Thermosensitive Chitosan–Hyaluronic Acid Hydrogel for Sustained Betamethasone and Levofloxacin Delivery in Corneal Wound Healing

Abstract

Purpose

This research designed a thermosensitive chitosan–hyaluronic acid (Ch-HA) hydrogel for the dual controlled release of betamethasone and levofloxacin to improve corneal wound healing and inhibit subsequent infections.

Methods

Ch (2% w/v) and HA (10% w/v) solutions were neutralized with varying concentrations of β-glycerophosphate. The optimal formulation was characterized for gelation time, swelling, degradation, morphology, cytocompatibility, drug release profiles, and antimicrobial activity. In vivo healing was assessed in a rabbit corneal burn model using fluorescein staining, histology, and cytokine analysis.

Results

Hydrogels with 35% β-glycerophosphate provided an 8-minute gelation time, low cytotoxicity, and favorable swelling/degradation profiles. In vitro release kinetics showed a release profile extending over 24 hours, reaching 49.6% for levofloxacin and 30.3% for betamethasone after 24 hours. The dual-loaded hydrogel exhibited effective antibacterial zones against Staphylococcus aureus (17.8 ± 1.5 mm), Escherichia coli (14.2 ± 1.8 mm), and Staphylococcus epidermidis (11.9 ± 0.8 mm). The antibacterial efficacy was comparable with the levofloxacin-only control for S. aureus and E. coli (P > 0.05), although slightly reduced for S. epidermidis (P < 0.05). In vivo, the Ch-HA-35-beta-levo group accelerated re-epithelialization (4.3 ± 0.6 days vs. 7.3 ± 0.6 days in untreated controls; P < 0.01). Histological analysis revealed significantly reduced epithelial thickening (56.5 ± 2.1 µm vs. 72.7 ± 2.1 µm; P < 0.001), and molecular analysis confirmed a significant downregulation of inflammatory markers, showing a 59% reduction in tumor necrosis factor-α and a 54% reduction in interleukin-6 expression (P < 0.05)

Conclusions

The thermosensitive Ch-HA hydrogel demonstrated promising physicochemical properties, controlled release kinetics, effective antimicrobial efficacy, and accelerated corneal healing, offering a combined therapeutic approach to simultaneously manage inflammation and infection in corneal injuries.

Translational Relevance

This instillable in situ gelling drug delivery system bridges laboratory innovation with clinical application by providing a sustained-release platform that simultaneously addresses inflammation and infection, offering a convenient once-daily regimen compared with the frequent instillation required by conventional eye drops that may improve patient compliance.

Introduction

Corneal disorders represent a significant global health burden, with the World Health Organization estimating that more than 10 million people suffer from corneal blindness worldwide.¹ The cornea's protective barriers significantly restrict drug penetration, with conventional eye drops delivering less than 5% of administered doses to target tissues.^2,3 Physiological mechanisms such as tear turnover and nasolacrimal drainage further diminish drug bioavailability.⁴

These challenges have driven research into advanced ophthalmic delivery platforms. Among these, thermosensitive hydrogels offer particular promise owing to their unique phase transition properties, converting from liquid to gel at a physiological temperature, thereby improving contact time and controlling drug release.^5–8

Chitosan (Ch), a positively charged polysaccharide, provides excellent mucoadhesion to corneal surfaces, biocompatibility, and versatile drug release capabilities.^9,10 When combined with hyaluronic acid (HA), Ch-based hydrogels exhibit enhanced potential for corneal wound healing through improved epithelial regeneration, corneal hydration, and patient comfort.^11,12

Dual drug delivery systems are particularly relevant for corneal injuries with concurrent inflammation and infection. Betamethasone mitigates inflammation that can impair healing, while levofloxacin prevents bacterial infections.^13,14 A single vehicle delivering both agents offers streamlined treatment protocols, reduced dosing frequency, and potentially synergistic effects.⁵

Thermosensitive Ch/β-glycerophosphate (β-GP) hydrogels have emerged as promising candidates for ocular delivery.¹⁵ However, their clinical application has been limited by a functional trade-off: high β-GP concentrations required for rapid gelation often induce cytotoxicity, whereas lower concentrations improve biocompatibility but result in slow sol-gel transition and rapid precorneal washout. Unlike photocross-linked systems¹⁶ that rely on external ultraviolet (UV) irradiation, thermosensitive variants offer patient-friendly administration but necessitate precise formulation optimization. Although recent investigations have explored poloxamer-based alternatives^17,18 or growth-factor-loaded scaffolds¹⁹ to enhance retention and healing, these systems often face challenges regarding rapid erosion or limited dual-drug encapsulation capacity. In this study, we developed an optimized Ch-HA hydrogel with 35% w/v β-GP, identifying a critical biocompatibility window that balances rapid in situ gelation with optimal safety. This system provides a trimodal therapeutic strategy, simultaneously delivering betamethasone and levofloxacin while leveraging the regenerative properties of HA to manage complex corneal chemical burns.

Materials and Methods

Medium-molecular-weight Ch (molecular weight, 210 kDa; determined by gel permeation chromatography using pullulan standards; degree of deacetylation, 86.3%, measured by pH titration) was sourced from Merck (Darmstadt, Germany).²⁰ β-GP, HA (25 kDa), thiazolyl blue tetrazolium bromide (MTT), and all additional compounds were also procured from Merck (Darmstadt, Germany). Cell culture substances, such as Dulbecco's Modified Eagle Medium and fetal bovine serum, were sourced from Gibco (Thermo Fisher Scientific, Waltham, MA, USA). All materials were of analytical quality (≥98% purity) and used without additional purification. Ultrapure water (resistivity exceeding 18.2 MΩ·cm at 25°C) was used throughout the study.

Preparation of Thermosensitive Hydrogels Containing HA

Thermosensitive Ch-HA hydrogels were synthesized by adapting previously established protocols.¹⁵ Briefly, 2 g of Ch was dissolved in 100 mL of 0.1 M acetic acid under continuous stirring at 25°C for 24 hours until a clear solution formed, yielding a final concentration of 2% w/v. Temperature was precisely controlled (± 0.5°C) using a thermostatically regulated water bath. The β-GP solution was sterilized by filtration through a 0.22-µm membrane. The Ch and HA precursor solutions were sterilized by UV exposure for 2 hours in a laminar flow before mixing. In parallel, 1.5 g of HA was dissolved in 15 mL of sterile deionized water, producing a 10% w/v solution.

Different concentrations of β-GP (10%, 15%, 20%, 25%, 30%, 35%, and 40% w/v) were prepared in deionized water, sterilized by 0.22 µm filtration, and cooled in an ice bath (0°C–4°C). The β-GP solution was added dropwise to the Ch solution (at a controlled rate of approximately 1 mL/min) while stirring on ice for 20 minutes, and the pH was adjusted to 7.40 ± 0.05 using a calibrated pH meter. The HA solution was then added dropwise over 15 minutes.

For the drug-loaded hydrogels, betamethasone and levofloxacin (10 mg/mL each) were introduced to the Ch-HA mixture under continuous stirring on ice for 30 minutes. All drug-loading procedures were performed under laminar flow (ISO class 5) to ensure sterility. The final formulations were kept at 4°C until further use.

The experimental formulations were prepared for different study purposes. Physicochemical tests used Ch, Ch-HA-30, Ch-HA-35, and Ch-HA-40; in vitro drug release tests used Ch-HA-35-beta, Ch-HA-35-levo, and Ch-HA-35-beta-levo; and in vivo/antibacterial studies included six groups: Ch, Ch-HA, Ch-beta, Ch-levo, Ch-beta-levo, and Ch-HA-35-beta-levo.

Physicochemical Characterizations of Hydrogel

Gelation Time

The gelation time was measured through the inverted tube technique, a widely accepted technique for evaluating sol gel transition.^21–23 All formulations were tested in triplicate (n = 3) under identical conditions to ensure reproducibility. Each sample (1 mL) was transferred into a test tube maintained at 37 ± 0.5°C in a temperature-controlled water bath. The test tube was tilted at a 45° angle every 30 seconds until no visually detectable flow occurred for at least 30 seconds. This end point was independently verified by two observers to minimize subjective bias. Samples failing to gel within 60 minutes were monitored up to 24 hours; if no transition occurred, they were designated as no gel. The method was validated using standard polymer solutions with known gelation properties to confirm reliability.

Chemical Composition Analysis of the Hydrogel

Hydrogel samples (5 mL each) were frozen at −80°C for 24 hours, then lyophilized for 24 hours using a freeze-dryer (Christ Alpha 1–2 LD plus, Martin Christ, Osterode am Harz, Germany) operating at 0.05 mbar. The resulting powders were mixed with spectroscopic grade KBr at a ratio of 1:100 w/w and compressed into transparent pellets using a hydraulic press (10 tons pressure for 2 minutes). Fourier transform infrared (FTIR) spectroscopy was performed using a Bruker Tensor 27 spectrometer in the range of 4000 to 400 cm⁻¹ at a resolution of 4 cm⁻¹, with 64 scans per sample. Spectra were baseline-corrected and normalized using OPUS software (Bruker, Billerica, MA, USA) to ensure accurate peak analysis.

Surface Morphology Analysis

Freeze-dried hydrogel specimens (Ch-HA-30, Ch-HA-35, and Ch-HA-40) were carefully sectioned using a sharp blade to minimize structural distortion. Samples were affixed to aluminum stubs with carbon tape and subsequently coated with gold using a sputter coater (Quorum Q150R ES, East Sussex, UK) set to 10 mA for 60 seconds, achieving a consistent coating thickness of approximately 15 nm. Samples were then examined under a scanning electron microscope (Seron Technologies Inc. AIS2300C, Korea) operating at an accelerating voltage of 8 Kv and a working distance of 13–16 mm. A systematic sampling approach was employed, with a minimum of five fields of view examined per sample at magnifications ranging from 150× to ensure comprehensive morphological assessment. Representative images were captured to qualitatively assess pore distribution and appearance. Blade sectioning at liquid nitrogen temperature was used for freeze-dried hydrogels to avoid fractures seen with microtomes. This method minimizes artifacts, is validated in prior studies,^24,25 and allows representative morphological analysis from multiple sample areas.

Swelling Measurement

Hydrogel samples were prepared and lyophilized as described in Chemical Composition Analysis of the Hydrogel to obtain dried hydrogels. This assay quantified water uptake/swelling from the dry, lyophilized state. Each dried hydrogel sample (W₀, approximately 50 mg, accurately weighed to ±0.1 mg) was immersed in 20 mL of phosphate-buffered saline (PBS) (pH 7.4) in sealed containers maintained at 37 ± 0.5°C in a thermostatically controlled environment. At specified intervals (0.5, 1, 1.5, 2, 4, 6, 8, 12, and 24 hours), samples were extracted, gently blotted with filter paper (Whatman No. 1; standardized 5 s gentle pressure) to remove surface liquid, and promptly weighed (W^t) using an analytical balance (±0.1 mg). The swelling ratio (%) was calculated as:

$$\begin{array}{c}\hfill \mathit{Swelling}\mathit{ratio}\left(\%\right)=([W^t-W_0]/W_0)\times 100.\end{array}$$

Measurements were performed in triplicate (n = 3) under identical conditions, and the results are shown as mean ± standard deviation (SD). The equilibrium swelling ratio was determined as the point at which three consecutive measurements showed less than 1% variation. The very early hydration transient that occurs immediately upon immersion may be faster than our first sampling time; therefore, this dry-state assay does not time resolve the initial hydration phase. All samples were processed identically under controlled blotting, bath volume, and temperature, ensuring valid comparisons within this intended dry state scope.

In Vitro Degradation

Hydrogel samples were prepared and lyophilized as described in Chemical Composition Analysis of the Hydrogel to obtain dried hydrogels. The dried hydrogel samples (approximately 50 mg each, accurately weighed to ±0.1 mg) were placed in preweighed, sterile 50-mL conical tubes and incubated in 20 mL of PBS (pH 7.4) at 37°C for 30 days. The PBS medium was replaced every 3 days to simulate physiological fluid exchange and prevent potential pH changes owing to degradation products. Samples were removed at predetermined intervals (days 1, 2, 4, 6, 8, 10, 15, 20, 25, and 30), carefully rinsed three times with distilled water to remove buffer salts, blotted to remove surface moisture, lyophilized under identical conditions as the initial preparation, and weighed using an analytical balance. Degradation (%) was calculated as:

$$\begin{array}{c}\hfill \mathit{Weight}\mathit{lost}\left(\%\right)=((W_0-W^t)/W_0)\times 100,\end{array}$$

where W₀ is the initial dry weight and W^t is the dry weight at time t. Each measurement was done in triplicate, with individual samples used for each time point to avoid reintroducing partially degraded samples into the test medium.

In Vitro Metabolic Activity Assessment (MTT Assay)

Mesenchymal stem cells (MSCs, passage 3–5) sourced from the Iranian Biological Resource Center in Tehran were cultured in Dulbecco's Modified Eagle Medium enriched with 10% fetal bovine serum, 100 U/mL penicillin, and 100 µg/mL streptomycin, under conditions of 37°C and 5% CO₂ in a humidified incubator. MSCs were selected for this initial biocompatibility screening owing to their availability and well-established protocols in our laboratory, as well as their recognized sensitivity to cytotoxic materials as recommended by ISO 10993-5 guidelines.²⁶ Cells were plated in 96-well plates at a density of 2 × 10⁴ cells per well in complete medium and incubated overnight at 37°C to facilitate adhesion.²⁷ Cell density was optimized through preliminary experiments to ensure logarithmic growth phase during the assay period.²⁸ Cytocompatibility was assessed on drug-free Ch-HA hydrogels to exclude pharmacological interference with the MTT assay.

Sterilized hydrogel formulations (Ch-HA-30, Ch-HA-35, and Ch-HA-40) were prepared under aseptic conditions. After 24 and 48 hours of incubation, the hydrogels were gently removed using sterile forceps. To avoid optical interference from detached gel fragments during the MTT colorimetric reaction, only a brief PBS rinse was applied to remove residual debris without disturbing the adherent cell layer. Subsequently, cells were incubated with 10% (v/v) MTT solution (5 mg/mL in PBS, sterile filtered) in serum-free medium for 4 hours at 37 °C to evaluate metabolic activity.²⁷

After incubation, the medium was discarded, and formazan crystals were solubilized in 100 µL of DMSO for each well. The plates underwent gentle agitation for a duration of 10 minutes to achieve complete dissolution. Optical density was subsequently measured at 570 nm using a microplate reader (BioTek Synergy HTX, Winooski, VT, USA), with a reference wavelength of 630 nm employed to adjust for background absorbance. Metabolic activity (%) was determined as the mean ± SD using the formula:

$$\begin{array}{c}\hfill \mathit{Metabolic}\mathit{activity}\left(\%\right)=\left(ODs/ODc\right)\times 100.\end{array}$$

In this context, ODs represented the optical density of cells treated with hydrogel, and ODc indicated the optical density of the untreated control group, which is considered to have 100% viability. A positive control (latex extract, expected to induce cytotoxicity) and a negative control (high-density polyethylene extract, biocompatible) were included to validate the assay, as per ISO 10993-5 recommendations. All experiments were conducted in triplicate (n = 3), with six technical replicates per condition.^26,29

In Vitro Drug Release

Drug-incorporated Ch-HA-35 hydrogel disks (8 mm diameter, 2 mm height, approximately 100 mg in weight) were prepared using custom-made silicone molds to ensure dimensional consistency. Each disk was immersed in 2 mL of PBS (pH 7.4) in sealed amber glass vials to protect light-sensitive components. Samples were incubated at 37 ± 0.5°C under constant agitation (100 rpm) in a thermostatically controlled orbital shaker.

At specified time points (0.5, 1, 2, 4, 6, 8, 12, and 24 hours) over a 24-hour period, 1 mL of the release medium was withdrawn using a calibrated autopipette and immediately replenished with an equal volume of fresh, prewarmed PBS to maintain sink conditions while preserving the constant volume of release medium. Sink conditions were verified by ensuring that drug concentrations in the release medium remained below 10% of their saturation solubility throughout the experiment. All sampling procedures were conducted in triplicate to ensure experimental reproducibility. Drug concentrations were determined by UV-visible spectrophotometry using PBS as blank where betamethasone was measured at approximately 234 nm and levofloxacin at approximately 288 to 292 nm, based on published spectra showing a strong absorption band for betamethasone near 240 to 242 nm with only a weak shoulder around 290 to 320 nm³⁰ and a principal absorption band for levofloxacin near 292 nm,³¹ minimizing cross-absorbance and confirming linear calibration in PBS (1–50 µg/mL; R² > 0.999).

Ch-HA-35 was selected for drug release studies based on its optimal gelation time (approximately 8 minutes), which provided sufficient handling time for clinical application while ensuring rapid solidification at a physiological temperature. Additionally, it demonstrated favorable swelling and degradation profiles that would facilitate controlled drug release. Drug release tests were performed for three formulations: Ch-HA-35-beta, Ch-HA-35-levo, and Ch-HA-35-beta-levo to compare single-drug and dual-drug release behavior.

Antibacterial Activity

Antimicrobial efficacy was assessed via the disk diffusion method against Escherichia coli (ATCC 25922), Staphylococcus aureus (ATCC 25923), and Staphylococcus epidermidis (ATCC 12228), selected to represent common ocular pathogens. Bacterial strains were cultured overnight in Mueller–Hinton broth at 37°C, then adjusted to the 0.5 McFarland standard (approximately 1–2 × 10⁸ CFU/mL), and confirmed through optical density measurements at 600 nm using a spectrophotometer. Disks of hydrogels (6 mm in diameter), created with a sterile biopsy punch, were positioned on Mueller–Hinton agar plates that had been inoculated with 100 µL of each standardized bacterial suspension, which was spread uniformly using a sterile glass spreader. Antibiotic disks containing 5 µg of levofloxacin were used as positive controls, whereas blank Ch-HA hydrogels devoid of any drugs functioned as negative controls. The cultured plates were maintained at 37°C for 24 hours under standardized conditions, after which the diameters of inhibition zones were precisely assessed with an electronic caliper at three different points per zone. To ensure experimental consistency, all tests were conducted in three independent replicates on separate days using freshly prepared bacterial cultures, and outcomes are expressed as averages with mean ± SD.

In Vivo Corneal Wound Healing Study

Animal Model and Grouping

Eighteen mature male New Zealand white rabbits (body weight range, 2–2.5 kg; 12–14 weeks of age) were obtained from the institutional animal facility and allowed to acclimatize for 1 week before experimentation. Animals were housed in individual standard cages within a controlled environment, featuring a temperature of 22 ± 2°C, relative humidity of 50% to 60%, and a 12-hour light/dark cycle, with continuous access to a standard diet and water. Health status was assessed daily by a veterinarian to ensure animal welfare throughout the study period. Rabbits were randomly allocated into six experimental cohorts, each comprising three animals. To mitigate bias, assignment concealment was implemented by an independent technician through the use of numerically labeled containers with group designations known only to the technician until the moment of intervention. The study protocol received formal approval from the Institutional Review Board and Ethics Committee of Tehran University of Medical Sciences (Approval Code: IR.TUMS.FARABIH.REC.1399.001). All procedures complied with the ARRIVE guidelines for reporting animal research.³² All animal procedures were conducted in strict accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.

Corneal Burn Induction and Hydrogel Application

Rabbits were anesthetized with intramuscular xylazine (10 mg·kg⁻¹) and ketamine (50 mg·kg⁻¹). Depth of anesthesia was confirmed by the absence of the corneal reflex and withdrawal response to toe pinch. Local anesthesia was performed using 1% w/v tetracaine eye drops administered 5 minutes before the procedure.

An 8-mm filter paper disk saturated with absolute ethanol (100%) was applied to the central cornea of the both eyes for exactly 20 seconds to create an ethanol-induced corneal burn.³³ The left eye received the hydrogel treatment, while the contralateral right eye served as an injured untreated control. To ensure reproducibility and minimize technical variability, the procedure was performed by a single blinded investigator using a consistent, gentle placement technique without applying excessive pressure. The exposure time was strictly monitored using a digital chronometer. The disk was removed, and the eye was immediately rinsed with 10 mL sterile saline to limit exposure of deeper ocular tissues to ethanol. The extent of initial damage was documented using fluorescein staining and slit lamp photography to ensure consistent baseline injury across all subjects. Each treatment group received a specific hydrogel formulation (Ch, Ch-HA, Ch-beta, Ch-levo, Ch-beta-levo, or Ch-HA-35-beta-levo, 50 µL volume) applied topically to the corneal surface using a micropipette with sterile tips, while the right eye served as an untreated control. The hydrogels were applied once daily throughout the study period, at the same time each day (09:00 ± 1 hour) to minimize circadian variations in healing responses. After hydrogel application (50 µL volume applied using a micropipette), eyes were gently held closed for 2 minutes to facilitate initial mucoadhesive interaction with the ocular surface and prevent immediate drainage, allowing the thermosensitive transition to proceed in situ. This frequency was specifically selected based on the hydrogel's strong mucoadhesive properties, which significantly extended precorneal residence time. Clinical observations confirmed that the gel remained macroscopically visible on the ocular surface for 6 to 8 hours after application. Although the in vitro release extends to 24 hours, this prolonged residence time (compared with drops) allows for substantial drug absorption and tissue accumulation, supporting a once-daily regimen. Postprocedure care included daily monitoring of general health parameters (food intake, body weight, activity level), and corneal evaluation for signs of excessive inflammation or infection. Elizabethan collars were used to prevent self-injury. Buprenorphine (0.05 mg/kg subcutaneously every 12 hours for the first 48 hours) was administered as analgesic support. To ensure unbiased clinical scoring, a blinded investigator evaluated corneal healing without knowledge of treatment group assignments.

Clinical Monitoring and Fluorescein Staining

Corneal wound closure was monitored with 1% fluorescein sodium staining every 24 hours (days 1–9). A standardized image acquisition protocol was established, with all images taken under consistent lighting conditions using a digital camera (Nikon D810, Tokyo, Japan) equipped with a macro lens and cobalt blue filter at a fixed distance from the corneal surface. The stained defect area was photographed, and digital images were transferred to a computer workstation for analysis. Quantification was performed using ImageJ software (National Institutes of Health, Bethesda, MD, USA; version 1.53) by a researcher blinded to treatment allocation. The fluorescein-stained area was outlined using the freehand selection tool, calibrated using a scale marker in each image, and measured in square millimeters. Three independent measurements were performed for each image, and the mean value was recorded. Complete healing was defined as absence of fluorescein uptake in the lesion area, indicating restoration of the corneal epithelial barrier. Day 2 and day 4 images were selected for presentation in Figure 8a, because they represented critical time points in the healing process, with day 4 showing complete healing in the Ch-HA-35-beta-levo group.

Figure 8.

In vivo assessment of corneal regeneration. (a) Clinical imaging of fluorescein-stained corneal lesions in rabbit eyes: Hydrogel-treated left eyes versus nontreated right eyes at 2- and 4-day postintervention intervals. Epithelial defects are marked by green fluorescence in central corneal regions. The treatment duration was defined as the post-treatment interval when full epithelial closure was observed in treated eyes. (b) Quantitative comparison of epithelial regeneration timelines across experimental groups. Data expressed as mean ± SD (N = 3). Statistical significance: *P ≤ 0.05.

Histological Analysis

On the ninth experimental day, designated rabbits were humanely euthanized with intravenous pentobarbital sodium (100 mg/kg) after sedation with ketamine/xylazine. Corneal specimens were surgically excised with a 2-mm scleral rim to maintain tissue orientation and structural integrity. The collected tissues underwent sequential processing: primary fixation in 10% neutral buffered formalin for 24 hours at room temperature, dehydration through a graded ethanol series (70%, 80%, 90%, and 100%, 30 minutes each), xylene-based clearing (two changes, 30 minutes each), and final embedding in paraffin blocks. Thin sections (5 µm thick) were prepared using a rotary microtome (Leica RM2255, Wetzlar, Germany) from the central corneal region.

Sections were adhered to positively charged glass slides, deparaffinized, and processed for hematoxylin and eosin staining in accordance with standard protocols. Quantitative analysis was conducted using ImageJ software (National Institutes of Health) by an experienced histopathologist blinded to treatment groups. For epithelial thickness measurement, five randomly selected fields per section were photographed at 400× magnification, and five measurements were taken at equal intervals across each field. Stromal keratocyte density was determined by counting cell nuclei in 10 randomly selected high-power fields (400×) per sample and expressing the result as cells per high-power field (HPF). All quantitative analyses were conducted in triplicate to guarantee the reliability and reproducibility of the results.

Gene Expression (Quantitative Reverse Transcription Polymerase Chain Reaction)

Corneal tissue specimens were excised on day 9 post-intervention, immediately flash-frozen in liquid nitrogen, and preserved at −80°C until analysis. Total RNA was isolated using the RiboEx extraction system (GeneAll Biotechnology, Seoul, Korea) in accordance with manufacturer specifications. RNA integrity and concentration were assessed via NanoDrop spectrophotometry, with samples exhibiting A260/A280 ratios between 1.8 and 2.0 selected for downstream applications. Reverse transcription was performed with the Parstous cDNA Synthesis Kit (Tehran, Iran) using 1 µg of purified RNA per reaction. A quantitative analysis of gene expression for the inflammatory markers tumor necrosis factor (TNF)-α and interleukin (IL)-6 was performed using a Rotor–Gene Q real-time polymerase chain reaction system from Qiagen (Hilden, Germany). The amplification reactions included SYBR Green Master Mix (Ampliqon, Odense, Denmark), gene-specific primers at a concentration of 10 pM/μL, and the template cDNA. The thermal cycling profile included an initial denaturation step at 95°C for 10 minutes, succeeded by 40 amplification cycles at 95°C for 15 seconds and 60°C for 60 seconds. The specificity of amplification was confirmed via melting curve analysis, incorporating suitable controls in every experimental run. Relative quantification was performed using the comparative threshold cycle method (2^−ΔΔCt), with GAPDH serving as the endogenous reference standard for normalization.

Statistical Analysis

Quantitative data are expressed as mean ± SD. Statistical analyses were conducted using GraphPad Prism version 9.0 (GraphPad Software, San Diego, CA, USA). The normality of data distribution was verified using the Shapiro–Wilk test. Intergroup comparisons were performed using one-way analysis of variance followed by Tukey's post hoc test for multiple comparisons, with statistical significance defined as a P value of less than 0.05. Although the in vivo sample size was limited to three in accordance with the 3R principles of animal welfare, the statistical robustness was validated by calculating the effect size (Cohen's d). Large effect sizes (d > 0.8) confirmed that the observed significant differences were driven by substantial treatment effects rather than random variation, ensuring the reliability of the findings, despite the minimized sample size.

Results

Characterization of Thermosensitive Hydrogel

Gelation Time

Hydrogels containing 10% to 15% w/v β-GP did not gel within 24 hours at 37°C. In contrast, formulations with greater than 30% w/v β-GP (35% and 40%) gelled in under 10 minutes, whereas the 30% formulation required approximately 15 minutes. Specifically, the Ch-HA-35 hydrogel displayed an optimal gelation time of approximately 8 minutes, striking a balance between manageable handling and rapid in situ gel formation (Fig. 1a). This moderate gelation time allowed for convenient application while ensuring prompt solidification at the ocular surface, which is crucial for maintaining therapeutic concentrations.

Figure 1.

Physicochemical characterization of hydrogel formulations. (a) Macroscopic observation of the sol-gel transition. The Ch-HA-35 formulation (left) flows as a solution at 25°C, while (right) it forms a stable, nonflowing hydrogel after incubation at 37°C. (b) FTIR spectra of HA and Ch-HA-35.

Chemical Structure Analysis of Hydrogel

FTIR analysis confirmed the successful integration of hydrogel components (Fig. 1b). The spectrum of HA displayed characteristic peaks at 1624 cm⁻¹ (amide I) and 1406 cm⁻¹ (carboxylate group). Following the formation of the Ch-HA hydrogel, a new peak appeared at 1077 cm⁻¹, confirming the incorporation of β-GP. Crucially, significant spectral changes indicated molecular interactions: the HA carboxylate peak shifted from 1406 cm⁻¹ to 1391 cm⁻¹. This shift provided direct evidence of strong electrostatic cross-linking between the protonated amine groups (–NH₃⁺) of Ch and the anionic carboxyl groups (–COO⁻) of HA. Additionally, although the N–H/O–H stretching vibration remained positionally stable (from 3428 cm⁻¹ to 3429 cm⁻¹), the band exhibited broadening, suggesting enhanced hydrogen bonding networks facilitated by the glycerol groups of β-GP, which contributed to the supramolecular stability of the hydrogel matrix.

Surface Morphology (Scanning Electron Microscopy [SEM])

SEM images showed an interconnected, porous structure in freeze-dried samples. Higher β-GP concentrations were associated with smaller, more closely packed pores, suggesting a denser crosslink network (¹Fig. 2). The Ch-HA-35 formulation exhibited an optimal balance of pore size and distribution, creating a network structure conducive to controlled drug diffusion while maintaining sufficient mechanical integrity. This interconnected porosity plays a crucial role in facilitating both adequate swelling for drug release and cellular interactions necessary for biocompatibility.

Figure 2.

SEM images of the Ch-HA-30, Ch-HA-35, and Ch-HA-40 formulations, which showed interconnected porous structure.

Cytotoxicity Assay

MSCs maintained high viability with Ch-HA-30 formulation (90% at 24 hours, 85% at 48 hours) and Ch-HA-35 formulation (85% at 24 hours, 80% at 48 hours), both exceeding the 70% ISO 10993 biocompatibility threshold.²⁶ However, Ch-HA-40 formulation demonstrated reduced viability (75% at 24 hours, 60% at 48 hours), falling below acceptable biocompatibility standards at 48 hours with a statistically significant difference from control (P < 0.01) (Fig. 3).²⁹

Figure 3.

The metabolic activity of MSCs after 24 and 48 hours of incubation with different Ch-HA hydrogels. Data are presented as mean ± SD, N = 3. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001.

Swelling Behavior

All hydrogels exhibited rapid initial swelling in the first 90 minutes, followed by a slower uptake over 24 hours. Ch-HA-35 reached an equilibrium swelling ratio of approximately 240% by approximately 1.5 hours and maintained a plateau up to 24 hours (Fig. 4). The presence of HA enhanced the hydrogel's water retention capacity compared with Ch-only formulations.

Figure 4.

Swelling ratio percentage of different Ch-HA hydrogels. All formulations reached swelling equilibrium after approximately 1.5 hours and then maintained a plateau up to 24 hours.

In Vitro Biodegradation Study

In PBS (pH 7.4, 37°C), the hydrogels showed a biphasic degradation profile over 30 days. An initial faster weight loss was observed during the first 4 days, followed by a slower rate. Ch-HA-35 lost about 50% of its original mass by day 8 and approximately 86% by day 30, reflecting a controlled degradation that can support sustained therapeutic delivery (Fig. 5). This degradation timeline aligned well with typical corneal wound healing processes, allowing for drug release during the critical inflammatory and proliferative phases while ensuring eventual clearance of the hydrogel matrix.

Figure 5.

In vitro degradation profiles of different Ch-HA hydrogels exposed to PBS (pH 7.4) at 37 °C.

In Vitro Drug Release

Drug release profiles from the Ch-HA-35 hydrogel were monitored over a 24-hour period (Fig. 6). The Ch-HA-35 hydrogel exhibited distinct release patterns for levofloxacin and betamethasone over 24 hours, with both drugs demonstrating continuous and sustained release kinetics. Levofloxacin consistently showed higher release rates than betamethasone throughout the observation period.

Figure 6.

The cumulative release profiles of betamethasone and levofloxacin from Ch-HA-35 hydrogel at different time intervals. Data are presented as mean ± SD, N = 3.

In the initial 6 hours, levofloxacin reached 15% cumulative release compared with betamethasone's 9%. The release accelerated during the 8- to 12-hour period, with levofloxacin attaining 33% and betamethasone reaching 21% by the 12-hour mark. Between 12 and 24 hours, levofloxacin continued steady release to achieve 49.6 ± 2.1% cumulative release, while betamethasone's release rate diminished toward a plateau, reaching 30.3 ± 1.8% at 24 hours.

Importantly, a comparative analysis revealed no statistically significant differences between single-drug and dual-drug formulations. Betamethasone release from Ch-HA-35-beta (32.1 ± 2.2%) was comparable with its release from Ch-HA-35-beta-levo (30.3 ± 1.8%). Similarly, levofloxacin release from Ch-HA-35-levo (51.8 ± 2.5%) was statistically equivalent to its release from the dual drug formulation (49.6 ± 2.1%). No significant difference was observed between single- and dual-drug release profiles for both betamethasone and levofloxacin under the applied UV–Vis conditions.

Antibacterial Activity

Quantitative measurements of inhibition zone diameters revealed distinct performance differences between formulations (Figs. 7a, 7b). Ch, Ch-HA-35, and Ch-beta formulations produced minimal inhibition zones (4–6 mm) across all tested bacterial strains. The Ch-levo formulation generated inhibition zones of 16.7 ± 1.2 mm against E. coli, 17.5 ± 0.9 mm against S. aureus, and 15.2 ± 0.7 mm against S. epidermidis. The Ch-beta-levo formulation produced zones measuring 13.1 ± 2.1 mm, 16.9 ± 1.3 mm, and 15.3 ± 0.9 mm against the respective pathogens. The Ch-HA-35-beta-levo formulation exhibited inhibition zones of 14.2 ± 1.8 mm, 17.8 ± 1.5 mm, and 11.9 ± 0.8 mm against the three bacterial strains.

Figure 7.

Antibacterial activity of different hydrogels on E. coli, S. aureus, and S. epidermidis. (a) Visualized inhibition zone (white scale bar, 10 mm). (b) Histogram of inhibition zone diameter. Data are expressed as mean ± SD, N = 3. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001 indicate significant differences when compared with the Ch group. ^#P ≤ 0.05 and ^##P ≤ 0.01 indicate significant differences when compared with the Ch-HA-35 group.

Statistical analysis showed significant differences (P < 0.001) between levofloxacin-containing formulations (Ch-levo, Ch-beta-levo, and Ch-HA-35-beta-levo) and formulations without antibiotics (Ch, Ch-HA-35, and Ch-beta). The differences between Ch-levo and the dual-drug formulations did not reach statistical significance for E. coli and S. aureus, while a significant difference (P < 0.05) was observed for S. epidermidis.

In Vivo Corneal Wound Healing Study

Fluorescein staining revealed the progression of corneal epithelial healing across different treatment groups (Fig. 8a). At day 2 post injury, all groups displayed extensive fluorescein uptake in the central cornea, indicating comparable epithelial defects at this timepoint. By day 4, the Ch-HA-35-beta-levo treated eyes exhibited complete absence of fluorescein staining, while all other treatment groups and untreated control eyes still showed varying degrees of fluorescein uptake, indicating incomplete re-epithelialization. Quantitative analysis of recovery time (Fig. 8b) revealed biological variability in the baseline healing rate of untreated control eyes across cohorts (ranging from 5.3 to 8.3 days). Consequently, efficacy was evaluated based on the relative acceleration of healing within each subject (treated vs. contralateral control). For instance, in the Ch-only group, controls healed in 8.3 ± 0.6 days versus 6.3 ± 0.6 days for treated eyes, while in the Ch-levo group, controls healed in 5.3 ± 0.6 days versus 4.0 ± 1.0 days for treated eyes. The Ch-HA-35-beta-levo formulation resulted in a recovery time of 4.3 ± 0.6 days, which was significantly faster than its specific control group (7.3 ± 0.6 days; P < 0.01). Although all treatment groups accelerated healing compared with untreated controls, the Ch-HA-35-beta-levo group showed comparable healing times with the Ch-beta-levo group (4.0 ± 1.0 days), with no statistically significant difference between these two dual-drug formulations. However, the HA-containing hydrogel significantly improved epithelial quality and reduced scarring, as evidenced by histological analysis.

Histological Studies

Hematoxylin and eosin staining on day 9 revealed a thinner, more uniform corneal epithelium and fewer inflammatory cells in the Ch-HA-35-beta-levo–treated corneas compared with untreated controls. Controls exhibited a thicker epithelium, edema, and more pronounced inflammatory infiltration, underscoring the improved tissue organization in dual-drug–treated eyes (Fig. 9a). The quantitative results, as shown in Figure 9b, indicated that corneas without hydrogel treatments displayed significant epithelial thickening. In contrast, corneas filled with Ch-HA-35-beta-levo hydrogel formed a thinner layer of epithelial cells. The epithelium thickness in Ch-HA-35-beta-levo was also significantly lower than other treatments. Furthermore, the keratocytes density in superficial stroma were significantly increased in the untreated groups compared with the Ch-HA-35-beta-levo treatment group (Fig. 9c).

Figure 9.

Histopathological evaluation of hydrogel-treated corneas at day 9 post surgery. (a) Hematoxylin and eosin (H&E)-stained cross-sections illustrating corneal morphology. The first column shows untreated injured corneas (control), and the second column shows the corresponding treated corneas for each group: (A) Ch, (B) Ch-HA, (C) Ch-beta, (D) Ch-levo, (E) Ch-beta-levo, and (F) Ch-HA-35-beta-levo. Key structures are labeled: epithelium, Bowman's layer, keratocytes (*), stroma, Descemet's membrane, and endothelium. (b) Quantitative assessment of epithelial layer thickness and (c) stromal keratocyte population density derived from hematoxylin and eosin (H&E) analysis. Data represent mean ± SD (N = 3 biological replicates). Statistical significance: *P ≤ 0.05, **P ≤ 0.01 (vs. untreated controls).

Comprehensive statistical analysis demonstrated that Ch-HA-35-beta-levo produced a significant reduction in epithelial thickness (56.5 ± 2.1 µm) compared with its specific untreated control (72.7 ± 2.1 µm; P < 0.001). This comparison highlighted the hydrogel's ability to prevent injury-induced epithelial thickening observed in the control group. Although Ch-beta-levo also showed efficacy (59.4 ± 2.4 µm), the addition of HA provided incremental structural benefits. Similar trends were observed for stromal keratocyte density, with the Ch-HA-35-beta-levo group showing reduced inflammatory cell infiltration (86.7 ± 1.5 cells/field) compared with its control (103.0 ± 6.1 cells/field; P < 0.05).

Molecular Studies

TNF-α and IL-6 gene expression levels were quantified in corneal tissues on day 9 (Fig. 10). For TNF-α (Fig. 10a), the relative messenger RNA expression levels were: Ch (3.2 ± 0.3), Ch-HA-35 (2.9 ± 0.1), Ch-beta (1.8 ± 0.4), Ch-levo (2.2 ± 0.3), Ch-beta-levo (1.6 ± 0.3), and Ch-HA-35-beta-levo (1.3 ± 0.2). For IL-6 (Fig. 10b), the relative messenger RNA expression levels were: Ch (3.7 ± 0.2), Ch-HA-35 (3.0 ± 0.5), Ch-beta (2.5 ± 0.4), Ch-levo (2.8 ± 0.2), Ch-beta-levo (2.2 ± 0.2), and Ch-HA-35-beta-levo (1.7 ± 0.3).

Figure 10.

Quantitative evaluation of proinflammatory cytokine expression. (a–b) Quantitative reverse transcription polymerase chain reaction (qRT-PCR)–based quantification of relative messenger RNA expression levels for proinflammatory markers (a) TNF-α and (b) IL-6 in experimental treatment groups. Data are expressed as mean ± SD from independent biological replicates (N = 3). Error bars explicitly represent the SD to visualize data dispersion. The asterisk (*) denotes a statistically significant difference (*P ≤ 0.05) compared with the Ch group, as determined by one-way analysis of variance.

Statistical analysis revealed significantly lower expression of both inflammatory markers in the Ch-HA-35-beta-levo group compared with the Ch group (P < 0.05) as indicated by the asterisk in the figure. All betamethasone-containing formulations demonstrated reduced expression of the pro-inflammatory cytokines compared with the Ch and Ch-HA-35 groups, with the Ch-HA-35-beta-levo formulation showing the lowest expression levels for both markers.

Discussion

The thermosensitive Ch-HA hydrogel designed in this study presents a promising framework for delivering betamethasone and levofloxacin simultaneously to manage corneal injuries. By incorporating 35% w/v β-GP, our optimized formulation gelled in approximately 8 minutes at 37°C, balancing manageability with rapid in situ formation. Crucially, although the complete sol-gel transition required approximately 8 minutes, the immediate retention of the formulation upon application was facilitated by the strong mucoadhesive properties of Ch and HA. This initial mucoadhesive anchoring prevents immediate precorneal washout during the lag time required for thermal gelation, allowing the hydrogel to solidify gradually and withstand shear forces from blinking. This property directly addresses the frequent dosing and poor bioavailability issues typically encountered with conventional eye drops, where less than 5% of the administered dose reaches target tissues⁴ Similarly, HA-based platforms have been reported for simultaneous delivery of a corticosteroid and levofloxacin to improve residence time and controlled release.³⁴

The SEM analysis revealed an interconnected porous structure that varied with β-GP concentration. The Ch-HA-35 formulation exhibited a well-balanced porous architecture characterized by uniform distribution of interconnected cavities, creating an appropriate microenvironment for controlled drug diffusion. This morphological feature is critical for the hydrogel's functionality, because it influences both mechanical properties and release kinetics. The network of interconnected pores facilitates fluid exchange while maintaining structural integrity, which are essential characteristics for an effective ocular drug delivery system. The qualitative assessment of the SEM images suggests that the Ch-HA-35 formulation presented an optimal balance between pore density and structural stability compared with Ch-HA-30 (which showed larger, less defined pores) and Ch-HA-40 (which exhibited smaller, more tightly packed pores).

The swelling behavior of Ch-HA-35 hydrogel demonstrated a biphasic pattern with rapid initial uptake followed by equilibrium, reaching an equilibrium swelling ratio of approximately 240% by approximately 1.5 hours and maintaining a plateau up to 24 hours. This controlled expansion represents a significant advantage for ocular applications, as excessive swelling could cause discomfort and altered vision, while insufficient hydration would limit drug dissolution and release. The incorporation of HA enhanced the water retention capacity compared with Ch-only formulations reported in previous studies,¹⁵ where typical swelling ratios ranged from 100% to 150%. This improvement can be attributed to HA's intrinsic hydrophilicity and its ability to form extensive hydrogen bonds with water molecules.

The degradation profile of Ch-HA-35 showed controlled mass loss (approximately 50% by day 8 and 86% by day 30), which aligns strategically with the timeline of corneal epithelial regeneration.³⁵ Importantly, this degradation timeline is particularly suitable for managing complex corneal injuries, where healing may extend beyond the typical 24- to 48-hour epithelial regeneration period seen in simple abrasions. Although normal corneal epithelial defects heal rapidly, our system is designed for complicated cases involving inflammation, infection risk, or delayed healing, where sustained drug delivery provides therapeutic benefit during the inflammatory and remodeling phases. The controlled degradation ensures drug availability throughout the critical healing period while preventing excessive accumulation or long-term foreign body presence. This stability contrasts favorably with other common in situ gelling systems, such as poloxamer-based formulations, which often suffer from rapid erosion owing to dilution by tear turnover.^8,17,36 The Ch-HA matrix demonstrated superior structural integrity, likely attributable to the robust ionic interactions between the amine groups of Ch and the carboxyl groups of HA, as verified by FTIR. However, a limitation of the present in vitro degradation assay is the absence of lysozyme or other relevant enzymes in PBS. As lysozyme is naturally present in physiological fluids such as tears and can accelerate the degradation of Ch-based hydrogels,³⁷ the actual in vivo degradation rate may be higher than observed in our in vitro setting.

The observed differential release kinetics, characterized by approximately 30% release of betamethasone and 50% release of levofloxacin after 24 hours, align well with the physicochemical properties of these drugs and their interactions within the hydrogel matrix. Obara et al.¹⁶ reported an 80% release of hydrophobic paclitaxel from Ch hydrogels over 96 hours, suggesting that our observation of 30% betamethasone release at 24 hours represents a controlled profile driven by hydrophobic interactions with the polymer network. Conversely, Ruel-Gariépy et al.³⁸ demonstrated that low-molecular-weight hydrophilic compounds typically release more than 80% within 24 hours, which indicates that our lower 50% levofloxacin release reflects a partial retention by the cross-linked hydrogel structure. This distinction highlights how water content and pore size primarily influence hydrophilic drug diffusion, whereas hydrophobic interactions within the polymer network retard the release of lipophilic compounds.³⁹ Crucially, the release profile indicates that a significant therapeutic dose (approximately 15%–33%) is delivered during the initial 6 to 8 hours of retention. This finding suggests that the hydrogel functions as a high-capacity reservoir, facilitating drug saturation and tissue uptake (depot effect) during the contact phase, which sustains therapeutic activity even after the macroscopic clearance of the hydrogel.

The antibacterial efficacy demonstrated against clinically relevant pathogens (S. aureus, S. epidermidis, and E. coli) highlights the potential of this system for preventing secondary infections, a common complication in corneal injuries.⁴⁰ Notably, betamethasone did not attenuate levofloxacin's antimicrobial activity, an important finding considering previous reports of steroid-induced suppression of antibiotic efficacy.⁴¹ The comparable antibacterial performance between Ch-beta-levo and Ch-HA-35-beta-levo formulations indicates that HA incorporation preserves antimicrobial functionality while contributing additional benefits to wound healing.

The enhanced therapeutic efficacy observed with our dual-drug delivery system can be attributed to several synergistic mechanisms at the molecular level. Betamethasone likely exerts its anti-inflammatory effects through binding to cytoplasmic glucocorticoid receptors, subsequently modulating the expression of inflammatory mediators by inhibiting the nuclear factor-κB pathways.^42,43 This result was evidenced by the pronounced reduction in IL-6 and TNF-α expression in our study. Concurrently, levofloxacin prevents bacterial colonization by inhibiting DNA gyrase and topoisomerase IV.⁴⁴ Although the ethanol burn model represents a sterile injury, the inclusion of levofloxacin serves a critical prophylactic role,⁴⁵ mimicking the clinical management of corneal abrasions or photorefractive keratectomy. By maintaining a sterile microenvironment, levofloxacin allows the regenerative properties of the hydrogel matrix to proceed unimpeded. Additionally, the Ch component contributes through its interaction with cell membranes and positive charge, which promotes adhesion to the negatively charged corneal surface.^46,47 Meanwhile, HA facilitates epithelial migration through CD44 receptor-mediated signaling pathways, potentially activating Rho GTPases that regulate cytoskeletal rearrangement necessary for cell movement during re-epithelialization.^48,49

The incorporation of HA into Ch/GP hydrogel resulted in significant therapeutic advantages as evidenced by our in vivo studies. Although the HA-containing formulation (Ch-HA-35-beta-levo) showed comparable re-epithelialization rates to the Ch-beta-levo group (4.3 ± 0.6 days vs. 4.0 ± 1.0 days), the inclusion of HA proved critical for the quality of healing. The statistical similarity in closure time suggests that while drugs drive the speed of closure, the HA matrix significantly mitigates the adverse inflammatory response and prevents epithelial thickening, as evidenced by the superior histological outcomes and lower cytokine levels. This superior performance can be attributed to HA's multifunctional properties, which our data suggest operate through multiple complementary mechanisms: promoting epithelial regeneration as evidenced by the more organized epithelial layer observed in histological analysis,⁵⁰ optimizing corneal hydration as reflected in the improved swelling characteristics,⁵¹ and reducing inflammatory processes demonstrated by the lower inflammatory cell infiltration and cytokine expression.⁵² HA incorporation may modulate the microstructure and hydration behavior of the composite hydrogel, which could influence mechanical performance; however, quantitative rheological testing is required.

Histological and molecular analyses further substantiate the enhanced therapeutic effects of the Ch-HA-35-beta-levo formulation. The significantly thinner, more organized epithelium (56.5 ± 2.1 µm vs. 72.7 ± 2.1 µm in untreated controls) and reduced keratocyte density in the superficial stroma indicate more efficient resolution of inflammation and reduced scarring. The variation in epithelial thickness among untreated control groups, particularly the hypertrophy observed in the Ch-HA-35-beta-levo control cohort, reflects the heterogeneous nature of the wound healing response, where unregulated inflammation can lead to varying degrees of epithelial hyperplasia. The dual-drug hydrogel significantly mitigated this response, maintaining epithelial thickness closer to physiological norms.

At the molecular level, the greater suppression of proinflammatory cytokines (TNF-α and IL-6) in the Ch-HA-35-beta-levo group compared with Ch-beta-levo (P < 0.05) provides compelling evidence for HA's additional anti-inflammatory benefits. These findings parallel existing evidence that HA promotes epithelial migration, while corticosteroids modulate cytokine pathways.^53,54

The ethanol burn model was selected for this study to create a standardized and reproducible epithelial defect, allowing us to directly evaluate the healing effects of our dual-drug hydrogel without the confounding influence of an active bacterial infection. This approach also reflects common clinical situations, such as refractive surgery (e.g., photorefractive keratectomy), where intentional epithelial defects are produced with ethanol and prophylactic antibiotics are routinely given to prevent secondary bacterial colonization. The observed enhancement in corneal healing in the dual-drug group can be explained by several synergistic mechanisms: the anti-inflammatory action of betamethasone, the preventive antimicrobial effect of levofloxacin, and the intrinsic regenerative properties of the hydrogel matrix, especially HA, which is known to support epithelial regeneration.¹⁹ Taken together, these factors likely contributed to the accelerated wound closure in our model. Future studies will further explore these mechanisms using infectious models to better simulate clinical conditions with active infection.

Despite the promising therapeutic outcomes, this study has limitations. First, regarding physicochemical characterization, although the inverted tube method confirmed practical gelation time, oscillatory rheometry was not used to quantify viscoelastic moduli. Also, drug quantification was performed via UV-Vis spectrophotometry based on distinct absorption maxima; while established for comparative release profiles, future pharmacokinetic studies using high-performance liquid chromatography would provide superior resolution. Additionally, the use of enzyme-free PBS likely underestimated degradation rates, as physiological tear lysozyme is known to accelerate Ch hydrolysis in vivo.

Furthermore, swelling behavior was evaluated using lyophilized samples to determine equilibrium capacity; however, this static method differs from the clinical sol-gel transition on the ocular surface, implying that in situ volume expansion may vary from the reported dry-state ratios. Second, biological evaluations were constrained by the use of MSCs instead of corneal epithelial cells and a small animal sample size (n = 3). Consequently, this work should be interpreted as a pilot study establishing proof-of-concept efficacy. However, the calculation of large effect sizes (Cohen's d > 0.8) for key histological and molecular outcomes confirms that the observed differences are biologically robust and statistically valid despite the limited sample number. The observed variation in healing times among the untreated control groups (Fig. 8b) underscores the heterogeneity of biological responses in rabbits. Future studies should use larger sample sizes to normalize baseline variability; however, the consistent significant improvement of treated eyes over their specific contralateral controls validates the therapeutic effect. Finally, the 24-hour in vitro release profile does not fully encompass the multiday in vivo healing duration, and although the combined therapy proved superior, a formal factorial design would be required to mathematically distinguish between additive and synergistic effects. To advance this technology toward clinical application, future research should establish in vitro–in vivo correlations for release kinetics, evaluate long-term safety, compare efficacy against clinical standards in controlled trials, and develop standardized production protocols; assessing performance in other ocular surface disorders may further expand its therapeutic applications.

Conclusions

The thermosensitive Ch-HA hydrogel with β-GP (35% w/v) loaded with betamethasone and levofloxacin demonstrated optimal physicochemical properties for ocular application, including an 8-minute gelation time at 37°C, effective antibacterial activity against common ocular pathogens, and controlled release of both drugs (50% levofloxacin and 30% betamethasone over 24 hours). In the rabbit corneal burn model, this formulation significantly accelerated wound healing compared with untreated controls and improved tissue architecture compared with drug-loaded Ch controls, with substantially reduced epithelial thickness (56.5 ± 2.1 µm vs. 72.7 ± 2.1 µm in controls) and decreased proinflammatory cytokine expression (59% reduction in TNF-α and 54% reduction in IL-6). By simultaneously addressing inflammation and infection within a single biocompatible scaffold, this system offers a promising approach to enhance therapeutic outcomes in corneal injuries while reducing dosing frequency and improving patient compliance. Further clinical trials are warranted to validate these preclinical findings for human application.