Lipid matrix composition-tuned nanoparticles for enhanced topical retinol delivery and anti-photoaging efficacy

Abstract

Retinol (Ret), a hydrophobic vitamin A derivative, is widely used for skin rejuvenation, inhibiting the degradation of major extracellular matrix components such as collagen. Ret has been incorporated into lipid nanoparticles, such as liposomes, to improve stability. However, the long-term stability of these formulations under temperature variations as well as systematic optimization remain unknown. To overcome these limitations, we developed a retinol-loaded nanostructured lipid carrier (Ret-NLC) by optimizing the solid-to-liquid ratio (SL ratio) to enhance long-term stability, cellular delivery efficacy, and skin permeation. The formulation was optimized at an SL ratio of 6:4, achieving a particle size of approximately 200 nm and acceptable size distribution (polydispersity index, PDI ≈ 0.3). Ret-NLC stored at 25 and 4 °C maintained favorable dispersion characteristics during 12 weeks of storage, and in vitro release studies showed sustained release of Ret. In addition, among the tested SL ratios, Ret-NLC (6:4) exhibited the highest skin-mimetic membrane permeation, with an approximately 4.5-fold increase compared with free Ret, indicating suitability for the delivery of hydrophobic bioactive compounds to the skin. Biological evaluation revealed that Ret-NLC effectively reduced reactive oxygen species (ROS) and suppressed matrix metalloproteinase-1 (MMP-1) expression by 63.6%, which are key factors associated with skin aging and extracellular matrix degradation. Finally, Ret-NLC was incorporated into a cream formulation for practical applications. Compared with the Ret cream, the Ret-NLC cream exhibited improved formulation stability, while maintaining its physicochemical properties. Overall, these findings support the potential of Ret-NLC as a nanoplatform for delivering hydrophobic bioactive compounds in biopharmaceutical applications.

Graphical Abstract

Highlights

• •Solid-to-liquid lipid ratio optimized for stable retinol-loaded NLC formulation.
• •Ret-NLC showed ∼200 nm size and maintained dispersion integrity over 12 weeks.
• •Ret-NLC enhanced skin permeation 4.5-fold by free retinol via Franz diffusion cell.
• •Ret-NLC most effectively suppressed UVB-induced ROS in HaCaT keratinocytes.
• •Ret-NLC cream maintained rheological stability, antioxidant activity for 30 weeks.

1. Introduction

Photoaging of the skin results from extrinsic environmental stressors, particularly ultraviolet (UV) radiation, which accelerates extracellular matrix degradation through oxidative mechanisms and induces matrix metalloproteinase (MMP) expression in human skin (Fisher et al., 2001). Retinoids, including retinol (Ret) and its derivatives, are among the most clinically validated agents for mitigating photoaging owing to their ability to regulate collagen homeostasis and suppress matrix-degrading pathways (Fisher et al., 1997). Ret is preferred over other retinoids in cosmetic and dermatological formulations because of its proven anti-aging efficacy and superior tolerability (Milosheska and Roškar, 2022). However, its conjugated polyene structure renders it highly susceptible to photooxidation, thermal degradation, and oxygen-mediated decomposition (Gonçalves et al., 2016). Moreover, its strong lipophilicity limits its dispersion in aqueous systems and may restrict controlled penetration across the highly organized lipid matrix of the stratum corneum (Prausnitz and Langer, 2008). These limitations necessitate the development of advanced delivery systems capable of enhancing stability and enabling controlled and efficient dermal transport.

Various delivery strategies have been explored to address these challenges. Conventional emulsions are widely used in topical products; however, they provide limited structural protection against oxidative stress and environmental exposure, and their interfacial instability may accelerate their degradation during storage (Zhong et al., 2024). Vesicular systems, such as liposomes, improve solubilization and skin affinity; however, encapsulation alone does not ensure durable stabilization. Phospholipid vesicles remain susceptible to lipid peroxidation, and Ret degradation has been reported even in the presence of antioxidants, indicating that vesicle architecture requires further structural optimization (Chmykh and Nadeau, 2020). Although multilamellar vesicles, hybrid vesicles, polymer-based particles, and silicone-based particles have been introduced to enhance stability, their increased structural complexity raises concerns regarding scalability and manufacturing reproducibility (Rahman et al., 2024; Shields IV et al., 2018). Collectively, these findings highlight the necessity for structurally tunable delivery platforms capable of integrating stability enhancement with controllable permeation behavior and accessibility to manufacturing.

Lipid nanoparticles, particularly nanostructured lipid carriers (NLC), represent a promising platform for retinoid delivery and stabilization. NLC were developed as a second-generation lipid nanoparticle system to overcome the limitations of solid lipid nanoparticles (SLN), which consist of a highly ordered solid lipid matrix. Although SLN provides enhanced physical stability, its high crystallinity can restrict the accommodation of bioactive compounds and promote their expulsion during storage (Müller et al., 2002). NLC has been developed by incorporating liquid lipids into a solid matrix, thus generating structural imperfections that reduce crystallinity and enhance the loading capacity (LC) of bioactive compounds (Katouzian et al., 2017). The performance of NLC systems is strongly influenced by the SL ratio, which governs the internal crystallinity, structural disorder, and bioactive compound accommodation space. Variations in the SL ratio have been associated with changes in encapsulation efficiency (EE), release behavior, and physicochemical stability (Pardeike et al., 2009; Park et al., 2018; Jeitler et al., 2024). In dermal applications, lipid nanoparticle systems have been reported to enhance bioactive compound retention and modulate permeation across the stratum corneum (Müller et al., 2007), thereby highlighting the importance of rational lipid matrix design. Several studies have been conducted on encapsulating hydrophobic bioactive compounds, such as sesquiterpenes and curcuminoids, into lipid nanoparticles, particularly NLC (Lee et al., 2025; Ghate et al., 2016). Building on these approaches, recent studies have applied to NLC systems for the delivery of retinoids. These systems have revealed improved chemical stability and skin delivery compared to conventional formulations such as liposomes (Chmykh and Nadeau, 2020; Ghate et al., 2016). However, the influence of lipid matrix composition on crystallinity, colloidal stability, and skin permeation of retinoid-loaded NLC remains insufficiently characterized. Moreover, a systematic investigation linking these physicochemical characteristics to antioxidant and anti-photoaging activities, including MMP-1 expression in UVB-irradiated skin cells, has yet to be established.

Systematic optimization of formulation variables can simultaneously improve drug entrapment, stability, release behavior, and permeability, highlighting the importance of formulation design in achieving desired delivery performance (Kim et al., 2020). Therefore, we hypothesized that an appropriately designed lipid matrix could improve the colloidal stability, release characteristics, and in vitro delivery performance of Ret-loaded lipid nanoparticles. To test this hypothesis, we systematically evaluated the effects of lipid matrix composition on particle size, size distribution, zeta potential, storage behavior, release characteristics, and Ret recovery using a comparative Strat-M membrane assay (Fig. 1). Based on the overall balance of these physicochemical and functional characteristics, a lead NLC formulation was selected for subsequent biological evaluation. Physicochemical characterization was performed through in vitro release analysis, and the biological performance of the formulations of UVB-irradiated HaCaT cells were determined by assessing intracellular reactive oxygen species (ROS) production and MMP-1 expression. The selected Ret-NLC was further incorporated into a cream formulation to evaluate its translational potential for cosmetic and dermatological applications.

Fig. 1.

Schematic representation of lipid nanoparticles. (A) Schematic representation of lipid screening for lipid nanoparticle formulation. (B) Schematic representation of the physicochemical properties of lipid nanoparticles. (C) Schematic representation of the bio-application of Ret-loaded lipid nanoparticles.

2. Materials and methods

2.1. Materials

Safflower seed oil, soybean oil, isopropyl myristate (IPM), squalane, jojoba oil, dimethyl sulfoxide (DMSO), Ret (>95.0%), Tween® 80, Span® 80, and coumarin-6 (C6), were purchased from Sigma-Aldrich (St. Louis, MO, United States). Precirol® ATO 5 (Pre) was kindly gifted by Gattefosse Co. (Saint-Priest, Cedex, France). Dulbecco's modified Eagle medium (DMEM), fetal bovine serum (FBS), penicillin-streptomycin solution, and phosphate-buffered serum (FBS) were purchased from Corning Inc. (Corning, NY, United States). High-performance liquid chromatography (HPLC) methanol and acetonitrile were purchased from Duksan Pure Chemical Co. (Ansan, Gyeonggi, South Korea). Hoechst-33,342 was obtained from Thermo Fisher Scientific (Waltham, MA, United States). Deionized water was prepared using a Milli-Q purification system (Millipore, Billerica, MA, USA). All remaining reagents were of analytical grade.

2.2. Screening of liquid lipids

The solubility of Ret in selected liquid lipids is crucial for obtaining lipid nanoparticles with high EE and LC. Following the equilibrium method (Joshi and Patravale, 2008), excess Ret was added to the oils in the microcentrifuge tubes and shaken at 25 °C for 24 h using a rocker-type shaker (CR95; FINEPCR, Gunpo, South Korea). The oversaturated sample was centrifuged at 1073 ×g for 10 min using a centrifuge (MiniSpin®; Eppendorf, Hamburg, Germany). The supernatant was diluted appropriately with ethanol for spectrophotometric analysis of Ret at 325 nm using a microplate reader (Synergy H1; BioTek Instruments, Winooski, VT, USA). Each experiment was performed in triplicate.

2.3. Synthesis of Ret-loaded lipid nanoparticles

Ret-loaded lipid nanoparticles were prepared using high-shear homogenization and ultrasonication techniques with minor modifications (Hyun et al., 2022). Briefly, the lipid phase, consisting of solid lipid, liquid lipid, 0.1% (w/v) Ret, and Span® 80, was melted at 75 ± 5 °C and stirred at 450 rpm using a hot-plate magnetic stirrer (HSD180; MTOPS, Yangju, South Korea) for 10 min. Simultaneously, an aqueous phase consisting of Tween® 80 and distilled water was heated and stirred. The total concentration of the surfactants (Tween® 80 and Span® 80) was fixed at 1% (w/v). For each SL ratio, the individual amounts of two surfactants were precisely calculated and incorporated based on their respective HLB values relative to the total lipid concentration. After the two phases were homogeneously mixed, the hot aqueous phase was added dropwise to the lipid phase to obtain a pre-emulsion. The emulsion was stirred continuously for 10 min at a constant temperature of 75 ± 5 °C and subjected to high-shear homogenization (HZ1, LABTron Co., Seoul, South Korea) at 8000 rpm for 5 min. This was followed by further dispersion using a probe sonicator (UP200S, Hielscher Ultrasonics GmbH, Teltow, Germany) for 5 min (amplitude power: 30%; work time: 4 s; rest time: 2 s; power specifications: 400–480 W). Finally, the resulting dispersion was allowed to cool to 25 °C to obtain the NLC, which were then stored in a cool environment until characterization. Blank NLC were synthesized without adding Ret to the lipid phase using the same synthesis process.

2.4. Physicochemical characterization

The particle size, PDI, and zeta potential of the lipid nanoparticles were determined using dynamic light scattering (DLS) and electrophoretic light scattering (ELS) using a particle size and zeta potential analyzer (ELS-2000ZS; Otsuka Electronics Co., Hirakata, Japan). Light scattering was monitored at 25 °C and an angle of 90°. Each sample was diluted with purified water (1:50, v/v), and all measurements were performed in triplicate. The Fourier transform infrared (FT-IR) spectra of Ret, Pre, lipid mixture, Ret-NLC, and Ret-SLN were obtained using an FT-IR spectrometer (VERTEX 70; Bruker Optics GmbH, Ettlingen, Germany) over 4000–400 cm⁻¹. For nanoparticle samples (Ret-NLC and Ret-SLN), dispersions were lyophilized prior to analysis. The lipid mixture sample consisted of a mixture of Pre and IPM, at a volumetric ratio of 2:1 in volume, heated to 75 $\pm$ 5 °C for 10 min to obtain a semi-solid mixture. Differential scanning calorimetry (DSC) was performed on Ret, Pre, Ret-nanoemulsion (NE), Ret-NLC, and Ret-SLN using a thermal analyzer (SDT Q600; TA Instruments, New Castle, DE, USA). Ret and the solid lipid were analyzed in their original solid state, whereas Ret-NE, Ret-NLC and Ret-SLN were analyzed after lyophilization. The temperature range was 30–230 °C, with a heating rate of 10 °C/min under a nitrogen gas flow of 100 mL/min. Transmission electron microscopy (TEM; JEM-F200; JEOL Ltd., Akishima, Japan) was used to examine the morphology of the Ret-loaded lipid nanoparticles at an acceleration voltage of 200 kV. The samples were loaded onto a carbon/Formvar-coated copper (Cu) grids. The particle sizes in the TEM images were measured using ImageJ software.

2.5. Encapsulation efficiency and loading capacity analysis

The EE and LC of the Ret-loaded lipid nanoparticles were determined using an ultrafiltration-centrifugation technique. An Amicon centrifugal filter device (MWCO 10 kDa; MilliporeSigma, Burlington, MA, USA) was used to remove unencapsulated Ret during centrifugation at 12,000 rpm for 60 min. The amount of unencapsulated Ret in the aqueous phase was quantified by measuring fluorescence intensity. A standard curve was constructed using Ret, and EE and LC were calculated using the following equations:

$$\mathit{EE}\left(\%\right)=\frac{W_{\mathit{total}}-W_{\mathit{free}}}{W_{\mathit{total}}}\times 100$$ (1)

$$\mathit{LC}\left(\%\right)=\frac{W_{\mathit{total}}-W_{\mathit{free}}}{W_{\mathit{total}}-W_{\mathit{free}}+W_{\mathit{lipid}}}\times 100$$ (2)

where W_total is the total weight of Ret initially incorporated into the system, W_free denotes the weight of the free (unencapsulated) Ret remaining in the aqueous phase after separation, and W_lipid denotes the total mass of lipid (solid plus liquid lipid; 250 mg in all formulations) and excludes surfactants.

2.6. Storage stability test

The stability of Ret-NE, Ret-NLCs (2:8, 4:6, 6:4, and 8:2), and Ret-SLN was evaluated over 12 weeks of storage at 25 °C by monitoring changes in colloidal properties, including hydrodynamic size, PDI, and zeta potential. For each formulation, samples from the same preparation batch were stored under the designated temperature condition throughout the 12-week study. At each predetermined time point, the stored samples were analyzed under the same measurement conditions described in Section 2.4. Visual appearance was also examined for macroscopic changes, including phase separation, creaming, precipitation, or aggregation. The experiments were carried out in triplicate for each sample, and the results were presented as the mean ± standard deviation (SD).

2.7. In vitro release profile

An in vitro Ret release study was performed using Ret-loaded lipid nanoparticles under sink conditions according to the dialysis bag method with a molecular weight cutoff (10 kDa) (Truong et al., 2022). Briefly, 5 mL of the Ret-loaded lipid nanoparticle dispersion was placed in treated dialysis bags and immersed in 100 mL of freshly prepared dissolution medium consisting of phosphate-buffered saline (PBS; pH 7.4) with 10% v/v ethanol with constant stirring at 200 rpm. Procedures involving Ret were conducted under light-protected conditions to prevent photodegradation. Aliquots (1 mL) of the medium were collected at predetermined times, and the same volume of fresh medium was added immediately thereafter. The Ret concentration of the samples was analyzed using UV–Vis spectrophotometry at 326 nm. Calibration in PBS (pH 7.4) containing 10% v/v ethanol was linear over 0–215 μg/mL at this wavelength. The analysis was performed in triplicate, and the results are expressed as the mean ± SD. The cumulative release percentage (CR%) of Ret released from each formulation was calculated using the following equation:

$$\mathit{CR}\left(\%\right)=\frac{V_e{\sum }_{i=1}^{n-1}{C}_{n-1}+V_0{C}_n}{m}\times 100$$ (3)

where V_e is the volume removed, V₀ is the total volume of the release medium, C_n is the concentration at the selected time, and m is the total milligram of Ret in Ret-loaded lipid nanoparticles.

2.8. Cell culture

HaCaT cells, a human epidermal keratinocyte cell line, were provided by Prof. K. Y. Choi from the Department of Biotechnology, Yonsei University, Korea. HaCaT cells were cultured in DMEM supplemented with 10% (v/v) fetal bovine serum (FBS) and 1% (v/v) antibiotic-antimycotic solution and maintained at 37 °C in a humidified atmosphere containing 5% CO₂.

2.9. Intracellular delivery analysis

The transport characteristics of Nile Red-labeled nanoparticles were studied at the cellular level using confocal laser scanning microscopy (CLSM). HaCaT cells were seeded in confocal dishes at a density of 3 × 10⁴ cells/dish. After 24 h of incubation, the cells were cultured with the samples for an additional 24 h, washed with PBS (pH 7.4), and fixed with 10% formalin solution for 15 min at 25 °C. Cells were stained with Hoechst-33,342 before analysis to determine the position of the nuclei. After washing the cells several times with PBS, the localization of the fluorescence signals from Nile Red and Hoechst-33,342 was visualized using a CLSM (LSM 900; Carl Zeiss AG, Oberkochen, Germany). To assess the endocytic pathway, HaCaT cells were seeded in 24-well culture plates (SPL Life Sciences, Pocheon, South Korea) at a density of 1 × 10⁵ cells/well. Cells were pretreated with each inhibitor in serum-free medium for 1 h, followed by treatment with the C6-labeled NE, NLC, SLN for 4 h. Chlorpromazine (2 μg/mL) and genistein (50 μg/mL) were used to inhibit clathrin- and caveolae-mediated uptake. Wortmannin (50 ng/mL) was used to inhibit macropinocytosis. Methyl-β-cyclodextrin (2 mg/mL) was used to deplete cholesterol for the inhibition of lipid raft-mediated uptake. ATP-dependent uptake was suppressed by incubating the cells at 4 °C. After treatment, fluorescence intensity was analyzed using flow cytometry (BD FACSymphony™ A5; Becton Dickinson, Franklin Lakes, NJ, USA).

2.10. In vitro skin penetration test

A Franz diffusion cell (Teledyne instrument, Phoenix RDS, effective area of 1.32 cm²) was used to evaluate the in vitro Ret permeation. The volume of the receptor phase was approximately 6.5 mL. The temperature of the assay was accurately controlled at 37 °C to mimic human skin. A PES membrane filter (Strat-M filter, diameter of 25 mm) was mounted between the donor and receptor compartments, and approximately 1 mL of Ret, Ret-NE, Ret-NLCs (2:8, 4:6, 6:4, and 8:2), and Ret-SLN were placed on the membrane. The acceptor medium consisted of PBS (pH 7.4) and was stirred at 600 rpm using a magnetic stir bar to maintain concentration uniformity and minimize stagnant boundary layers. The membrane filter was removed from the Franz diffusion cell after 24 h, and the fluid (1 mL) from the receptor chamber was collected to quantify Ret. The membrane filter was rinsed with distilled water to recover surface retained Ret, and the Ret concentration in the solution was analyzed using HPLC. For comparative analysis, combined Ret recovery was calculated as the sum of Ret detected in the receptor phase and residual Ret recovered by rinsing the Strat-M membrane after the 24 h experiment.

2.11. Antioxidant effect analysis

Intracellular ROS production was detected by staining the cells with the fluorescent probe 2′,7′-dichlorofluorescin diacetate (DCFH-DA). This dye is a stable compound that readily diffuses into cells and is hydrolyzed by intracellular esterases to yield non-fluorescent intermediate 2′,7′-dichlorofluorescin (DCFH), which is trapped within the cells. HaCaT cells were seeded in either 24-well plates (1 × 10⁵ cells/well) for fluorescence microscopy or confocal dishes (3 × 10⁴ cells/dish) for CLSM and incubated for 24 h before UVB irradiation. After 24 h of incubation following UVB irradiation, the cells were treated with free Ret, Ret-NE, Ret-NLC, or Ret-SLN at a Ret concentration of 15 μM and further incubated for 24 h. The cells were incubated with 20 μM DCFH-DA at 37 °C for 30 min in the dark and washed with PBS. For CLSM, the cells were also fixed with 10% formalin solution for 15 min at 25 °C and stained with Hoechst-33,342 to visualize cell nuclei. Fluorescence signals from DCF were visualized using a fluorescence microscope (IX71; Olympus Corp., Tokyo, Japan), whereas DCF and Hoechst-33,342 fluorescence signals were observed using a CLSM (LSM 900; Carl Zeiss).

2.12. Anti-aging effect analysis

The anti-photoaging effect was measured through UVB-induced MMP-1 downregulation (Yang et al., 2024). HaCaT cells were seeded in six-well plates at a density of 3.5 × 10⁵ cells/well and incubated for 24 h. The cells were then irradiated with UVB at 30 mJ/cm² using a UV crosslinker (XL-1000 UV Crosslinker; Spectroline, Westbury, NY, USA). They were treated with Ret at 15 μM using free Ret, Ret-NE, Ret-NLC, and Ret-SLN and incubated for 24 h. Western blotting was performed to analyze MMP-1 expression. Briefly, the cells were lysed in RIPA buffer and denatured by heating at 95 °C for 5 min. Proteins were transferred to nitrocellulose membranes and blocked with 5% skim milk for 1 h. The membranes were incubated for 18 h at 4 °C with 1:1000 and 1:500 diluted specific primary antibodies for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and MMP-1 proteins (Santa Cruz Biotechnology, Dallas, TX, USA), respectively. After washing thrice with Tris-buffered saline containing Tween 20 (TBST), the membranes were incubated for 2 h at 24 °C with 1:3000 diluted enzyme-linked secondary antibodies. The protein bands were detected using an enhanced chemiluminescence (ECL) system and visualized using a gel documentation system (Gel Doc XR+ System; Bio-Rad Laboratories, Hercules, CA, USA). The western blotting results were quantified using ImageJ software.

2.13. Ret-NLC cream formulation and analysis

The cream base was prepared using an oil-in-water (O/W) emulsification method. For the aqueous phase, purified water, polyols, and a thickening agent (xanthan gum) were mixed using a homogenizer at 1000–2000 rpm for approximately 20 min. For the oil phase, caprylic/capric triglyceride was combined with solid lipids, including stearic acid, glyceryl stearate, sorbitan stearate, and butter components, and the mixture was stirred at 500 rpm for 40 min at 75 °C. After adjusting the temperature of the aqueous phase to 75 °C, the oil phase was gradually added, and the mixture was emulsified at 1500–2500 rpm for 40 min. The resulting emulsion was then cooled to 35 °C in the manufacturing vessel. Finally, preservatives and Ret-NLC (7.5%) were added, followed by stirring for 10 min. To evaluate physicochemical stability, the pH, rheological properties, and visual appearance of the control, Ret, and Ret-NLC cream formulations were evaluated. Digital images of the Control, Ret, and Ret-NLC cream formulations stored at 4 °C and 25 °C were obtained 6 weeks after preparation. The pH was measured using a pH/ion meter (SevenCompact pH/Ion meter; Mettler Toledo, Greifensee, Switzerland). Rheological analyses were conducted at 25 °C using a rheometer (MCR 702e; Anton Paar GmbH, Graz, Austria) equipped with a 25-mm parallel plate at a 1 mm gap, over a shear rate ranging from 10⁻² to 10² s⁻¹. For long-term stability testing, formulations were stored at 4 °C and 25 °C for 12 weeks. pH and phase separation were evaluated at 1-week intervals during the first month and at 2-week intervals thereafter up to 12 weeks. Viscosity at 4 °C was measured immediately after preparation and at week 12. For harsh-condition accelerated testing, 300 mL of each cream was filled into pump containers and stored at 45 °C and 70% relative humidity for 30 weeks. Samples were collected at 0, 3, 7, 14, and 210 d and assessed against baseline values to evaluate compliance with predefined specification criteria. The antioxidant activities of the formulations were evaluated using a 2,2-diphenyl-1-picrylhydrazyl (DPPH) radical-scavenging assay in 96-well plates. Ret and Ret-NLC creams were diluted in ethanol and mixed with an equal volume of 0.2 mM DPPH solution (Cho et al., 2023). After incubation in the dark for 30 min at 25 °C, the absorbance was measured at 517 nm using a microplate reader. Radical scavenging activity (%) was calculated as.

$$\text{Radical scavenging activity}\left(\%\right)=\left[\left({\mathrm{A}}_{\text{control}}-{\mathrm{A}}_{\text{sample}}\right)/{\mathrm{A}}_{\text{control}}\right]\times 100$$ (4)

where A refers to absorbance, using standard methods to compare the antioxidant capacity of each formulation.

2.14. Statistical analysis

In vitro experiments were conducted in triplicate, and the data are presented as the mean ± SD. Group comparisons were performed using one-way analysis of variance (ANOVA), followed by Tukey's post-hoc test using IBM SPSS Statistics, version 29. Statistical significance is indicated as * P < 0.05, ** P < 0.01, and *** P < 0.001.

3. Results and discussion

3.1. Selection of lipids for enhanced stability of lipid nanoparticles

To efficiently encapsulate Ret within lipid nanoparticles and evaluate its biological efficacy, suitable lipids must be selected. Solid and liquid lipids were carefully selected based on their physicochemical properties and solubility. Owing to safety considerations and the potential for practical application in the cosmetics industry, only lipids listed as FDA inactive ingredients were used.

During solid lipids preparation, lipid nanoparticles in the pre-emulsion state undergo recrystallization during high-shear homogenization. Particle inhomogeneity and batch-to-batch variations may occur when lipid memory effects occur. Therefore, solid lipids with lower melting points were preferred to facilitate synthesis at temperatures 10 °C above the melting point of the lipid (Jores et al., 2004). To prevent the lipid memory effect, Precirol ATO 5 (Pre) was selected based on its thermal and structural characteristics. Pre exhibited a melting point of approximately 55 °C, which is approximately 10 °C lower than that of Ret (62–64 °C). Selecting a lipid with a melting point lower than that of Ret allows processing at milder temperatures, facilitating the homogeneous incorporation of Ret into the lipid matrix, while minimizing thermal degradation. In addition, Pre consists of mixed fatty acid chains of different chain lengths (C16 and C18), resulting in a more disordered crystalline structure and weaker intermolecular interactions than those of pure fatty acids (Guo et al., 2015). This structural disorder hinders the formation of a highly ordered and stable crystalline lattice during recrystallization, thereby reducing the lipid memory effect and improving retention of the bioactive compounds within the lipid matrix. For liquid lipid selection, higher solubility of bioactive compound in the liquid lipid is generally associated with increased loading efficiency in NLC (Joshi and Patravale, 2008). To select the lipid matrix of the Ret-loaded lipid nanoparticles, a total of five candidate liquid lipids were selected for initial screening, drawing from a previous study in which these oils were used in lipid nanoparticle synthesis (Lee et al., 2025). These candidate oils were systematically evaluated to identify the most compatible fluidizing agent capable of maximizing cargo incorporation. The saturation solubility of Ret in each oil was determined using an equilibrium solubility method. Among the various oils tested, IPM was ultimately selected because it exhibited the highest saturation solubility value of 154.6 ± 44.2 mg/mL (Fig. S1).

Nonionic surfactants were evaluated based on their hydrophilic-lipophilic balance (HLB) values to identify surfactants suitable for stabilizing the selected solid and liquid lipids. Combinations of high- and low-HLB surfactants can be selected to match the required HLB of a given lipid phase (Prinderre et al., 1998). Based on this approach, Tween 80 and Span 80 were selected as suitable surfactants for the solid and liquid lipids. To evaluate a suitable surfactant-to-lipid ratio for stable nanoparticle formation, particle size and zeta potential were evaluated at total lipid concentrations of 1% and 2% (w/v) with varying surfactant concentrations (Fig. S2). At a total lipid concentration of 2% (w/v), formulations prepared without surfactants exhibited particle sizes exceeding 2000 nm, indicating severe aggregation and insufficient stabilization. In contrast, the addition of 0.5% (w/v) surfactant concentration markedly reduced the particle size by approximately five-fold, revealing the critical role of surfactants in stabilizing high-lipid formulations. Surfactant concentrations of >1% (w/v) produced zeta potential values below −30 mV, indicating enhanced electrostatic stabilization against aggregation (Teng et al., 2019). However, excessively high surfactant concentrations may induce micelle formation and formulation in homogeneity (Bouyer et al., 2012). Therefore, 1% (w/v) surfactant was considered an appropriate concentration for maintaining sufficient stabilization, while minimizing excessive surfactant-associated effects.

3.2. Lipid matrix composition optimization of Ret-loaded lipid nanoparticles

The total lipid concentration was optimized to determine the appropriate lipid matrix composition by evaluating the particle size, zeta potential, and PDI. For this purpose, formulations were screened at a fixed SL ratio of 7:3 within the canonical NLC range reported in previous studies (Pardeike et al., 2009). The comparatively smaller particle size of Ret-NLC avoids the cellular uptake and permeation limitations associated with particles exceeding 200 nm (Lee et al., 2025), thereby enhancing stratum corneum interaction and improving skin permeation for effective topical delivery (Schäfer-Korting et al., 2007). As the total lipid concentration increased, the particle size increased owing to insufficient surfactant coverage, which limited efficient droplet stabilization (Fig. 2A). Notably, at total lipid concentrations of ≥1%, the particle size exceeded the 200 nm threshold (Table S1). Consistent with these results, the absolute zeta potential values decreased as the total lipid concentration increased, indicating reduced electrostatic repulsion between particles and an increased risk of particle aggregation (Fig. 2B). In addition, all formulations, except for the 4% (w/v) group, exhibited PDI values below 0.3, indicating relatively homogeneous particle distributions (Fig. 2C). Therefore, formulations containing 1%, 2%, and 4% (w/v) total lipids were excluded. Although both the 0.25% and 0.5% (w/v) formulations satisfied the particle size criterion, the 0.5% (w/v) was selected as the optimal condition because the higher lipid concentration was expected to provide a larger lipid matrix for encapsulation and retention of lipophilic active ingredients, while maintaining acceptable colloidal stability. Furthermore, formulations with higher lipid concentrations are generally advantageous for high-energy processes (Gordillo-Galeano and Mora-Huertas, 2018).

Fig. 2.

Physicochemical properties of the Ret-loaded lipid nanoparticles. (A) Particle size, (B) zeta potential, and (C) PDI of blank NLC with varying total lipid concentration at an SL ratio 7:3. (D) Particle size and (E) zeta potential of total lipid 0.5% (w/v) blank NLC with varying SL ratios. (F) Digital images of Ret-NE, Ret-NLC, and Ret-SLN. (G) Particle size, zeta potential, (H) EE, and (I) LC of Ret-NE, Ret-NLC, and Ret-SLN. The results are presented as the mean ± SD (n = 3).

Because liquid lipids enhance the loading efficiency of bioactive compounds by introducing imperfections into the lipid matrix, whereas solid lipids contribute to structural stability, the SL ratio is a critical formulation in NLC group. Although conventional NLC systems are commonly prepared within a limited SL ratio range, the optimal composition may vary depending on the physicochemical properties of the ingredient. Therefore, the SL ratio was systematically ranged 0:10–10:0, deliberately extending beyond the conventional NLC window to bracket the NLC compositions with pure liquid NE and pure SLN as mechanistic reference systems (Figs. 2D and E). Among the tested ratios, formulations with 4:6 to 8:2 ratios provided an optimal balance between process viscosity and applied shear, enabling efficient particle size control, while promoting homogeneous molecular dispersion of the bioactive compound within the lipid matrix (Table S2). This may be attributed to reduced crystallization-induced bioactive compound expulsion and enhanced defect accommodation (Mehnert and Mäder, 2012). These results revealed that the SL ratio substantially influences the physicochemical properties of lipid nanoparticles; accordingly, the 6:4 ratio was selected as a leading NLC candidate for further evaluation, enabling direct comparison with NE and SLN.

Based on these considerations, Ret-loaded lipid nanoparticles were synthesized and classified into three formulations: NE, NLC, and SLN (Fig. 2F). The hydrodynamic size increased with the SL ratio. In contrast, the zeta potential exhibited the opposite trend, with Ret-NE exhibiting the lowest absolute value (Fig. 2G). The EE decreased as the solid lipid ratio increased because the Ret solubility in liquid lipids was much higher than that in solid lipids (Fig. 2H). LC followed a similar trend, with the highest EE (65.6%) and LC (11.6%) were observed for Ret-NE (Fig. 2I). The Ret-NLC showed intermediate EE and LC, reflecting the ability of its mixed lipid matrix to retain the loaded bioactive compound, while maintaining structural stability. In general, smaller nanoparticles exhibit a greater surface area and closer contact with the skin surface, facilitating efficient permeation. In addition to particle size, a zeta potential below −30 mV plays a crucial role in maintaining colloidal stability by preventing aggregation in lipid nanoparticles through electrostatic repulsion (Bhattacharjee, 2016). Based on these initial physicochemical properties, a total lipid concentration of 0.5%, surfactant concentration of 1%, and SL ratio of 6:4 emerged as the leading candidates and were advanced for further evaluation in subsequent experiments.

3.3. Physicochemical characterization of Ret-loaded lipid nanoparticles

To investigate the long-term storage stability of the Ret-loaded lipid nanoparticles with different SL ratios, their hydrodynamic size, zeta potential and PDI were evaluated for 12 weeks at 25 °C (Fig. S3, Table S3). The formulations exhibited formulation-dependent temporal changes in hydrodynamic size, with more pronounced size increases observed for Ret-NLC (8:2) and Ret-SLN at later storage times. Some non-monotonic changes in particle size were observed and should not necessarily be interpreted as physical shrinkage or dissociation, as DLS measures an ensemble-averaged hydrodynamic diameter that can vary with sample polydispersity and size-distribution characteristics (Stetefeld et al., 2016). Zeta potential values varied non-monotonically during storage at 25 °C (Doane et al., 2012). The decrease of zeta potentials observed in some formulations up to week 8 followed by an increase at week 12 is more reasonably interpreted because of time-dependent changes in lipid packing and interfacial organization instead of spontaneous restoration of the original particle surface (Iskandar et al., 2024). Moreover, in formulations containing nonionic surfactants such as Tween 80 and Span 80, changes in the magnitude of zeta potential do not necessarily correspond directly to changes in colloidal stability. This is because sterically stabilized particles can exhibit variations in electrokinetic potential depending on the surrounding interfacial and ionic environment (Owen et al., 2009). The PDI values of all formulations showed minor fluctuations below 0.3 throughout storage. Given the non-monotonic nature of these changes and the absence of time-resolved structural analyses during storage, they were not attributed to a specific structural rearrangement. Instead, colloidal stability was evaluated collectively based on hydrodynamic size, PDI, zeta potential, and visual appearance. Visual examination showed that Ret-NE and Ret-NLC formulations remained macroscopically homogeneous without apparent phase separation or precipitation, whereas Ret-SLN showed more pronounced aggregation during refrigerated storage (Fig. S4). Collectively, these results indicated that Ret-NLC (6:4) did not exhibit a progressive loss of dispersion homogeneity during storage. Therefore, the 6:4 formulation, which had been identified as a leading NLC candidate based on the initial physicochemical screening, was retained for subsequent evaluation of delivery performance.

The morphological properties of the Ret-loaded lipid nanoparticles were analyzed using TEM. A schematic diagram was included to illustrate the differences in particle size and internal lipid matrix compactness among the three formulations. TEM analysis showed spherical particles with average sizes of 172.0 ± 9.1 nm (Ret-NE), 226.1 ± 46.8 nm (Ret-NLC), and 340.8 ± 57.0 nm (Ret-SLN) (Fig. 3A). Ret-NE displayed internally heterogeneous “lipid pocket” structures, whereas Ret-NLC and Ret-SLN exhibited more uniform and denser internal morphologies, consistent with solidified or semi-solid lipid matrices. The slightly darker peripheral regions observed in Ret-NLC and Ret-SLN indicate surfactant-enriched interfacial layers and peripheral bioactive compound distribution patterns in lipid nanoparticles.

Fig. 3.

Physicochemical characterization of Ret-loaded lipid nanoparticles. (A) Schematic representation and morphological analyses of Ret-NE, NLC, and SLN, conducted using TEM. Scale bars: 200 nm. (B) FT-IR spectra for Ret, Pre, lipid mixture, Ret-NLC, and Ret-SLN. (C) DSC and (D) PXRD patterns of Ret, Pre, Ret-NE, NLC, and SLN. The results are presented as the mean ± SD (n = 3).

The structural, thermal, and crystallinity characteristics of Ret-NE, Ret-NLC, and Ret-SLN were analyzed using FT-IR spectroscopy, DSC, and powder X-ray diffraction (PXRD). The FT-IR spectrum of Ret showed characteristic O—H stretching near 3400 cm⁻¹ and C—H stretching at 2800–3000 cm⁻¹. In contrast to Ret, the other lipid components exhibited characteristic ester bond stretching peaks at approximately 1739 and 1178 cm⁻¹ (Shrestha et al., 2021). Consistent with this observation, the FT-IR spectra of Ret-NLC and Ret-SLN did not display new functional peaks relative to the Ret and lipid components, indicating that no chemical bonding or degradation occurred during nanoparticle preparation (Fig. 3B). DSC was used to analyze the melting points and crystallinity of the formulations. Pure Ret and the solid lipid (Pre) displayed well-defined melting transitions in their thermograms. No distinct melting peaks were observed for Ret-NE, which was composed solely of liquid lipids. In contrast, Ret-NLC and Ret-SLN exhibited broadened and reduced-intensity endothermic peaks, indicating a less ordered crystalline structure (Fig. 3C). This decrease in the melting point may be attributed to the incorporation of liquid lipids and surfactants, which can disrupt the crystalline packing of the lipid matrix and lead to a more disordered structure (Lin et al., 2022). Notably, the melting temperature of Ret-SLN (51.9 °C) was lower than that of Ret-NLC (58.5 °C). This may be attributed to the different distribution of Ret between the lipid phases. In Ret-NLC, Ret can partition into the liquid lipid, enabling a smaller amount of Ret to dissolve within the solid lipid domains; in Ret-SLN, no liquid lipid is present, and the entire Ret load must be accommodated within the solid lipid. Ret dissolved in the crystal lattice acts as a lattice impurity and depresses the melting temperature, which is consistent with the greater melting point depression observed for Ret-SLN. Notably, the melting temperature reflects crystal perfection instead of the degree of crystallinity; because the formulations differ in composition, the crystallinity indices calculated from the melting enthalpies are not directly comparable between them. The PXRD patterns further supported the DSC results (Fig. 3D). Pure Pre exhibited sharp diffraction peaks associated with its crystalline structure, whereas Ret-NLC and Ret-SLN showed weakened and broadened peaks with slight peak shifts, indicating the expansion of the lipid lattice and partial amorphization (Müller et al., 2000). The shift toward lower diffraction angles reveals an increased interplanar spacing due to the incorporation of liquid lipids and Ret, which disrupts the ordered packing of the solid lipid matrix. These PXRD results provide evidence of the lower melting point of Ret-SLN compared to that of Ret-NLC. Overall, the physicochemical analyses confirmed that Ret-NLC formed a partially amorphous, low-crystallinity lipid matrix with a stable nanoscale morphology, supported efficient Ret encapsulation, and was suitable for hydrophobic bioactive compound delivery.

3.4. Intracellular delivery of Ret-loaded lipid nanoparticles

To assess the delivery performance of the lipid nanoparticles with different SL ratios, their cellular uptake, release profile, endocytic uptake mechanisms and the combined amount of permeated and retained Ret were evaluated. For the cellular uptake study, the Ret concentration was maintained below 20 μM, whereas the lipid concentration was also maintained within a range confirmed to be non-cytotoxic to the cells (Fig. S5). CLSM analysis revealed that intracellular fluorescence was observed within the cellular boundaries, indicating successful cellular uptake instead of surface adsorption (Fig. 4A). Notably, compared to the group treated with Nile Red (NR) alone, the fluorescence signals in the carrier-loaded groups were more distinctly aggregated, confirming successful intracellular delivery in the form of nanoparticles. Ret-loaded lipid nanoparticles were subjected to in vitro release experiments using equilibrium dialysis (Fig. 4B). The release profiles of NE and NLC were comparable, consistent with previous reports showing that release behavior is strongly influenced by lipid composition, drug–lipid affinity, matrix organization, and intraparticle drug localization (Wu et al., 2021; Mitri et al., 2011). This behavior may result from redistribution of lipophilic compounds toward less crystalline or interfacial regions during lipid crystallization (Gordillo-Galeano et al., 2022). The initial burst release behavior, similar to previous results for primaquine-loaded NE, NLC, and SLN, may be attributed to the unentrapped bioactive compounds that are adsorbed on the surface of the lipid nanoparticles instead of the bioactive compounds entrapped in the lipid phase (Zheng et al., 2025). Sustained release of Ret from NLC within the extracellular matrix may provide a supplementary contribution, allowing passively released Ret to diffuse into the surrounding cells (Wang et al., 2024). Furthermore, Ret-NLC exhibited a cumulative intracellular release exceeding 80%, which was higher than that of Ret-SLN. The relatively lower release rate observed for Ret-SLN compared with Ret-NLC is consistent with previous studies, indicating that the fluidity of the liquid lipid matrix enhances bioactive compound diffusion toward the particle surface, thereby accelerating bioactive compound release (Gordillo-Galeano and Mora-Huertas, 2018). Also, this trend was attributed to the higher crystallinity and more compact lipid packing of the SLN matrix, highlighting the critical role of lipid-matrix organization in controlling drug release (Nafee et al., 2018). These results further revealed that lipid-matrix organization modulates Ret release, with Ret-NLC exhibiting an intermediate release behavior between Ret-NE and Ret-SLN.

Fig. 4.

In vitro delivery performance of Ret-loaded lipid nanoparticles. (A) CLSM images of HaCaT cells treated with Nile Red (NR) and NR-lipid nanoparticles. Scale bar: 5 μm. (B) The bioactive compound release profile was conducted at 37 °C, showing a sustained-release form. (C) Uptake mechanism of Ret-loaded lipid nanoparticles. (D) The cumulative absorption of Ret was determined by combining the permeated and residual amounts. Results are presented as the mean ± SD (n = 3). Statistical significance was determined as *p < 0.05, **p < 0.01, and ***p < 0.001. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

To evaluate the uptake mechanisms of Ret-loaded lipid nanoparticles, cells were pretreated with pharmacological inhibitors and analyzed using flow cytometry (Fig. 4C, Table S4). For NE, caveolae- (∼24%), ATP- (∼26%), and cholesterol- (∼35%)-mediated uptake pathways were the most prominently inhibited. The pronounced inhibition of NE cellular uptake following caveolae/lipid raft disruption is consistent with the well-established cholesterol-sensitive nature of this endocytic pathway (Mayor and Pagano, 2007). In contrast, owing to their more rigid and structured lipid matrices, NLC and SLN may influence the cellular uptake behavior, potentially leading to the involvement of different endocytic pathways (Izza et al., 2022; Gurnani et al., 2022). For NLC, caveolae- (∼44%), cholesterol- (∼76%), and ATP- (∼84%)-mediated uptake was identified as the primary internalization mechanism. This indicates that NLC is preferentially internalized through energy-dependent lipid raft/caveolae-associated endocytosis, which is known to facilitate efficient intracellular transport of lipid nanoparticles. Although SLN uses cholesterol- (∼43%) and ATP- (∼59%) dependent pathways, macropinocytosis has also been found to be partially involved (Chai et al., 2016). This may be attributed to the relatively larger particle size of SLN as well as to its rigid solid lipid matrix and particulate nature, which are less conducive to direct membrane interaction and fusion, thereby promoting uptake through endocytic pathways (Balenzano et al., 2025). Instead, SLN is likely recognized as a particulate entity, triggering actin-driven membrane ruffling and subsequent internalization through macropinocytosis (Chai et al., 2016). Collectively, these observations revealed that the mechanical and structural states of the lipid matrix modulate the balance of endocytic pathways.

In vitro skin penetration test was conducted to evaluate the effect of the SL ratio on formulation-dependent Ret recovery after 24 h (Fig. 4D). Strat-M has been used as a synthetic membrane model for comparative assessment of formulation-dependent permeation behavior (Haq et al., 2018). Combined Ret recovery, calculated as the sum of receptor-phase Ret and residual Ret recovered from the membrane, increased as the SL ratio increased from 0:10 to 6:4 and subsequently decreased at higher solid-lipid ratios (8:2 and 10:0). Ret-NLC (6:4) showed the highest combined Ret recovery among the tested formulations and an approximately 4.5-fold increase relative to free Ret. This non-monotonic dependence on the SL ratio reveals that lipid matrix composition influenced Ret disposition in the comparative Strat-M assay. Notably, this assay was used as one functional screening endpoint instead of as a direct measure of Ret deposition within skin layers or systemic transdermal delivery. Based on the overall balance of physicochemical characteristics, storage behavior, release behavior, and the Strat-M assay results, Ret-NLC (6:4) was selected for subsequent biological evaluation.

3.5. Antioxidant and anti-aging effects of Ret-loaded lipid nanoparticles

Ret metabolism to retinoic acid in cells subsequently activates RAR/RXR signaling, leading to the upregulation of antioxidant gene expression (Fisher and Voorhees, 1996). To evaluate the antioxidant effects of the formulations, intracellular ROS levels were measured using the DCFH-DA assay, in which fluorescence intensity corresponds to intracellular ROS generation (Murphy et al., 2022). A UVB dose of 30 mJ/cm², which minimally impacted cell viability, was selected for subsequent experiments (Fig. S6). The cells were then stained with DCFH-DA, followed by fluorescence microscopy and confocal imaging to quantitatively and qualitatively assess ROS scavenging, respectively. Confocal imaging confirmed that ROS generation was reduced in the groups treated with Ret or Ret-loaded lipid nanoparticles compared with UVB-treated HaCaT cells (Fig. 5A). UVB irradiation at 30 mJ/cm² significantly elevated DCF fluorescence compared to that of the untreated control (***p < 0.001), as quantified through ImageJ analysis of fluorescence microscopy images (Fig. 5C, Fig. S7). This is consistent with previous reports that UVB induces the rapid generation of superoxide anions, hydrogen peroxide, and hydroxyl radicals in keratinocytes, contributing to photoaging and photocarcinogenesis (Gokce et al., 2012). Treatment with all Ret formulations reduced DCF fluorescence in UVB-treated HaCaT cells. UVB irradiation alone increased intracellular ROS levels by approximately 3.8-fold compared to the untreated control. In comparison, treatment with free Ret reduced DCF fluorescence by 1.6-fold in the UVB group, whereas Ret-NE and Ret-SLN exhibited a 2.3-fold reduction.

Fig. 5.

Evaluation of oxidative stress and MMP-1 expression in HaCaT cells treated with Ret-loaded lipid nanoparticles. (A) ROS production measured using CLSM. Scale bar: 20 μm. (B) Scheme showing the mechanism of action of Ret. (C) Fluorescence intensity of DCFH fluorescence quantified using ImageJ. (D) Western blot analysis of MMP-1 expression in HaCaT cells irradiated with UVB, (E) followed by Ret and Ret-loaded lipid nanoparticle treatment and quantitative analysis of relative MMP-1 protein expression. The results are presented as the mean ± SD (n = 3). Statistical significance was determined as *p < 0.05, **p < 0.01 and ***p < 0.001.

Notably, Ret-NLC produced the lowest measured DCF fluorescence among the tested formulations, corresponding to an approximately 4.6-fold reduction relative to the UVB-treated group. This favorable antioxidant response may be associated with formulation-dependent differences in Ret incorporation, release behavior, and cellular delivery. These findings revealed that Ret-NLC can provide a favorable intracellular antioxidant response under UVB-induced oxidative stress conditions. The reduction in DCF fluorescence is unlikely to result from the direct interference of Ret with probe chemistry, indicating that Ret mitigates intracellular ROS levels, potentially through its antioxidant activity (Palace et al., 1999). The greatest inhibition of DCF fluorescence was observed for Ret-NLC, followed by Ret-NE and Ret-SLN. This may be attributed to formulation-dependent differences in EE, release kinetics, and cellular delivery, supporting Ret-NLC as the most effective formulation for Ret-mediated antioxidant protection against UVB irradiation among the lipid nanoparticle systems.

UVB exposure accelerates photoaging by activating transcription factors such as AP-1, which upregulate MMP-1, a key collagen-degrading enzyme responsible for dermal extracellular matrix breakdown (Fig. 5B) (Yang et al., 2024). To evaluate the anti-photoaging activity of the Ret formulations, HaCaT cells were treated with Ret-loaded lipid nanoparticles following UVB irradiation. Western blot analysis revealed a 2.8-fold increase in MMP-1 expression in UVB-treated cells compared to untreated controls (Fig. 5D). These findings are consistent with the results of the DCFH-DA assay, indicating that UVB-induced increases in intracellular ROS contribute to the upregulation of MMP-1 expression. The effects of equivalent Ret concentrations delivered through different lipid nanoparticle systems were evaluated. Treatment with free Ret, Ret-NE, Ret-NLC, and Ret-SLN reduced UVB-induced MMP-1 expression by approximately 60.8%, 66.9%, 63.6%, and 61.2%, respectively (Fig. 5E). Although MMP-1 suppression was comparable among the Ret-containing formulations, the order of MMP-1 suppression among the lipid nanoparticle formulations (Ret-NE > Ret-NLC > Ret-SLN) paralleled their relative Ret release profiles, indicating that formulation-dependent release behavior may contribute to the observed biological response. These findings revealed that all Ret-containing formulations retained anti-photoaging activity, with comparable suppression of UVB-induced MMP-1 expression.

3.6. Ret-loaded nanostructured lipid carrier as cream formulation

Several physicochemical parameters including temperature influence the storage stability of emulsified cream formulation (Yang et al., 2023). To evaluate the long-term stability of the Ret-NLC cream formulation, control (blank), Ret, and Ret-NLC creams were stored at 4 and 25 °C for 12 weeks, during which their physicochemical properties were periodically monitored. Digital images obtained after 6 weeks of storage revealed that all cream formulations maintained a homogeneous appearance without noticeable phase separation at both temperatures (Fig. 6A). The Ret-NLC cream also exhibited appropriate viscosity, showing no significant difference from the control formulation. Furthermore, the Ret-NLC cream displayed a favorable spreading ability, forming a smooth and thin layer upon topical application, indicating suitable rheological properties for dermal use. In addition, the degree of phase separation was monitored at specific intervals during 12 weeks of storage period and was systematically divided into three grades (Table S5). At 25 °C, a typical storage condition for cosmetic formulations, all cream formulations maintained stability without detectable changes. In contrast, minor visual changes appeared after 4 weeks of storage at 4 °C, with no significant differences observed. The pH of emulsions can significantly influence their stability through various factors such as surface charge, emulsifier solubility, and interfacial tension (Xiao et al., 2025). Furthermore, deviation from the slightly acidic physiological pH range of the skin can increase the risk of irritation by disrupting the epidermal barrier and elevating transepidermal water loss (TEWL) (Rajkumar et al., 2023). During the storage period, the pH of all cream formulations remained within the mildly acidic range and exhibited no significant changes (Fig. 6B). This indicated that the Ret-NLC cream formulation exhibited satisfactory long-term stability and had no measurable effect on the physiological pH of the skin upon application. In general, lower viscosity facilitates better spreadability during application, whereas higher viscosity at rest prevents phase separation during storage. Therefore, the shear-thinning behavior, which is commonly observed in cosmetic creams, is considered highly favorable (Franco-Gil et al., 2024). Rheological analysis of the formulations revealed that all creams exhibited typical shear-thinning behavior immediately after preparation, which was maintained even after 12 weeks of storage (Fig. 6C). In addition, no significant changes in viscosity were observed over time, indicating that the Ret-NLC cream formulation was suitable for long-term storage and cosmetic use.

Fig. 6.

Physicochemical characterization and antioxidant efficacy of a Ret-NLC-loaded cream formulation. (A) Long-term stability of Ret-NLC cream formulation during a 12-week period at 4 and 25 °C observed by digital images. (B) pH of Ret-NLC cream formulation during a 12-week period at 4 and 25 °C. (C) Viscosity of Ret-NLC cream formulation immediately after preparation and after 12 weeks of storage. (D) Long-term stability of Ret-NLC cream formulation after 30 weeks at 45 °C/70% RH. (E) DPPH radical scavenging activity of Ret-NLC cream formulation after 30 weeks. The results are presented as the mean ± SD (n = 3).

To further evaluate the stability of cream formulations, long-term stability tests were conducted under harsh conditions. Previous long-term stability studies have reported formulation-dependent degradation of retinoids, with significant declines observed under high-temperature conditions (Temova Rakuša et al., 2021). When the formulations were stored under harsh conditions of 45 °C and 70% relative humidity for 2 weeks, all sample groups maintained their stability. However, after prolonged storage for 30 weeks, phase separation was observed in the control and Ret creams, indicating destabilization of the emulsion system (Fig. 6D). In comparison, the Ret-NLC cream formulation maintained its structural integrity without visible phase separation, preserving a stable oil–water interface throughout the storage period. This instability observed in the control and Ret cream formulations is consistent with the inherent tendency of conventional emulsions to undergo droplet coalescence and interfacial disruption over time (Bibette et al., 1999). The enhanced stability of the Ret-NLC formulation may be attributed to the solid-liquid lipid matrix, which physically restricts movement of the dispersed phase and minimizes interfacial remodeling (Lee et al., 2025). Furthermore, the mixed lipid matrix likely provides a hydrophobic microenvironment that limits oxygen diffusion and molecular mobility, thereby reducing Ret degradation and enhancing matrix-mediated stabilization (Müller et al., 2002).

To assess the retention of antioxidant activity of Ret and Ret-NLC incorporated into cream formulations, a DPPH radical scavenging assay was performed. The antioxidant activities of Ret and Ret-NLC creams were comparatively evaluated to determine whether the cream formulation affected the efficacy of Ret (Fig. 6E). Although the DPPH radical-scavenging activities of the Ret and Ret-NLC creams were comparable, the Ret-NLC cream retained substantial antioxidant activity and exhibited superior physical stability under harsh storage conditions. These results revealed that the NLC matrix may contribute to maintaining the Ret-associated antioxidant activity in the cream formulation, which is consistent with previous studies on retinoid-loaded NLC systems (Ghate et al., 2016). The mixed solid–liquid lipid matrix provides a hydrophobic microenvironment that limits oxygen diffusion and molecular mobility, thereby reducing Ret degradation and contributing to matrix-mediated stabilization.

4. Conclusions

In this study, we systematically revealed that lipid matrix composition is a key determinant of the physicochemical performance, colloidal stability, release behavior, and in vitro delivery performance of Ret-loaded lipid nanoparticles. Among the tested NLC compositions, Ret-NLC with an SL ratio of 6:4 was selected as the preferred formulation based on its overall balance of physicochemical and delivery characteristics; it exhibited a nanoscale particle size of approximately 200 nm and an acceptable size distribution (PDI ≈ 0.3). Ret-NLC (6:4) maintained acceptable dispersion characteristics during 12 weeks of storage without undergoing progressive macroscopic destabilization. In vitro release studies revealed a characteristic biphasic release profile consisting of an initial burst release followed by sustained release, with cumulative bioactive compound release following the order Ret-NE > Ret-NLC > Ret-SLN. Notably, in the comparative Strat-M assay, Ret-NLC (6:4) showed the highest combined Ret recovery among the tested SL ratios, with an approximately 4.5-fold increase relative to free Ret. This non-monotonic response further revealed that lipid matrix composition influenced Ret recovery and membrane transport behavior in the skin-mimetic screening model. In UVB-irradiated HaCaT cells, Ret-NLC (6:4) produced the lowest intracellular ROS levels among the tested formulations, whereas MMP-1 suppression was comparable among the Ret-containing formulations.

Collectively, the present study provides several key findings: (1) Ret-NLC maintained acceptable dispersion characteristics during 12 weeks of storage without progressive macroscopic destabilization. (2) In the comparative Strat-M assay, Ret-NLC showed the highest combined Ret recovery among the tested SL ratios, approximately 4.5-fold higher than free Ret. (3) Ret-NLC exhibited a distinct energy-dependent lipid raft/caveolae-associated cellular uptake profile, highlighting the influence of lipid matrix organization on cellular internalization. (4) Ret-NLC produced the strongest intracellular antioxidant response among the tested lipid nanoparticle systems, whereas MMP-1 suppression remained comparable among the Ret-containing formulations. (5) Following its selection based on the preceding evaluations, Ret-NLC was successfully translated into a cream formulation, maintaining acceptable physicochemical properties, physical stability, and Ret-associated antioxidant activity. This nanoplatform provides a novel avenue for developing effective and reliable anti-aging formulations in pharmaceutical fields.

CRediT authorship contribution statement

Hae Kyung Shin: Writing – original draft, Visualization, Formal analysis, Data curation. Eun Yong Lee: Writing – original draft, Visualization, Investigation. Seunghyeon Lee: Writing – review & editing, Visualization, Methodology, Data curation. Chaelin Min: Writing – review & editing, Visualization, Investigation, Data curation. Minyoung Yi: Methodology, Data curation. Sung Yeon Cho: Visualization, Methodology. Sang Hyeon Bok: Methodology, Data curation. Young Hoon Roh: Writing – review & editing, Supervision, Project administration, Funding acquisition, Conceptualization.

Funding

This research was supported by the National Research Foundation (NRF) funded by the Korean government (MSIT) (No. RS-2025-16071710). This research was supported by the Yonsei University Research Fund of 2024 (2024-22-0543). The work was supported in part by Brain Korea 21 (BK21) FOUR program.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary material

Data availability

Data will be made available on request.