Development and Characterization of a Photocrosslinkable, Chitosan-Based, Nerve Growth Factor–Eluting Hydrogel for the Ocular Surface

Abstract

Purpose: Recombinant human nerve growth factor (rhNGF; cenegermin-bkbj, OXERVATE) is the first and only U.S. Food and Drug Administration–approved treatment for moderate to severe neurotrophic keratopathy. The aim of this study was to determine the feasibility of incorporating a version of rhNGF in a mucoadhesive hydrogel capable of sustained drug release to the ocular surface.

Methods: Hydrogels loaded with rhNGF were synthesized by conjugating chitosan with azidobenzoic acid (Az-Ch), adding rhNGF, and exposing the solution to ultraviolet (UV) radiation to induce photocrosslinking. Az-Ch hydrogels were evaluated for physical properties and rhNGF release profiles. Cytocompatbility of Az-Ch was assessed using immortalized human corneal limbal epithelial (HCLE) cells. TF1 erythroleukemic cell proliferation and HCLE cell proliferation and migration were used to assess the bioactivity of rhNGF released from Az-Ch hydrogels.

Results: Az-Ch formed hydrogels in <10 seconds of UV exposure and demonstrated high optical transparency (75–85 T%). Az-Ch hydrogels exhibited good cytocompatibility with no demonstratable effect on HCLE cell morphology or viability. rhNGF was released gradually over 24 hours from Az-Ch hydrogels and retained its ability to induce TF1 cell proliferation. No significant difference was observed between rhNGF released from Az-Ch and freshly prepared rhNGF solutions on HCLE cell proliferation or percent wound closure after 12 hours; however, both were significantly better than control (P < 0.01).

Conclusions: rhNGF-loaded Az-Ch hydrogels exhibited favorable physical, optical, and drug-release properties, as well as retained drug bioactivity. This drug delivery system has the potential to be further developed for in vivo and translational clinical applications.

Translational Relevance: Az-Ch hydrogels may be used to enhance rhNGF therapy in patients with NK.

Introduction

Neurotrophic keratopathy (NK, also known as neurotrophic keratitis) is a challenging ocular surface disease caused by damage to the trigeminal nerve and/or its ophthalmic branches, most commonly from herpetic viral infections, intracranial masses, and surgical trauma.¹ Impaired corneal innervation results in diminished blink reflex, reduced tear production, depleted limbal epithelial stem cells, and inhibited wound-healing abilities.^2,3 Clinically, these effects may lead to persistent epithelial defects, stromal ulceration, and ultimately corneal perforation. Although NK is a rare disease, the preponderance of vision- and eye-threatening complications and the lack of directed therapies make the condition particularly burdensome. Part of the pathophysiology of NK may be attributed to the reduced levels of protective, neurotrophic factors typically released by functional nerves.^4,5 Fortunately, a topical recombinant human nerve growth factor (rhNGF) eye-drop formulation (cenegermin-bkbj; brand name OXERVATE) is now commercially available. 

Cenegermin-bkbj is the first U.S. Food and Drug Administration (FDA)-approved therapy that directly addresses the pathophysiology of NK, and its safety and efficacy in corneal epithelial defect healing have been demonstrated in phase I and II clinical trials in patients with moderate to severe NK.^6,7 However, cenegermin-bkbj presents several challenges in application and preservation. As a large protein that dimerizes in its active form (molecular weight, ∼26 kDa), rhNGF is unstable,⁸ necessitating a new vial of cenegermin-bkbj to be assembled each day by the patient to preserve the bioactivity of the protein. In addition, eye drops are typically inefficient drug-delivery vehicles due to the limited capacity of the tear film and the blinking mechanism, with only an estimated 5% bioavailability.⁹ 

To improve the efficiency of unstable medications such as cenegermin-bkbj, sustained-release vehicles may be employed, potentially reducing application frequency and increasing medication stability.¹⁰ Due to their tunable properties and high water content, hydrogels—insoluble networks of crosslinked hydrophilic polymers—are an appealing option for this application. However, designing such a hydrogel that allows for the diffusion of large molecules, such as rhNGF, while maintaining the bioactivity of the drug is challenging. 

To accomplish this task, we used a modified chitosan hydrogel loaded with rhNGF to achieve sustained release of the protein. We investigated the feasibility of using this hydrogel to deliver rhNGF by studying the physical properties, drug release characteristics, in vitro activity, and in vitro efficacy on corneal epithelial wound healing. 

Materials and Methods

Synthesis of Az-Ch

Photocrosslinkable chitosan (with azidobenzoic acid, Az-Ch) was synthesized as previously described,¹¹ with modifications. Briefly, N,N,N′,N′-tetramethylethane-1,2-diamine (TEMED; 300 µL) was added to a solution of 4-azidobenzoic acid (ABA, 80 mg; Thermo Fisher Scientific, Waltham, MA) in dimethyl sulfoxide (DMSO, 1 mL; Sigma-Aldrich, St. Louis, MO) and vortexed. This mixture was added to a solution of 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC, 139.5 mg; Thermo Fisher Scientific) in deionized water (1 mL), vortexed, and added to a solution of low-molecular-weight chitosan (75%–85% deacetylation, 400 mg; Sigma-Aldrich) in a water and DMSO solution (1:1, 200 mL). The pH of the mixture was adjusted to 5.0, and the reaction was stirred at room temperature overnight in darkness. Unreacted ABA and chitosan were removed by centrifugation at 11,500 g for 3 hours. The supernatant was aggregated, and the pH of the mixture was adjusted to 9.5 to precipitate Az-Ch, followed by centrifugation at 15,200 g for 10 minutes. The pellet was dissolved in acidified water, and the pH was adjusted to 5.0. This process of alkaline precipitation and re-dissolution was repeated five times¹² to purify the product and remove reactants. The product was lyophilized for 48 hours and stored at −80°C until further use. Hydrogen-1 nuclear magnetic resonance (¹H-NMR) spectroscopy (Unity INOVA 500 MHz; Varian, Palo Alto, CA) was used to determine the degree of modification of azide groups with ABA by comparing the spectra of ABA, unmodified chitosan, and Az-Ch. Throughout synthesis and long-term storage, the Az-Ch was protected from light. 

Physical Properties

Hydrogels were formed by preparing an aqueous solution of Az-Ch (40 mg/mL) and exposing it to ultraviolet (UV) radiation (100 mW/cm², Loctite Zeta 7401; Loctite Corporation, Rocky Hill, CT) for 10 seconds. Ten seconds of UV exposure was found to be sufficient to produce an insoluble hydrogel. Optical transparency was assessed by preparing hydrogels (n = 4) in 96-well plates and measuring light transmittance between 400 nm and 800 nm with a spectrophotometer, with phosphate-buffered saline (PBS) serving as a blank. To determine the effect of extended exposure to an aqueous environment (such as the tear film) on optical transparency, hydrogel light transmittance was measured after 1 hour, 1 day, and 1 week of incubation in PBS. 

To understand the physical characteristics of Az-Ch hydrogels, viscoelastic properties were evaluated using a rotating rheometer (AR-G2; TA Instruments, New Castle, DE) equipped with a parallel plate geometry (d = 20 mm) with a fixed gap at 1.0 mm. Crosslinked hydrogels were used to determine shear moduli, and a UV curing accessory was used (OmniCure S2000, 50 mW/cm²; Excelitas, Waltham, MA) to dynamically assess the viscoelastic properties of the hydrogel during UV-induced photocrosslinking. Samples were subjected to oscillatory measurements. Frequency sweeps were conducted at 25°C in the range of 0.1 to 10 Hz, with a constant strain rate at 1%, which is in the linear viscoelastic range, as predetermined by strain sweeps at 1 Hz. 

To determine swelling ratio, Az-Ch (n = 4) hydrogels were prepared as above, dried overnight in an oven (37°C), and weighed to obtain the dry weight (W₀). Then, the hydrogels were hydrated in PBS at 37°C for predetermined periods of time up to 24 hours, and the wet weight (W₁) was measured. The swelling ratio was determined as follows: swelling ratio = (W₁ – W₀)/W₀. 

Drug Release, Quantification, and Bioactivity

To investigate the drug release characteristics of Az-Ch, rhNGF-loaded Az-Ch (rhNGF-loaded), hydrogels were prepared as follows. First, lyophilized rhNGF (R&D Systems, Minneapolis, MN) was dissolved in PBS to a concentration of 50 µg/mL and added to aqueous Az-Ch (40 mg/mL). A micropipette was then used to transfer 30 µL of the solution to 5-mL glass vials, and finally the solution was exposed to UV radiation for 10 seconds to form hydrogels. Az-Ch hydrogels without rhNGF (blank hydrogels) were prepared as controls. The volume (30 µL) was chosen to reflect typical ophthalmic solution drop volumes.¹³ After UV treatment, the vials containing the rhNGF-loaded or blank hydrogels (n = 3) were filled with 1 mL PBS (drug-release media) and placed on a rotational shaker at 100 rpm and 37°C. At predetermined time points, the drug-release media (rhNGF-loaded and blank) were exchanged with fresh PBS and stored at –20°C until ready for drug quantification, which was determined by an enzyme-linked immunosorbent assay (Invitrogen NGF beta Human ELISA Kit; Thermo Fisher Scientific). To determine the effect of incorporating a growth factor–binding molecule to reduce burst release, 0.25 or 0.5 mg/mL heparin sodium salt (grade I-A, derived from porcine intestinal mucosa; Sigma-Aldrich) was added to aqueous rhNGF solutions prior to mixing with Az-Ch. The concentrations of heparin were similar to those that have been previously demonstrated to facilitate corneal wound healing.¹⁴ To assess the effect of proteases present in the tear film on the drug release kinetics of the hydrogel, lysozyme was added to the drug-release media at physiologic concentration (2 mg/mL).¹⁵ 

Retained drug bioactivity was assessed using a human TF1 erythroleukemic cell assay.¹⁶ Briefly, TF1 cells (American Type Culture Collection, Manassas, VA) were grown in RPMI 1640⁺ Medium (RPMI 1640 supplemented with 10% fetal bovine serum and 2 ng/mL of recombinant human granulocyte-macrophage colony-stimulating factor [rhGM-CSF]; R&D Systems) in a humidified chamber at 37°C and 5% CO₂. Cells were seeded in 96-well plates at a density of 1 × 10⁴ cells/well in 100 µL RPMI 1640⁺ media. To the media in each well, 50 µL of PBS (control), rhNGF-loaded Az-Ch drug-release media (diluted in sterile PBS to ∼10.5-nM rhNGF, confirmed by ELISA), or fresh rhNGF solution (10.5-nM rhNGF in PBS) was added. The drug-release medium was obtained from hydrogels incubated at 37°C for 24 hours. Cells were incubated for 40 hours before quantification using an MTS assay (CellTiter 96 AQueous One Solution Cell Proliferation Assay; Promega, Madison, WI). Treatment groups were tested in triplicate with five internal replicates. The rhNGF concentration used in this assay (10.5 nM) was chosen to align with the dose-dependent response range of TF1 cells.^17,18 

HCLE Culture Conditions

An immortalized human corneal limbal epithelial (HCLE) cell line originally sourced from Ilene Gipson, PhD¹⁹ (Schepens Eye Research Institute of Massachusetts Eye and Ear, Boston, MA) was generously donated by Vickery Trinkhaus-Randall, PhD (Boston University School of Medicine, Boston, MA) to be used for cytocompatibility, proliferation, and wound-healing assays. HCLE cells were maintained in complete media comprised of Gibco Keratinocyte-SFM (Thermo Fisher Scientific) supplemented with recombinant human epidermal growth factor (rhEGF, 5 ng/mL), bovine pituitary extract (BPE, 50 µg/mL), and 1% penicillin–streptomycin. The cells were grown in a humidified chamber at 37°C and 5% CO₂. 

Cytocompatibility and In Vitro Proliferation

To assess the cytocompatibility of hydrogels, HCLE cells were seeded in 12-well plates at 1 × 10⁴ cells/well, incubated overnight, and then exposed to complete media supplemented with 30 µL of either blank Az-Ch hydrogel or sterile PBS. Two concentrations (10 and 40 mg/mL) of Az-Ch were tested. To ascertain the independent effect of the Az-Ch hydrogel material, heparin was not used in formulations for cytocompatibility testing or for cell proliferation and wound healing assays. After 1 and 3 days of growth, cells were either fixed for morphological analysis (n = 3) or quantified (n = 5) using a resazurin-based colorimetric assay (PrestoBlue HS Cell Viability Reagent; Thermo Fisher Scientific). 

Fixation was performed with 4% paraformaldehyde (15 minutes), followed by washing with PBS, permeabilization with 100% methanol (10 minutes), and blocking with 2% BSA in PBS (1 hour). HCLE cells were then immunostained with Invitrogen beta Actin Monoclonal Antibody (MA1-140; Thermo Fisher Scientific) for 1 hour and Invitrogen cyanine3-conjugated polyclonal secondary antibody (A10521; Thermo Fisher Scientific) for 2 hours. Nuclear counterstaining was achieved with Invitrogen NucBlue Reagent (Hoechst 33342; Thermo Fisher Scientific). Cells were then imaged with a DMi8 fluorescence microscope (Leica Microsystems, Wetzler, Germany) and evaluated for morphology. 

For cell viability quantification, HCLE cells were washed with PBS, incubated with a solution of PrestoBlue Reagent and Keratinocyte-SFM (1:10) for 1 hour at 37°C and measured for absorbance at 570 nm and 600 nm. The percent reduction in the PrestoBlue Reagent (corresponding to total cell viability) was determined per the manufacturer’s instructions. 

To assess the effect of rhNGF-loaded drug-release media on corneal epithelial cell proliferation, HCLE cells were seeded in 96-well plates at 3 × 10³ cells/well, incubated overnight, and then starved for 24 hours using basal media (Keratinocyte-SFM without BPE or rhEGF). Cells were then washed with PBS and exposed to basal media supplemented with fresh rhNGF solution (100 nM), rhNGF-loaded drug-release media (∼100 nM) with or without an NGF receptor inhibitor (Ro 08-2750, 10 µM; R&D Systems), or PBS (control). At the selected concentration, Ro 08-2750 selectively inhibits rhNGF binding to both of its primary receptors, TrkA and p75^NTR. The rhNGF concentration was selected because it was found to produce optimal proliferative response (lowest concentration with highest rate of proliferation; data not shown). Complete media served as a positive control. Cells were incubated for 24 hours and quantified using PrestoBlue Reagent as described above (n = 8 wells per group). Experiments were repeated independently three times. 

Wound Healing

To assess wound-healing response, an in vitro scratch assay was performed. HCLE cells were seeded in 12-well plates at 1 × 10⁵ cells/well and grown to 80% to 90% confluence. Cells were then starved for 24 hours in basal media, and a linear scratch wound was created with a 10-μL pipette tip in the confluent monolayer of cells. After wounding, cells were washed and incubated in basal media supplemented with either fresh rhNGF solution (100 nM), rhNGF-loaded drug-release media (∼100 nM), or PBS (control). Live-cell imaging was used to monitor wound closure, and phase contrast images were obtained with the Leica DMi8 fluorescence microscope every 5 minutes for 12 hours. These images were processed using ImageJ 1.53k (National Institutes of Health, Bethesda, MD) to measure the remaining wound area at each time point. The wound areas at 4, 8, and 12 hours were compared to the wound area at onset to determine percent wound closure. In addition, the overall rate of change in the wound area over the 12-hour period was calculated (n = 4 wells per group). Experiments were repeated independently three times. 

Statistical Analysis

Summary data are shown as mean ± standard deviation or n (%). Statistical analyses included Mann–Whitney and Kruskal–Wallis tests, as appropriate. P < 0.05 was considered statistically significant. 

Results

Physical and Chemical Characterization of Az-Ch Hydrogel

The functionalization of chitosan with ABA was confirmed by the appearance of aromatic peaks at 7.2 and 7.8 ppm (representing ABA) in the ¹H-NMR spectra (Fig. 1A, arrows). Using these proton signals and the peak at 4.8 ppm, which represents chitosan (Fig. 1A, asterisk), the degree of conjugation was estimated to be 4.6%. Az-Ch hydrogels exhibited similar visible light transmittance profiles when compared to PBS throughout 1 week of storage (Fig. 1B). Both pre-crosslinked Az-Ch (Fig. 1C) and post-crosslinked Az-Ch (Fig. 1D) demonstrated a high level of optical clarity. 

Figure 1.

 

Chemical and optical properties of Az-Ch hydrogel. (A) ¹H-NMR spectrum of Az-Ch (chitosan conjugated with ABA). Peaks for ABA (7.2 and 7.8 ppm, arrows) confirm conjugation with chitosan (4.8 ppm, asterisk). (B) Light transmittance of Az-Ch hydrogel across the visible light spectrum. Light transmittance of Az-Ch hydrogels was assessed for freshly prepared hydrogels (0 hour) and hydrogels stored for 1 hour, 1 day, and 1 week (n = 4). (C) Schematic of the intended application of the Az-Ch hydrogel to the cornea. A bandage contact lens may be placed atop the hydrogel to aid with comfort and stability. (D, E) Photographs of pre-crosslinked hydrogels (D) and post-crosslinked hydrogels (10-second UV exposure) (E) demonstrate a high degree of optical clarity.

The intrinsic physical properties of the 10-second photocrosslinked Az-Ch hydrogels were assessed by rheometry. In amplitude/strain sweeps at 1 Hz, Az-Ch hydrogels exhibited characteristic hallmarks of classical hydrogels (Fig. 2A). The storage modulus G′ (126.5 Pa) surpassed the loss modulus G′′ (26.9 Pa), and they were linear and constant across the entire strain deformation spectrum (0.1%–100%), demonstrating the retained integrity of the hydrogel throughout applications of stress. Furthermore, the frequency sweep at 1% strain (Fig. 2B) consistently revealed a higher G′ than G′′, signifying frequency dependence across the tested frequency range (0.1–10 Hz). 

Figure 2.

 

Rheological characteristics of Az-Ch hydrogel. (A) Amplitude sweep and (B) frequency sweep of crosslinked Az-Ch hydrogels (10-second exposure at 100 mW/cm²). G′ (storage modulus) was higher than G′′ (loss modulus) across all strains and frequencies tested, indicating preserved hydrogel integrity throughout the experiment. (C) Shear moduli were measured continuously on aqueous Az-Ch solutions that were subjected to photocrosslinking with 50-mW/cm² UV curing. Longer exposure time resulted in a stiffer hydrogel, presumably as a result of increased crosslinking density. (D) Swelling ratios of Az-Ch hydrogels over time (n = 4), demonstrating a high degree of swelling over the first 8 hours.

Ten seconds of crosslinking in the 100-mW/cm² UV chamber used for most experiments was enough to produce a hydrogel. Rheological assessments of Az-Ch during crosslinking were performed using an in situ rheo-optical instrument at a power of 50 mW/cm², which permits simultaneous oscillatory shear measurements and UV light irradiation (Fig. 2C). The experiment demonstrated a fast time-dependent emergence of elastic response during gelation upon UV exposure. G′ and G′′ were initially equal followed by a rapid increase in G′, reaching a steady state (after 100 seconds) with up to a 300-fold magnitude change. This transition from viscous to elastic behavior, marked by a crossover between G′ and G′′ at very early stages, defines the gel time (t_gel), which revealed an instantaneous gelation after UV exposure using the 50-mW/cm² power lamp. Az-Ch hydrogels exhibited high swelling ratios (Fig. 2D) over time and reached a peak at approximately 6 hours. This characteristic provides insight into the impact of swelling on the rhNGF release kinetics. 

In Vitro Drug Release Characteristics

rhNGF-loaded Az-Ch hydrogels released 1.5 µg of drug (equivalent to the loading dose) over 24 hours, indicating no significant loss of rhNGF through the hydrogel production process (Fig. 3A). Release of rhNGF from hydrogels occurred rapidly during the first 8 hours and then became gradual for the remainder of the time (Fig. 3A). Drug release beyond 24 hours was minimal (data not shown). The incorporation of heparin (both 0.25 and 0.5 mg/mL) within the hydrogel reduced the burst release up to 8 hours (P < 0.05) without significantly affecting the duration or total amount released (P > 0.05). Lysozyme in the drug-release media had no significant effect on the cumulative drug release at each time point (data not shown). The rhNGF released from the Az-Ch hydrogels induced proliferation of TF1 erythroleukemic cells to a similar degree as fresh rhNGF solution (Fig. 3B), demonstrating retained bioactivity of the released rhNGF. 

Figure 3.

 

The rhNGF release characteristics of Az-Ch and heparin-modified Az-Ch hydrogels. (A) Quantification of rhNGF released from hydrogels loaded with 1.5 µg of rhNGF after incubation in drug release media (PBS). The addition of heparin at both 0.25 mg/mL (Hep-0.25) and 0.5 mg/mL (Hep-0.5) resulted in a slower and more gradual release profile than Az-Ch without heparin (Hep-0) over the first 8 hours. *P < 0.05; ns, P > 0.05 (n = 3 hydrogels). (B) The effect of rhNGF released from hydrogels (Az-Ch-NGF) on TF1 erythroleukemic cell proliferation was measured and compared to freshly prepared rhNGF solution (NGF, P > 0.05) and PBS (negative control, P = 0.004; n = 3 per condition). Data indicate mean ± SD.

Cytocompatibility, Cell Proliferation, and Migration Assays

HCLE cells grown in the presence of blank Az-Ch hydrogel (Fig. 4B) demonstrated typical polygonal morphology and preserved adherence, grossly unchanged from controls (Fig. 4A). Furthermore, growth in the presence of blank Az-Ch hydrogels did not significantly reduce the proliferation of HCLE cells (P > 0.05) (Fig. 4C). To assess the ability of drug-release media from rhNGF-loaded Az-Ch to promote corneal epithelial cell health and wound healing, HCLE cell proliferation and migration assays were performed. In the cell proliferation assay (Fig. 5A), HCLE cells grown in drug-release media from rhNGF-loaded Az-Ch hydrogels (Az-Ch-NGF) proliferated at a rate similar to those grown in fresh solutions of rhNGF (NGF), and both were higher than basal media (BM) control: 128.0% ± 34.4% (P = 0.002) and 125.4% ± 37.6% (P = 0.006), respectively, normalized to control. The presence of an NGF receptor inhibitor completely reversed the effect of the hydrogel drug-release media on cell proliferation. In the HCLE scratch assay, accelerated wound healing was observed by 4 hours in those treated with Az-Ch-NGF or NGF and persisted throughout the 12-hour observation (Fig. 5B). After 12 hours, wounds had healed to a greater extent when treated with Az-Ch-NGF (84.9% ± 13.4%; P = 0.001) (Figs. 5B–5D) or NGF (80.3% ± 13.2%; P = 0.001) (Fig. 5B) compared to controls treated with BM (62.0% ± 13.7%) (Figs. 5B, 5E, 5F). The overall wound closure rates were significantly higher in each of the treatment groups (Az-Ch-NGF: 39.8 ± 6.3 × 10³ µm²/hr, P = 0.001; NGF: 37.6 ± 6.2 × 10³ µm²/hr, P = 0.001) compared with the BM control (29.0 ± 6.4 × 10³ µm²/hr). There was no significant difference between the treatment groups in overall wound closure rates or percent wound closure at each time point. 

Figure 4.

 

Cytocompatibility of Az-Ch hydrogels. (A, B) Representative images showing cellular morphology and immunolabeling of cell nuclei (blue) and β-actin (green) in HCLE cells after growth in complete growth media (CM) supplemented with PBS (A) or Az-Ch hydrogels (B), demonstrating similar morphological characteristics after 3 days. Scale bars: 50 µM. (C) HCLE cell viability after incubation in CM supplemented with sterile PBS (control) or Az-Ch hydrogels (10 and 40 mg/mL) for 1 or 3 days. Reduction of the PrestoBlue Reagent is correlated with total viable cell count. (ns: P > 0.05; n = 4 per condition). Scale bars: 100 µm.

Figure 5.

 

HCLE cell proliferation and migration in response to rhNGF released from Az-Ch hydrogels. (A) HCLE viable cell proliferation after 24 hours in various growth conditions: nutrient-depleted BM, BM supplemented with freshly prepared rhNGF aqueous solution (NGF), BM supplemented with media containing rhNGF released from Az-Ch (Az-Ch-NGF; with or without an NGF receptor inhibitor), and CM as a positive control. Both NGF (P < 0.01) and Az-Ch-NGF (P < 0.01) induced HCLE cell proliferation to a greater degree than BM, with no significant difference between the treatment groups (P > 0.05). Data are expressed as mean ± SD (n = 8 wells per group). Experiments were repeated independently three times, and the data are normalized to the outcome of the BM group in each experiment. (B) Wound closure after 4, 8, and 12 hours, expressed in healed area relative to initial wound area. P < 0.01 at each time point for each treatment group. (C–F) Representative images of HCLE wound closure assay at start (C, E) and after 12 hours (D, F). After 12 hours, cells incubated with Az-Ch-NGF (D) or NGF (not shown) exhibited greater wound closure than BM only (control, F). Scale bars: 100 µm.

Discussion

Most current therapies for NK are aimed at protecting the ocular surface (e.g., artificial tears, punctal occlusion, bandage contact lens). Cenegermin-bkbj is the first FDA-approved therapy that directly targets the underlying pathophysiology of NK. A standard treatment course of cenegermin-bkbj is 8 weeks (6×/day), but patients often require an additional course.²⁰ Additionally, given recent evidence that cenegermin-bkbj may induce corneal nerve regeneration in the long term,^21,22 chronic use of cenegermin-bkbj may have continued benefits. As eye drops tend to have poor bioavailability and require frequent administration, the Az-Ch hydrogel described in this study could potentially be used to make more practical and efficient long-term use of the medication. 

Several properties make chitosan an ideal polymer for use in ocular drug delivery. Chitosan is highly biocompatible and biodegradable, making it a popular base for drug delivery applications.^23,24 Moreover, chitosan is known to have intrinsic wound-healing and antimicrobial properties.^24,25 The net positive charge of chitosan interacts with negatively charged residues in the glycocalyx, making the polymer mucoadhesive.^23,26 Finally, the abundance of amine groups makes chitosan readily modifiable to achieve a wide range of characteristics. In this study, we functionalized chitosan amine groups with ABA, whose azide group converts to a highly reactive nitrene-containing intermediate upon exposure to UV radiation, resulting in covalent photocrosslinking.^11,27 

As a hydrogel, Az-Ch encapsulates hydrophilic molecules such as rhNGF and allows the sustained release of drugs via diffusion through the hydrogel mesh.²⁸ Az-Ch hydrogels released rhNGF over the course of 24 hours, retaining the corneal epithelial wound-healing properties of the released drug in vitro; however, after 8 hours, the release rate gradually decreased. One strategy to reduce the release rate (and shift the release profile toward the ideal zero-order controlled release) is to incorporate a component that increases the effective drug size. Heparin electrostatically binds certain growth factors, such as NGF, with a relatively weak affinity,²⁹ and our experiments demonstrated its potential use to modulate drug release. 

The stability of rhNGF through the manufacturing process of Az-Ch was an important consideration. Az-Ch photocrosslinks in under 10 seconds, thus limiting the duration of UV exposure and mitigating rhNGF degradation during the manufacturing process. To reliably quantitate the dose-dependent NGF bioactivity, the TF1 proliferation assay was used because TF1 erythroleukemic cell line³⁰ exhibits factor-dependent proliferation in vitro.¹⁸ Furthermore, the proliferation and migration of human corneal epithelial cells has been demonstrated to be enhanced by NGF exposure,³¹ and our study revealed that rhNGF released from loaded Az-Ch hydrogels maintained this effect. 

In its commercial formulation, approximately 1.5 µg of cenegermin-bkbj is administered daily (20 µg/mL,³² 6×/day). Az-Ch hydrogels release ∼1.5 µg of rhNGF per day (per 30 µL hydrogel). A lower total amount of rhNGF was chosen given the theoretical benefits of using a sustained-release hydrogel. In the case of an eye drop, the majority of the eye drop is lost almost instantaneously through blinking, conjunctival absorption, and tear meniscus overflow.³³ As a result, the bioavailability of topically administered eye drops is typically reported to be less than 10%.³⁴ By releasing the drug continuously from a hydrogel adhered to the ocular surface, we expect a higher degree of bioavailability and less drug required to achieve equivalency.³⁵ The hydrogel may be applied and adhered to the affected area, with a persistent concentration gradient, promoting diffusion to the ocular surface. 

Although we expect no significant difference in properties or functionality, the rhNGF used in our study (derived from a mouse myeloma cell line, NS0) is produced differently from the rhNGF used in commercial applications (derived from Escherichia coli).³⁶ To facilitate adaptation to a commercial therapeutic agent, the active pharmaceutical ingredient in this hydrogel may be substituted with cenegermin-bkbj. 

Limitations of this study include the use of in vitro systems for testing. Animal models are necessary to ensure successful translation to in vivo systems. The extensive swelling of Az-Ch is a potential translational challenge; although likely allowing for sufficient oxygenation for underlying tissue, the hyperhydrous environment of the hydrogel may impede cell attachment.^11,37 The hydrogel is also limited by its duration of drug release—24 hours. We suspect that a longer duration on the ocular surface may result in rhNGF degradation, negating the benefit of a more extended-release duration, but this hypothesis must be tested. 

In summary, we have presented characterizations of a biocompatible, photocrosslinkable, chitosan-based hydrogel for the sustained release of rhNGF to the ocular surface that retains drug bioactivity. This hydrogel could potentially be placed once daily by a patient, achieving all-day drug delivery and thereby potentially improving patient adherence and allowing for more practical long-term use of the medication. Finally, the hydrogel material may be further developed and adapted for use as a delivery vehicle for other growth factors or proteins. 

Acknowledgments

Supported by the Harvard-Vision Clinical Scientist Development Program, National Eye Institute, National Institutes of Health (5K12EY016335-17), by a Schepens Eye Research Institute Core Grant (P30EY003790), and by the National Science Foundation–funded Harvard Materials Research Science and Engineering Center (DMR-1420570) 

Disclosure: L.N. Kanu, None; A.E. Ross, None; W. Farhat, None; S.V. Mudigunda, None; N. Boychev, None; L. Kuang, None; A.E.K. Hutcheon, None; J.B. Ciolino, None