-NPs. (D and E) Quantitative analysis of the amplitude of a wave and b wave in ERG detection (n = 5). (F and G) Schematic diagram of mouse visual quality detection and its quantitative analysis (n = 5), cpd=circles per degree. (H and I). Immunofluorescence image of mouse retinal flat mount and its quantitative analysis of RBPMS (n = 5). (J and K). Immunofluorescence image of mouse retinal flat mount and its quantitative analysis of βIII-tubuin (n = 5). ns p > 0.05; ** p < 0.01; *** p < 0.001.

Protective Effects Against NMDA-Induced RGC Injury of MY-NPs

To evaluate the protective effects of MY-NPs, mice were divided into three groups: control, NMDA, and NMDA+ MY-NPs. Retinal dysfunction caused by NMDA injury was indirectly evidenced by a significant decline in both a-wave and b-wave amplitudes in ERG and a marked reduction of visual acuity compared to the control group. Notably, treatment with MY-NPs effectively alleviated these deficits, leading to improvement in both a-wave and b-wave amplitudes and the restoration of visual acuity (Figures 1D–G and S2).

Further confirmation of NMDA-induced RGC degeneration was obtained through retinal flat-mount immunohistochemistry, which revealed a significant decrease in RBPMS-positive and βIII-tubulin-positive cells compared to controls (Figure 1H–K). In contrast, MY-NPs administration preserved RGC populations, resulting in a higher density of RBPMS⁺ and βIII-tubulin⁺ cells relative to the NMDA group, indicating a neuroprotective effect (Figure 1H–K).

Additionally, histological analysis of retinal sections showed that NMDA exposure led to decreased cell nuclei in the ganglion cell layer (Figure 2A and D), increased expression of the inflammatory cytokine TNF-α (Figure 2B and E), and an elevation in apoptotic signals, as indicated by TUNEL staining in the ganglion cell layer and inner nuclear layer (Figure 2C and F). However, treatment with MY-NPs significantly reversed these changes. In addition, it markedly reduced the number of TUNEL-positive cells in the inner nuclear layer following NMDA exposure (Figure 2C), indicating both anti-inflammatory and anti-apoptotic effects.

Figure 2.

Anti-inflammatory and anti-apoptotic effects of MY-NPs. (A) HE staining of mouse eyeball sections. (B) Immunohistochemistry of TNF-α. (C) Immunofluorescence of Tunel. (D) Quantitative analysis of cell nuclei in ganglion cell layer according to the HE staining (n = 4). (E) Fluorescence intensity of TNF-α (n = 6). (F) Fluorescence intensity of Tunel in ganglion cell layer (n = 5). * p < 0.05; ** p < 0.01; *** p < 0.001.

MY-NPs May Inhibit the Inflammatory Signaling Pathways Activated by NMDA

To explore the underlying mechanisms of MY, RNA sequencing was performed on retinal tissues from the control and NMDA and NMDA+MY-NPs groups. The results indicated that the NMDA model primarily influences pathways related to cytokine-cytokine receptor interaction, JAK-STAT signaling, and TNF signaling, et al, and MY-NPs can mitigate this change to some extent (Figure 3A and B). Additionally, qPCR analysis of retinal tissues revealed a significant upregulation of inflammation-, Chemokine- and apoptosis-related genes, including Myd88, TNF-α, MMP3, and Caspase-3, in the NMDA group, whereas MY-NPs intervention significantly downregulated their expression (Figure 3C–K).

Figure 3.

Sequencing analysis suggests possible mechanism of action of MY-NPs. (A) Gene expression heat map of Ctrl group, NMDA group and NMDA + MY-NPs group (n = 3). (B) KEGG enrichment analysis of differentially expressed genes between Ctrl group and NMDA group (NMDA vs Control). (C–K). qPCR analysis of the relative mRNA expression of TNF-α (C), IL-1β (D), Myd88 (E), MMP3 (F), CD86 (G), CCL2 (H), CCL4 (I), COX2 (J), and Caspase3 (K) in retinal tissues (n = 3). * p < 0.05; ** p < 0.01; *** p < 0.001.

MY-NPs Attenuate Microglial Activation

Considering that microglial activation is a key trigger of retinal inflammation, we further examined the microglial activation marker IBA1. In the NMDA group, a significant increase in IBA1-positive cells was observed in both Layer 1 and 2, along with a larger cell body proportion and shorter branches, indicating substantial microglial activation. In contrast, MY-NPs treatment effectively reduced the number of IBA1-positive cells, decreased the cell body proportion, and promoted branch elongation (Figure 4A–G).

Figure 4.

MY-NPs may play a role by inhibiting microglial activation. (A) IBA1⁺ immunofluorescence of retinal tissue, confocal microscope micrographs at different z-axis levels. (B and C) Relative cell body diameter, branch length in Layer1. (D) Number of IBA1⁺ cells in Layer1 (n = 4). (E and F). Relative cell body diameter, branch length, in Layer2. (G) Number of IBA1⁺ cells in layer2 (n = 4). (H) Immunofluorescence images of CD86, IBA1, and DAPI I. Immunofluorescence images of INOS. (J and K) Quantitative analysis of CD86 and INOS (n = 4). (L and P) qPCR detection of IL-6, CD86, CCL2, CCL4, and INOS expression in cells (n = 4). ns p > 0.05; * p < 0.05; ** p < 0.01; *** p < 0.001.

To further investigate these effects at the cellular level, BV2 microglial cells were cultured and activated using LPS+IFN-γ. Immunofluorescence staining showed that the expression of activation markers CD86 and iNOS was significantly increased in activated microglia but was notably suppressed following MY-NPs intervention (Figure 4H–K). The qPCR analysis revealed a significant increase in the expression of IL-6, CD86, CCL2, CCL4, and iNOS in the LPS/IFN-γ-stimulated group. Importantly, MY-NPs treatment effectively suppressed these inflammatory responses (Figure 4L–P).

MY-NPs Reduce Inflammatory Mediator Production by Activated Microglia and Attenuate R28 Cell Apoptosis

Analysis of the BV2 cell supernatant showed a substantial increase in inflammatory cytokines such as IL-6 and TNF-α, along with elevated NO levels in the LPS+IFN-γ-treated group (Figure 5A–C). MY-NPs administration resulted in a marked reduction of inflammatory responses, although the lowest dose tested (0.2 mg/mL) failed to significantly decrease TNF-α levels. (Figure 5B). Pairwise comparisons among the MY-NP concentrations revealed different concentration-response patterns across the inflammatory endpoints. For IL-6 and TNF-α, 0.4 mg/mL MY-NPs produced significantly greater inhibition than 0.2 mg/mL, whereas no significant difference was observed between 0.4 and 0.8 mg/mL, suggesting that suppression of these cytokines approached a plateau at concentrations ≥0.4 mg/mL. In contrast, inhibition of NO production increased progressively across the three concentrations, with 0.8 mg/mL showing the greatest effect. To evaluate the effects of activated microglia on retinal cells, the retinal ganglion cell line R28 was treated with supernatants from BV2 cells under different conditions. The groups included R28 treated with supernatant from resting BV2 cells (R28-BP), activated BV2 cells (R28-BL), and activated BV2 cells treated with MY-NPs at concentrations of 0.2, 0.4, or 0.8 mg/mL (R28-BL + MY-NPs). The analysis revealed a significant upregulation of the Caspase-3 activity (Figure 5D), and an increase in Annexin V-FITC levels in the R28-BL group (Figure 5E and F). However, 0.4 mg/mL MY-NPs and 0.8 mg/mL MY-NPs treatment markedly reversed those changes in R28 cells, which means MY-NPs might inhibit the early apoptosis of R28 induced by inflammatory factors in activated microglia supernatant. Although 0.8 mg/mL showed the strongest effect on NO production and caspase-3 activity among the concentrations tested, the present data do not establish a single optimal therapeutic concentration. In addition, we did not find that MY-NPs had a direct protective effect on R28 (Figure S3).

Figure 5.

Composition of supernatant of BV2 treated with MY-NPs and its anti-apoptotic effect. (A–C) ELISA detection of IL-6 (n = 4), TNF-α (n = 4) and NO (n = 4) in supernatant of activated BV2 cells treated with different concentrations of MY-NPs (0.2mg/mL, 0.4mg/mL, 0.8mg/mL MY-NPs). (D) Quantification of caspase 3 activity of R28, R28-BP = R28 treated with supernatant of inactivated BV2 culture, R28-BL = R28 treated with supernatant of activated BV2 culture, R28-BL+MY-NPs = R28 treated with supernatant of activated BV2 culture with MY-NPs (n = 3). (E) Flow cytometry results of Annexin V-FITC/PI. The red box shows the percentages of cells in Q2 and Q3. (F) Quantification of Annexin V-FITC (n = 3). ns p > 0.05; * p < 0.05; ** p < 0.01; *** p < 0.001.

Biosafety and Potential Targeting Ability of MY-NPs

To evaluate the biosafety of MY-NPs and facilitate their potential clinical application, a preliminary biosafety analysis was conducted. After intravitreal administration for 7 and 14 days, no significant differences were observed between the treatment and control groups in OCT imaging, visual function assessments, or pathological sections, indicating good ocular tolerance of the MY-NPs delivery system (Figure 6A and B).

Figure 6.

Biosafety and potential targeting of MY-NPs. (A) Fundus photographs, vascular imaging, HE sections of healthy mice and mice treated with MY-NPs at 7 and 14 days. (B) Visual function testing of healthy mice and mice treated with MY-NPs at 7 and 14 days (n = 3). (C and D), In vivo biodistribution of MY in retina (n = 3) and vitreous cavity (n = 5) after 8h, 24h and 72h injection of 2μL MY-NPs. (E and F). Fluorescence imaging and quantitative analysis of Nile red content on the cell surface after Nile red-loaded NPs intervened in activated BV2 cells. ns p > 0.05; * p < 0.05; ** p < 0.01; *** p < 0.001.

In addition, an ocular biodistribution and retention assessment was performed over 72 h following intravitreal administration of MY-NPs. Myricetin was detectable in both the vitreous and retina at 8 h after injection. Its concentration decreased over time; however, measurable levels remained detectable in both ocular compartments at 72 h. These findings indicate persistence of myricetin in ocular tissues for at least 72 h following MY-NP administration but do not constitute a complete pharmacokinetic characterization (Figure 6C and D).

It is well established that activated microglia upregulate CD44 receptors (Figure S4), which serve as specific binding sites for hyaluronic acid (HA). Leveraging this property, the potential targeting ability of MY-NPs was assessed using Nile red-labeled nanoparticles (Nile-NPs). Fluorescence analysis revealed that activated microglia exhibited significantly higher Nile red intensity on their surface at 0.5 and 1 h post-treatment compared to non-activated cells (Figure 6E and F). These findings indicate enhanced nanoparticle uptake by activated microglia and suggest a possible contribution of HA-CD44-associated interactions; however, they do not establish microglia-specific targeting.

Discussion

In this study, we developed HA-PEG-PLGA nanoparticles encapsulating myricetin (MY-NPs), a natural flavonoid with largely unexplored potential for RGC protection. Our results demonstrate that MY-NPs attenuate NMDA-induced RGC loss, preserve visual function, and are associated with reduced microglial activation and inflammatory responses.

Most of the RGC protective drugs are administered intravitreally. The first barrier encountered by intravitreal injection of NPs is the vitreous humor. The vitreous is a highly hydrated gel-like matrix composed of a network of collagen fibers and negatively charged hyaluronic acid, forming a mesh-like structure with pore sizes ranging from approximately 500 nm to 1 μm. The mobility of NPs within the vitreous largely depends on their diffusion and interaction with the vitreous gel. Two key factors influencing this process are surface charge and particle size. Generally, negatively charged nanoparticles with a diameter smaller than 100 nm experience minimal diffusion resistance.²⁸ The NPs used in this study meet these criteria. The HA-PEG-PLGA nanoparticles were designed to overcome two major challenges limiting the clinical application of myricetin: its poor aqueous solubility and rapid systemic clearance. Encapsulation of myricetin within a PLGA core, stabilized by PEG and HA, resulted in a remarkable improvement in solubility and allowed for sustained release over 4 days in vitro. In contrast, conventional carriers like micelles exhibited burst release and instability in the vitreous humor due to dilution effects.²⁹ The HA coating provided potential active targeting through CD44 receptors, which are upregulated on activated microglia.^30,31 These combined features may provide advantages for ocular delivery compared with conventional formulations, although direct comparative studies are required.

Myricetin exhibits excellent antitumor, anti-inflammatory, and antioxidant bioactivities. It is reported that myricetin can exert inhibitory effects in tumors by regulating the PI3K/AKT/MAPK signaling pathways to induce cell cycle arrest and apoptosis.³² Myricetin also involves mechanisms such as antioxidant stress and inhibition of NF-κB-mediated inflammatory responses.³³ However, there were nearly no reports on the mechanisms of myricetin in ophthalmology. Therefore, we conducted a preliminary exploration through retinal tissue sequencing and found that it may primarily regulate cytokine-cytokine receptor interaction, JAK-STAT signaling, and TNF signaling. These pathways might be related to the activation of microglial cells, especially M1 microglial cells, which play a strong pro-inflammatory role in immune responses and can secrete a large number of pro-inflammatory factors to amplify the immune response.³⁴ Moreover, cytokines such as TNF-α and IL-6, along with activation of the JAK-STAT signaling pathway, promote M1 microglial polarization,³⁵ thereby perpetuating a vicious cycle of neuroinflammation. Excessive expression of pro-inflammatory factors can cause RGC damage and apoptosis, while IFN-γ–mediated activation of the JAK-STAT signaling pathway may further exacerbate inflammation and promote RGC apoptosis.³⁶ In this study, MY-NPs were associated with reduced pro-inflammatory microglial activation, decreased expression of inflammatory mediators, and attenuation of RGC apoptosis. These findings suggest that suppression of microglia-associated inflammatory responses may contribute to the neuroprotective effects observed following MY-NP treatment. However, because the present study did not comprehensively characterize distinct microglial phenotypic states, the observed changes should not be interpreted as definitive evidence of altered microglial polarization.

NMDA-induced retinal injury should not be considered exclusively microglia-dependent. Excessive NMDA receptor activation can directly initiate excitotoxic stress in retinal neurons, while neuronal injury and associated danger signals can simultaneously activate retinal glial cells and amplify tissue damage through secondary inflammatory pathways. Thus, the final extent of RGC loss likely reflects the combined contribution of primary excitotoxic injury and subsequent non-cell-autonomous inflammatory amplification.

Within this context, MY-NP treatment was associated with reduced pro-inflammatory microglial activation, decreased inflammatory mediator production, and improved RGC survival. Furthermore, conditioned medium from activated BV2 microglia promoted apoptosis and increased caspase-3 activity in R28 cells, whereas treatment of activated microglia with MY-NPs attenuated these effects. In contrast, MY-NPs did not demonstrate a clear direct protective effect on R28 cells under our experimental conditions. Collectively, these findings support the possibility that attenuation of microglia-associated inflammatory responses contributes, at least in part, to the neuroprotective effects of MY-NPs, rather than indicating that microglial regulation is the sole mechanism responsible for RGC protection.

Despite its encouraging findings, this study has several limitations that should be acknowledged. Firstly, to translate this research into optic nerve neuroprotection for glaucoma, the NMDA-induced acute excitotoxicity model effectively replicates RGC apoptosis—a key feature of glaucomatous neurodegeneration—but it does not fully capture chronic optic nerve axon degeneration or elevated intraocular pressure (IOP). Future studies should validate these findings in chronic IOP-dependent models (eg, microbead occlusion or DBA/2J mice) to assess their impact on axonal integrity. Secondly, while RNA-seq identified cytokine-related and JAK-STAT pathways, the upstream mechanisms connecting myricetin to pro-inflammatory microglial activation remain unclear. Single-cell sequencing could help identify the specific microglial subpopulations influenced by MY-NPs. And transcriptomic changes do not necessarily indicate corresponding protein-level pathway activation. Additional validation using protein-based approaches, such as Western blotting or phospho-specific immunostaining, will be required to confirm the involvement of these signaling pathways. Thirdly, the biosafety assessment in this study was limited to a period of 14 days. Longer-term monitoring is necessary to evaluate potential cumulative toxicity or immune responses from repeated injections. Fourthly, CD44 is not a microglia-specific receptor. In this study, it only suggests the potential targeting capability of MY-NPs. And the cellular distribution and microglial targeting specificity of MY-NPs were not fully conclusively demonstrated. Further in vivo investigations utilizing cell-type-specific markers or genetic tracing techniques would be valuable to more definitively characterize the biodistribution and cell-type selectivity of MY-NPs within retina tissues. Fifthly, the nanoparticle preparation was evaluated using a single formulation batch. Although the preparation procedure was standardized, independent batch-to-batch reproducibility was not systematically assessed. Future studies should evaluate multiple independently prepared batches and compare physicochemical parameters, including particle size, PDI, zeta potential, drug loading, and encapsulation efficiency, to establish the robustness and scalability of the MY-NP formulation. Another limitation is the absence of a free-myricetin treatment group. Therefore, although the present study demonstrates the protective effects of MY-NPs against NMDA-induced retinal injury, it cannot determine the extent to which nanoparticle encapsulation itself enhances the therapeutic efficacy of myricetin. A direct comparison between free MY and MY-NPs, using appropriately matched doses and vehicle controls, will be required in future studies to clarify the contribution of the delivery system to ocular pharmacokinetics and neuroprotective efficacy.

To facilitate clinical translation, future studies should focus on optimizing nanoparticle pharmacokinetics through methods such as tracer isotopes or fluorescence imaging. Additionally, exploring non-invasive delivery methods, such as suprachoroidal injection, could enhance patient compliance. Combining MY-NPs with IOP-lowering agents, like Rho kinase inhibitors, should also be investigated for potential synergistic effects.

Conclusion

This study demonstrates that HA-PEG-PLGA nanoparticles provide a feasible delivery platform for myricetin and are associated with protection against NMDA-induced excitotoxic retinal injury. MY-NP treatment preserved retinal function, reduced RGC loss, and attenuated microglia-associated inflammatory responses in an acute injury model. These findings support MY-NPs as a potential experimental neuroprotective strategy for retinal neurodegenerative conditions. However, because the present study was conducted in an acute NMDA-induced injury model, further validation in chronic glaucoma-relevant models, together with long-term safety and mechanistic studies, will be necessary to determine the future translational potential of this approach.

Funding Statement

This work was supported by National Natural Science Foundation of China (82201222, 82501281); Hunan Natural Science Foundation Project (grant numbers 2025JJ90261, 2026JJ81525). Foundation of HNXJ PHILANTHROPY (grant numbers KY25014). Aier Eye Hospital Group Research Fund (AMK2501D37, AMF2501D38).

Abbreviations

MY, myricetin; NP, nanoparticle; MY-NPs, myricetin-loaded nanoparticles; IOP, intraocular pressure; HA, hyaluronic acid; PEG, Polyethylene Glycol; PLGA, Poly (lactic-co-glycolic acid); NMDA, N-methyl-D-aspartate; RGCs, retinal ganglion cells; DMSO, dimethyl sulfoxide; DMF, dimethylformamide; ERG, Electroretinography; TEM, transmission electron microscopy.

Data Sharing Statement

The data presented in this study are available on request from the corresponding author.

Ethics Approval and Consent to Participate

The ethics approval has been obtained (AEI20230079) by Hunan Aier Eye Institute Animal Welfare and Ethics Committee.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

There is no conflict of interest in this study.