Targeted Neuroprotection of Retinal Ganglion Cells Via AAV2-hSyn-NGF Gene Therapy in Glaucoma Models

Abstract

Purpose: The purpose of this study was to evaluate the neuroprotective effects of delivering nerve growth factor (NGF) to retinal ganglion cells (RGCs) through adeno-associated virus serotype 2 (AAV2) carrying a neuronal-specific human synapsin (hSyn) promoter.

Methods: AAV2-hSyn-NGF was injected intravitreally in three glaucoma models: optic nerve crush (ONC), microbead-induced ocular hypertension (MB), and genetic glaucoma model (DBA). Quantitative polymerase chain reaction (qPCR) and enzyme-linked immunosorbent assay (ELISA) determined the optimal injection concentration of AAV vector. Flow cytometry monitored immune responses. Transduction efficiency was quantified using green fluorescent protein (GFP) co-localization with RGC-specific marker RNA-binding protein with multiple splicing (RBPMS). The RGCs’ density, retinal nerve fiber density, ganglion cell complex thickness, and positive scotopic threshold response (pSTR) were measured to assess structural and functional outcomes of the RGCs. Non-parametric Mann-Whitney U tests or Kruskal-Wallis tests were utilized to ascertain the statistical significance (P < 0.05).

Results: The optimal concentration of AAV vector for intravitreal injection was determined to be 1 × 10¹⁰ vector particles (VPs) per eye. The use of the hSyn promoter significantly enhanced targeting specificity to RGCs, resulting in a transduction efficiency of 46.64% ± 2.18%. Administration of AAV2-hSyn-NGF effectively preserved the RGCs’ density, nerve fiber layer integrity, and the thickness of ganglion cell complex, while maintaining the RGCs’ function across three glaucoma models. Furthermore, this gene delivery system did not elicit detectable immune responses or structural damage to the retina.

Conclusions: The AAV2-hSyn-NGF gene therapy offers a safe and effective neuroprotective strategy for RGCs across multiple glaucoma models, making it a promising candidate for future clinical trials in patients with glaucoma.

Glaucoma is a prevalent ocular neurodegenerative condition characterized by the death of retinal ganglion cells (RGCs) and axonal degeneration.^1–3 The primary treatment approach involves using targeted methods such as medication and surgery to mitigate the damage caused by elevated intraocular pressure (IOP) to the retinal nerves.^4,5 However, lowering IOP dose not consistently prevent the progressive death of RGCs and subsequent visual impairment.^6,7 Currently, the mechanism underlying RGCs’ death remains elusive, with potential causes including ischemia, oxidative stress, mitochondrial dysfunction, activation of apoptotic signals, and neurotrophic factor deprivation.^8–13 Because regenerating lost RGCs poses challenges, safeguarding against RGCs’ death induced by glaucoma remains a paramount concern.⁶ 

Neuroprotective strategies independent of IOP primarily fall into three categories: cell therapy, non-cellular therapy, and gene delivery-based neuroprotective therapy.¹⁴ In 2017, the introduction of Voretigence Neparvovec (Luxturna), the first adeno-associated virus (AAV)-based eye gene therapy drug, showcased the significant potential of gene therapy in treating ocular diseases.¹⁵ Notably, extensive research efforts to develop gene therapies for protecting RGCs were actively underway. For instance, brain-derived neurotrophic factor (BDNF), erythropoietin (EPO), ATP-binding cassette transporter A1 (ABCA1), and nuclear factor erythroid 2-related factor 2 (NRF2) were transduced into retinal cells using AAVs to mitigate the loss of RGCs in animal models of glaucoma.^16–20 However, the models utilized in these studies inherently leaned toward acute RGC injury and deviated from the pathological changes seen in chronic neurodegenerative diseases like glaucoma. Despite promising results in animal models, this discrepancy still obstructed its advancement toward clinical trials.^21,22 In our study, we used three animal models of glaucoma, with the longest observation period exceeding 5 months. 

Nerve growth factor (NGF) is a frequently utilized neurotrophic factor (NTF) in clinical nerve injury treatment, extensively applied in neurotrophic keratopathy and optic nerve diseases.^23,24 Studies have demonstrated that the absence of NTFs could impede axonal transport, ultimately resulting in cell apoptosis, which is also a significant pathogenic mechanism of glaucoma.^25,26 However, the dual effect of neurotrophins (NTs) influences their protective role against RGCs in glaucoma. Researches have demonstrated that NTFs, such as NGF, can bind to tropomyosin-related kinase (Trk) receptors on RGCs to promote neuronal survival, but can also indirectly induce apoptosis of RGCs by binding to the p75 NT receptor (p75^NTR) on Müller cells.^27–29 In our study, we used the neuronal-specific human synapsin (hSyn) promoter, known for its high specificity to RGCs.³⁰ This targeted expression design limited off-target NGF diffusion, aiming to minimize p75^NTR-mediated apoptosis of RGCs. 

In this study, AAV2-NGF carrying a specific hSyn promoter demonstrated substantial protective effects on RGCs in all three glaucoma models and did not induce additional immune responses, holding potential for future clinical trials in patients with glaucoma. 

Methods

Animals

Eight-week-old C57BL/6 and DBA/2J mice were obtained from Beijing Vital River Laboratory Animal Technology Co., Ltd. (Beijing, China). They were housed under standard humidity and temperature conditions and maintained on a 12-hour light/dark cycle at the Animal Center of Shandong Eye Institute, with ad libitum access to water and food. Three distinct glaucoma animal models were utilized in this study: the optic nerve crush (ONC) model, the microbead-induced ocular hypertension (MB) model, and the genetic glaucoma model (DBA). For each model, animals were divided into four groups: the wild type (WT) group, the glaucoma group (ONC, MB, and DBA), the glaucoma with AAV2-hSyn-NGF intravitreal injection group (ONC + NGF, MB + NGF, and DBA + NGF), and glaucoma with AAV2-hSyn-control (Con) intravitreal injection group (ONC + Con, MB + Con, and DBA + Con), with 45 mice in each group. All experimental procedures were approved by the Ethics Committee of Qingdao Eye Hospital of Shandong First Medical University and conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Inducible ONC Model of Glaucoma

The ONC model was established in C57BL/6 mice following the established protocol.³¹ In brief, the surgical procedure was performed under a ZEISS surgical microscope using sterilized microsurgical instruments. Mice were anesthetized via intraperitoneal injection of sodium pentobarbital (50 mg/kg) and topically anesthetized with 0.5% proparacaine hydrochloride eye drops. The optic nerve was exposed by opening the bulbar conjunctiva to separate the orbital soft tissue. Subsequently, the optic nerve was crushed 1 to 2 mm posterior to the optic disk for 10 seconds using reverse microscopic self-closing forceps (Dumont #N7, Roboz, Cat #RS5027). After the procedure, the tissue was returned to its original position. Then, the mice were placed on a 37°C heating pad for recovery and returned to their cages after natural awakening. The same anesthesia protocol and postoperative care measures were applied to the other procedures in this study. 

Inducible MB Model of Glaucoma

In order to reduce microbead deposition in the visual axis, the MB glaucoma model in C57BL/6 mice was optimized based on the method described by Urcola et al.³² Briefly, a 10 µL Hamilton syringe with a 36-gauge needle was used to puncture into the posterior chamber from 1 mm behind the limbus. After entering the anterior chamber through the pupil, 2 µL of 4.5 µm polystyrene microbeads (BaseLine Chromtech Research Center, Tianjin, China) were injected into the iridocorneal angle. The injected microbeads accumulated in the trabecular meshwork and Schlemm's canal, mechanically obstructing the physiological outflow pathway of aqueous humor. 

Monitoring of IOP

IOP was measured using the iCare TONOLAB tonometer (Colonial Medical Supply, Espoo, Finland). The tonometer automatically eliminated erroneous values and calculated the final result by averaging five measurements. To minimize measurement errors, all IOP measurements were conducted by the same ophthalmologist in mice without anesthesia and stress. IOP measurements were conducted between 14:00 and 16:00 Beijing Standard Time. To ensured that subsequent experimental analyses were physiologically consistent, the models that exhibited significant IOP fluctuations were systematically eliminated from further investigation. 

AAV Vector Injection

AAV-NGF and AAV-Con vectors containing the neuronal specific hSyn promoter were provided by OBiO Technology Co., Ltd. (Shanghai, China). The AAV-NGF contained NGF (Mouse, NM_013609) under the control of a hSyn promoter, with green fluorescent protein (GFP) fused (AAV2-hSyn-Ngf-3 × Flag-IRES2-GFP-WPRE). AAV carrying only the GFP (AAV2-hSyn-MCS-3 × Flag-IRES2-GFP-WPRE) was used as a negative control (AAV-Con). Multiple cloning site (MCS) allowed the insertion of any gene of interest, such as NGF, which could be tagged with a 3 × Flag sequence for protein detection and purification. The internal ribosome entry site 2 (IRES2) sequence facilitated the independent translation of the inserted gene and GFP, with GFP serving as a reporter for tracking gene expression. The woodchuck hepatitis virus post-transcriptional regulatory element (WPRE) enhanced mRNA stability and translation efficiency, optimizing the overall expression of the transgene. AAV-CMV-NGF shared the same functional components as AAV-hSyn-NGF except for the promoter and was used solely as a control in Western blot to detect retinal cell apoptosis. A 1.5 µL volume of AAV vector (1 × 10⁷−1 × 10¹⁰ viral particles [VPs] per eye) was slowly injected into the vitreous cavity using a 10 µL syringe with a 36-gauge needle positioned 3 mm behind the limbus. After injection, the needle was left in the vitreous cavity for 30 seconds to facilitate uniform diffusion of the carrier solution. All injection procedures were performed by the same ophthalmologist. Three weeks after the AAV-NGF injection, a 1 × 1 mm² field of view from the central region (1–2 mm from the optic disc) of each quadrant was selected to assess transduction efficiency. The transduction efficiency of AAV-NGF was determined by counting the proportion of GFP-positive RGCs relative to the total RGC population. The RGC-specific gene RNA-binding protein with multiple splicing (RBPMS) was utilized to quantify the number of RGCs.^33,34 

Optical Coherence Tomography Image Acquisition

The anesthesia method was the same as previously described, and pupil dilation was achieved using tropicamide phenylephrine eye drops (Mydrin-P). The fully anesthetized and mydriatic mouse was positioned on the operating table. While ensuring the eye surface remained moist, retinal images around the optic papilla were captured using an optical coherence tomography (OCT) device, the Small Animal Ophthalmic Ultramicroscopic Imaging System (Optoprobe Science Ltd., United Kingdom). Subsequently, the built-in software was used to automatically segment the retinal image and measure the thickness of the ganglion cell complex and the whole retinal layer. The ganglion cell complex, which comprised the retinal nerve fiber layer (RNFL), ganglion cell layer (GCL), and inner plexiform layer (IPL), was widely used for evaluating the pathological status of the retina in glaucoma models.^35,36 

Electroretinogram and Visual Evoked Potential Recording

We allowed the mouse to acclimate in a dark room overnight before data collection. Under dim red light, the fully anesthetized and mydriatic mouse was placed on a rodent examination table equipped with a constant temperature heating pad. The recording electrodes were positioned, along with the reference electrodes, and grounding electrodes on the cornea surface, postauricular skin, and tail skin, respectively. Scotopic flash electroretinogram (ERG) was recorded by Hand-held Multi-species Electroretinography (HMsERG; OcuScience, Henderson, NV, USA) from −5.0 log to −2.0 log with a 0.5 log increment in flash intensity. A detectable positive scotopic threshold response (pSTR) was recorded at −4.0 log. The pSTR was regarded as the signal that most accurately reflected the function of RGCs.^37,38 During flash visual evoked potential (VEP) acquisition, a subcutaneous recording electrode was positioned at the midline scalp region equidistant from the bilateral auditory meatus. Contralateral eyes were occluded with blackout goggles to ensure monocular stimulation. Full-field flashes (20 ms duration, 3 cd.s/m² intensity) were delivered at 1 hertz (Hz) using a Ganzfeld stimulator. Thirty responses were averaged to obtain the final waveform. 

Quantitative Real-Time PCR

Mice were euthanized using an overdose of anesthetic. Retinal tissue was collected 3 weeks after intravitreal injection of AAV-NGF for qPCR analysis. Retinal total RNA was extracted using the TransZol Up Plus RNA Kit (ER501-01-V2; TransGen Biotech, Beijing, China). Subsequently, complementary DNA (cDNA) was synthesized with the PrimeScript RT Reagent Kit (RR037A; TaKaRa Bio Inc., Shiga, Japan). The forward primer for NGF was 5′-GCCAAGGACGCAGCTTTCTA-3′, and the reverse primer was 5′-TTCAGGGACAGAGTCTCCTTCTG-3′. Quantitative PCR analysis was conducted using the QuantStudio 5 Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). 

Enzyme Linked Immunosorbent Assay

Mouse retina was harvested 3 weeks post-intravitreal injection of AAV vectors. The concentration of NGF/TrkA/p75^NTR protein was quantified using a Mouse NGF/TrkA/p75^NTR ELISA Kit (Jiangsu Meimian Industrial Co. Ltd., Jiangsu, China) following the manufacturer's protocol. Absorbance was measured at 450 nm using a microplate reader (Molecular Devices, Sunnyvale, CA, USA). 

Western Blot

Retina was obtained for Western blot analysis 3 weeks following the AAV vector injection. A volume of 100 µL NP40 Lysis Buffer was added to the homogenized retinal tissue and incubated on ice for 30 minutes. Following centrifugation at 12,000 rpm for 20 minutes at 4°C, the supernatant was collected, and the protein concentration was measured using an Enhanced BCA Protein Assay Kit (P0010; Beyotime, China). The sample was then mixed with 5 × loading buffer and heated at 95°C for 10 minutes. Subsequently, electrophoresis was performed on an SDS-PAGE gel. Following the transfer of proteins onto a PVDF membrane (Millipore, Bedford, MA, USA), the membrane was blocked with 5% skim milk and incubated overnight with cleaved caspase-3 antibody (1: 1000, #9661; Cell Signaling Technology) at 4°C. The membrane was subsequently incubated with an HRP-conjugated secondary antibody (1:3000, 20000798; Proteintech) for 1 hour. Immunoreactive bands were visualized using the ChemiDoc Touch imaging system (Bio-Rad, USA). Protein bands were quantified using ImageJ grayscale image peak analysis. The GAPDH antibody (1:4000, LK9002; Tianjin Shanggen Biotechnology Co., Ltd.) served as a loading control. 

Flow Cytometry

Flow cytometry was conducted on retinal tissue following previously established protocols.³⁹ Retina tissue was enzymatically digested in RPMI 1640 medium containing 2 mg/mL Collagenase A (Roche Diagnostics, Germany) for 30 minutes to achieve a single-cell suspension. The resulting single-cell suspension was filtered through a 35 µm cell strainer (Falcon, Corning Inc., Corning, NY, USA) and washed twice with PBS. Samples were incubated with APC-conjugated anti-mouse Ly-6G antibody (Cat #164505; BioLegend), APC-conjugated anti-mouse CD45 antibody (Cat #103111; BioLegend), APC-conjugated anti-mouse CD19 antibody (Cat #115511; BioLegend), APC-conjugated anti-mouse CD8a antibody (Cat #100712; BioLegend), APC-conjugated anti-mouse CD4 antibody (Cat #100411; BioLegend), APC-conjugated anti-mouse F4/80 antibody (Cat #123116; BioLegend), APC-conjugated Rat IgG2a, κ Isotype Ctrl antibody (Cat #400511; BioLegend), and APC-conjugated Rat IgG2b, κ Isotype Ctrl antibody (Cat #400611; BioLegend) at 4°C in the dark for 30 minutes, respectively. Subsequent to washing away excess antibodies, a total of 10,000 cells per retinal sample were acquired using a CytoFLEX flow cytometer (Beckman Coulter, Indianapolis, IN, USA) and analyzed using FlowJo V10.6 software (Tree Star, Woodburn, OR, USA). Gating strategies were established based on relevant isotype controls. 

Immunofluorescence Staining

The excised eyeball was fixed in 4% PFA at room temperature for 2 hours. Subsequently, the isolated retina was permeabilized with PBS containing 1% Triton X-100 for 2 hours and blocked with PBS containing 0.1% Triton X-100 and 1% BSA overnight. The retina was then incubated with anti-RBPMS antibody (1:400, Cat #GTX118619; GeneTex) at 4°C for 2 days. Following 3 washes in PBS, the retina was incubated with anti-rabbit Alexa Fluor 594 (1:400, Cat #ab150080; Abcam) at room temperature in the dark for 2 hours. The retina was then thoroughly washed with PBS, followed by even division into four lobes, and observed using a Zeiss LSM880 confocal microscope. A 1 × 1 mm² field of view from the central region (1–2 mm from the optic disc) of each quadrant was chosen for counting stained positive cells, with the resulting average of the four quadrants considered as the RGC density of the whole retina. The counting operation was independently reviewed by three separate ophthalmologists who were unaware of the experimental groups using Image J software. Similarly, retinal nerve fibers were stained with Alexa Fluor 594 anti-tubulin beta 3 antibodies (1:400, Cat #801208; BioLegend). 

Toluidine Blue Staining and Axon Quantification

Optic nerve tissues were harvested from euthanized mice and immediately fixed in 2.5% glutaraldehyde solution for 24 hours at 4°C. Following post-fixation with 1% osmium tetroxide, the specimens were dehydrated through a graded ethanol series and embedded in epoxy resin. Semi-thin cross-sections (1 µm thickness) were cut using an ultramicrotome (Leica EM UC7) and stained with 1% toluidine blue solution for precise myelin visualization. 

For systematic axon counting, stained sections were imaged under 100 × oil immersion objective using a light microscope (Leica DM4 B) equipped with a digital camera. Four non-overlapping fields from both peripheral and central regions of each nerve were randomly selected for analysis to ensure representative sampling. Axon quantification was performed using Image J software. All counts were conducted by three blinded investigators. 

Hematoxylin and Eosin Staining

Mouse eyes were fixed in 4% paraformaldehyde solution for 24 hours at 4°C to preserve intraocular structures. After paraffin embedding, sagittal sections (4 µm thickness) were obtained using a rotary microtome (Leica RM2235). Sections were deparaffinized in xylene, rehydrated through graded alcohols, and stained with Mayer's hematoxylin for 5 minutes to visualize nuclei. Differentiation was performed in 0.5% acid alcohol, followed by bluing in 0.2% ammonium hydroxide solution. Cytoplasmic staining was achieved with eosin Y for 1 minute. Stained sections were dehydrated and mounted with neutral balsam for observation. 

Statistics

Statistical analysis was performed using SPSS 23.0. Measurement results were expressed as mean ± standard deviation (SD). The comparisons between the two groups were conducted using the non-parametric Mann-Whitney U test, and the non-parametric Kruskal-Wallis test was utilized for comparisons between multiple groups. Statistical significance was defined as a P value < 0.05. 

Results

AAV-NGF Carrying an hSyn Promoter Specially Targeting RGCs

To target RGCs and induce overexpression of NGF, we designed an AAV2 vector carrying an hSyn promoter and NGF coding gene (AAV-NGF; Fig. 1A). Retinal tissue was collected 3 weeks after the intravitreal injection of AAV-NGF to assess transduction efficiency and NGF expression (Fig. 1B).^20,30 In comparison to the injection of 1 × 10¹⁰ VPs per eye, lower concentrations of AAV-NGF could only achieve limited transduction of RGCs, resulting in relatively low NGF expression levels (Figs. 1C–E). Consequently, a dose of 1 × 10¹⁰ VPs/eye was adopted in subsequent experiments. Although AAV vectors carrying the hSyn promoter still induced apoptosis upon injection, they demonstrated reduced retinal cell apoptosis compared to the CMV promoter 3 weeks after intravitreal injection (Figs. 1F, 1G). Within 9 weeks following the AAV-NGF injection, the expression levels of TrkA receptors and p75^NTR were upregulated in the retina tissue (Figs. 1H, 1I), a finding consistent with prior reports.^40,41 Immunofluorescence analysis demonstrated that with the hSyn promoter, the AAV vector achieved a specific transduction rate of up to 46.64% ± 2.18% for RGCs (Fig. 1J). Furthermore, wholemount 3D retinal images showed that the expression of GFP carried by the AAV vector was concentrated in the RGC layer (Fig. 1K, Supplementary Fig. S1). 

Figure 1.

 

Induction of NGF overexpression in the retina through intravitreal injection of AAV-NGF. (A) Key functional components of AAV vectors. (B) Timeline of experimental procedures. (C) GFP expression in the retina after intravitreal injection of AAVs (n = 3). (D) Relative NGF expression in the retina with varying concentrations of AAV-NGF injection (n = 3). (E) NGF protein concentration in the retina with varying concentrations of AAV-NGF injection determined by ELISA (n = 3). (F) Expression of apoptosis marker cleaved caspase-3 in the retinal tissue following intravitreal injection of WT control, AAV-Con, AAV-hSyn-NGF, and AAV-CMV-NGF groups (n = 3). (G) Quantitative analysis of cleaved caspase-3 expression. (H) Time-dependent changes in TrkA receptors expression in retinal tissue after AAV-NGF injection for 3, 6, and 9 weeks (n = 3). (I) Time-dependent changes in p75^NTR receptors expression in retinal tissue after AAV-NGF injection for 3, 6, and 9 weeks (n = 3). (J) GFP distributions in the RGC layer after injecting AAV vectors of 1 × 10¹⁰ VP/eye. White arrows indicate co-localization of GFP with RGCs (n = 3). (K) Representative 3D images of GFP (green), RBMPS (red), and DAPI (blue) expression in the retina after injecting AAV vectors of 1 × 10¹⁰ VP/eye. White arrows indicate the RGC layer (n = 3). Statistical analysis was performed using the Mann-Whitney U test. *P < 0.05.

No Detectable Immune Responses or Structural Damage After Intravitreal Injection of AAV-NGF

Flow cytometry was conducted to assess the safety of AAV-NGF after 3 weeks of intravitreal injection by detecting the infiltration of immune cells into the retina. No significant differences were detected in all the immune markers among the WT, AAV-Con, and AAV-NGF groups, including CD45⁺ leukocytes (0.62% ± 0.19% in WT, 0.88% ± 0.18% in AAV-Con, and 0.55% ± 0.03% in AAV-NGF), F4/80⁺ macrophages (0.38% ± 0.18%, 0.49% ± 0.09%, and 0.30% ± 0.04%), Ly6G⁺ neutrophils (0.32% ± 0.09%, 0.53% ± 0.13%, and 0.38% ± 0.02%), CD19⁺ B cells (0.41% ± 0.14%, 0.46% ± 0.14%, and 0.43% ± 0.07%), CD4⁺ T cells (0.33% ± 0.11%, 0.41% ± 0.12%, and 0.26% ± 0.17%), and CD8a⁺ T cells (0.50% ± 0.07%, 0.50% ± 0.13%, and 0.51% ± 0.06%; Figs. 2A, 2B). Consistent with the immunological findings, hematoxylin and eosin (H&E) staining of ocular sections revealed intact retinal architecture across all experimental groups. All illustrated in Figure 2C, no histopathological abnormalities were observed in the WT, AAV-Con, or AAV-NGF groups. 

Figure 2.

 

Flow cytometry analysis of major immune cells and histological evaluation of retinal morphology following intravitreal injection of AAVs. (A) Representative fluorescence-activated cell sorter (FACS) plots showing immune cell populations in the retina of WT, AAV-Con, and AAV-NGF groups (n = 3). (B) Quantitative analysis of the proportions of CD45⁺, F4/80⁺, Ly6G⁺, CD19⁺, CD4⁺, and CD8a⁺ cells in the retina (n = 3). (C) Representative H&E staining images of retinal sections from WT, AAV-Con, and AAV-NGF groups (n = 3). Statistical analysis was performed using the Kruskal-Wallis test. ns = not significant.

AAV-NGF Injection Reduced RGC Loss in the ONC Model

The ONC model was induced 3 weeks after intravitreal injection of AAV vector, followed by the collection of retinal tissue for analysis 2 weeks later (Fig. 3A).⁴² For the ONC group, the retinas were collected for analysis 2 weeks after injury.^42,43 Comparatively, the density of RGCs significantly decreased in mice subjected to optic nerve crushing for 10 seconds (261.67 ± 20.50/mm²), as opposed to WT mice (3152.33 ± 98.53/mm², P < 0.05). Inducing the ONC model after pre-injection of AAV-NGF significantly increased the survival of RGCs (564.67 ± 22.57/mm²; P < 0.05). Conversely, pre-injection of AAV-Con failed to prevent the loss of RGCs following optic nerve crushing (293.67 ± 15.96/mm²; Figs. 3B, 3C). This finding indicated a protective effect of AAV-NGF on RGCs in the ONC-induced glaucoma model. 

Figure 3.

 

AAV-NGF conferred protection against RGC loss and functional impairment in ONC-induced glaucoma. (A) Timeline of experimental procedures. (B) Representative images of RBPMS-positive RGCs (red) in WT, ONC, ONC + NGF, and ONC + Con groups (n = 3). (C) Quantification of RBPMS-positive RGCs (n = 3). (D) Immunofluorescent staining of Beta III Tubulin (red) in mice retina (n = 3). (E) Representative OCT images of retinal structure centered on the optic disk. The ganglion cell complex (RNFL and GCL and IPL) spanned between the red and blue lines, whereas the whole retinal layer extended between the red and yellow lines (n = 3). (F) Quantitative analysis of the average thickness of the ganglion cell complex (n = 3). (G) Quantitative analysis of the average thickness of the whole retinal layer (n = 3). (H) Representative pSTR electroretinogram traces in WT, ONC, ONC + NGF, and ONC + Con groups (n = 3). (I) Quantification of pSTR amplitude (n = 3). (J) Quantification of pSTR latency (n = 3). Statistical analysis was performed using the Mann-Whitney U test and Kruskal-Wallis test. *P < 0.05, ***P < 0.01. ns = not significant.

Another notable pathological change in glaucoma was the density of retinal nerve fibers, evaluated via Beta III Tubulin staining.^43,44 Optic nerve crushing significantly reduced the density of retinal nerve fibers. However, injecting AAV-NGF 3 weeks prior considerably reduced the damage caused by optic nerve crushing, whereas AAV-Con did not exhibit the same effect (Fig. 3D). Furthermore, this observation was validated in live mice through the acquisition of retinal OCT images (Fig. 3E). Optic nerve crushing caused a significant reduction in both the thickness of the ganglion cell complex (RNFL and GCL and IPL; 62.13 ± 0.54 µm) and the whole retina (219.37 ± 3.41 µm) compared to WT mice (70.63 ± 0.29 µm and 238.13 ± 4.25 µm, P < 0.05). AAV-NGF administration effectively prevented the reduction in total retinal thickness (238.23 ± 4.13 µm) induced by optic nerve crushing. However, although the injection of AAV-NGF did not entirely prevent the decrease in the thickness of the ganglion cell complex (68.83 ± 0.58 µm), the decrease was significantly smaller than that observed in the ONC and ONC + Con (61.57 ± 0.25 µm) groups (P < 0.05; Figs. 3E, 3F, 3G). 

To further assess the protective effects of AAV-NGF on optic nerve integrity, axon density was quantified using optic nerve cross-sections. Consistent with previous findings, optic nerve crushing significantly reduced axon density compared with the WT groups (2.09 × 10⁵ ± 0.30 × 10⁵ vs. 6.13 × 10⁵ ± 0.38 × 10⁵ axons/mm², P < 0.05). AAV-NGF treatment mitigated this loss, resulting in a significantly higher axon density (3.20 × 10⁵ ± 0.31 × 10⁵ axons/mm²) compared to the untreated ONC group (P < 0.05). In contrast, AAV-Con injection failed to preserve axonal integrity, with axon density (1.96 × 10⁵ ± 0.29 × 10⁵ axons/mm²) comparable to the ONC group (Supplementary Figs. S2A, S2B). 

Apart from structural alterations, AAV-NGF injection significantly preserved the function of RGCs against impairment induced by glaucoma. Optic nerve crushing significantly reduced the amplitude (18.00 ± 0.75 µV) and prolonged the latency (186.43 ± 4.56 ms) of pSTR, compared to the WT group (48.13 ± 0.45 µV, 142.33 ± 8.21 ms), indicating a significant impairment of RGC function (P < 0.05). Relative to the AAV-Con group (18.03 ± 0.54 µV, 194.6 ± 9.02 ms), AAV-NGF intravitreal injection significantly ameliorated the reduction in amplitude (39.46 ± 0.77 µV) and prolonged latency (147.63 ± 3.06 ms) induced by optic nerve crushing, bringing it close to the WT level (P < 0.05; Figs. 3H, 3I, 3J). In addition, VEP measurements revealed a marked reduction in N2P2 wave amplitude in the ONC group (4.21 ± 0.20 µV) compared to the WT group (20.37 ± 1.34 µV, P < 0.05). However, intravitreal injection of AAV-NGF led to a substantial recovery in amplitude (14.03 ± 0.70 µV, P < 0.05), bringing it closer to the WT level. In contrast, the amplitude in the ONC + Con group (4.37 ± 0.37 µV) did not show significant improvement compared to the ONC group, indicating that only AAV-NGF specifically provided a protective effect (Supplementary Figs. S2C, S2D). In summary, our AAV-NGF system efficiently restored both the structural integrity and the function of RGCs after optic nerve injury. 

AAV-NGF Injection Rescued RGCs in the MB Model

The MB model is a well-established pressure-dependent glaucoma model capable of rapidly experiencing elevated IOP and subsequent loss of RGCs within a brief period.³⁵ The MB model was induced 1 week after intravitreal injection of AAVs.^16,37 Four weeks later, comprehensive evaluation of RGCs was conducted regardless of whether AAV injection was performed or not (Fig. 4A).^16,37 The injected microbeads obstructed the outflow pathway of aqueous humor, resulting in sustained IOP of 20.17 ± 1.96 millimeters of mercury (mm Hg) in the subsequent weeks (Fig. 4B). Following successful induction of glaucoma for 4 weeks, a significant reduction in the number of RGCs (1851.00 ± 95.94/mm²) was observed (P < 0.05), consistent with findings from previous studies.^16,45,46 Pre-injection of AAV-NGF before inducing the MB model significantly reduced the loss of RGCs (2701.33 ± 55.63/mm²; P < 0.05). Conversely, pre-injection of AAV-Con had no effect on the survival of RGCs (1888.67 ± 39.24/mm²; Figs. 4C, 4D). 

Figure 4.

 

AAV-NGF injection rescued RGCs in microbead-induced glaucoma. (A) Timeline of experimental procedures. (B) Fluctuation in IOP within 4 weeks after microbead injection (n = 3). (C) Representative images of RBPMS-positive RGCs (red) in the WT, MB, MB + NGF, and MB + Con groups (n = 3). (D) Quantification of RBPMS-positive RGCs (n = 3). (E) Immunofluorescent staining of Beta III Tubulin (red) in mice retina (n = 3). (F) Representative OCT images of retinal structure centered on the optic disk. The ganglion cell complex (RNFL and GCL and IPL) spanned between the red and blue lines, whereas the whole retinal layer extended between the red and yellow lines (n = 3). (G) Quantitative analysis of the average thickness of the ganglion cell complex (n = 3). (H) Quantitative analysis of the average thickness of the whole retinal layer (n = 3). Statistical analysis was performed using the Mann-Whitney U test and Kruskal-Wallis test. *P < 0.05, ***P < 0.01. ns = not significant.

Immunofluorescence staining of retinal nerve fibers confirmed the neuroprotective effect of AAV-NGF in the MB-induced glaucoma model. Although quantitative analysis was limited by the staining background from the retina, the application of AAV-NGF visually attenuated the reduction in the density of retinal nerve fibers caused by elevated IOP (Fig. 4E). Furthermore, compared with the MB (63.37 ± 0.42 µm) and MB + Con (64.10 ± 0.79 µm) groups, OCT images showed a significantly thicker ganglion cell complex after pre-injection of AAV-NGF (68.80 ± 1.20 µm, P < 0.05; Figs. 4F, 4G). However, although differences were observed in the thickness of the whole retinal layer among the MB (225.90 ± 1.70 µm), MB + NGF (233.37 ± 5.24 µm), and MB + Con (221.20 ± 3.54 µm) groups, there was no statistical difference (Fig. 4H). 

The neuroprotective effect of AAV-NGF was further confirmed by assessing the density of optic nerve axons in the MB model. Compared to the WT group (6.06 × 10⁵ ± 0.34 × 10⁵ axons/mm²), the MB induction significantly reduced axonal density to 4.76 × 10⁵ ± 0.27 × 10⁵ axons/mm² (P < 0.05). However, AAV-NGF pre-treatment restored axonal counts to 5.70 × 10⁵ ± 0.14 × 10⁵ axons/mm², significantly higher than the MB group (P < 0.05) and reaching 92.3% of the WT baseline level. In contrast, the MB + Con group (4.84 × 10⁵ ± 0.26 × 10⁵ axons/mm²) exhibited no significant difference in axon density compared to the MB group (P > 0.05; Supplementary Figs. S3A, S3B). 

AAV-NGF Injection Restored RGCs in Genetic Glaucoma Model

DBA/2J mice, serving as a model of genetic glaucoma with slower IOP elevation and RGC loss progression, more accurately simulate key aspects of glaucomatous neurodegeneration.⁴⁷ Retinal structure and function analysis were performed on DBA/2J mice without AAV vector injection at 3 and 12 months of age, respectively, as the control groups (DBA 3 months and DBA 12 months). Intravitreal injection of the AAV vector was performed in DBA/2J mice at 5.5 months of age, and evaluation was conducted at 12 months of age (Fig. 5A), according to a previous study.⁴⁸ Over a 10-month observation period, the IOP of DBA/2J mice began to increase at 7 months, reaching its peak at 10 months (Fig. 5B). Due to the continuous increase in IOP, the density of RGCs in 12-month-old DBA/2J mice decreased to 1335.33 ± 36.63/mm², representing a decrease of 56.52% ± 3.37% of RGCs compared to 3-month-old DBA/2J mice (3084.00 ± 171.23/mm²). Injection of AAV-NGF (2621.67 ± 117.58/mm²) significantly reduced the loss of RGCs caused by elevated IOP (P < 0.05), whereas injection of AAV-Con (1189.00 ± 151.63/mm²) failed to achieve this protective effect (Figs. 5C, 5D). Compared with 3-month-old DBA/2J mice, the density of retinal nerve fibers significantly decreased in 12-month-old DBA/2J mice, consistent with the pathological changes observed in glaucoma.^49,50 Intravitreal injection of AAV-NGF in DBA/2J mice resulted in increased retinal nerve fiber density at 12 months of age compared to AAV-Con injection (Fig. 5E). Furthermore, the decline in the thickness of the ganglion cell complex observed in 12-month-old DBA/2J mice (64.20 ± 0.42 µm) also mimics glaucoma feature. DBA + NGF group (68.73 ± 0.77 µm) exhibited a significantly thicker ganglion cell complex at 12 months of age compared to the DBA + Con group (62.93 ± 0.29 µm, P < 0.05), with no significant difference observed in the thickness of the whole retina layer (Figs. 5F–H). As the number of RGCs and the density of retinal nerve fibers decreased, the amplitude of pSTR in 12-month-old DBA/2J mice (18.30 ± 0.36 µV) significantly decreased, accompanied by prolonged latency (205.07 ± 9.20 ms, P < 0.05). However, the amplitude and latency of pSTR in 12-month-old DBA/2J mice injected with AAV-NGF (44.07 ± 1.93 µV, 158.60 ± 11.47 ms) were closer to those of mice at 3 months (47.50 ± 1.44 µV, 138.10 ± 4.24 ms, P < 0.05), indicating the protective effect of AAV-NGF on RGC function (Figs. 5I–K). 

Figure 5.

 

AAV-NGF conferred protection against RGCs loss and functional impairment in DBA/2J model of glaucoma. (A) Timeline of experimental procedures. (B) Fluctuation in IOP of DBA/2J mice within 12 months (n = 3). (C) Representative images of RBPMS-positive RGCs (red) in the DBA 3 months, DBA 12 months, DBA + NGF, and DBA + Con groups (n = 3). (D) Quantification of RBPMS-positive RGCs (n = 3). (E) Immunofluorescent staining of Beta III Tubulin (red) in mice retina (n = 3). (F) Representative OCT images of retinal structure centered on the optic disk. The ganglion cell complex (RNFL and GCL and IPL) spanned between the red and blue lines, whereas the whole retinal layer extended between the red and yellow lines (n = 3). (G) Quantitative analysis of the average thickness of the ganglion cell complex (n = 3). (H) Quantitative analysis of the average thickness of the whole retinal layer (n = 3). (I) Representative pSTR electroretinogram traces in the DBA 3 months, DBA 12 months, DBA + NGF, and DBA + Con groups (n = 3). (J) Quantification of pSTR amplitude (n = 3). (K) Quantification of pSTR latency (n = 3). Statistical analysis was performed using the Mann-Whitney U test and Kruskal-Wallis test. *P < 0.05, ***P < 0.01. ns = not significant.

Discussion

In our study, we used an innovative approach by utilizing the neuronal-specific hSyn promoter, which enabled precise targeting of RGCs by the AAV2 vector. Subsequently, we validated the neuroprotective efficacy of AAV-NGF in three distinct glaucoma models. Intravitreal administration of AAV-NGF displayed efficient protection to both the retinal structure and functions without inducing obvious immune responses, underlying a solid foundation for subsequent clinical trials. 

Intravitreal injection allows direct transduction of RGCs by AAV vectors.⁵¹ However, low transduction efficiency and poor specificity pose an obstacle to RGC gene therapy. AAV vector carrying a CAG promoter exhibited a transduction efficiency of 21% on RGCs in an experimental optic neuritis model.²⁰ AAVs with CBA or CMV promoters showed only 15% or 13% transduction efficiency of RGCs.³⁰ Additionally, not all AAV-based RGC protection studies provided detailed transduction efficiency of AAV vectors in RGCs.^16–20 In this study, we used the neuronal-specific hSyn promoter, achieving a transduction efficiency of up to 46.64% ± 2.18% for RGCs, significantly higher than that of other types of promoters. 

It is widely recognized that NGF exerts a dual effect on RGCs.^52–54 NGF binding to TrkA receptors on RGCs can promote RGC survival, while binding to p75^NTR on Müller cells indirectly leads to RGC apoptosis.^27–29 Currently, gene therapy for neuroprotection in glaucoma models primarily utilizes broad-spectrum promoters, such as CMV, which exhibit a generalized expression in various retinal cell types. There is a report that only 45% of retinal cells transfected with the AAV vector carrying the CMV promoter are RGCs, with the remaining 55% consisting of amacrine, Müller, and horizontal cells.³⁰ In our study, the hSyn promoter facilitates robust and specific transduction to RGCs, avoiding transducing Müller cells, thus effectively reducing the potential side effect from Müller cell-mediated apoptosis of RGCs.³⁰ 

Previous studies have shown that NTFs, such as BDNF, can protect RGCs in glaucoma models.^16–18 However, long-term BDNF treatment paradoxically reduces TrkB receptor levels in the retina, thereby diminishes its therapeutic benefits and impairing RGC survival over time.¹⁶ In contrast, our study demonstrates that NGF delivery promotes sustained upregulation of TrkA receptors, a response consistent with its neuroprotective effects mediated through TrkA signaling. This distinct receptor modulation underscores a critical advantage of NGF in promoting endogenous trophic support to preserve RGC functionality. 

NGF is a widely used neurotrophic medication in clinical practice. Previous studies have demonstrated that topical administration of NGF eye drops protects RGCs and optic nerve axons from degeneration and inhibits RGC apoptosis induced by high IOP and ONC in animal models.^55–58 Subsequent clinical trials have also confirmed that short-term, high-dose recombinant human NGF (rhNGF) eye drops are safe and tolerable in patients with glaucoma.⁵⁹ Because glaucoma is a chronic neurodegenerative disease requiring long-term intervention, our aim was to simplify the drug administration process and achieve prolonged neuroprotection. In our study, a single intravitreal injection of AAV-NGF yielded promising results in all 3 glaucoma models, with the longest observation period lasting 5 months. Furthermore, AAV-NGF injection did not induce any detectable adverse reaction in the eye, laying the foundation for subsequent clinical trials. 

In our study, MB-induced glaucoma was modified based on the published method.³² A puncture site was created 1 mm posterior to the limbus using an 11-0 suture needle. A 36-gauge needle attached to a 10 µL syringe was then inserted through the puncture into the posterior chamber. After passing through the pupil, microbeads were injected into the anterior chamber angle.⁶⁰ To minimize damage to the anterior lens capsule, the bevel of the needle was oriented toward the lens and kept close to the iris during the injection. This optimized procedure reduced microbead accumulation in the central corneal stroma and endothelium caused by punctures in the central transparent cornea. Over time, the microbeads in the anterior chamber migrated deeper into the angle, partially restoring transparency in the visual axis. This facilitated the acquisition of high-quality in vivo fundus photographs and OCT images in mice. 

Despite the promising neuroprotective effects of AAV-NGF across multiple glaucoma models, several limitations warrant consideration. First, although our study demonstrated sustained neuroprotective effects for up to 5 months, glaucoma remains a chronic, lifelong condition. The long-term efficacy and safety of AAV-NGF therapy remain to be fully elucidated. Second, AAV-NGF administration was performed before inducing glaucoma injury in all experimental groups. Although preventive neuroprotection is often prioritized for evaluating mechanistic efficacy, this approach differs significantly from clinical glaucoma management strategies. Third, although we assessed structural preservation and functional outcomes, higher-order visual functions, such as contrast sensitivity or visual field preservation, were not evaluated. These metrics will be incorporated as important evaluation criteria in subsequent clinical trials, as they directly correlate with the patients’ quality of life. 

In conclusion, by enhancing the specificity and transduction efficiency of AAV2-hSyn-NGF targeting RGCs, we observed significant neuroprotective effects in all three glaucoma models, with no detectable adverse reactions. Moving forward, we will continue to validate AAV2-hSyn-NGF in mammal models and advance the process toward clinical trials. 

Acknowledgments

Supported by the National Natural Science Foundation of China (82101091 to B.N.Z.), the Shandong Provincial Key Research and Development Program (2021ZDSYS14 to L.X.), the Academic Promotion Program of Shandong First Medical University (2019ZL001 and 2019RC008 to L.X.). 

Author Contributions: X.Z., B.N.Z., and L.X. designed research; X.Z., B.Q., Z.R., and L.C. performed experiments; X.Z. wrote the manuscript; X.P., Q.Z., B.N.Z., and L.X. provided suggestions to the manuscript. 

Data Availability: The data that support the findings of this study are available from the corresponding author upon request. 

Disclosure: X. Zhu, None; B. Qi, None; Z. Ren, None; L. Cong, None; X. Pan, None; Q. Zhou, None; B.N. Zhang, None; L. Xie, None