Improved Physicochemical Stability and Retinoblastoma Anticancer Activity of Salt-Caged Acid–Base Neutralized Melphalan Loaded in L-Glutathione and pH Responsive Fattigated Nanoparticles

Abstract

Purpose

The clinical uses of melphalan (MEL) for ocular retinoblastoma (RB) delivery are limited by its poor aqueous stability and insufficient ocular bioavailability. Intravitreal injection of MEL is also invasive and less common for patients. A novel MEL-loaded formulation was developed to improve the stability and RB cellular uptake of MEL using the acid–base neutralization (ABN) and the enzyme-responsive hyaluronic acid (HA)–cystamine (CYS)–oleic acid (OA) nanoparticles (HON).

Methods

An amphiphilic HA–CYS–OA conjugate (HOC) with an L-glutathione (GSH)-responsive CYS linker was synthesized via a two-step coupling method and subsequently self-assembled to form HON. Neutralized MEL (NM) was prepared using ABN, loaded into HA (NM–HA) and HON (NM-HON), and lyophilized to obtain solid-state powders.

Results

ABN reduced the hydrolysis of MEL by forming salt-induced cage nanoaggregates. The ionic complexation of NM with HA further improved the stability of the solution. The NM-HON exhibited improved chemical and solid-state stabilities through synergistic ionic complexation and hydrophobic encapsulation. HONs demonstrated nanoscale size, negative zeta potential, and high entrapment efficiency based on dynamic light scattering (DLS), field-emission scanning electron microscopy (FE-SEM), and field-emission transmission electron microscopy (FE-TEM), whereas fourier-transform infrared spectroscopy (FT-IR), differential scanning calorimetry (DSC), and high-power X-ray diffraction (HP-XRD) indicated amorphization and strong drug–polymer interactions. NM-HON and NM-HA exhibited enhanced in vitro permeability compared to free MEL and NM. Most importantly, NM-HON exhibited controlled drug release, pH sensitization, and GSH responsiveness via CYS linker. NM-HON exerted improved anticancer activity and the lowest inhibitory concentration (10.52 ± 2.24) in RB cells (WERI-Rb-1) with statistical significance as compared with free MEL (7.53 ± 0.35 µg/mL) while preserving low toxicity in normal cells (ARPE-19).

Conclusion

The synergistic strategy of ABN to prepare salt-caged NM and fattigated HOC to form self-assembled nanoparticles offers a promising way to simultaneously improve the stability, permeability, and pH- and GSH-responsive controlled release of MEL for ocular chemotherapy.

Graphical Abstract

Introduction

Retinoblastoma (RB) is reported to be the most frequently occurring primary intraocular tumor in childhood, requiring early detection and appropriate treatment not only for saving life but also for preserving the globe and visual functions.¹ Among diverse therapeutic strategies available, local chemotherapy using intra-arterial or intravitreal injection of melphalan (MEL) has become a mainstay because it enables high intraocular drug concentrations while reducing systemic exposure by bypassing the blood-eye barrier and delivers directly to the posterior segment of the eye.^2,3 Intravitreal MEL has been reported to induce the regression of vitreous seeds with high rates of ocular salvage, with intra-arterial infusion of MEL emerging as the cornerstone of RB management.⁴ However, these procedures are invasive, require specialized techniques, and are associated with complications, such as vascular occlusion, vitreous hemorrhage, and endophthalmitis.

MEL is highly unstable in aqueous solutions because of its rapid hydrolysis and oxidative degradation, limiting its clinical application as an ophthalmic formulation for RB.^5,6 It is prone to rapid hydrolysis and therefore has an extremely short half-life in aqueous media. However, solid-state MEL exhibits excellent stability compared to its rapid degradation in aqueous solutions. Unopened powder is stable for 2 to 3 years when stored at room temperature and protected from light.^6,7 Ocular bioavailability after topical administration of free-diffusing systemic drugs is highly limited by pre-corneal clearance, tear circulation, and restricted corneal permeability by blood-retinal barrier. Therefore, researchers utilize several advanced pharmaceutical strategies to safely achieve therapeutic levels at the ocular fundus by overcoming blood-retinal barrier.^8,9 Therefore, developing noninvasive and patient-friendly formulations with enhanced MEL stability in aqueous environments that can effectively deliver MEL to ocular tumor tissues like RB cells is highly challenged.

Acid–base neutralization (ABN) is a simple yet effective strategy for enhancing the aqueous stability of weakly acidic or basic drugs such as MEL and rebamipide. For example, neutralization improves both the solubility and stability of rebamipide by forming salt-caged ion pairs and reducing the susceptibility to hydrolysis.¹⁰ Similarly, the formation of ionic salt-caged MEL aggregates between the protonated amine groups and counterions can decrease water accessibility and mitigate hydrolytic degradation. Furthermore, hydrophilic macromolecules, such as hyaluronic acid (HA), offer additional stabilization and functional benefits via electrostatic polymeric complexation with MEL. The negatively charged carboxyl groups in HA electrostatically interact with cationic drugs, forming drug–polymer complexes that retard the diffusion of water and reactive oxygen species. Moreover, HA is a well-known ligand for CD44 receptors, which are overexpressed in RB and several tumor cells, providing a basis for receptor-mediated tumor targeting.^11–13 Furthermore, polymeric-based nano-systems offer a promising solution for delivering ophthalmic drugs across blood-retinal barrier, making localized or targeted nanocarrier delivery systems essential for treating eye diseases.¹⁴

Fattigation that confers fatty acid conjugation by covalently bonding hydrophilic polymers, such as HA, peptides, and biomolecules, with different types of fatty acids have been widely utilized as promising pharmaceutical strategies.^15–17 Such fattigated polymers can self-assemble into nanoparticles owing to their amphiphilic nature, encapsulating model drugs to improve their physicochemical and biopharmaceutical properties and in vivo performances, such as pharmacokinetics and cancer targeting.^11,12,18,19 Recently, we developed an amphiphilic HA–CYS–OA conjugate (HOC) with an L-glutathione (GSH)-responsive disulfide-containing CYS linker to form self-assembled nanoparticles (HON), which are challenging to deliver MEL with therapeutic limitations as a novel delivery system.^10,12 A CYS linker is a widely used, redox-responsive chemical crosslinker formed by two cysteamine molecules linked by a disulfide bond (-S-S-). It is primarily designed to respond to GSH, which is the body’s primary intracellular antioxidant and reducing agent. It is well-known that GSH is highly concentrated in the cell’s cytosol and in tumor environments (typically 2–10 mM), compared to low extracellular spaces (2–20 μM).^12,20 RB cells also exhibit higher intracellular GSH levels several-fold higher than normal cells, enabling them to resist oxidative damage and apoptosis for rapid tumor growth.^21,22

In the presence of GSH, the disulfide bond of the CYS linker undergoes rapid reductive cleavage. The GSH molecule donates a hydrogen/electron, breaking the -S-S- bond and separating the linker into two thiol components. Furthermore, HA utilized as a backbone of fatty acid-conjugated polymer has an excellent ability to bind to the CD44 (acidic cell surface adhesion protein) receptor that is overexpressed in various types of cancer cells.^{8,11,12,20,23}

For these reasons, HON possesses a hydrophilic HA shell and a hydrophobic OA core with GSH redox-responsive disulfide bonds of CYS that can be cleaved in GSH-rich intracellular environments, such as tumors. HON can physically encapsulate MEL or NM via ABN within a hydrophobic core and provide enhanced stability, HA-mediated mucoadhesion, and CD44 targeting to improve accumulation at the ocular surface and RB-related cells by enabling drug release in acidic and GSH-responsive tumor microenvironments.^13,24 Thus, a combination of ABN-induced stabilization and HA-based fattigated nanoparticles with CYS linker as an improved tumor-sensitizing biomaterial is hypothesized to synergistically protect MEL from degradation and enhance anticancer activity against RB cells by controlling its permeability and release rate, thereby providing a promising topical chemotherapy platform for RB.^10,12,25,26

We investigated the improvement of aqueous-labile stability and RB anticancer activity of MEL by challenging the ABN and fattigation platforms. We synthesized an amphiphilic HOC conjugate and characterized it using instrumental analysis. Next, we prepared neutralized MEL (NM) using ABN and subsequently optimized it by varying the salt concentration and HA ratios. Next, NM was loaded into HA and HON and designated as NM–HA and NM–HON, respectively. Their physicochemical properties, aqueous and solid-state stabilities, and in vitro release behavior under different pH conditions with or without GSH enzyme were investigated. The permeability was also determined using a permeation barrier membrane (PB-M). Finally, the anticancer activity of formulations against normal retinal (ARPE-19) and RB (WERI-Rb-1) cells was evaluated.

Materials and Methods

Materials

Melphalan (base: 94 %) was purchased from Fisher Scientific (Waltham, USA). Sodium hyaluronate (MW = 10 kDa) was purchased from Lifecore Biomedical (Chaska, MN, USA). Cystamine dihydrochloride (CYS·2HCl, purity: 96 %), oleic acid (OA, purity: ≥ 98.5 %), L-glutathione reduced (GSH), MES hydrate (MES), Trizma base (TRIS), trypsin–EDTA (0.25 %) solution and phosphate-buffered saline (PBS) tablets were obtained from Sigma-Aldrich (St. Louis; MO, USA). Further, N-hydroxysuccinimide (NHS), 1-ethyl-3(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC·HCl) was supplied by Thermo Fisher Scientific (Seoul, Republic of Korea). Anhydrous sodium acetate was purchased from Daejung Chemicals (Siheung; Gyeonggi-do, Republic of Korea). Sodium hydroxide (NaOH beads, 98 %) and hydrochloric acid (HCl, 35–37 %) were purchased from Samchun Chemical (Seoul, Republic of Korea). Acetonitrile (ACN) and methanol (MeOH) for high-performance liquid chromatography (HPLC) were purchased from Honeywell Burdick and Jackson (Muskegon, MI, USA). All the chemicals and reagents were of analytical grade.

HOC Synthesis

A two-step synthetic procedure was used to prepare the HOC conjugate. The HA–CYS conjugate was synthesized using a previously reported method.^11,12 First step, HA (2 g) was dissolved in 200 mL of 0.1 M MES buffer (pH 4.5) at 25 °C. Next, EDC·HCl (4.8 g) and NHS (2.9 g) were added to the HA solution, and the mixture was stirred at 500 rpm for 1 h to activate the carboxyl groups of HA. The pH was subsequently adjusted to 7.0 using 0.1 M PBS solution (pH 8.5). CYS·2HCl (5.6 g) was added to the HA solution under continuous stirring at 500 rpm, and the pH was adjusted to 7.0 using 5N NaOH. After 6 h of the reaction at 25 °C, the pH was raised to 8.6, using a 1 M TRIS solution to stop the reaction. Next, the solution was transferred into a dialysis tube (molecular weight cut-off of 3.5 kDa) and dialyzed successively against 0.1 M NaCl solution, 25 % ethanol (EtOH) solution, and deionized water (DW) each for 24 h to retain the HA–CYS conjugate and remove low molecular weights (< 3.5 kDa). The dialyzed solution was filtered through a 0.45 μm polyvinylidene fluoride (PVDF) membrane (Durapore^®; Merck Millipore Ltd., Carrigtwohill, Ireland) and lyophilized at −80 °C under a vacuum of 5 mTorr (Freeze Dryer FD8508; IlShinBioBase Co. Ltd., Gyeonggi, Republic of Korea) for 48 h to obtain the HA–CYS conjugate.

For the second step, OA (200 μL), EDC·HCl (188 mg), and NHS (110 mg) were dissolved in 20 mL of DMF and stirred for 1 h at 25 °C to activate the carboxyl groups of OA. Separately, HA–CYS (500 mg) was dissolved in 30 mL of 0.01 M PBS (pH 7.4). After 1 h, the OA solution was slowly added to the HA–CYS solution under stirring at 50 °C. The pH of the reaction solution was adjusted to 8.0 using 5 N NaOH, and the reaction was maintained for 6 h. Acetone was added to induce precipitation, and the precipitates were collected by filtration and washed several times with warm methanol. Next, the precipitates were dispersed in DW and dialyzed against 50 % EtOH, 0.1 M NaCl, and DW (each for 24 h). The final sample was freeze-dried at −80 C under a vacuum of 5 mTorr for 48 h to obtain the HOC conjugate. A scheme for synthetic procedure for HA-CYS and HOC is drawn in Figure S1.

Preparation of Salt-Caged NM

Effect of Salt Concentrations

The compositions of the formulations used to prepare NM, NM–HA, and NM–HON using ABN are listed in Table 1. First, the salt concentration was optimized. Briefly, MEL (50 mg) was dissolved in different concentrations (0.1, 0.2, and 0.3 M) of 5 mL of HCl to obtain a transparent solution. HA (50 mg) was dispersed in 5 mL of NaOH solution (0.1, 0.2, or 0.3 M). The acidic phase was rapidly poured into the basic phase with stirring (500 rpm) at 25 °C to form a colloidal suspension. Finally, the suspension was frozen at − 80 °C to obtain a neutralized drug powder.

Table 1.

Formulation Compositions for the Preparation of NM, NM-HA, and NM-HON by ABN

Code	Acidic Phase		Basic Phase
	MEL (mg)	5 mL of HCl (M)	HA (mg)	HOC (mg)	5 mL of NaOH (M)
MEL	50	–	–	–	–
NM	50	0.1	–	–	0.1
NM1-50	50	0.1	50	–	0.1
NM2-50	50	0.2	50	–	0.2
NM3-50	50	0.3	50	–	0.3
NM1-100 (NM-HA)	50	0.1	100	–	0.1
NM1-150	50	0.1	150	–	0.1
NM-HON	50	0.1		100	0.1

Effect of HA Concentrations

The salt concentration was first optimized, and comparisons were made by varying the proportion of HA. Briefly, MEL (50 mg) was dissolved in 5 mL of 0.1 M HCl solution to obtain a transparent solution. In addition, HA (0, 50, 100, and 150 mg) was dispersed in 5 mL of 0.1 M NaOH solution. Next, the acidic phase was rapidly added to the basic phase with stirring (500 rpm) at 25 °C to obtain a colloidal suspension. Finally, the suspension was frozen at −80 °C to obtain a neutralized drug powder.

Preparation of NM-Loaded Formulations

NM were loaded into HA or HON. Briefly, MEL (50 mg) was dissolved in 5 mL of 0.1 M HCl solution to obtain a transparent solution. HA or HOC (100 mg) was dispersed in 5 mL of 0.1 M NaOH solution at an optimal ratio of 1:2. The acidic phase was then rapidly added to the basic phase with stirring (500 rpm) at 25 °C to obtain a colloidal suspension. The nanosuspensions thus obtained were freeze-dried at −80 °C to collect NM, NM–HA, and NM–HON in a powder form.

Encapsulation Efficiency (EE) and Drug Loading (DL) of NM–HON

The amount of MEL loaded onto NM–HON was quantified by HPLC. The predetermined amount of NM–HON was dispersed in DW, followed by adding MeOH (1:1, v/v), sonicated for 5 min, and filtered using a 0.45 μm PVDF membrane (Hyundai micro; Seoul, Republic of Korea) to collect the extracted MEL solution. The MEL solution was diluted 10-fold with acetonitrile (ACN), and the MEL concentration was measured using HPLC (Waters e2695; Waters Corp., Milford, MA, USA). All experiments were performed in triplicate.

The EE and DL of NM–HON were calculated using the following equation:

Physicochemical Characterization of NM, NM-HA, and NM–HON

Fourier Transform-Infrared Spectroscopy (FT–IR)

The molecular interactions between MEL, HA, and HOC were confirmed using FT-IR spectroscopy (Nicolet iS50; Thermo Fisher, Waltham, MA, USA). Binding and interactions were verified through peak analysis of the MEL, HA, and HOC powders. All samples were measured in the ATR mode with a resolution of 2 cm⁻¹ at a wavelength range of 500 to 4000 cm⁻¹.

Field Emission Scanning Electron Microscopy (FE–SEM)

The morphology of each formulation and raw material was measured using FE–SEM (JSM-9700F; JEOL, Tokyo, Japan). All samples were diluted in DW to a concentration of 1 mg/mL. Each sample was dropped onto a copper grid with carbon tape, followed by vacuum drying and platinum coating for 60s before observation at 5.0 kV and 25 °C.

Field Emission Transmission Electron Microscope (FE–TEM)

The nanoparticle morphology of NM–HON and blank (HON) in the solution was analyzed by FE–TEM (Tecnai G2 F30 S-Twin; FEI, Netherlands). All samples were diluted in DW to a concentration of 1 mg/mL. Each sample was dropped onto a 200-mesh copper grid, stained with 2 % PTA solution (pH 7.2), and allowed to dry overnight before the analysis.

Differential Scanning Calorimetry (DSC)

DSC (DSC 200 F3 Maia, Netzsch, Selb, Germany) was used to measure the thermal behaviors of MEL, NM, NM–HA, and NM–HON. Thermograms of the free MEL, HA, HOC, NM, NM–HA, and NM–HON of each formulation were obtained using a differential scanning calorimeter by scanning from 25 to 200 °C at 2 °C/min. Nitrogen was used as the purge gas, and each sample (approximately 5 mg) was placed on a standard open aluminum pan with reference to an empty pan.

High Power X-Ray Diffraction (HP–XRD)

The crystallinity of the MEL formulations in different polymers (HA and HOC) compared to free MEL was investigated using a HP-XRD (D/max-2500V/PC; Rigaku, Japan). The measurements were performed at 40 kV and 150 mA using Cu–Kα radiation. All samples were analyzed at two angles, from 5 degrees to 60 degrees, with a step size of 0.02 degrees.

Dynamic Light Scattering (DLS)

The particle size, polydispersity index (PDI), and zeta potential of blank HONs and NM–HONs were analyzed using a DLS (ELSZ-2000 S; Otsuka Electronics, Osaka, Japan) at a scattering angle of 90 degrees. Different lyophilized samples (equivalent to 0.05% MEL, w/v) were reconstituted in artificial tear fluid (ATF, pH 7.2) composed of 6.7 g/L sodium chloride, 2.0 g/L sodium bicarbonate, and 0.08 g/L calcium chloride dihydrate in DW and adjusted to pH 7.4 and subsequently vortexed for further analysis.⁷ Each formulation was diluted 10 times with DW to reduce the influence of salt in the ATF on the comparison of zeta potentials among samples. The measurements were performed in triplicate at 25 °C.

High Performance Liquid Chromatography (HPLC)

The quantity of MEL was evaluated using HPLC (Waters Corp., Milford, MA, USA) with a reverse-phase column (Hypersil GOLD™, 250 mm × 4.6 mm, 5 µ, C₁₈). The mobile phase consisted of methanol, deionized water, and acetic acid in a ratio of 49:49:2 (v/v/v) at a flow rate of 1 mL/min. The analytical column was controlled at 25 ± 0.5 °C, and the UV/Vis detector was set at 260 nm. The loading volume of each sample was 20 µL. The mobile phase was filtered through a 0.45 µm PVDF membrane filter and degassed before use.

Stability Study

For liquid-state stability, the freeze-dried powders were dissolved and exposed to 10 mL of PBS (pH 7.4), equivalent to 0.2 mg/mL MEL concentration. At the designated time points, 700 µL of each solution was collected using a 1 mL syringe. An NM solution prepared without the polymer was used as a control. The collected samples were filtered through a 0.45 µm polytetrafluoroethylene (PTFE) syringe filter and subsequently diluted 10 times with the mobile phase before HPLC.

We next assessed solid-state stability, for which the powdered samples were stored at 25 °C and 70% relative humidity (RH). The samples were drawn at 7, 14, 21, and 28 d and dissolved in 1 mL of mobile phase, equivalent to 5 mg/mL MEL concentration. After complete dissolution, the solution was diluted 10 times with the mobile phase. Next, solutions were filtered using a 0.45 μm PVDF syringe filter. The filtered solutions were analyzed thrice for each formulation using HPLC (Waters e2695; Milford, MA, USA) (n = 3).

In Vitro Permeability Study

Each formulation (NM, NM-HA, and NM-HON) and free MEL were analyzed using a DHC-6TD permeability testing system with a PB-M (Logan Instruments, Somerset, NJ, USA). PB-M is a synthetic cellulosic membrane providing acidity and permeability similar to human epidermis to evaluate the in vitro release and permeation of topical and transdermal drugs and provides highly efficient detection of the passive mass transfer (mass transfer screening) of drugs regardless of the polarity of the substance.^19,26 The pose size of PB-M is not applicable because it is made non-porous.

Briefly, each formulation and free MEL were prepared with the same amount of MEL (0.05 %) in 10 mL of ATF (pH 7.2). The receptor chamber contained 20 % EtOH (5 mL) in PBS buffer (pH 7.4) to maintain the sink conditions and was stirred at 600 rpm. A volume of 300 µL from each sample was placed on the surface of the PB-M membrane (25 mm diameter, surface area: 38 mm²). The entire receptor chamber was collected at predetermined time points (0.5, 1, 2, 4, and 6 h) and replaced with fresh medium. All collected samples were filtered before HPLC analysis. Each formulation was tested in triplicate (n = 3).

In Vitro Release Study

We performed dialysis using Slide-A-Lyzer MINI dialysis devices (15 mL conical tubes, 10 kDa MWCO inserts, Thermo Scientific, Rockford, IL, USA) to assess the release behavior of MEL from NM, NM–HA, and NM–HON compared to free MEL. Drug-release experiments were conducted in three solutions: PBS (pH 7.4), 10 mM sodium acetate buffer (ABS, pH 5.2), and PBS (pH 7.4) with 10 mM GSH to confirm the characteristics of HON. Predetermined amounts of lyophilized NM, NM–HA, and NM–HON (equivalent 0.5 % MEL) were suspended in artificial tear fluid (ATF, pH 7.4). Next, 1 mL of each sample was loaded into the donor compartment and placed in a conical tube that acted as the receiver compartment and filled with 13.5 mL of PBS (pH 7.4), ABS (pH 5.2), or PBS (pH 7.4) with 10 mM GSH. The dialysis devices were incubated in a water bath shaker at 37 °C and 100 rpm. Afterward, 1 mL of each sample was withdrawn at different time points (0.5, 1, 2, 4, 6, and 8 h) and replaced with an equal volume of fresh medium in a receiver chamber. The sample was diluted 10 times with the mobile phase before HPLC analysis. Each formulation was tested in triplicate (n = 3).

Cell Culture and Maintenance

Human RB cancer cell line (WERI-Rb-1) and human normal retinal pigment epithelial cell line (ARPE-19) were obtained from ATCC (Manassas, VA, USA). Roswell Park Memorial Institute (RPMI) 1640 medium, Dulbecco’s modified Eagle’s medium F12 (DMEM/F12) media, fetal bovine serum (FBS), and 1 % penicillin/streptomycin (PS) were supplied by Gibco (Grand Island, NY, USA). RB cells (WERI-Rb-1) were grown in RPMI 1640 medium containing 10 % FBS and 1 % PS. ARPE-19 cells were grown in DMEM/F12 media containing 10 % FBS and 1 % PS.²⁷ The cell counting kit (CCK-8) was obtained from Dojindo (Kumamoto, Japan). The cells were maintained in a humidified incubator at 37 °C with 5 % CO₂.

In Vitro Cytotoxicity Study

WERI-Rb-1 cells were cultured in the RPMI 1640 medium containing 10 % FBS and 1 % PS. ARPE-19 cells were cultured in DMEM/F12 medium containing 10 % FBS and 1 % PS. ARPE-19 cells were seeded at a density of 5×10³ cells/well in a 96-well plate and cultured for 24 h to allow for attachment. WERI-Rb-1 cells were seeded at a density of 1×10⁴ cells/well in a 96-well plate coated with poly-D-lysine (1 mg/mL) and cultured for 24 h.²⁸ After 24 h incubation, WERI-Rb-1 medium was centrifuged at 125 × g for 5 min to ensure effective cell attachment. Cells were washed with serum-free media and then treated with 100 μL of the sample containing different concentrations of each formulation (3.125, 6.25, 12.5, 25–50,100 μg/mL) via serial dilution, and 100 μL of fresh medium were added to the culture medium. After 24 h, 20 μL of the WST-1 (EZ-Cytox) solution was added to the culture medium and incubated for 1 h in an incubator. Finally, the absorbance was measured at 450 nm using a microplate reader (ELISA; Synergy H1, Biotek^TM, Santa Clara, CA, USA).

Statistical Analysis

Statistical analysis was performed using analysis of variance (ANOVA) in SigmaPlot software (version 14.0; Systat Software, San Jose, GA, USA). A two-tailed unpaired Student’s t-test and one-way ANOVA were run to compare differences between two groups and among multiple groups, respectively. The statistical analysis was considered to be significant if the p value is <0.05 (ns = no significance, *p < 0.05, **p < 0.01, ***p < 0.001. ****p<0.0001). The data are expressed as the mean ± standard deviation.

Results and Discussion

Synthesis of HOC

Amphiphilic HOC conjugates were prepared using a two-step synthetic process. CYS, a pH- and glutathione (GSH)-responsive linker, was used to couple the carboxyl groups of HA and OA via amide bond formation as shown in Figure S1. The degree of substitution (DS) was varied by the amount of CYS (13.5~52.4 %) and OA (4.6~12.2 %) per HA. The low DS levels of HA-CYS and HA-CYS-OA (HOC) was used for this work and quantified as 13.5 % and 4.6 %, respectively. The FT-IR spectra of HA, CYS, OA, HA-CYS, and HOC are compared in Figure 1. The FT-IR spectrum of HA–CYS revealed that the carbonyl band of HA at 1609 cm⁻¹ and the amine band of CYS at 1593 cm⁻¹ shifted to 1607 and 1557 cm⁻¹, respectively, indicating the formation of an amide (–NH–CO–) linkage between HA and CYS.^11,29 The HOC spectrum exhibited asymmetric -CH₂ and symmetric -CH₂ peaks at 2922 cm⁻¹ and 2853 cm⁻¹, corresponding to the alkyl chain of OA after combining HA–CYS with OA. In addition, the bands at 1604 and 1556 cm⁻¹ demonstrated increased intensity, further supporting additional amide bond formation between the carboxyl groups of OA¹⁷ and the remaining amino groups of CYS. Collectively, these findings confirm that HA and OA were successfully conjugated via the CYS linker.

Figure 1.

Fourier-transform infrared (FT-IR) spectra of hyaluronic acid (HA), cystamine (CYS), oleic acid (OA), HA-CYS, and HA-CYS-OA conjugate (HOC).

Optimization of MEL Stability

Effect of Acid/Base Neutralizing Concentration

The stability of MEL was optimized by varying the salt concentration and HA ratio. Effect of salt concentrations on MEL stability in pH 7.4 PBS and their related FT-IR spectra are shown in Figure 2. Time-dependent degradation of MEL was monitored for 24 h, For salt screening, the HA was fixed at 50 mg, and NM1-50, NM2-50, and NM3-50 differed only in the HCl and NaOH molarity (0.1, 0.2, and 0.3 M, respectively. MEL stability was strongly influenced by the ionic strength of the formulation (Figure 2a). After 2 h, the residual MEL content decreased in the order NM1-50 (47.86 ± 0.43 %) > NM2-50 (35.57 ± 0.32 %) > NM3-50 (29.38 ± 0.38 %). After 24 h, NM1-50 still retained approximately 30 % of its initial MEL, whereas NM2-50 and NM3-50 had degraded by approximately 3 % and 2 %, respectively. These results indicate that an appropriate ionic environment is required to maintain the ion-pair complex between protonated MEL (–NH₃⁺) and HA carboxylates (–COO⁻), together with ancillary hydrogen bonding, thereby forming a local “salt cage” that restricts water access and mitigates hydrolytic degradation. This interpretation was supported by FT–IR analysis in NM1-50 (0.1 M NaCl), the HA carbonyl band around 1609 cm⁻¹ was broadened and the MEL amine band at 1581 cm⁻¹ was markedly attenuated, indicating strong ionic and hydrogen-bonding interactions between MEL and HA (Figure 2b). In contrast, NM2-50 and NM3-50 demonstrated a shift in the main band to 1611–1612 cm⁻¹ with partial recovery of the amine region, suggesting weakened MEL–HA complexation because of ionic screening at higher salt concentrations. Altogether, the kinetic and spectroscopic results demonstrate that excessive salt loading compromises MEL stability, whereas 0.1 M NaCl provides an optimal balance between neutralization and ionic shielding. Therefore, we selected it as the optimal ionic strength for subsequent studies.

Figure 2.

Effect of acid/base neutralizing concentrations on MEL stability in pH 7.4 PBS (a) and their related FT-IR spectra (b).

Effect of HA Amounts

Optimized ionic strength (0.1 M NaCl) of HCl and NaOH molarity for HA ratio screening were kept while NM, NM1-50, NM1-100, and NM1-150 differed only in the HA amount (0, 50, 100, and 150 mg, respectively). Figure 3 gives the effect of HA amounts on MEL stability in pH 7.4 PBS at 25 °C and their related FT-IR spectra. Although MEL concentration decreased over time in all formulations, the degradation rate was markedly dependent on the HA ratio (Figure 3a). At 2 h, the remaining MEL content decreased in the order NM1-100 (84.53 ± 0.92 %) > NM (67.08 ± 0.88 %) > NM1-150 (53.21 ± 1.07 %) > NM1-50 (47.86 ± 0.43 %). Among these, NM1-100 exhibited the greatest stabilizing effect, retaining approximately 85 % of its initial MEL concentration at 2 h and approximately 37 % at 24 h, whereas NM, NM1-50, and NM1-150 decreased to below 30% after 24 h. Thus, an optimal HA concentration is required for MEL stabilization. At a low HA ratio (1:1), the HA matrix is likely insufficient to engage all MEL molecules via electrostatic and hydrogen-bonding interactions, leaving a fraction of the drug more exposed to the aqueous phase and susceptible to hydrolysis. However, an excessive HA ratio (1:3) may induce ion shielding or a localized increase in viscosity because of surplus free –COO⁻ groups, thereby limiting efficient drug–polymer complex formation. The FT-IR spectra support this interpretation (Figure 3b). Free MEL demonstrated characteristic bands at 1581 cm⁻¹ (N–H bending) and 1517 cm⁻¹ (aromatic C=C), and HA exhibited a carbonyl band at 1609 cm⁻¹. The carbonyl region in NM1-100 shifted to 1612 cm⁻¹ with broad O–H/N–H stretching and attenuation of the MEL amine band, indicating strong ionic and hydrogen-bonding interactions between MEL and HA and the formation of a compact MEL–HA network. Similar shifts were observed for NM1-50 and NM1-150, but with less pronounced broadening and better separation of MEL-related bands, suggesting weaker or partially incomplete complexation at these ratios. Altogether, these kinetic and spectroscopic findings confirm that the HA ratio of 1:2 provides the most efficient MEL–HA complexation and protective microenvironment that suppresses MEL hydrolysis. Based on this screening, NM1-100 was selected and hereafter denoted as NM-HA; NM, NM-HA, and NM-HON were also used as representative formulations for subsequent comparisons.

Figure 3.

Effect of HA amounts on MEL stability in pH 7.4 PBS at 25 °C (a) and their related FT-IR spectra (b).

Stability of Optimized NM Formulations

Stability of NM, NM-HA, and NM-HON

The chemical stability of MEL, neutralized MEL (NM) and NM-loaded formulations (NM-HA, and NM-HON) in pH 7.4 PBS for 24 h is compared in Figure 4. The optimized MEL formulations were established under ABN conditions (0.1 M NaCl, HA:MEL = 1:2). The remaining MEL concentration gradually decreased in all formulations; however, the degradation rate differed markedly depending on the ABN conditions. After 2 h, NM-HA and NM-HON retained more than 80 % of the initial MEL (84.53 ± 0.92 % and 81.31 ± 0.50 %, respectively), whereas NM and free MEL decreased to 67.08 ± 0.88 % and 55.79 ± 1.91 %, respectively. At 24 h, the residual MEL content followed the order NM-HON (46.75 ± 0.24 %) > NM-HA (37.59 ± 0.64 %) > NM (21.33 ± 0.22 %) > MEL (7.88 ± 0.07 %). Notably, NM-HON and NM-HA retained approximately 50 % and 40 % of the initial MEL, respectively, after 24 h, whereas free MEL underwent rapid degradation, with less than 10 % remaining after 8 h. These findings indicate that the combination of ABN and OA-based nanoparticle assembly markedly enhanced the MEL stability, with an additional contribution from polymer complexation. Neutralization promoted the formation of ion-paired aggregates between protonated MEL (–NH₃⁺) and oppositely charged counterions, thereby decreasing water accessibility and delaying hydrolytic degradation. The neutralized MEL in NM-HA was further stabilized through electrostatic and hydrogen-bonding interactions between the MEL amino groups and HA carboxylate groups, creating a viscous polymeric microenvironment that hindered the diffusion of water and other reactive species. The amphiphilic HON in NM-HON underwent spontaneous self-assembly into nanoparticles with a hydrophobic OA-rich core and a hydrophilic HA shell, providing dual protection. The HA shell reduced water ingress at the particle surface, whereas the OA core limited direct contact between the MEL and the aqueous phase. Consequently, the self-assembled NM-HON system effectively stabilized the MEL by synergistically combining HA-mediated ion complexation with hydrophobic encapsulation within the OA domains. Although NM-HON and NM-HA showed the similar stabilization effect, NM-HON is more challenging for pH- or GSH-responsive release behaviors that are more favorable for tumor targeting hereafter.

Figure 4.

Comparative stability of MEL, neutralized MEL (NM) and NM-loaded formulations (NM-HA, and NM-HON) in pH 7.4 PBS (n = 3). **p < 0.01, ***p < 0.001. ****p<0.0001 One-way ANOVA compared with MEL.

Solid-State Stability

The short-term solid-state stability of freeze-dried MEL, NM, NM-HA, and NM-HON powders was assessed over 4 weeks at 25 °C and 70 % relative humidity (RH) (Figure 5). The MEL contents of MEL, NM, NM-HA, and NM-HON at the beginning of storage were 98.00 ± 1.34 %, 99.41 ± 3.19 %, 100.85 ± 2.90 %, and 100.76 ± 1.13 %, respectively, which were all comparable to the theoretical content in a short-term storage condition. The MEL content in free MEL and NM gradually declined as storage time increased, reaching 81.78 ± 0.29 % and 83.80 ± 2.74 %, respectively, in week 4. In contrast, NM-HA maintained a relatively higher MEL content of 88.44 ± 3.01 % over the same period. Notably, NM-HON preserved more than 95% of the initial MEL content throughout the 4-week study; the value at week 4 (98.56 ± 1.32 %) was nearly identical to the initial level, indicating the highest solid-state stability among the test formulations.

Figure 5.

Solid-state stability of freeze-dried NM, NM-HA, and NM-HON at 25 °C and 70 % RH (n = 3). *p<0.05, **p < 0.01, ****p<0.0001 One-way ANOVA compared with MEL.

Physicochemical Properties of HON and NM-HON

The physicochemical properties of HON and NM-HON before and after lyophilization are listed in Table 2. Blank HON nanoparticles exhibited an average particle size of 249.57 ± 6.95 nm and a zeta potential of −31.74 ± 0.38 mV, confirming that the amphiphilic HOC spontaneously self-assembles into a stable nanostructure. The NM-HON nanoparticles displayed an average particle size of 307.60 ± 2.46 nm and a similar polydispersity index (PDI) of 0.28 ± 0.04 following MEL encapsulation, indicating successful drug loading without significant aggregation. Nanoparticles (NPs) with size range from 10 nm to 1000 nm improved topical passage of large, poorly water-soluble molecules through the biological barriers of ocular system although larger particles up to 1000 nm aid in tissue adherence but can cause eye irritation. However, optimal nanoparticle size for ocular delivery ranging from 10 to 200 nm enhances ocular bioavailability, bypasses the blood-retinal barrier, and improves sustained drug release compared to traditional drops.^30,31 The current particle size was slightly larger but tolerable than the optimal size ranges due to the high molecular weight of HOC to form self-assembled HON.

Table 2.

Physicochemical Properties of HON and NM-HON Before and After Lyophilization (n = 3)

Nanoparticle	Particle Size (nm)	Polydispersity Index (PDI)	Zeta Potential (mV)	EE (%)	DL (%)
HON	249.57 ± 6.95	0.24 ± 0.04	−31.74 ± 0.38	–	–
NM-HON (before freeze-drying)	290.40 ± 5.48	0.24 ± 0.03	−11.15 ± 0.74	94.36 ± 0.84	–
NM-HON (after freeze-drying)	307.60 ± 2.46	0.28 ± 0.04	−28.52 ± 2.76	93.43 ± 1.97	25.95 ± 0.55

Abbreviations: EE, encapsulation efficiency; DL, drug loading.

The zeta potential of NM-HON was −28.52 ± 2.76 mV, demonstrating sufficient negative charge to maintain colloidal stability via electrostatic repulsion between HA carboxyl groups. The EE and DL of NM-HON were 93.43 ± 1.97 % and 25.95 ± 0.55 %, respectively, highlighting the strong affinity of MEL for the HON nanoparticle system. This high efficiency can be attributed to non-covalent interactions, including the electrostatic attraction between the protonated amine group of MEL and the negatively charged HA backbone, hydrophobic interactions of the aromatic and alkyl groups within the OA region, and stabilization of disulfide bonds by the CYS linker.^11,32,33 These interactions contributed to efficient drug encapsulation within the hydrophobic core, limiting drug diffusion into the aqueous phase during nanoparticle formation.

The morphological characteristics of NM-HON were examined using FE-TEM (Figure 6). NM-HON displayed well-defined spherical nanoparticulate structures that were uniformly dispersed, which is consistent with the hydrodynamic size distribution obtained from the DLS analysis. The small particulates observed on the surface of the NM-HON were presumed to be NaCl crystals generated during ABN according to our previous investigation using powder x-ray diffraction.¹⁰ In contrast, the control group HON nanoparticles exhibited slightly smaller particle sizes than NM-HON while maintaining a similar nanoparticle morphology, confirming that MEL encapsulation increased the particle size. Overall, HON successfully self-assembled into a stable nanocarrier capable of efficiently encapsulating MEL, with a combination of electrostatic, hydrophobic, and disulfide bond interactions, contributing to its structural stability and high drug-loading capacity.

Figure 6.

FE-TEM images of HON (a) and NM-HON (b).

Physicochemical Characterization of NM, NM-HA, and NM-HON

FE-SEM

The morphological analysis using FE-SEM of MEL, HA, HOC, NM and NM-loaded formulations (NM-HA, and NM-HON) is shown in Figure 7, revealing clear differences in the solid-state morphologies of the formulations. Free MEL (Figure 7a) appeared as large, dense aggregates composed of tightly packed plate-like crystals, whereas HA (Figure 7b) formed a continuous matrix-like structure, and HOC (Figure 7c) exhibited a nanoparticulate surface morphology. Following ABN, NM (Figure 7d) demonstrated a more dispersed crystalline arrangement, whereas NM-HA (Figure 7e) presented a matrix-like structure in which MEL was finely dispersed and adhered within the HA network, indicating the disruption of the native MEL crystal lattice and polymer-assisted dispersion. In contrast, NM-HON (Figure 7f) exhibits a nanoparticle structure with a rough surface, which is consistent with the formation of HON-based nanoparticle structures after freeze-drying.

Figure 7.

FE-SEM images of MEL (a), HA (b), HOC (c), NM (d), NM-HA (e), and NM-HON (f).

FT-IR Analysis

Instrumental analysis of MEL, HA, HOC, NM, NM-HA, and NM-HON were recorded to elucidate the molecular interactions between MEL and HA. In FT-IR spectra (Figure 8a), free MEL displayed characteristic bands at 1581 cm⁻¹ (–NH bending of primary amines) and 1517 cm⁻¹ (aromatic C=C stretching), whereas HA exhibited prominent peaks at 3285 cm⁻¹ (O–H stretching) and 1609 cm⁻¹ (carboxyl C=O-related stretching). HOC displayed a similar band near 1609 cm⁻¹ and an additional peak at 2922 cm⁻¹ corresponding to the C–H stretching of the OA chains, indicating successful hydrophobic conjugation. New bands appeared at 3288 cm⁻¹ (broad O–H/N–H stretching) and 1556 cm⁻¹ (amide-related vibration) in the NM formulation, and the MEL amine band at 1581 cm⁻¹ shifted slightly to a lower wavenumber with decreased intensity, suggesting the formation of an ion-pair microenvironment between protonated MEL and the counterion. The HA carbonyl-related band at 1609 cm⁻¹ shifted to 1611 cm⁻¹ in the NM-HA spectrum, whereas the MEL aromatic C=C band appeared at 1518 cm⁻¹. The MEL amine band was no longer resolved as a separate peak and appeared to be superimposed on the HA band, yielding a single band at 1611 cm⁻¹, indicative of hydrogen bonding and ionic interactions between MEL amino groups and HA carboxyl groups. The NM-HON spectrum displayed a further slight shift of this band to 1612 cm⁻¹ and a broad O–H/N–H stretching band at 3292 cm⁻¹, suggesting reinforced hydrogen bonding and amide-type interactions in the nanoparticle system. The C–H stretching peak of the OA chains remained around 2920–2922 cm⁻¹, confirming retention of the hydrophobic OA segments within the HON nanoparticles. Overall, these spectral features support the incorporation of MEL into the hydrophobic core of NM-HON and its stabilization via electrostatic and hydrogen bonding interactions with the HA–CYS shell, thereby contributing to the enhanced stability of MEL in NM-HA and NM-HON.

Figure 8.

Fourier transform-infrared (FT-IR) spectra (a), differential scanning calorimetry (DSC) thermograms (b), and high power X- ray diffraction (HP-XRD) patterns (c) of MEL, NM, NM-HA, NM-HON, and raw materials.

DSC Analysis

The solid-state characteristics of the formulations were evaluated by DSC, and the thermograms are shown in Figure 8b. Free MEL exhibited a sharp endothermic peak at 70 °C, corresponding to its melting point, and a distinct exothermic event at 169.8 °C, indicative of a highly crystalline structure with a dense molecular packing. In contrast, the NM formulation exhibited a shift in the melting endotherm to a lower temperature and a decrease in the intensity of the exothermic peak, which also shifted to 160.4 °C. These changes suggest that the formation of the salt cage network disrupted the native MEL lattice, resulting in a more dispersed solid-state stabilized by ionic and hydrogen-bonding interactions. The melting endotherm in NM-HA further decreased to 51.1 °C, with a broadened peak and reduced intensity of the degradation-related exothermic peak around 169.9 °C. This behavior is consistent with molecular or microcrystalline dispersion of MEL within the HA matrix, driven by electrostatic interactions between the protonated amino groups of MEL and the carboxylate (–COO⁻) groups of HA, and reflects progression toward an amorphous-like state.³⁴ Among the formulations, NM-HON displayed the lowest and broadest melting transition (56.9 °C) and a markedly attenuated exothermic peak at 167.4 °C, indicating the greatest disruption of MEL crystallinity. These findings support that MEL is efficiently encapsulated within HON nanoparticles, limiting drug–drug contacts and hindering recrystallization. Overall, the pronounced reduction in the crystallinity of NM-HA and NM-HON suggests enhanced amorphization and nanodispersion of MEL, which are expected to contribute to improved thermal and chemical stability in the solid state.

HP-XRD Analysis

HP-XRD patterns of MEL, HA, HOC, NM, NM-HA, and NM-HON were recorded to investigate the crystalline state of MEL in each formulation (Figure 8c). Free MEL exhibited multiple sharp diffraction peaks in the 10–35 degrees range, confirming its highly crystalline nature. In contrast, the characteristic crystalline reflections of MEL were absent in NM-HA and NM-HON, indicating that MEL underwent amorphization because of interactions with HA or HON during ABN. In NM-HA, this halo-like pattern suggests that MEL was molecularly dispersed within the HA matrix via electrostatic interactions and hydrogen-bonding, which is consistent with the DSC data demonstrating the disappearance of melting endotherms. Furthermore, the HP-XRD pattern of NM-HON indicated that MEL was encapsulated within the hydrophobic HON nanoparticles and existed in a uniform amorphous state. The loss of crystalline peaks supported the formation of strong intermolecular interactions between MEL and the HON conjugate, confirming successful nanoparticle encapsulation. In addition, the distinct peaks observed at approximately 30 degrees and 45 degrees in NM, NM-HA, and NM-HON can be attributed to the crystalline NaCl formed during ABN.^10,35 The presence of these salt reflections suggested that NaCl was at least partially adsorbed onto the negatively charged HA or HON surfaces. This interpretation agrees with the FE-TEM observations, in which particulate deposits on the NM-HON surface were consistent with surface-associated salts, forming an ionic layer that may reduce water accessibility to the encapsulated MEL core, thereby contributing to the overall stabilization of the formulation.

In Vitro Permeability Study

The cumulative permeability profiles of MEL, NM, NM-HA, and NM-HON using PB-M in pH 7.4 PBS as a receptor medium are shown in Figure 9. PB-M membrane, a synthetic cellulosic membrane was used in a preliminary study although PB-M is not directly related with ocular tissue such as cornea, sclera, or conjunctiva. The 20 % EtOH in receptor phase is relatively high concentration but was used to maintain the sink condition of poorly water-soluble MEL. This 20 % EtOH does not influence the PB-M membranes and the solubility of MEL-loaded nanoparticles in donor phase. Both the NM-HA and NM-HON demonstrated significantly enhanced permeability compared to free MEL at all time-points. The cumulative permeation of free MEL was 43.76 ± 0.67 %, NM was 59.83 ± 0.66 % after 6 h, whereas NM-HA and NM-HON exhibited 82.00 ± 0.32 % and 78.93 ± 1.68 %, respectively. Free MEL exhibited the lowest permeability rate owing to its limited solubility in aqueous solutions and instability at physiological pH. In contrast, the NM formulation demonstrated improved permeability compared to free MEL, which was attributed to the partial ionization and enhanced solubility of MEL during neutralization. The permeability of the formulations incorporating polymers (HA or HOC) was more than twice that of free MEL, with no significant difference observed between NM-HA and NM-HON. Overall, the loading of MEL onto HON nanoparticles or the interaction between MEL and HA (via ionic or hydrogen bonding) significantly enhanced drug permeability.

Figure 9.

In vitro permeability profiles of MEL, NM, NM-HA, and NM-HON using permeation barrier membrane (PB-M) in pH 7.4 PBS (n = 3). **p < 0.01, ***p < 0.001. ****p<0.0001 One-Way ANOVA compared with MEL.

In Vitro Release Study

The dialysis membrane method is the most common and widely utilized traditional method as compared with other methods such as Sample and Separate Method and Continuous Flow/Flow-Through Cell although the physical separation is difficult due to the minimal size difference between the nanoparticles and the released free drug.^36,37 The release profiles of MEL from each formulation were evaluated under different physiological conditions that simulated normal and tumor-like environments: PBS (pH 7.4), ABS (pH 5.2), and PBS (pH 7.4) with 10 mM GSH (Figure 10). The cumulative release of MEL over 8 h under PBS (pH 7.4) conditions were as follows: MEL (73.50 ± 3.95%) > NM (61.91 ± 0.60 %) > NM-HA (51.40 ± 0.17 %) > NM-HON (32.77 ± 0.25 %) (Figure 10a). Free MEL exhibited the highest release rate, whereas NM demonstrated enhanced solubility in its neutral form. NM-HA demonstrated an intermediate diffusion rate owing to ionic and hydrogen bonding interactions with HA. In contrast, NM-HON exhibited the slowest and most sustained release, remaining at approximately 33 % even after 8 h, suggesting that MEL was encapsulated within the hydrophobic OA core of the nanoparticles formed by HON. The experiment at pH 5.2 was conducted to investigate the release behavior in the mildly acidic range (pH 5.0–5.5), which mimicked the lysosomal and endosomal environment within cancer cells, and evaluate the efficiency of anticancer drug delivery.¹⁶ The release profile at pH 5.2 displayed free MEL at 70.29 ± 2.45 %, NM showed 65.52 ± 1.33 %, and NM-HON showed 67.20 ± 0.64 %, with approximately 65 to 70 % released over 8 h (Figure 10b). However, NM-HA exhibited a relatively low release of 57.45 ± 1.24 %, possibly because of NM-HA exhibiting stronger hydrogen bonding and electrostatic interactions between the amine groups of HA and MEL under acidic conditions, delaying drug release as diffusion takes longer to occur.

Figure 10.

In vitro release profiles of MEL, NM, NM-HA, and NM-HON in pH 7.4 PBS (a), pH 5.2 ABS (b), and pH 7.4 PBS with 10 mM GSH (c) (n = 3). *p < 0.05, **p<0.01, ***p < 0.001. ****p<0.0001 One-way ANOVA compared with MEL.

Interestingly, NM-HON exhibited faster release under acidic conditions because of its pH-sensitive linker (CYS)-binding properties, demonstrating differentiated formulation stability and release control characteristics in acidic environments.^11,12 In addition, under 10 mM GSH conditions mimicking the intracellular reduced state of tumor cells, free MEL, NM, and NM-HA exhibited release patterns similar to those observed at pH 7.4 and pH 5.2, at 73.47 ± 3.17 %, 66.55 ± 1.98 %, and 49.43 ± 0.22 %, respectively, over 8 h (Figure 10c). However, the NM-HON formulation exhibited an increased release rate of 62.99 ± 0.51 % at 8 h. This demonstrates that the CYS linker undergoes a reduction in the presence of GSH that breaks the disulfide bond (-S-S-), thereby accelerating drug release. This suggests that NM-HON can selectively release more drugs under conditions that mimic an intracellular reducing environment. Overall, the NM-HON enabled the controlled release of MEL by regulating its pH and redox reactivity. The controlled release of the nano-system at physiological pH and accelerated release under reducing tumor microenvironments demonstrate the potential for local and sustained ocular delivery.

In vitro Cytotoxicity Study

The in vitro cytotoxicity of free MEL, NM, NM-HA, and NM-HON was evaluated in ARPE-19 normal retinal epithelial cells and WERI-Rb-1 RB cells after 24 h (Figure 11). All formulations in ARPE-19 cells caused a concentration-dependent decrease in cell viability (Figure 11a). NM-HON exhibited the highest survival rate compared to the controls at low and intermediate concentrations, demonstrating excellent initial safety for normal cells. Even at 25 µg/mL, NM-HON (66.24 ± 1.97 %) remained higher than NM (60.54 ± 11.17 %) and NM-HA (61.15 ± 2.74 %). However, differences between formulations diminished as concentrations increased (50–100 µg/mL), and NM-HON exhibited the lowest survival rate. This is interpreted as the result of the enhanced cell entry and drug exposure of HON, demonstrating safety at low concentrations owing to the protective effects of gradual release and surface HA, but revealing toxicity at high concentrations owing to accumulation and overexposure. Therefore, NM-HON was confirmed to have advantages in terms of normal retinal cell safety up to concentrations ≤ 25 µg/mL within the therapeutic range. In WERI-Rb-1 cells, all the formulations resulted in a concentration-dependent decrease in cell viability (Figure 11b). NM-HON exhibited the highest cytotoxicity at 12.5 µg/mL. The cell viability rate was 27.07 ± 1.00 %. NM-HA and NM reduced cell viability more than free MEL, at intermediate and high concentrations, indicating that simple neutralization or HA complexation alone enhanced the anticancer effect. NM-HON induced a marked loss of viability in WERI-Rb-1 cells at 12.5 µg/mL, whereas ARPE-19 cells maintained substantially higher viability at the same concentration, suggesting a degree of preferential cytotoxicity toward WERI-Ib-1 cells. This trend is likely attributable to the increased effective exposure of MEL within tumor cells via HA-mediated CD44 targeting, reductive CYS-linker-driven release from NM-HON, and improved aqueous MEL stability following ABN.¹² The disulfide bonds in HOC are susceptible to being cleaved in the reductive tumor microenvironment; resulting in OA and NM being rapidly released together, thus killing cancer cells synergistically while reducing toxicity to normal cells.

Figure 11.

In vitro cytotoxicity of MEL, NM, NM-HA, and NM-HON in ARPE-19 normal cells (a) and WERI-Rb-1 cancer cells (b) after incubation for 24 h (n = 3). *p < 0.05, **p<0.01, ***p < 0.001. ****p<0.0001 One-way ANOVA compared with MEL.

Overall, NM-HON exhibited the most pronounced anticancer efficacy among the tested formulations. The half-maximal inhibitory concentration (IC₅₀) of MEL, NM, NM-HA, and NM-HON in ARPE-19 cells were 19.38 ± 7.35, 18.18 ± 8.58, 22.86 ± 3.28, and 24.57 ± 5.77 µg/mL, respectively. The corresponding IC₅₀ in WERI-Rb-1 cells were 10.52 ± 2.24, 11.78 ± 3.13, 7.89 ± 1.67, and 7.53 ± 0.35 µg/mL, respectively. The NM-HON formulation was safer than the other formulations in ARPE-19 cells. Conversely, NM-HON exhibited the lowest IC₅₀ value in WERI-Rb-1 cells. These data confirmed that NM-loaded HON enhanced the antitumor activity of MEL while maintaining its safety profile in normal cells. The CD44 blocking experiment of NM-HON to support receptor-mediated claims will be needed if necessary.

Conclusions

This study demonstrates that combining ABN and fattigated nanoparticles with CYS linker significantly enhances the aqueous and solid-state stability, permeability, tumor-sensitizing controlled release and anticancer activity of MEL compared to the free drug. Neutralized MEL formulations (NM, NM-HA, and NM-HON) with optimized salt concentration and HA ratio significantly reduced MEL degradation compared to the free drug, with NM-HON demonstrating the greatest chemical and solid-state stability through the synergistic effects of ion complexation and hydrophobic encapsulation. HON-based nanoparticles exhibited appropriate size, surface charge, and high drug entrapment. Spectroscopic/thermal and diffraction analyses confirmed strong drug–polymer interactions and partial amorphization of MEL in the nanoparticles. The optimized NM-HON formulation exhibited pH- and GSH-responsive release, improved permeability across an artificial membrane, and enhanced cytotoxicity against WERI-Rb-1 RB cells while maintaining relatively lower toxicity toward ARPE-19 normal retinal cells. Table S1 compares commercial injectable with current investigative NM-HON for pharmaceutical and therapeutic needs. Although these early-stage in vitro results are promising as a noninvasive intraocular delivery system, further studies including in vivo ocular pharmacokinetics, tolerability, and efficacy in orthotopic RB models are necessary to evaluate the translational potential of this formulation for ocular chemotherapy.

Data Sharing Statement

All raw data required from this study are available from the corresponding author upon reasonable request.

Disclosure

The authors declare that they have no competing financial interests or personal relationships that may have influenced the work reported in this study.