Enzymatic Disruption of the Internal Limiting Membrane Enhances Transplanted Retinal Ganglion Cell Survival

Abstract

Purpose

In mammals, retinal ganglion cells (RGCs) lack the ability to regenerate after injury, and RGC transplantation is a potential approach for cell replacement therapy. However, the internal limiting membrane (ILM) in the retina serves as a significant barrier to donor cell integration, highlighting the need for strategies to disrupt it. We hypothesize that plasmin can enzymatically disrupt the ILM, thereby enhancing the survival of transplanted donor RGCs.

Methods

In vivo, 8- to 10-week-old mice were intravitreally injected with phosphate-buffered saline (PBS) or plasmin (2.3 or 4.6 µg). Immunohistology was performed to verify ILM disruption and assess microglial reactivity and RGC survival. Hematoxylin and eosin (H&E) staining was used to evaluate immune cell infiltration in the vitreous, electrophysiology to assess retinal function, and optical coherence tomography to evaluate retinal structure. Donor RGCs were obtained by differentiating human embryonic stem cells into retinal organoids (ROs). The RO-derived RGCs were transplanted into plasmin-treated eyes 2 weeks after plasmin treatment. Transplanted RGCs were identified using tdTomato fluorescence.

Results

Our study identified the optimal dose of plasmin (2.3 µg) that effectively disrupts the ILM without triggering microglia reactivity or immune infiltration in the vitreous. In addition, neither the RGC number nor RGC function was affected after plasmin treatment, whereas increased survival of transplanted cells was observed in the plasmin-treated retinas.

Conclusions

Our findings provide a translational insight on plasmin-mediated ILM disruption, demonstrating that an optimal dose can effectively degrade the mouse ILM without compromising RGC integrity, offering a promising strategy to enhance transplanted cells’ survival.

Retinal ganglion cells (RGCs) are the first neurons in the visual pathway, solely responsible for transmitting visual signals from the retina to the brain via the optic nerve.¹ Damage or degeneration of RGCs can result in vision impairment and loss, as observed in conditions like glaucoma and optic neuropathies. Various strategies are being explored to restore RGC function, including cell replacement therapy and gene therapy. Gene therapy involves delivering exogenous genetic material into cells using viral or non-viral vectors to modify or restore the function of a specific gene, with the goal of enhancing the survival and regeneration of RGCs.² In cases where a significant portion of RGCs is lost, cell replacement therapy also emerges as a potential treatment avenue. Given the limited availability of human RGCs, inducing the differentiation of human stem cells into RGCs offers a means of generating replacement cells for transplantation into the impaired retina.^3,4 Although several studies have demonstrated the successful transplantation of human stem cell-derived RGCs into various animal models, the challenge of low donor cell survival post-transplantation persists.^5,6 Consequently, devising methods to enrich transplantable RGCs⁷ and enhance donor cell survival and integration within the host eye remains a critical requirement for advancing cell replacement therapy.

A significant challenge encountered when transplanting cells into the host eye is the presence of the internal limiting membrane (ILM), a basement membrane primarily composed of extracellular matrix proteins like laminin and collagen IV, which acts as a significant barrier.^8,9 This barrier obstructs the migration of transplanted stem cells, viral vectors, therapeutic compounds, and nanoparticles, thereby limiting the engraftment of RGCs.^10–12

Removal of the ILM is considered an important procedure for enhancing RGC survival and integration after transplantation.^8,9 To address this challenge, a research team mechanically disrupted the ILM in a rat model to create a more receptive environment for donor cells, in a process designed to mimic ILM peeling in surgeries of human patients, such as in macular hole repair. The in vivo experiments showed improved access of intravitreally delivered vectors to the inner neurosensory retina and the ex vivo results revealed enhanced integration of transplanted RGCs within the host neurosensory retina.¹³

Surgical peeling of the ILM is a common procedure beneficial for repairing traction maculopathies affecting the vitreoretinal interface, such as macular pucker,¹⁴ retinal detachment,¹⁵ and macular holes.¹⁶ However, vitrectomy combined with ILM peeling may lead to various complications, such as changes in intraocular pressure, progression of cataracts or dislocation of lenses, iatrogenic retinal tears or detachment, alterations in retinal pigment epithelium, visual field impairments, vitreous hemorrhage, and endophthalmitis.^17–19 In response to these challenges, the process of enzymatic digestion, which hydrolyzes the ILM, presents itself as a potential alternative method for ILM removal, offering a solution that mitigates the risk of inducing complications commonly associated with surgical removal techniques. This enzymatic approach holds promise in providing a safer means of addressing ILM-related concerns.

A different group of researchers developed an approach using both ex vivo and in vivo models, focusing on enzymatic digestion of the ILM with Pronase E. More specifically, it was shown that Pronase E effectively disrupted the ILM in organotypic retinal cultures, without detectable changes in gross retinal viability, and also that this method significantly improved donor RGC survival, enhanced their localization within the retinal layers, and promoted proper dendritic organization in the inner plexiform layer.^20,21 However, the immune response to enzyme treatment was not assessed, nor were the survival or functional outcomes of host RGCs following ILM digestion evaluated. Because Pronase E is a nonspecific protease mixture, there is a potential risk of off-target proteolysis leading to inflammatory pathways.

To explore a different clinically available alternative, the efficacy and immune response of plasmin were examined in a rodent model. Plasmin is a vitreolytic enzyme analogous to the clinically used Ocriplasmin. Marketed under the brand name Jetrea, ocriplasmin is a recombinant protease that targets fibronectin and laminin, which are integral components of the vitreoretinal interface. Approved by the US Food and Drug Administration (FDA) in 2012,²² it was utilized in patients to address symptomatic vitreomacular adhesion by breaking down the proteins responsible for tethering the vitreous to the macula.²³ This action facilitates the detachment of the vitreous from the retina.²⁴ While the use of ocriplasmin may lead to adverse effects, such as vitreous floaters and photopsia, the majority are mild to moderately severe, and transient in nature.^25,26 Although ocriplasmin was discontinued in 2020, plasmin—a nonspecific serine protease with fibrinolytic properties—remains a key enzyme with similar biological activity. In addition to its role in fibrinolysis, plasmin can hydrolyze various glycoproteins, including laminin and fibronectin. This proteolytic function was first reported in 1981, laying the foundation for pharmacologic vitreolysis as a potential therapeutic approach for vitreoretinal diseases.²⁷ Beyond its role in ILM proteolysis, plasmin also exhibits the capacity to liquefy the vitreous humor,²⁸ potentially fostering improved attachment of donor cells to the host retina.

Our findings identify an efficient plasmin treatment dose that effectively disrupts the ILM without inducing immune cell infiltration, microglial reactivity, or damage to the RGCs. Furthermore, our results provide a translational perspective on plasmin-mediated ILM disruption highlighting its potential to enhance donor cell survival in RGC replacement therapy.

Methods

Cryosection Immunostaining

For the preparation of retinal cryosections, mouse eye globes were incubated in 4% PFA at 4°C overnight for fixation. Subsequently, they were washed 3 times (15 minutes each) with PBS and sequentially incubated in 15% sucrose, followed by 30% sucrose both at 4°C overnight. After those steps, they were embedded in optimal cutting temperature (Fisher Scientific; 23-730-571) mounting medium and cryosections of 12-µm thickness were cut. Consequently, sections were permeabilized with 0.1% Triton X-100 (Sigma Aldrich; X100) in PBS for 30 minutes and blocked with 5% normal goat serum (NGS; Invitrogen; 10000C) and 0.1% Triton X-100 in PBS for 1 hour at room temperature. They were then incubated overnight at 4°C with different primary antibodies, such as mouse anti-Brn3a (1:100, Millipore Sigma; MAB1585), rabbit anti-laminin antibody (1:100, PA1-16730; Thermo Scientific), or rabbit anti-Iba1 (1:100, 019-19741; Fujifilm). Following 3 washes with PBS (15 minutes each), the sections were incubated overnight at 4°C with nuclear stain 4′,6-diamidino-2-phenylindole (DAPI; 1:5000; Sigma Aldrich; D9542) and secondary antibodies Alexa Fluor 488 goat anti-mouse (1:500, #A32723; Invitrogen), Alexa Fluor 555 goat anti-rabbit (1:500, #A32732; Invitrogen) or Alexa Fluor 488 goat anti-rabbit (1:500, #A-11034; Invitrogen). Finally, imaging was conducted using an Olympus Life Science IX83 inverted microscope.

Retina Flatmount Immunostaining

To create the retina flatmount, following dissection and fixation in 4% PFA for 1 hour, mouse retinas were flattened with 4 relaxing cuts dividing them into 4 quadrants. Blocking was performed using 5% NGS and 1% Triton X-100 in PBS for 1 hour. Furthermore, retinas were incubated overnight at 4°C with the mouse anti-Brn3a (1:100; Millipore Sigma; MAB1585) or the rabbit anti-RBPMS (1:100; Sigma Aldrich; ABN1362). After undergoing 3 washes with PBS of 15 minutes each, the retina flatmount samples were incubated overnight with Alexa Fluor 488 goat anti-mouse secondary antibody (1:500, #A32723; Invitrogen) or Alexa Fluor 488 goat anti-rabbit (1:500, #A-11034; Invitrogen) along with DAPI (1:500, #D9542; Sigma Aldrich). Three more washes of 15 minutes each in PBS were done, and then the retinas were covered with 1.5-mm coverslips using anti-fade mounting medium (ProLong Gold; Life Technologies). Imaging of the samples was conducted using the Leica Microsystems DMi8 automated inverted microscope, which has the ability to spirally scan slides and take photographs of the whole flatmount retina. The Olympus Life Science IX83 confocal inverted microscope was used to create z-stack images.

Histopathology

To evaluate the immune infiltration in the vitreous cavity, 8- to 10-week-old mice were injected with PBS or plasmin (2.3 µg or 4.6 µg; EMD Millipore, 527621-10U). Two weeks later, the mice were euthanized, and their eyes were enucleated and fixed in Davidson’s (Hartmann's) Fixative (Electron Microscopy Sciences; 64133-10) at room temperature for 24 hours. The ocular globes were then embedded in paraffin, and 5-µm sections were cut and stained with hematoxylin and eosin (H&E). The vitreous and retina were examined under light microscopy.

For the lipopolysaccharide (LPS)-induced uveitis reference picture, 8- to 10-week-old mice were intravitreally injected with Escherichia coli (E. coli) LPS O111:B4 (250 ng in 2 µL; Sigma Aldrich; L2630). The mice were euthanized 24 hours after injection, and their eyes were enucleated and fixed in Davidson’s (Hartmann's) Fixative at room temperature for 24 hours. The ocular globes were then embedded in paraffin, and 5-µm sections were sectioned and stained with H&E.

Human Retinal Organoid Differentiation

Human retinal organoids were differentiated from a transgenic human embryonic stem cell line Brn3b-tdTomato-Thy1.2 following our laboratory’s protocol.²⁹ The stem cell line is originally from WiCell Research Institute. It was generously gifted to us from Dr. Don Zack's laboratory at Johns Hopkins University and we thank him for his generous donation.

Briefly, stem cells maintained in StemFlex medium (Gibco; #A3349401) were passaged by adding 0.5 mM trypsin-ethylenediaminetetraacetic acid (EDTA; Gibco; 25-300-120) to the culture medium for 5 minutes at 37°C and resuspended in a 3:1 ratio of StemFlex and neural induction medium (NIM) on day 0 to initiate embryonic body (EB) formation. NIM consisted of 1X N2 supplement (100X; Gibco; 17502-048), 1X MEM non-essential amino acids solution (100X; Gibco; 11140050), 2 µg/mL heparin (0.2%; Stemcell Technologies; 07980) and 1X Antibiotic-Antimycotic (100X; Gibco; 15240062) in DMEM/F12 (1:1; Gibco; 11-330-032). On day 1, the EBs were transferred to a 1:1 ratio of StemFlex and NIM medium, followed by a 1:3 ratio of StemFlex and NIM medium on day 2. On day 3, the EBs were resuspended in NIM and seeded on Matrigel-coated wells (Corning; CLS354277), with media change occurring every other day.

After 16 days of differentiation, the medium was changed to retinal differentiation medium (RDM), consisting of 57.6% DMEM/F12 (1:1), 38.4% DMEM (Gibco; 11885084), 1X B-27 supplement (minus vitamin A; 50X; Gibco; 12587010), 1X MEM non-essential amino acids solution, and 1X antibiotic-antimycotic (100X). On day 28, the cells were gently scraped in a single 360-degree motion using a cell scraper and transferred to a 6-well flat-bottom ultra-low attachment dish (Genesee Scientific; #25-105). The cell medium was switched to RC2, composed of 50% DMEM/F12, 35% DMEM basic, 1X B-27 Supplement, 10% fetal bovine serum (Gibco; 10437028), 1X GlutaMax (100X; Gibco; 35050061), 100 µM Taurine (100 mM; Sigma; T-0625), 1X MEM non-essential amino acids solution, and 1X antibiotic-antimycotic (100X). Additionally, 1 µM retinoic acid (Sigma Aldrich; R2625) was supplemented every other day for 2 weeks after scraping to facilitate maturation. RC2 medium was changed every 2 days, and human retinal organoids were maintained in this medium until the indicated experimental time points.

Donor Cell Transplantation

Eight- to 10-week-old C57BL/6J mice (either sex; Jackson Laboratory) were randomly separated into 3 groups (PBS, 2.3 µg, and 4.6 µg of plasmin) and given ad libitum access to food and water. This research was conducted in compliance with the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research. Mice were anesthetized with ketamine/xylazine (40–55 mg/kg ketamine and 4–5.5 mg/kg xylazine). Proparacaine (0.5%; Sandoz; PO-6543) as a local anesthetic was also applied topically to the cornea. The non-operated eye was kept moist with Genteal artificial tear drops (Alcon; NDC 0065-0426-36). Then, 2.3 µg or 4.6 µg of plasmin (EMD Millipore; 527621-10U) or PBS was injected intravitreally into the mice eyes using a 30-gauge needle on a 5-µL glass Hamilton syringe. Care was taken to prevent damage to the lens and to ensure that the retinal blood supply remained unaffected. Two weeks after plasmin injection, each mouse received an intravitreal injection of 20,000 cells in 2 µL of medium. One week after transplantation, the mice were euthanized by a rising concentration of CO₂ and perfused/fixed intracardially with PBS and 4% paraformaldehyde (PFA). Eyes were surgically dissected, immersed in 4% PFA overnight at 4°C, and flatmounts were created and stained accordingly.

For cell transplantation, the hROs were dissociated with Accumax solution (Sigma Aldrich; A7089) at 37°C for 30 minutes. After incubation, the hRO derived-cells were centrifuged and resuspended with NB-SATO growth medium containing a Rho-kinase (ROCK) inhibitor (Sigma Aldrich; Y-27632). Growth medium was made by first preparing SATO. The SATO (100X) was made by adding 400 mg Transferrin (Sigma Aldrich; T-1147), 400 mg bovine serum albumin (BSA; Sigma Aldrich; A-4161), 10 µL progesterone (2.5 mg/100 mL EtOH; Sigma Aldrich; P8783), 64 mg putrescine dihydrochloride (Sigma Aldrich; P7507), 400 µL sodium selenite (4 mg/100 µL 0.1N NaOH and 10 mL neurobasal; Sigma Aldrich; S5261). Consequently, it was filtered through a 0.22-µm filter (Sigma Aldrich; SLGV013SL). The final NB-SATO growth medium was prepared by adding 200 µL penicillin-streptomycin (Gibco; 15140122), 200 µL insulin (5 µg/mL; Sigma Aldrich; I6634), 200 µL sodium pyruvate (100 mM; Gibco; 11360070), 200 µL triiodothyronine (T3; 100X; Sigma Aldrich; T6397), 200 µL 1 mM L-glutamine (100X; Gibco; A2916801), 400 µL B27 (50X; Gibco; 17-504-044), 200 µL SATO (100X) and 20 µL N-acetyl-L-cysteine (NAC; 5 mg/mL, 1000X) in 20 mL Neurobasal medium (Gibco; 21103049). NAC was made by dissolving 50 mg N-acetyl-L-cysteine (Sigma Aldrich; A8199) in 10 mL neurobasal medium. When all ingredients of NB-SATO were combined it was filtered through a 0.22-µm filter. After filtration, 20 µL BDNF (1000X, 50 µg/mL; PeproTech; 450-02), 20 µL CNTF (1000X, 10 µg/mL; PeproTech; 450-13), forskolin (1000X, 5 mM; Sigma Aldrich; 6866), and 40 µL FGF (500X, 20 µg/mL; PeproTech; 100-18B) were added.

Electroretinography

The whiskers of 8- to 10-week-old C57BL/6 wild-type mice were trimmed, followed by an overnight adaptation period. They were then anesthetized with an intraperitoneal injection of a ketamine/xylazine mixture (100 mg/mL) at a dose of 5 mL/kg under dim red light. Pupils were fully dilated using 2.5% phenylephrine combined with 1% tropicamide eye drops and their cornea was anesthetized by 0.5% proparacaine drops. Following anesthesia, the mice were placed on a heated platform. GenTeal severe lubricant eye gel (Alcon; NDC 0078-0429-47) was applied to their eyes for lubrication. Electroretinograms were recorded with the Celeris system (Diagnosys, LLC). Pattern electroretinograms (pERGs) were recorded using one pattern stimulator along with one full field stimulator as the fellow eye reference electrode. Full-field electroretinograms (ffERG) were recorded using two full field simulators. The electrodes were lubricated with GenTeal to enhance conductance and placed in contact with each cornea and aligned with the center of the pupils. Before presenting the stimuli, electrode impedance was confirmed to be below 10 KΩ. Under this arrangement, in the pERG, stimuli were presented to one eye while the other, the unstimulated eye, acted as the reference. The pERG response was elicited by a contrast reversing bar stimulus with a spatial frequency of 0.059 cycles/degree, mean luminance of 50 cd/m², and reversal rate of 2.1/seconds. A total of 400 sweeps were averaged for analysis. The full pERG amplitude was determined from the positive peak at P1 to the negative trough at N2, often termed as P1N2. The pERG responses were recorded 2 weeks after plasmin injections, 1 month after plasmin injections, and 2 weeks after 6-aminohexanoic acid injections (Sigma Aldrich; 07260). In the ffERG, white flashes of 0.5 cd·s·m⁻² were delivered to elicit retinal responses. All readings were recorded using the Espion software from Diagnosys.

Optical Coherence Tomography

Optical coherence tomography (OCT) scans were obtained 2 weeks post-treatment. Imaging was conducted using a visible light OCT (vis-OCT) prototype developed at the UPMC Vision Institute. Retinal vis-OCT scans were acquired at an A-line sampling rate of 70 kilohertz (kHz). The system utilized a light spectrum ranging from 510 to 610 nm, generated by a laser (EXU-6 OCT; NKT Photonics, Boston, MA, USA), providing an axial resolution of 1.2 µm in tissue. The incident power on the cornea was maintained at 0.8 mW. Volumetric raster scans, consisting of 500 × 2 × 500 A-lines near the optic disc, were acquired for retinal analysis and RGC axon bundle assessment.

To ensure boundary accuracy, retinal layers were segmented automatically with manual correction. Retinal layer thickness measurements were performed to assess retinal degeneration following treatment, specifically ganglion cell complex (GCC) between the ILM and the inner plexiform layer (IPL), as well as whole retina between the ILM and Bruch's membrane (BM). Data were categorized into three experimental groups: eyes treated with PBS, eyes treated with 2.3 µg of plasmin, and eyes treated with 4.6 µg of plasmin.

Quantification and Statistical Analysis

The number of mice used in each experiment is indicated in the figures and/or the figure legends. Linear percentage coverage of laminin was calculated by measuring the average of laminin coverage at 20 × magnification from 3 fields of view from 3 sections for each eye. Cell counting for Iba1+ and Brn3a+ was conducted from one field of view from one section. For retinal wholemounts, each retina was divided into four quadrants. In each quadrant, Brn3a⁺ cells were counted from a single randomly selected digital micrograph taken from the peripheral region, located 1.5 mm from the optic nerve head. TdTomato+ cells after transplantation were counted manually from the whole retina of each sample. Quantification of cell numbers measured in retinal sections was normalized to liner micrometer of retinal cell layer, using central sections near the optic nerve head as a standard to select slides for cell counting. Cells were manually counted in a masked manner. Statistical analysis was performed by calculating the mean ± SEM. Unpaired t-tests and 1-way ANOVA followed by post hoc t-tests with Tukey's correction was conducted for data analysis. P < 0.05 was considered as significant. Graphs were created using Prism 10 software (GraphPad, La Jolla, CA, USA).

Results

Intravitreal Administration of the Optimal Plasmin Dose Does Not Cause Immune Cell Infiltration or Microglia Activation

In a pig model, treatment with ocriplasmin did not induce inflammatory responses.³⁰ To determine whether this holds true in mice, we intravitreally injected PBS, 2.3 µg, or 4.6 µg of plasmin into mouse eyes. Two weeks post-treatment, histological examination of the eyes revealed no immune cell infiltration in either the 2.3 or 4.6 µg groups (Fig. 1A), confirming that plasmin treatment did not trigger inflammatory responses in the vitreous cavity.

Figure 1.

Intravitreal administration of the optimal plasmin dose does not cause immune cell infiltration or microglia activation. (A) No immune cell infiltration was observed in the vitreous cavity of the plasmin-treated eyes. A representative image at the bottom of A serves as a reference, illustrating lipopolysaccharide-induced immune cells infiltration. (B) Intravitreal injection of 4.6 µg of plasmin resulted in a significant increase in the number of microglia cells in the retina, especially in the ganglion cell layer, in contrast to the 2.3 µg dose which induced no microglia reactivity. Mean ± standard error of the mean (SEM) shown; ns, not significant; ****P < 0.0001, ***P < 0.001, **P < 0.01 by 1-way ANOVA and post hoc t-test with Tukey correction (N = 4 per group A N ≥ 9 per group B).

We further investigated the impact on resident retinal immune cells, specifically microglia, by preparing cryosections of PBS-, 2.3 µg-, and 4.6 µg-treated retinas and staining them with the microglia marker Ionized Calcium-Binding Adaptor Molecule 1 (Iba1). Although no significant difference was observed between the PBS and 2.3 µg groups, the 4.6 µg group showed a notable increase in microglia, particularly in the ganglion cell layer (Fig. 1B). These findings suggest that the 2.3-µg dose is better tolerated, as it mitigates the microglial reactivity observed at the 4.6-µg dose, making it a more suitable candidate for further experiments.

Plasmin Treatment Is Capable of Disrupting the Internal Limiting Membrane Without Damaging RGCs

After investigating the effects on immune response, we sought to determine whether plasmin could effectively disrupt the ILM. For this reason, laminin staining was performed to visualize the ILM. Our data showed that both 2.3 µg and 4.6 µg of plasmin treatment disrupts the structural integrity of the ILM, as proven by the interruption of laminin expression, which is a major component of the ILM. The immunofluorescence of laminin at the ILM was quantified using the quantitative method by Zhang et al.²⁰ Linear percentage coverage of the retinal surface in cryosections was measured showing that plasmin after 2 weeks reduces the coverage of ILM by 40% (Fig. 2A). Both doses of plasmin demonstrated the same percentage of disruption. Consequently, and because the anatomic location of the RGCs is right beneath the ILM, it is important to overcome the ILM barrier without damaging them. To evaluate that, the RGC marker Brn3a was used to immunostain retinal cryosections and wholemounts after 2 weeks of plasmin treatment. Our results demonstrated that neither the 2.3 µg nor the 4.6 µg group showed a significant reduction in RGC numbers (Figs. 2B, 2C). Overall, these results show that intravitreal plasmin treatment is capable of disrupting the ILM, whereas maintaining the RGC population numbers.

Figure 2.

Plasmin treatment is capable of disrupting the internal limiting membrane without damaging RGCs. (A) Representative pictures showing ILM immunostained using laminin, disrupted by both 2.3 µg and 4.6 µg of plasmin treatment. A representative image at the top of A illustrates a non-injected negative control. (B) Retinal sections demonstrating that the number of Brn3a+ RGCs was not reduced by either the 2.3 µg or the 4.6 µg plasmin treatment. (C) Retinal flatmounts showing that the Brn3a+ RGC population was maintained across all three groups. Mean ± SEM shown; ns, not significant; ****P < 0.0001 by 1-way ANOVA and post hoc t-test with Tukey correction (A and B; N ≥ 4 per group).

Retinal Function Is Reduced in the 4.6 µg Plasmin Treatment Group, While Ganglion Cell Complex Thickness Demonstrated no Changes

Apart from assessing RGC numbers, it was essential to determine whether RGCs maintained their functional integrity following plasmin treatment. To assess this, we conducted pERG recordings on PBS- and plasmin-treated mice 2 weeks after treatment (Fig. 3A). The pERG is an effective noninvasive method for evaluating RGC function, as its generation relies on the physiological response of existing RGCs to a reversing black-and-white checkerboard pattern, with the response diminishing in cases of RGC degeneration.¹⁵ There was no significant difference in total amplitude between the plasmin 2.3 µg and PBS groups, indicating that RGCs remained physiologically viable and that 2.3 µg plasmin treatment did not cause measurable functional damage. In contrast, the total amplitude in the plasmin 4.6 µg group was significantly decreased, demonstrating impairment of function. Additionally, the analysis of the P1 wave showed significant decrease in the 4.6 µg group compared to the PBS. To ensure that our findings were not attributable to the 6-aminohexanoic acid present as a stabilizing component in the commercial plasmin preparation, we assigned mice to 3 groups: PBS, 0.5 mM 6-aminohexanoic acid (equivalent to the concentration present in the 2.3 µg plasmin dose), and 1 mM 6-aminohexanoic acid (equivalent to the concentration in the 4.6 µg plasmin dose). Analysis of pERG recordings revealed no differences in overall response amplitudes among the groups, indicating that the observed effects are not driven by 6-aminohexanoic acid itself (Fig. 3B). Moreover, to assess the long-term effects of plasmin, pERG was performed 1 month after injection, revealing that the 4.6 µg group continued to show a significant reduction in RGC function (Fig. 3C). Next, the photoreceptors’ functional integrity was tested using ffERG (Fig. 3D), which records the summed electrical response of the retina to uniform light flashes and is widely used to evaluate overall retinal function.³¹ Scotopic a- and b-waves were analyzed from the ffERG and the a-wave amplitude was significantly reduced in the 4.6 µg group compared with both the PBS and 2.3 µg groups, indicating decreased photoreceptor activity, predominantly from rods. In contrast, scotopic b-wave amplitudes showed no statistically significant differences among the three groups, suggesting preserved bipolar cell function. Additionally, OCT scans, an imaging modality for in vivo, noninvasive, and longitudinal monitoring of the retina for both rodents and humans,³² were performed 2 weeks after the injections (Fig. 3E). Although no apparent differences in retinal structure were observed among the groups, the texture of the RGC axon bundle in the plasmin 4.6 µg-treated eyes appeared less distinct compared with the PBS and plasmin 2.3 µg groups, suggesting that 4.6 µg of plasmin may induce a certain degree of damage to the RGCs. We also assessed the impact of plasmin treatment on retinal thickness. Two weeks after administering PBS or plasmin, we measured the GCC thickness, from the ILM to the IPL, as well as the total retinal thickness from the ILM to BM. No significant differences were observed among the groups. Oscillatory potentials exhibited a trend toward reduction between the PBS and 4.6 µg groups (P = 0.08) but did not reach statistical significance, implying maintained inner retinal activity, particularly from bipolar and amacrine cells (Fig. 3F). Overall, our data demonstrate that the 2.3 µg of plasmin is better tolerated than the 4.6 µg dose, as it neither reduces the number of RGCs nor impairs their function.

Figure 3.

RGC function is reduced in the 4.6 µg plasmin treatment group, whereas ganglion cell complex thickness demonstrated no changes. (A) Graph illustrating the generated pERG responses from the plasmin-injected and PBS-injected eyes 2 weeks after treatment. The use of 4.6 µg of plasmin treatment caused reduction of RGCs and cone function while with the 2.3 µg dose RGCs maintained their functional integrity. (B) The pERG responses 2 weeks after 6-aminohexanoic acid injections showing no significant difference in total amplitude indicating that the observed effects are not driven by 6-aminohexanoic acid itself. (C) One-month post-treatment, pERG revealed a consistent reduction in RGC function in the 4.6 µg group. (D) Scotopic a-wave analysis revealed a significant reduction in the 4.6 µg group compared to both other groups function, whereas scotopic b-wave analysis showed no statistically significant differences. (E) OCT images collected 2 weeks post-treatment showing retinal structure (top) and the corresponding RGC axon bundles (bottom). GCC and whole retina thickness demonstrated no significant changes in the plasmin treatment groups compared to PBS. (F) Oscillatory potentials showed no significant differences between groups, with a nonsignificant trend toward reduction in the 4.6 µg group. Mean ± SEM shown; ns, not significant; *P < 0.05 by 1-way ANOVA and post hoc t-test with Tukey correction (N ≥ 5 per group).

The Survival of Transplanted Cells Is Increased One Week After Transplantation in the Plasmin-Treated Eyes

Last, we wanted to investigate whether ILM disruption following plasmin-treatment would have impact on donor cell survival. To obtain human RGCs, we differentiated human BRN3B-H9-tdTomato-Thy1.2 human embryonic stem cells into human retinal organoids (hROs). In our study, we observed tdTomato fluorescence in the center of the day 60 retinal organoids indicating RGC formation (Fig. 4A). To transplant donor cells, we dissociated day 60 retinal organoids and intravitreally injected approximately 20,000 cells into mouse eyes. Dissociated human retinal organoids were transplanted in PBS and plasmin 2.3 µg-treated retinas. One week after donor cell transplantation more tdTomato+ human RGCs were observed on the plasmin- compared to the PBS-injected retinas suggesting increased donor cells survival (Fig. 4B). However, 1 month after cell transplantation no significant difference in cell survivability was observed between the two groups (Fig. 4C). By using 3D Z-projection image analysis, tdTomato+ donor neurites were observed in the ganglion cell layer 1 week after transplantation (Fig. 4D). Later, the immunostaining data of 1 month transplantation experiments showed that donor tdTomato+ cells were stained with RBPMS on wholemount retinas, as the same layer of endogenous RBPMS+ cells, confirming their RGC identity (Fig. 4E). These data suggest that the transplanted cells potentially migrated in the host retina's ganglion cell layer.

Figure 4.

The survival of transplanted cells is increased 1 week after transplantation in the plasmin-treated eyes. (A) Day 60 human retinal organoid expressing tdTomato fluorescence in the center, indicating Brn3b expression and RGC presence. (B) More tdTomato+ cells were observed on plasmin-treated retinas 1 week after transplantation indicating increased donor cell survivability. (C) No difference in donor cell survivability was observed between the two groups 1 month after transplantation. (D) Z-projections and 3D representations of a plasmin-treated retina, 1 week after transplantation, showing tdTomato+ donor cells extending their axons in the same layer where other non-donor DAPI+ cells are. (E) Representative images of a plasmin-treated retina, 1 month after transplantation, showing colocalization of tdTomato fluorescence and RBPMS staining in donor cells confirming their RGC identity. Mean ± SEM shown; **P < 0.01 by unpaired Student's t-test (N ≥ 4 per group).

Discussion

Although plasmin was investigated in the eyes of other animal models,^33,34 the effects of plasmin on mouse ILM remain unknown. In this study, we utilized plasmin to induce ILM digestion in mice in vivo.

Remarkably, our findings reveal that the application of 2.3 µg of plasmin effectively disrupts the ILM without inducing significant ocular inflammation, reducing RGC numbers, or impairing their function. OCT scans conducted 2 weeks after plasmin administration showed no changes in GCC or total retinal thickness. This was consistent in both the 2.3 µg and the 4.6 µg plasmin groups. Additionally, our study unveiled that the survival of donor cells is increased 1 week after transplantation in plasmin treated-retinas and demonstrated the potential integration of donor cells within the host retina, albeit without significant observation of extensive neurite growth.

In rodent models, pluripotent stem cell-derived RGCs have been successfully transplanted into mouse retinas with donor cells surviving for up to 12 months. These cells developed axonal processes that co-aligned with host axons and extended into the optic nerve head, demonstrating structural integration.³⁵ Additionally, studies using primary RGC transplants in adult rat retinas have demonstrated light-evoked responses in donor cells, providing direct functional evidence of integration.³⁶ These findings indicate that functional incorporation is feasible when donor cells survive and integrate appropriately. In our study, enhanced donor cell survival following plasmin treatment may thus increase the likelihood of such light-responsive outcomes.

However, by 1 month after transplantation, donor cell survival no longer differed significantly between plasmin-treated and PBS-treated retinas. A plausible explanation is the absence of immunosuppression, which may have permitted host immune rejection and subsequent loss of the transplanted cells.

The current study has several limitations that should be acknowledged. First, a limitation of this study is that donor cells were not visualized on retinal flatmounts in areas where the ILM was disrupted, preventing direct illustration of the spatial relationship between regions of ILM disruption and donor RGC attachment. Although we did not directly visualize transplanted RGCs within regions of plasmin-mediated laminin disruption, several mechanisms may explain the improved survival we observed. Plasmin activity can remodel the extracellular matrix beyond the immediate injection site, altering the biomechanical and biochemical properties of the surrounding retinal environment in ways that indirectly support donor cell maintenance. Moreover, plasmin-induced vitreous liquefaction, through cleavage of laminin, fibronectin, and other adhesion molecules, may reduce mechanical stress on transplanted cells, enhance nutrient and oxygen diffusion, and facilitate the distribution of survival-promoting factors. Collectively, these effects provide a plausible explanation for the enhanced survival of donor RGCs, even when the cells are not located within or in close proximity to the primary sites of laminin disruption. Additionally, it is well-documented that certain proteins may transfer from cell to cell and complicate experimental design as well as interpretation in all cell transplantation studies.^37–39 In a retinal cell transplantation model, intercellular material transfer (MT) has been reported between transplanted and host photoreceptors.^40–42 To investigate whether MT also occurs between donor and host cells following human stem cell-derived RGC transplantation, a previous study transplanted tdTomato+ RGCs into GFP-expressing mice. Their data demonstrated that MT occurred from donor RGCs to recipient Müller glia in retinas with ILM disruption.²¹ MT was not investigated in the present study; however, these findings highlight the potential importance of evaluating MT in future in vivo studies of RGC transplantation, particularly following ILM disruption. Moreover, future studies should investigate the survival and integration of transplanted cells in damaged retinal models, such as the silicone oil-induced ocular hypertension glaucoma model, to determine whether human embryonic stem cell-derived RGCs can effectively attach and integrate within such an environment. In our last evaluation of integration and neuron extension of donor cells in the host retina, no neurite outgrowth was observed beyond the ganglion cell layer. This limitation in neurite extension may be attributed to the relatively brief period of transplantation preceding the collection of host eyes. Longer time windows need to be investigated to demystify that. As research continues into RGC replacement, several challenges endure. These obstacles encompass ensuring the smooth integration and proper functioning of transplanted cells, addressing immune rejection, and guaranteeing their long-term safety and efficacy. Nevertheless, advancements in this area offer promising prospects for developing innovative therapies for conditions characterized by RGC loss.

Conclusions

Our findings overall provide a translational insight on plasmin-mediated ILM disruption, demonstrating that the optimal dose can effectively digest the mouse ILM without compromising RGC integrity. With further extended investigation, this strategy could represent a promising approach to improve donor cell survival and advance cell replacement therapy, although additional studies are needed to confirm its efficacy.