Development and characterization of a topical ferrochelatase inhibitor nanoemulsion for choroidal neovascularization therapy

Abstract

Choroidal neovascularization (CNV) is a hallmark of neovascular age-related macular degeneration (nAMD). We previously identified the heme synthesis enzyme ferrochelatase (FECH) as a promising therapeutic target. This study aimed to develop and characterize topical ophthalmic nanoemulsions (NEs) of a novel FECH inhibitor, SH-17023, for CNV therapy, to avoid the intravitreal injections needed for standard-of-care anti–vascular endothelial growth factor (anti-VEGF) biologics. SH-17023-loaded NEs were prepared by spontaneous emulsification and optimized using D-optimal mixture design-based Quality-by-Design to obtain nanometric globule size (Z_avg), low polydispersity index (PDI), and highest drug loading capacity (% LC). Therapeutic efficacy was assessed in the laser-induced CNV (L-CNV) mouse model with fundus imaging, optical coherence tomography (OCT), fluorescein angiography, and ex vivo vasculature staining. The optimized formulation was transparent with a globule size of 32.832 ± 2.125 nm, PDI 0.201 ± 0.003, spherical morphology, and LC of 7.436 ± 0.035%. It showed zeta potential of −29.2 ± 0.45 mV, sustained drug release, and robust accelerated and kinetic stability. Attenuated total reflectance-Fourier transform infrared spectroscopy and X-ray diffraction revealed excellent drug–excipient compatibility. Ex vivo goat cornea permeation of NEs showed significantly higher drug transport (permeability coefficient (K_p) = 0.050 ± 0.012 h⁻¹.cm⁻², steady-state flux (J_ss) = 37.51 ± 9.29 μg.h⁻¹.cm⁻²) than the drug in solution (K_p = 0.002 ± 0.0004 h⁻¹.cm⁻² and J_ss = 1.575 ± 0.488 μg.h⁻¹.cm⁻²) with an enhancement ratio of 23.815. Optimized NEs significantly reduced L-CNV assessed by OCT and ex vivo staining by >45% compared to blank NEs, without ocular irritation or toxicity, indicating their promise for nAMD therapy.

Graphical abstract

Highlights

• •Ferrochelatase inhibitors are promising antiangiogenic agents for ocular disease.
• •We made topical ophthalmic nanoemulsions (NE) of a novel ferrochelatase inhibitor.
• •We used QbD-based development, with optimization of NEs by D-optimal mixture design.
• •An optimized NE had excellent ocular tolerability and ex-vivo ocular permeability.
• •The NE was therapeutic in a murine laser-induced choroidal neovascularization model.

1. Introduction

Age-related macular degeneration (AMD) is a complex degenerative disease causing irreversible central vision loss, mainly in persons 60 years or above (Wong et al., 2014). Worldwide, about 200 million people suffer from some form of AMD with an anticipated incidence of 288 million by 2040 (Fleckenstein et al., 2024; Wong et al., 2014). AMD mainly affects the macula of the eye, causing progressive damage to visual ability. There are two major types of AMD: dry AMD (degenerative) and wet AMD (exudative AMD or neovascular AMD/nAMD). nAMD is less common (10% of cases) compared to dry AMD (90%), but is responsible for 90% of blindness caused by AMD (Holekamp, 2019). The nAMD involves macular neovascularization (MNV), with new blood vessels developing from the choroid (choroidal neovascularization, CNV) or the deep retinal vasculature to the retinal pigment epithelium (RPE) or through the RPE to the subretinal region. The development and progression of this abnormal vasculature results in hemorrhage, exudations, scarring, and RPE detachment, leading to rapid vision loss (Hobbs et al., 2024).

Intravitreal injections of anti-vascular endothelial growth factor (anti-VEGF) drugs, such as faricimab, bevacizumab, brolucizumab, ranibizumab, conbercept, and aflibercept, are the current standard of care for the treatment of nAMD (Fei et al., 2024; Muniyandi et al., 2025; Sarkar and Dyawanapelly, 2021). These anti-VEGF therapies have a high cost, show limited efficacy in many patients, can lead to tachyphylaxis, have a risk of ocular and systemic side effects, and require frequent intravitreal injections with significant patient and health system burden and access challenges (Fei et al., 2024; Muniyandi et al., 2025). Our survey study revealed that 80% of patients with ocular neovascular disease would prefer topical or oral therapies if available (Jacobs et al., 2021). Hence, identification of alternative molecular targets and therapeutics with suitable delivery systems is needed for effective treatment of nAMD.

We previously identified ferrochelatase (FECH), a key enzyme in heme biosynthesis, as a promising alternative antiangiogenic target (Babu et al., 2020; Basavarajappa et al., 2017; Sishtla et al., 2022). FECH is a mediator of neovascularization in the eyes and is upregulated in murine and human eyes undergoing CNV (Basavarajappa et al., 2017). FECH catalyzes the insertion of ferrous ion into protoporphyrin IX (PPIX) in the final step of heme biosynthesis (Hamza and Dailey, 2012). The protoheme serves as a cofactor for proteins important for angiogenesis, including mitochondrial complex IV and endothelial nitric oxide synthase, thereby promoting angiogenesis (Basavarajappa et al., 2017; Shetty et al., 2020). Mutation or knockdown of FECH blocks angiogenesis in vitro and in vivo (Basavarajappa et al., 2017). N-methylprotoporphyrin (NMPP) is a potent inhibitor of FECH (Dailey and Fleming, 1983) and the antifungal drug griseofulvin produces NMPP in vivo as an off-target effect via alkylation of the heme group of cytochrome P450 (Liu et al., 2015), hence both have been used in preclinical studies for the inhibition of FECH. Both NMPP and griseofulvin selectively block the growth of endothelial cells without causing death in vitro. The intravitreal and oral administration of griseofulvin reduces CNV (Basavarajappa et al., 2017; Chobisa et al., 2024). However, NMPP does not have properties required for drug development due to its bulkiness and poor solubility. Further, the unusual mechanism of griseofulvin hinders the development of novel griseofulvin analogues for FECH inhibition (Sishtla et al., 2022). We previously conducted a high-throughput screen followed by structure-activity relationship studies to identify small-molecule inhibitors of FECH. The study identified triazolopyrimidinone derivatives having strong activity for FECH inhibition, notably a potent compound SH-17023 (Sishtla et al., 2022). SH-17023 inhibits FECH and causes precursor (PPIX) accumulation in human retinal endothelial cells (HRECs) (Sishtla et al., 2022). It demonstrated in vitro anti-angiogenic activity, inhibiting HREC proliferation, tube formation, and migration and showed in vivo therapeutic activity against ocular neovascularization in the laser-induced choroidal neovascularization (L-CNV) model when delivered intravitreally as a proof of concept (Sishtla et al., 2022). Given the shortcomings of this delivery route, topical administration is appealing. However, the topical administration of SH-17023 is greatly limited due to its poor aqueous solubility (log P = 5.28). Hence, a suitable formulation is needed for effective delivery to treat CNV.

Topical administration offers pain-free administration, improved patient adherence, low cost, site-specific drug delivery, and ultimately reduces side effects (Mehrandish and Mirzaeei, 2021; Yang et al., 2022). However, topical drug delivery through traditional dosage forms (eye drop, drug solution, ointments, gels, and suspensions) faces several challenges, such as limited drug permeation across complex ocular barriers (pre-corneal, corneal, and conjunctival barriers) and physiological events (induced lacrimation, tear turnover, conjunctival absorption, drug metabolism, drainage, efflux etc.) leading to poor ocular bioavailability (Henostroza et al., 2020; Mohapatra et al., 2025). Further, they offer pulsed delivery (an initial high concentration during instillation followed by rapid decline due to drainage), failing to provide sustained drug release or targeted delivery, leading to less than ideal therapeutic outcomes (Dhaval et al., 2020; Mohapatra et al., 2025).

Advanced therapeutic platforms use novel drug-delivery systems (NDDS) for ocular disease therapy. Among various NDDS, nanoemulsions (NEs) are identified as a promising formulation for effective ocular permeation (Mohapatra et al., 2025). NEs, also called miniemulsions, are isotropic colloidal dispersions having two immiscible phases (oil and water) emulsified by surfactants/co-surfactants or co-solvents having a globule size of 20–500 nm (Fernandes et al., 2021; Singh et al., 2020; Smail et al., 2021). Due to their nanometric globule size, they appear as a transparent or translucent dispersion and offer long-term stability (Smail et al., 2021). Based on the dispersion/external medium and dispersion/internal phase, they are classified into w/o, o/w, and bi-continuous types (Choradiya and Patil, 2021; Singh et al., 2020). Among them, the o/w type can be used for ocular drug delivery due to the aqueous external phase, which can interact with aqueous tear fluid, and the internal oil phase solubilizes and loads hydrophobic drugs, thereby allowing prolonged ocular retention and improved ocular permeation (Fernandes et al., 2021; Singh et al., 2020). The oily excipients of NEs interact with the lipid layer of the tear film, offering prolonged ocular retention and serving as drug depots (Singh et al., 2020). NEs enhance the aqueous solubility of hydrophobic drugs, reduce contact angles, provide improved contact with ocular tissue, prolong ocular retention, reduce dosing frequency, offer sustained/controlled drug release, improve ocular permeation, and enhance therapeutic outcomes in pre-clinical and clinical contexts (Attia et al., 2024; Choradiya and Patil, 2021; Fernandes et al., 2021; Kishore et al., 2024; Singh et al., 2020).

In this work, ophthalmic NEs of the novel FECH inhibitor, SH-17023, were developed for CNV therapy. The NEs were formulated by spontaneous emulsification and optimized using a D-optimal mixture design-based statistical Quality by Design (QbD) approach to obtain a nanometric globule size, low polydispersity index, and the highest drug loading capacity. Further, the optimized NEs were characterized for physicochemical and pharmaceutical properties as well as ex vivo ocular permeability, and in vivo irritancy. Therapeutic efficacy was shown in the L-CNV murine model.

2. Materials and Methods

2.1. Materials

Active pharmaceutical ingredient (API): SH-17023 (a novel FECH inhibitor) was synthesized by WuXi AppTec, Hong Kong following our previously-described protocol (Sishtla et al., 2022).

Internal standard: The certified reference material midazolam was obtained from Cerilliant®, Sigma-Aldrich, USA.

Oils: Corn oil and safflower oil were obtained from Sigma-Aldrich, USA. The lavender oil, lemon oil, and sesame oil were obtained from Bush, Canada. Peppermint oil, sunflower oil, castor oil, orange oil, eucalyptus oil, anise oil, almond oil, rose oil, emu oil, and bergamot oil were obtained from Medisca Inc., Canada. Avocado oil was obtained from A.Vogel, Canada; citronella oil was obtained from Fern & Petal, Canada. Jojoba oil, green tea oil, cinnamon oil, clove oil, and pine oil were obtained from New Directions Aromatics, Canada.

Surfactants and co-surfactants: Tween® 20 (polyoxyethylene (20) sorbitan monolaurate), Tween® 40 (polyoxyethylene (40) sorbitan monolaurate), Tween® 80 (polyoxyethylene (80) sorbitan monooleate), and Span® 80 (sorbitan monooleate) were obtained from Sigma-Aldrich, USA. Labrafac™MC60 (glycerol monocaprylocaprate, liquid, Hydrophilic-Lipophilic Balance (HLB): 5, water insoluble surfactant), Labrafil®M 1944 CS (oleoyl polyoxyl-6 glycerides, HLB: 9), Labrasol® (caprylocaproyl polyoxyl-8 glycerides, HLB: 12), Capryol®90 (propylene glycol monocaprylate, oily liquid, HLB: 5, nonionic water-insoluble co-surfactant), Capryol®PGMC (propylene glycol monocaprylate, oily liquid, HLB: 6, nonionic water-insoluble surfactant used as cosurfactant) were received from Gattefosse, Canada as gift samples.

Solvents and co-solvents: Propylene glycol (PG), polyethylene glycol 300 (PEG 300), and polyethylene glycol 400 (PEG 400) were obtained from TCI, USA. Transcutol® P (diethylene glycol monoethyl ether) was provided as a gift sample by Gattefosse, Canada. Glycerol was obtained from Sigma-Aldrich, USA. Optima™ LC-MS grade acetonitrile, methanol, & water were obtained from Fisher Scientific, USA. Millipore and other analytical-grade ultrapure solvents were used throughout the work as per the requirements.

Consumables: Polyvinylidene fluoride (PVDF) sterile syringe filter (0.22 μm, Milex®-GV) was procured from Sigma Aldrich, USA, and used for sterilization of formulation and filtration of samples for LC-MS analysis. A carbon-coated copper grid for HRTEM analysis was purchased from Ted Pella, USA. Cellulose dialysis tubing (MWCO 12–14 kDa) was obtained from Bartovation, USA.

Biologicals and animals: Freshly excised goat eyes were collected from a local slaughterhouse (Iqbal Foods), Canada. C57BL/6J mice of both sexes were procured from Jackson Laboratory, USA.

Reagents and Chemicals: Sodium lauryl sulfate (SLS), 4′,6-diamidino-2-phenylindole (DAPI), sodium azide, sodium bicarbonate (NaHCO₃), calcium chloride (CaCl₂), and potassium chloride (KCl) were purchased from Sigma Aldrich, USA. Sodium hydroxide (NaOH) was purchased from Caledon Laboratories Ltd., Canada. Sodium chloride (NaCl) was purchased from BioShop Canada Inc., Canada. Isotonic saline was purchased from Fresenius Kabi Canada Ltd., Canada, and fluorescein sodium was procured from Chem-Impex International, USA. Ketamine hydrochloride injection (Narketan, 100 mg/mL) and xylazine (Rompum, 20 mg/mL) were obtained from Vetoquinol SA, France. Tropicamide eye drop (Alcon Mydriacyl Solution) was obtained from Alcon Laboratories, Inc., USA, and lubricating carbomer gel (Optixcare Eye Lube Lubricating gel) was obtained from Aventix Animal Health Corp., Canada. Atipamezole was purchased from Cayman Chemicals, USA. Fluoromount-G mounting medium was procured from SouthernBiotech, USA. Biotin-conjugated Griffonia simplicifolia-isolectin B4 (GS-IB4) was procured from Invitrogen, USA, and DyLight 488-conjugated streptavidin was procured from Invitrogen, USA. Paraformaldehyde was obtained from Thermo Fisher Scientific, USA. Bovine serum albumin (BSA) was procured from Bioshop Inc., Canada. Triton X-100 was purchased from Promega Corporation, USA.

2.2. Liquid chromatography-Mass spectrometry (LC-MS) and validation

The LC-MS analysis was performed on an Agilent LC system (Agilent 1260 Infinity, Agilent, USA) furnished with an autosampler (1260 HIP ALS), thermostatically controlled column chamber (1260 TCC), binary pump (1260 Bin pump), degasser (1260 HiP degasser), reverse phase column (Poroshell 120, EC-C18, 2.1 × 50 mm, 2.7 μm particle size), with triple-quadrupole mass spectrometer (Thermo Scientific™ TSQ Endura™), and Thermo Scientific Xcalibur software (4.4). The column was kept in a thermostatically controlled chamber with a temperature of 40 °C. The gradient elution was carried out with Optima™ LC/MS Grade water (Fisher Scientific) containing water as mobile phase A (A) and Optima™ LC-MS grade acetonitrile (Fisher Scientific) as mobile phase B (B) at a flow rate of 0.3 mL/min. For the gradient elution, the mobile phase was 90% A and 10% B (0.00–1.30 min), followed by a linear gradient of 80% A and 20% B (1.30–2.0 min), and switch to 10% A and 90% B (2–6 min), then returned to 80% A and 20% B (7.0–7.5 min), followed by the equilibrium gradient of 90% A and 10% B (7.5–10.0 min) (Supplementary file Section 1.1, Table S1, and Fig. S1. A triple-quadrupole mass spectrometer (Thermo Scientific™ TSQ Endura™) equipped with Heated-electrospray ionization (H-ESI) was used for the MS analysis. Then, the data were acquired in positive ionization mode with selected reaction monitoring (SRM) mode at the precursor-to-product ion transitions (Q1/Q3) for SH-17023 and midazolam (as internal standard) at m/z of 425.2/281.1 and 326.1/290.5, respectively.

Working standards (1–1000 ng/mL) of SH-17023 containing 2 ng/mL of midazolam (as internal standard) were prepared in LC-MS grade methanol. The sample injector speed was maintained at a drawing and ejecting speed of 200 μL/min. Five microliters of each working standard were injected into the LC-MS system using an autosampler, and the data were analyzed using Thermo Xcalibur Qual Browser software. The absence of any carryover effect was ensured by injecting blank samples (LC-MS grade methanol, Optima™, Fisher Scientific) intermittently after every two injections. The LC-MS analysis was carried out in triplicate, and the calibration curve was made using the area of the analyte normalized to the area of the internal standard vs. the concentrations (ng/mL). The optimized LC-MS method was validated as per International Council for Harmonisation (ICH) guidelines for linearity and range, accuracy, precision, limit of detection (LOD), limit of quantification (LOQ), and robustness. The detailed LC-MS methodology, including estimation of mass of precursor to product ion, preparation of calibration standards, and validation as per ICH guidelines are presented in the supplementary file (Section 1.1.1). The validated LC-MS method was used for the solubility study, estimation of drug content, loading capacity, stability, release study and ocular permeation studies.

2.3. Screening of oil, surfactant, co-surfactant, and co-solvent by solubility study

The solubility of SH-17023 in various excipients (oils, co-surfactants, surfactants, and co-solvents) was evaluated by saturation solubility study with the use of a validated LC-MS method. Various oils (almond oil, anise oil, avocado oil, bergamot oil, castor oil, cinnamon oil, citronella oil, clove oil, corn oil, emu oil, eucalyptus oil, green tea oil, jojoba oil, lavender oil, lemon oil, orange oil, peppermint oil, pine oil, rose oil, safflower oil, sesame oil, sunflower oil), surfactants & co-surfactants (Caproyl®90, Caproyl® PGMC, Labrafac™ MC60, Labrafil® M 1944 CS, Labrasol®, Span® 80, Tween® 40, Tween® 20, Tween® 80), and co-solvents (glycerol, propylene glycol, PEG 400, PEG 300, Transcutol® P) were used for solubility determination. Briefly, an excess amount (5 mg) of SH-17023 was added to 50 μL of various oils, surfactants, co-surfactants, and co-solvents, vortexed (BV 1000 vortex mixer, Benchmark Scientific Inc., USA), bath sonicated for 10 min (Ultrasonic bath, Fisher Scientific, USA), and placed in a thermal mixer (Thermo Scientific, USA) at 37 ± 1 °C with 250 rpm for 24 h. Then the samples were centrifuged (Eppendorf 5417R, USA) at 10,621 ×g for 10 min, and 10 μL of the supernatant was collected. The collected supernatants were extracted and suitably diluted with LC-MS grade methanol, centrifuged at 10,621 ×g, filtered via a 0.22 μm syringe filter (Milex®-GV, Sigma Aldrich, USA), and kept in screw capped amber coloured vials for the drug quantification by the validated LC-MS method.

2.4. Construction of pseudo-ternary phase diagrams

The pseudo-ternary phase diagrams were constructed with the selected oils, surfactants, co-surfactants, and cosolvents having maximum drug solubility using the water titration method (Cardoso et al., 2022). Briefly, various ratios of surfactants or surfactant with cosurfactants (1:1) or surfactant with cosurfactants & co-solvents (1:1:1, S_mix) were mixed with selected oil phase at 1:9, 2:8, 3:7, 4:6, 5:5, 6:4, 7:3, 8:2, and 9:1 (v/v) into transparent glass vials (Attia et al., 2024). Millipore water was mixed dropwise, vortexed until the formation of a homogeneous phase, and visualized for optical clarity after each water addition. The volume of water required to change the clarity/attainment of turbidity was considered the point of water titration, and the samples were kept tightly sealed with Parafilm at 37 °C for 48 h to observe clarity changes. Then the percentage of each phase (%v/v) was calculated at the point of critical change in clarity/turbidity/phase separation. The pseudo-ternary phase diagrams were plotted using Origin Pro 2025 software (Origin Lab Corporation, USA) and the monophasic zones were identified for the development of nanoemulsions.

2.5. Formulation and development of SH-17023-loaded nanoemulsion by D-optimal mixture design

2.5.1. Response surface analysis

The pseudoternary diagrams were used to formulate nanoemulsions with the use of appropriate oil, surfactant/co-surfactant, and water. The impact of each formulation component on the formulation and the optimization was carried out by a quality-by-design approach. The influence of critical material attributes (CMAs) on critical quality attributes (CQAs) of the NE was investigated by response surface methodology (RSM). The % v/v of rose oil (X₁), surfactant-cosurfactant-cosolvent mixture or S_mix (Labrasol® + Tween® 80 + Transcutol® P) (X₂), and water (X₃) were considered as CMAs, and their impact on CQAs, such as hydrodynamic globule size (Y₁ = Z_avg), polydispersity index (Y₂ = PDI), and loading capacity (Y₃ = LC) was investigated (Table 1). The response surface analysis and formulation optimization were carried out using a D-optimal mixture design using Design-Expert® software (Stat-Ease Inc., USA). The lower and higher levels of X₁, X₂, and X₃ were chosen from the pseudo-ternary phase plots and pilot formulations.

Table 1.

Coded levels, real values for each factor under experiment, and CQAs.

	Levels
	Coded Values
	Low		High
Independent variables (CMAs)
Rose oil (X1, % v/v)	10	≤	30
Smix (Labrasol® + Tween® 80 + Transcutol® P) (X2, % v/v)	40	≤	80
Water (X3, % v/v)	10	≤	30
Dependent variables (CQAs)		Goal
Hydrodynamic vesicular size (Y1 = Zavg = nm)	To minimize
Polydispersity index (Y2 = PDI)	To minimize
Loading capacity (Y3 = %)	To maximize

Using the D-optimal mixture design, 16 batches of nanoemulsion were prepared. Briefly, the oil phase was mixed with the surfactant and vortexed for complete solubilization. Then the exact amount of water, as per the design of experiments (DOE) batch (Table 2, D-optimal mixture design), was added and vortexed at room temperature to develop an NE.

Table 2.

D-optimal mixture design-based batches of NE formulations with their corresponding responses.

	Factor 1 (X1)	Factor 2 (X2)	Factor 3 (X3)	Response 1 (Y1)	Response 2 (Y2)	Response 3 (Y3)
Formulation	A: Rose oil	B: Smix	C: Water	Vesicle Size (Zavg)	Polydispersity index (PDI)	Loading Capacity
	% v/v	% v/v	% v/v	nm	–
1	30	60	10	38.744 ± 0.1749	0.311 ± 0.0012	6.300 ± 0.0816
2	17.4142	52.5858	30	35.544 ± 0.1029	0.163 ± 0.0012	5.357 ± 0.0422
3	10	65	25	30.888 ± 0.1640	0.190 ± 0.0008	6.448 ± 0.1224
4	30	40	30	52.300 ± 0.1414	0.412 ± 0.0014	4.505 ± 0.0820
5	30	52.2731	17.7269	43.311 ± 0.1640	0.294 ± 0.0012	5.469 ± 0.0495
6	25.7562	47.8452	26.3985	39.177 ± 0.1368	0.218 ± 0.0009	5.302 ± 0.0817
7	30	60	10	38.513 ± 0.1643	0.309 ± 0.0012	6.400 ± 0.0816
8	21.5526	59.3152	19.1322	36.086 ± 0.0837	0.179 ± 0.0009	6.313 ± 0.0837
9	20.493	69.507	10	33.753 ± 0.2041	0.199 ± 0.0012	7.259 ± 0.1231
10	11.4118	58.5882	30	32.783 ± 0.1649	0.192 ± 0.0009	6.221 ± 0.0872
11	11.4118	58.5882	30	33.211 ± 0.2114	0.199 ± 0.0008	6.128 ± 0.0908
12	30	40	30	53.089 ± 0.2061	0.401 ± 0.0009	4.358 ± 0.1230
13	20.493	69.507	10	34.110 ± 0.1639	0.197 ± 0.0009	7.325 ± 0.0891
14	12.2338	71.5354	16.2309	38.400 ± 0.1632	0.221 ± 0.0009	7.025 ± 0.1273
15	10	80	10	55.488 ± 0.2114	0.373 ± 0.0014	8.179 ± 0.1295
16	10	80	10	55.744 ± 0.2042	0.388 ± 0.0008	8.360 ± 0.1234

The dependent variables (Z_avg, PDI, and LC) of each batch were measured in triplicate, and the mean outcomes were studied by the DOE software using ANOVA to investigate model significance (p < 0.005) and adequacy. The model parameters (model p-value, model F-value, lack of fit F-value, lack of fit p-value, the difference between adjusted and predicted R² value, correlation coefficient (R²), and adequate precision) were investigated for adequacy of the used DOE model. The goodness of fit of the experimental design was investigated by diagnostic plots. To predict the best NEs, the statistical model was fitted to various models (linear, cubic, quadratic, special quadratic, and special cubic). The impact of each factor on the selected responses was analyzed using 2D and 3D response surface plots.

2.5.2. Optimization and checkpoint analysis

Through numerical and graphical optimization methods, the best CMAs required for the formulation of the desired nanoemulsion with low Z_avg, PDI, and high LC were obtained. By formulating 3 checkpoint batches, the predictability of the D-optimal mixture design was validated using Eq.1 (Kassem et al., 2022).

$$\%P\mathit{rediction}\mathit{error}=\frac{\mathit{Experimental}\mathit{value}-P\mathit{redicted}\mathit{value}}{\mathit{Predicted\; value}}\times 100$$ (1)

2.6. Characterization of NE

The optimized NE was further characterized for the following parameters.

2.6.1. Organoleptic properties

The optimized NE was observed with the naked eye for the signs of transparency, clarity, precipitation, phase separation, and colour.

2.6.2. pH

The pH of the optimized NE was evaluated using a calibrated pH meter (AB15 pH meter, Fisher Scientific, USA) at room temperature to ensure the ocular compatibility (Ismail et al., 2020).

2.6.3. Conductivity

Conductivity of the NEs was measured at room temperature using a calibrated (with a standard KCl solution having the conductivity of 1413 μS/cm) digital conductivity meter (PC 700, Eutech Instrument, Singapore).

2.6.4. Viscosity

The viscosity of NEs was measured at 25 °C by a parallel plate Rheometer (MCR 72, EDU Edition, Anton Paar, Austria) furnished with RheoCompass™ software using a 25 mm spindle. The viscosity was measured in triplicate. After each viscosity measurement, the plate was cleaned and dried.

2.6.5. Percentage transmittance

The %T was measured by UV–visible spectroscopy using a Multi-Mode Reader (BioTek Synergy H1 Hybrid, Agilent, USA) at 600 nm using water as a blank. The undiluted NE was taken in a 96-well plate, and the %T was measured in triplicate.

2.6.6. Refractive index and isotropy

The RI of the optimized NE was estimated using an Abbe digital refractometer at 25 ± 1 °C (Ismail et al., 2020). Briefly, 3 to 4 drops of undiluted NE were spread uniformly on a prism, and the RI was noted. The isotropy of the NE was verified by measuring and comparing the RI values of 3 samples collected from different locations of the NE.

2.6.7. Dilutability

The dilutability of the optimized NE was estimated by diluting with water, PBS (pH = 7.4), and simulated tear fluid (0.22 g NaHCO₃, 0.68 g NaCl, 0.14 g KCl, 0.008 g CaCl₂·2H₂O, and water to 100 mL) at multiple dilutions (1:10, 1:50, 1:100, and 1:1000 times v/v). The samples were stored at room temperature for 24 h and examined for the signs of precipitation, clarity, and phase separation (Kassem et al., 2022).

2.6.8. Globule size, polydispersity, and zeta potential

The globule size (Z_avg), polydispersity index (PDI), and zeta potential (ζ) of the optimized NE were evaluated by a dynamic light scattering (DLS) instrument (Zetasizer Nano series, Malvern Panalytical Ltd., UK) at 25 °C with backscattering of 173° and equilibrium time of 120 s (Ismail et al., 2020). Polystyrene disposable cuvette (DTS0012, Malvern) was used for Z_avg and PDI, and disposable folded capillary cells (DTS1070, Malvern) were used for ζ measurement. The NEs were diluted ∼670 times with HPLC-grade water, filtered through a 0.22 μm polyvinylidene fluoride (PVDF) syringe filter (Milex®-GV, Sigma-Aldrich, USA), and analyzed by the instrument in triplicate.

2.6.9. High-resolution transmission electron microscopy (HRTEM) and selected area electron diffraction (SAED)

HRTEM analysis was performed to examine the morphology of the optimized NEs, and SAED was used to evaluate their crystallinity. Briefly, the optimized NEs were diluted 100 times with HPLC-grade water, and TEM samples were prepared by the drop-casting method. Briefly, 20 μL of the diluted NEs was dropped onto a 400-mesh carbon-coated copper grid (Ted Pella, USA), and the excess sample was absorbed with filter paper and dried overnight at room temperature to remove traces of water. Then the sample was subjected to the HRTEM instrument (TECNAI G²20, FEI, USA) having a high-angle annular dark field detector for morphology and crystallinity analysis at 200 kV. Then the average globule size and size distribution were analyzed from the HRTEM photomicrographs using ImageJ software (National Institutes of Health (NIH), USA).

2.6.10. Drug content

The drug content of the optimized NEs was evaluated using the validated LC-MS method. Briefly, the optimized NEs were diluted with LC-MS-grade methanol for the extraction of the contained drugs, followed by centrifugation at 10,621 ×g for 10 min, filtration of the supernatant (0.22 μm syringe filter), and estimation of the drugs using the LC-MS calibration curve. Then, the drug content was evaluated using Eq.2. (Kassem et al., 2022; Mohapatra et al., 2025).

$$\mathit{Drug\; content}\left(\%\right)=\frac{\mathit{Estimated\; quantity\; of\; drug}}{\mathit{Total\; amount\; of\; drug\; added\; in\; formulation}}\times 100$$ (2)

2.6.11. Entrapment efficiency

For the entrapment efficiency (EE) measurement, the optimized NE was subjected to ultracentrifugation in an Optima L-100 XP (Beckman Coulter) at 100,000 ×g for 30 min at 25 °C. The free drug/unencapsulated drug in the supernatant was separated from drug-loaded NE droplets by the ultracentrifugation. Then, the % EE was calculated using Eq.3 (Mohapatra et al., 2025).

$$\mathit{Entrapment\; efficiency}\left(\mathit{EE}\right)=\frac{\mathit{Am}o\mathit{unt\; of\; loaded\; drug}}{\mathit{Total\; amount\; of\; added\; drug}}\times 100$$ (3)

2.6.12. Loading capacity (LC)

This was estimated by measuring the amount of loaded drug to the total amount of formulation (drug and excipients) and represented by a percentage. The percentage LC was calculated using Eq.4 (Mohapatra et al., 2025).

$$\mathit{Loading\; capacity}\left(\%\right)=\frac{\mathit{Amount\; of\; loaded\; drug}}{\mathit{Total\; amount\; of\; formulation}}\times 100$$ (4)

2.6.13. In vitro drug release and release kinetics

The in vitro drug release of SH-17023 from the optimized NEs and solution was studied using a dialysis bag method. Cellulose dialysis tubing (MWCO 12–14 kDa, Bartovation, USA) was soaked overnight in deionized water to swell the bag and remove the preservatives. Based on preliminary experiments, phosphate-buffered saline (PBS, pH = 7.4) containing 5% sodium lauryl sulfate (SLS) was used as the release medium for the in vitro release study. Accurately, 100 μL of optimized NEs was added to 500 μL of release media, placed in a dialysis bag, hermetically sealed at both ends with threads, and placed vertically in a beaker containing 50 mL of release media. An equivalent quantity of SH-17023 solution in Transcutol P was used as a control for comparison. The receptor medium was maintained at ocular surface temperature of 34 ± 0.5 °C (Kassaee et al., 2023; Mahboobian et al., 2020; Mahran et al., 2021) with continuous stirring at 150 rpm by a temperature-controlled magnetic stirrer (Fisher Scientific, USA). At predetermined time intervals, a 1 mL aliquot was collected from the beaker and immediately replenished with the same volume of fresh blank medium to preserve sink conditions. The release study was continued up to 72 h. The quantity of drug released was quantified by the validated LC-MS method. Results are presented as the cumulative percentage of SH-17023 released over time. The cumulative percentage of drug released was calculated using Eq.5.

$$\mathit{Cumulative\; percentage\; drug}\mathit{rel}\mathrm{e}\mathit{ased}\left(Q\right)=\frac{C_{n}V+\sum _{i=1}^{i=n-1}{C}_i\times V_i}{W_0}\times 100$$ (5)

Where W₀ is the total amount of drug in optimized NEs added in the dialysis bag, C_n is the concentration of drug in the release medium at t_n, V is the total volume of release medium (50 mL), V_i is the aliquoted volume (1 mL) at t_i, C_i is the concentration of drug at t_i (Ma et al., 2022; Shi et al., 2023; Yang et al., 2022; Zhang et al., 2023).

Then the in-vitro drug release data were fitted to various mathematical kinetic models (zero order, first order, Higuchi, Korsmeyer-Peppas, and Hixson-Crowel cube root) to predict the release kinetics of drug release from the optimized NEs (Attia et al., 2024; Kassem et al., 2022). The linear regression analysis was performed to analyze the release data, and the correlation coefficient (R²) in each case was estimated to predict the best-fitting model. The mechanism of drug release was studied by fitting the first 60% of the in-vitro drug release data to the Korsmeyer-Peppas model and estimating the release exponent (n) (Kassem et al., 2022). The following mathematical kinetic models were used.

$$\mathit{Zero}-\mathit{order\; kinetics},C_0-C_t=K_{0}t$$ (6)

Where C₀, Ct, and K₀ are the drug concentration at time (t = 0), drug released at time t, and zero-order release constant.

$$\mathit{First}-\mathit{order\; kinetics},\mathit{LogC}=\mathit{Log}{C}_0-\frac{K_{1}t}{2.303}$$ (7)

Where C is the % drug remaining at time t, C₀ is the initial drug concentration and K₁ is the first-order rate constant.

$$\mathit{Higuchi\; model},Q=K_H{t}^{1/2}$$ (8)

Where Q is the quantity of drug released at time t and K_H is the Higuchi constant.

$$\mathit{Korsmeyer}-\mathit{Peppas\; model},\frac{M_t}{M_{\infty }}=k_{\mathit{kp}}{t}^n$$ (9)

Where M_t is the quantity of drug released at time t, M_∞ is the quantity of drug released at ∞ time, M_t / M_∞ is the fraction of drug released at time t, k_kp is the Korsmeyer rate constant, and n is the diffusional exponent.

$$\mathit{Hixson}-\mathit{Crowel\; cubic\; root}\mathit{law},\sqrt[3]{W_0}-\sqrt[3]{W_t}=K_{\mathit{HC}}t$$ (10)

Where W₀ is the initial amount of drug remaining at time 0, W_t is the amount of drug remaining to be released at time t, and K_HC is the Hixson-Crowell rate constant.

2.6.14. Drug-excipient compatibility

The drug-excipient compatibility was studied by attenuated total reflectance-Fourier transform infrared spectroscopy (ATR-FTIR) analysis (Porto et al., 2024) and X-ray diffraction studies.

2.6.14.1. ATR-FTIR study

Briefly, in the ATR-FTIR analysis, the vibrational spectra of SH-17023, rose oil, Labrasol®, Tween® 80, Transcutol® P, physical mixture, blank/placebo NE, and optimized NE were compared to predict the possible interaction among SH-17023 and excipients of NEs using an ATR-FTIR spectrophotometer (Nicolet™ iS50, Thermo Scientific, USA) equipped with an ATR accessory. Data acquisition was made using OMNIC Paradigm software (Version 2.8 HF2) within the wavenumber range of 4000 to 400 cm⁻¹ with 40 scans, and 4 cm⁻¹ spectral resolution.

2.6.14.2. X-Ray diffraction study

Similarly, the diffraction peaks of SH-17023, rose oil, Labrasol®, Tween® 80, Transcutol® P, physical mixture, blank/placebo NE, and optimized NE were compared to predict the possible interaction among SH-17023 and excipients of NEs using a diffractometer (Rigaku Miniflex 600 Japan) using Cu Kα radiation (λ = 1.5406 A°) at 15 mA current and 40 kV voltage. The diffraction was acquired using MiniFlex Guidance software (Version 2.0.2.1) at 25 °C from 2 Theta of 10 to 100° with a step size of 0.02°/step. Since most of the formulation components are viscous liquids, the samples were prepared by drop casting on a glass plate and drying at room temperature.

2.6.15. Ex vivo ocular permeation study

An ex vivo trans-corneal permeation study was conducted using goat eyes. Freshly excised goat eyes were collected in ice-cold PBS, transported at 4 °C, and dissected to obtain the cornea. The transcorneal diffusion study was carried out under non-occlusive conditions using a six-station vertical jacketed Franz diffusion apparatus (PermeGear Inc., USA) with a 3 mL receptor volume with a spherical joint (5 mm orifice diameter, 12 mm spherical diameter, and an effective diffusional area of 0.2 cm²). Bubbles between the cornea and receptor media were eliminated. The cornea was sandwiched between the donor and receptor chambers with the corneal epithelium towards the donor compartment and endothelium towards the receptor compartment containing fresh phosphate-buffered saline (pH 7.4 with 5% SLS) as diffusion media. Sodium azide (0.025% w/v) was added to the medium to avoid microbial growth (Attia et al., 2024). The Franz-diffusion cells were placed in black anodized aluminum cell holders, and the medium was stirred at 600 rpm using a temperature-controlled Franz cell magnetic stirrer (PermeGear V-series, V6-CB-01 Stirrer). The receptor compartment was maintained at ocular surface temperature (34 ± 0.5 °C) (Mahboobian et al., 2020) using a heater circulator (Julabo CORIO CD-BC4, 4.5 L). Before the sample application, the system was equilibrated for 30 min at 34 ± 0.5 °C to achieve corneal hydration and ocular surface temperature. Optimized NE and SH-17023 solution (in Transcutol) (15 μL) in triplicate were uniformly applied onto the corneal surface, which was in contact with receptor medium. Aliquots of 0.2 mL were collected at various time intervals (0, 0.083, 0.25, 0.5, 1, 2, 3, 4, 5, 6, 7, 9, 14, 16, 19, 24, 26, 28, 30, 36, 48, 52, 56, 63, and 72 h) from receptor compartment and replenished with the same volume of fresh diffusion medium to maintain the sink condition. Collected samples were diluted suitably with LC-MS grade methanol, added with internal standard, vortexed for 2 min, centrifuged at 10,621 ×g for 10 min, filtered through the 0.22 μM membrane filter, and the supernatants were analyzed for the amount of drug permeated by the validated LC-MS method. Then the cumulative amount of SH-17023 permeated through a unit area of cornea (μg/cm²) at a predetermined time (t) was calculated using Eq.11 and plotted (Ma et al., 2022; Youssef et al., 2022; Zhang et al., 2022). Then the permeability coefficient (K_p), steady-state flux (J_ss), and enhancement ratio (E_r) of optimized NEs and drug solution were calculated using Eqs. (12), (13), (14), respectively (Abd-Elaty et al., 2024; Attia et al., 2024; Kalam et al., 2016; Youssef et al., 2022).

$$\mathit{Cumulative\; amount}\left(\mathit{Qi}\right)=\frac{C_{i}V+\sum _1^{i-1}{C}_{i-1}\times V_i}{A}$$ (11)

Where, where Qi is the cumulative amount of drug permeated through the cornea, A is the effective diffusional area (0.2 cm²); V represents the volume of the receptor medium (3 mL); C_i represents the drug concentration (0.75 mg/15 μL) at a predetermined time (i) in receptor medium, V_i represents the volume after each sampling time; $\sum _1^{i-1}{C}_{i-1}$ denotes the summed total of the previously measured concentrations.

The permeability coefficient (k_p) was calculated using Eq.12.

$$\mathit{Permeability\; coefficient}\left(k_p\right)=\frac{\frac{\mathit{dQ}}{\mathit{dt}}}{A\times C_d}$$ (12)

Where A is the diffusion area, dQ/dt = amount of drug/time or rate of drug permeation, and C_d is the drug concentration in the donor compartment.

The drug permeation rate (flux) at steady state (J_ss) was calculated by dividing the slope of the linear portion of the graph by the area of the diffusion cell (Eq.13).

$$\mathit{Steady\; state\; Flux}\left(J_{\mathit{ss}}\right)=\frac{\frac{\mathit{dQ}}{\mathit{dt}}}{A}=k_p\times C_d$$ (13)

Where dQ/dt is the slope of the linear portion, and A is the diffusion area.

The enhancement ratio (E_r) was estimated using Eq.14.

$$\mathit{Enhancement\; ratio}\left(E_r\right)=\frac{J_{\mathit{ss}}\mathit{of}\mathit{NE}}{J_{\mathit{ss}}\mathit{of\; solution}}$$ (14)

2.6.16. Stability studies

The stability studies were carried out under accelerated conditions (heating-cooling, freeze-thaw, and centrifugal) and at room temperature (25 °C) up to 6 months.

2.6.16.1. Heating-cooling cycle

In the heating-cooling stability studies, the optimized NE was exposed to three cycles of heating and cooling within 40 and 4 °C each for a period of 48 h followed by examination for the signs of instability (cracking, creaming, and phase separation), colloidal properties (globule size, zeta potential), drug content, and chemical integrity etc. (Kassaee and Mahboobian, 2022; Mahboobian et al., 2020; Mehrandish and Mirzaeei, 2021; Mohapatra et al., 2025). The colloidal properties were evaluated using DLS, whereas the drug content was evaluated using the validated LC-MS method. Chemical integrity was evaluated by assessing shifts or disappearance of functional group peaks in the optimized NEs using ATR-FTIR spectroscopy.

2.6.16.2. Freeze-thaw cycle

This involved exposure of optimized NE to three cycles of freeze-thaw (−20 °C and 25 °C) each for a period of 48 h followed by evaluation of the signs of instability (cracking, creaming, and phase separation), colloidal properties (globule size, zeta potential), drug content, and chemical integrity (Kassaee and Mahboobian, 2022; Mahboobian et al., 2020; Mehrandish and Mirzaeei, 2021; Mohapatra et al., 2025).

2.6.16.3. Centrifugation study

The centrifugal stability was evaluated by subjecting the optimized NEs to centrifugation (Eppendorf, 5417 R) for 30 mins at 17950 ×g, followed by evaluation of signs of instability, colloidal properties, and chemical integrity (Kassaee and Mahboobian, 2022; Mahboobian et al., 2020; Mehrandish and Mirzaeei, 2021; Mohapatra et al., 2025).

2.6.16.4. Kinetic stability/ long-term stability study

Kinetic stability was assessed by storing the optimized NEs for up to 6 months at room temperature (25 ± 2 °C) and evaluating signs of instability, chemical integrity, and colloidal properties (Mohapatra et al., 2025).

2.7. Animals

All animal experiments were approved by the Animal Care Committee of the Centre for Phenogenomics, Mount Sinai Hospital, Toronto, Canada under Animal Use Protocol (AUP) no. 28-0428H. C57BL/6J mice (wild type, 7 weeks), male and female mice were housed and treated in accordance with the Association for Research in Vision and Ophthalmology (ARVO) Guidelines on Use of Animals in Ophthalmic and Visual Research. The animals were housed in cages with open access to standard food and water and maintained in a housing room at 25 ± 1 °C, 55 ± 5% RH, with a 12 h dark/light cycle. Before starting the experiments, they were acclimatized to the laboratory environment for 1 week. Mice were anesthetized by intraperitoneal injection of ketamine hydrochloride (100 mg/kg, Narketan, and xylazine (10 mg/ kg, Rompum). Pupils were dilated using 0.5% tropicamide eye drop (Alcon Mydriacyl Solution) and lubricated with carbomer gel (Optixcare Eye Lube Lubricating gel).

2.8. Ocular irritation/tolerability and toxicity studies

The ocular irritation of the optimized NEs was studied on the mice eyes using modified Organisation for Economic Co-operation and Development (OECD-405) guidelines for acute eye irritation/corrosion (OECD, 2023). Each animal was thoroughly examined for pre-existing corneal injury, signs of eye irritation, or ocular defects 24 h before starting the ocular irritation study. The in vivo ocular irritation studies were carried out using the Draize technique and histopathological evaluations (Ismail et al., 2020). Further, the eyes were thoroughly examined in detail by slit lamp biomicroscopy, spectral-domain optical coherence tomography (SD-OCT), and fundus fluorescein angiography (FA) using a Micron® 5 intraocular imager (Phoenix Micron, Inc., USA) before and after 7 days of twice daily topical ocular administration.

2.8.1. Animal experimental design for irritation and toxicity study

A total of 16 mice (8 adult male and 8 adult female) were divided into 4 groups (n = 4), so that each group contained 2 male and 2 female mice. Isotonic saline (0.9% NaCl) was selected as a non-irritant negative control, and 1 N NaOH was chosen as a standard irritant/positive control (Nayak and Misra, 2020). The right eye of each mouse received various treatments (0.9% NaCl, 1 N NaOH, blank NE, and optimized NE) for up to 7 days, whereas the left eye served as an untreated control. All samples were filtered (0.22 μm, syringe filter) before application to the eye. The eyes were visually evaluated according to the Draize criteria at 1, 2, 4, 6, 8, and 24 h on the first day, followed by every 24 h thereafter for up to 7 successive days. Ocular anterior and posterior imaging with the Micron® 5 intraocular imager was performed initially and after 7 days of administration. The data were analyzed by an investigator masked to the groups.

• •Group-I (negative control/ 0.9% NaCl): Mice topically treated with 15 μL of 0.9% NaCl b.i.d.
• •Group-II (positive control/ 1 N NaOH): Mice topically treated with 1 N NaOH (15 μL of 1 N NaOH soaked filter paper) once in the entire study.
• •Group-III (Test solution/ Blank NE): Mice topically treated with 15 μL of blank NE b.i.d.
• •Group-IV (Test solution/ Optimized SH-17023-loaded NE): Mice topically treated with 15 μL of optimized SH-17023-loaded NE b.i.d.

2.8.2. Draize technique

Each animal was thoroughly observed for any signs of toxicity, survival, or behavioural alteration, or change in body weight. The Draize irritation scores were assigned by visual evaluation of changes in ocular structures immediately and at 1, 2, 4, 6, 8, and 24 h after first administration, followed by assessments performed twice daily up to 7 days (before the first morning dose and in the evening). The degree of ocular irritation was investigated by scoring lesions of the cornea and conjunctiva, and chemosis at specific intervals as per Draize's scoring. Individual eyes of each experimental group were scored for each irritation parameter, and the average score for each parameter was calculated. Then, the total ocular irritation scores (OIS) for all irritation parameters were calculated by summing the average results for each parameter across treatments. The ocular irritation is evaluated from the average total ocular irritation score from 0 to 16, where an average score of 0–3 represents non-irritant, 4–8 represents slightly irritant, 9–12 represents moderately irritant, and 13–16 represents severely irritating or corrosive nature of the applied product (Fardous et al., 2022; Mohapatra et al., 2025). The irritation score of 2 or 3 in any evaluation or more than 4 for the total ocular irritation index (sum of all clinical irritation scores) was considered a meaningful irritant (Mahboobian et al., 2020). Since the eyes are very small and black in the case of C57BL/6J mice, the iris was not evaluated. The average ocular irritation score for cornea, conjunctiva, and chemosis was calculated for each group and interpreted for possible ocular irritation (Mohapatra et al., 2025). Further, the eyes were observed for any signs of tearing/allergic reaction, or inflammation.

2.8.3. Slit-lamp biomicroscopy, fundoscopy, SD-OCT, and FA

Slit-lamp biomicroscopy, SD-OCT, and FA were carried out on the initial day (Day 0) and at the end of the study (Day 7). Slit-lamp biomicroscopy was carried out to study the anterior segment of the eye using both narrow and wide beam exams. The narrow beam slit lamp exam was carried out to investigate complications in the cornea or in the anterior chamber, such as corneal swelling, corneal inflammation, corneal roughness or a change in the clarity of aqueous humor, whereas the wide beam biomicroscopy was carried out to investigate the entire eye surface and identify any changes or inflammation in cornea, conjunctiva, or the eyelid (Ibrahim et al., 2020). Further, the eyes were topically dosed with 0.5% (2 μL/eye) sodium fluorescein and investigated under fluorescent light on the slit lamp for identifying corneal surface abnormalities/ smoothness (Fardous et al., 2022).

Spectral domain OCT with guidance of bright-field live fundoscopy was performed using the Micron® 5 intraocular imager. The SD-OCT of the cornea and retina was carried out to study the abnormalities in the cornea and retina. Both horizontal and vertical cross-sectional OCT were done for a detailed observation of various ocular structures. Several OCT images of the cornea and retina were captured across the entire eye. Fundoscopy was performed to examine the retina, optic disc, macula, fovea, and retinal vessels.

Fundus FA was carried out to observe the ocular vasculature by i.p. administration of sodium fluorescein (2.5% prepared in sterile PBS) to anesthetized mice at a dose of 2.5 mL/kg. The FA photomicrographs were captured after 3 min of dye administration using a fluorescent imaging filter (Barrier: GFP/FA-B, and Excite: GFP/FA-E) of the Micron® 5 intraocular imager.

2.8.4. Histopathology

At the end of the study, the animals were euthanized by CO₂, and the eyes were carefully enucleated, washed with ice-cold saline, and fixed in 10% aqueous formalin overnight for histopathological observations. The eyes were submitted to The Centre for Phenogenomics Pathology Core, Canada for histopathology. The samples were processed in a Tissue Tek VIP 6 tissue processor (Sakura Finetek, USA), paraffin-embedded, and 5 μm-thick sections were collected onto charged slides (Assure, Epic Scientific, USA). The sections were deparaffinized and stained with Hematoxylin and Eosin in a Tissue Tek Prisma (Sakura Finetek, USA), and #1 glass coverslips were mounted with Glas G2 Mounting Media in a Tissue Tek Glas G2 automated coverslipper (Sakura Finetek, USA). Two animals (4 eyes) from each group were used for histopathological examination, and the remaining eyes of two animals from each group were used for an ATR-FTIR corneal interaction study via spectroscopy (see below). The sections were optically observed under an Olympus BX51 microscope (Evident Scientific, USA) and multi-mode reader (Cytation™, BioTek, USA) at 4× and 20× magnification.

2.8.5. ATR-FTIR study for NEs eye interaction evaluation

The ATR-FTIR analyses were carried out to evaluate the corneal interaction of NEs (Mohapatra et al., 2025; Porto et al., 2024). On day 7, the eyes were enucleated in PBS, and the cornea was immediately separated using dissection tools under a stereo microscope (Motic SMZ-171-TLED, China). Then the corneal tissues were subjected to an ATR-FTIR study. The spectra were acquired over the frequency range 4000–400 cm⁻¹, with 40 scans at a resolution of 4 cm⁻¹, using an ATR-FTIR spectrophotometer (Nicolet™ iS50, Thermo Scientific, USA) equipped with an ATR accessory and OMNIC Paradigm software.

2.9. Laser-induced choroidal neovascularization (L-CNV) model

CNV was developed by laser-induced rupture of Bruch's membrane (Chobisa et al., 2024; Gong et al., 2015; Muniyandi et al., 2025; Sishtla et al., 2022). L-CNV was performed on both eyes using the laser injector of the Micron® 5 intraocular imager at four locations/eye (12, 3, 6, and 9, o’clock or 1, 4, 7, and 10 o’clock to avoid hitting major blood vessels) with a 50 μm spot size, 270 mW pulse power, for 70 ms duration via an ophthalmic argon green laser (wavelength 532 nm) to rupture Bruch’s membrane. The fundus orientation was made by adjusting the mouse stage (up-down or left-right), focus, and aperture in such a way that the optic nerve was positioned at the centre of the view field, and the nerve fibers were uniform in all quadrants, as even focus is necessary to produce consistent CNV lesions (Gong et al., 2015). Before the laser was used, the beam intensity was measured several times using a laser power meter, with a 50 mW power setting for a 2500 ms pulse, producing a power level of 13–25 mW (24.5 in our case) as per the manufacturer’s instructions. Only intact eyes without noticeable ocular abnormalities (cataract, visible defects, and anomalous structures) were used for the L-CNV model (Gong et al., 2015). The uniformity and the diameter of the aiming beam on the fundus were observed before laser shooting. After the laser application, the eyes were lubricated with lubricating gel and placed on a pre-warmed heating pad, and atipamezole (1 mg/kg) was administered i.p. for reversal of anesthesia.

2.10. Animal experimental design for L-CNV

The mice were randomly assigned to treatments, and the data were analyzed by masking the groups. The mice were divided into 3 groups (n = 6, 1:1 male and female)

• •Group-I (Blank NE): L-CNV mice, receiving blank NE topically twice a day (15 μL/eye)
• •Group-II (Optimized NE): L-CNV mice, receiving SH-17023-loaded NE topically twice a day (15 μL/eye, equivalent to 0.75 mg SH-17023/eye).
• •Group-III (Sham): Mice without induction of L-CNV

2.11. In vivo CNV progression study

The in vivo CNV progression was evaluated after 7 days of laser treatment using the Micron® 5 intraocular imager equipped with Micron® 5 camera head, interchangeable objective lenses (LT 2 lens), temperature-controlled animal stand, and Micron Software Suite (Version: MSS 2.0.1). Mice were anesthetized, pupils were dilated, and eyes were lubricated as above. The fundoscopy, spectral-domain optical coherence tomography (SD-OCT or OCT), and fluorescein angiography (FA) were carried out to study the in vivo CNV progression (Chobisa et al., 2024).

2.11.1. Fundoscopy and SD-OCT

SD-OCT with bright-field live fundoscopy was performed using the Micron® 5 intraocular imager. Vertical and horizontal cross-sectional OCT imaging was carried out through the center of each laser-induced lesion. Several OCT images were captured across the lesion, and the largest cross-sectional lesion was used to calculate the lesion volume. The OCT analysis was performed by treating the lesions as ellipsoids and measuring their maximum dimensions in all three planes from OCT photomicrographs. The CNV lesion volumes were quantified from Eq.15 (Jensen et al., 2022; Sulaiman et al., 2015) using ImageJ software (1.52a, NIH, USA) with consideration of the exclusion criteria, such as (i) hemorrhage grade 1 (lesion excluded) and hemorrhage grade 2 (whole eye excluded), (ii) pre-existing retinal damage (whole eye excluded), (iii) the lesion was the only usable lesion in an eye, (iv) lesion with irregular shapes instead of circular shape in the fundus, (v) lesion fused with another lesion (both lesions excluded), (vi) suspicion of lack of Bruch’s membrane rupture, and (vii) imaging does not allow for adequate OCT scans.

$$\mathit{OCT}\mathit{volume}=\frac{4}{3}\mathit{\pi abc}$$ (15)

Where a, b, and c are semi-axes (half the lengths) along the x-axis, y-axis, and z-axis, respectively. The values of “a” and “c” are taken from the larger diameter of horizontal OCT and vertical OCT, respectively; whereas “b” is taken from whichever OCT image has the larger value.

2.11.2. Fundus fluorescein angiography

The FA was carried out to study vascular leakage after fundoscopy and OCT imaging by i.p. administration of sodium fluorescein (2.5% prepared in water for injection) to anesthetized mice at a dose of 2.5 mL/kg. The FA photomicrographs were captured after 3 min of dye administration using a fluorescent imaging filter (Barrier: GFP/FA-B, and Excite: GFP/FA-E) of the Micron® 5 intraocular imager. The FA images of the CNV lesions were used as a semi-quantitative measure of the leakage of lesions. Then, lesion severity was graded on a discrete scale from 0 (no fluorescent lesion) to 3 (saturated fluorescent lesion). Since lesion grading is a subjective evaluation, inter-personnel grading was performed by three independent observers, with masking of the experimental groups (Chobisa et al., 2024).

2.12. Ex vivo immunostaining and lesion volume quantification

The ex vivo immunostaining was carried out to visualize microvasculature in the choroidal flat mount as per our previous protocol with slight modifications (Chobisa et al., 2024). After the FA imaging, mice were euthanized using CO₂ (3–4 L/min for 7 min), and the eyes were enucleated. The collected eyes were pierced with a needle and were fixed in 4% paraformaldehyde overnight. After fixation, the eyes were dissected to remove the cornea from the posterior cup (retina/choroid). The posterior cups were further dissected to collect retinal pigment epithelium (RPE)-choroid. Then, the RPE-choroid was washed thrice (10 min/wash) at room temperature and incubated with 2.5% of bovine serum albumin (BSA) containing 0.3% of Triton X-100 in phosphate-buffered saline (PBS) for 1.5 h to avoid nonspecific staining. The blocking solution was removed and further incubated with lectin (biotin-conjugated GS-IB4) at a dilution of 1:250 in a solution (0.5% BSA in 0.3% Triton X-100 in PBS) for 48 h at 4 °C in an orbital shaker. After incubation, the RPE-choroid was washed thrice with PBS (20 min/wash) and incubated with detection reagent (DyLight 488 Streptavidin, 1:400 dilution in 0.5% BSA in 0.3% Triton X-100 in PBS solution) in a dark environment overnight at 4 °C using an orbital shaker. The non-specific staining was evaluated by considering no lectin control with the addition of the detection reagent. Afterwards, RPE-choroids were washed thrice with PBS (20 min/wash), stained with DAPI (diluted 1:10000 in PBS) for 10 min, and then washed twice in PBS (10 min/wash). The choroidal tissues were flat-mounted, covered with Fluoromount-G mounting medium along with glass coverslips, and stored in the dark at 4 °C.

The choroidal flatmounts were observed at 5× (all lesions) and 20× (individual lesions) magnifications. The z-stack photomicrographs of choroidal lesions were captured (3 stacks per lesion, a gap of 3 μm, 20× magnification) using an inverted LED fluorescence phase contrast motorized microscope (AXIO observer, Carl Zeiss, GmbH, Germany). The L-CNV lesion volumes were calculated from the lesion area of choroidal flat mounts measured by ImageJ software (1.52a, NIH, USA) by multiplying the total depth of the image (6 μm) by the average of the stained area of each lesion (Sishtla et al., 2022). Lesions are included in the analysis considering exclusion criteria (Gong et al., 2015; Poor et al., 2014), such as (i) inability to locate or identify lesion, (ii) presence of dissection damage, (iii) presence of inadequate dissection such as intact portions of the retina that was stained on top of choroid, (iv) linear lesion instead of circular, (v) lesion had fused with another lesion (both lesions excluded), (vi) the lesion had a size indicating it was an outlier lesion (“too big,” is more than 5 times larger than the next biggest lesion in the eye; “too small,” that is, it was less than 1/5 the area of the next smallest lesion in the eye; “too different,” that is, after all of the lesions in a specific treatment group were measured, the lesion’s area was 5-fold greater than the mean for that group), and (vii) the lesion was the only usable lesion in an eye. Additional exclusion criteria for CNV and FA (no FA leakage) were also considered.

2.13. Statistical analysis

The mean and standard deviation (SD) for all experiments was calculated and expressed as mean ± SD. Statistical analysis was performed with Student’s t-test at P < 0.05 using GraphPad Prism 10 (GraphPad Software, Inc., USA).

3. Results and discussion

3.1. LC-MS analysis and validation

The detailed validation results of the developed LC-MS method are provided in the supplementary file (Section 3.1, Fig.S2-S4, and Table S2-S5). Very sharp and distinct peaks of SH-17023 and midazolam as internal standards were obtained at run time (Rt) of 4.22 and 4.53, respectively (Fig.S2). SH-17023 showed the linearity within 5–1000 ng/mL with the regression equation of y = 1.179 x + 19.741 (n = 3) with an R² value of 0.998 (supplementary file, section 3.1.1). The percentage recovery was found to be within 98.303–100.091% at 50, 100, and 150% of standard addition (supplementary file, section 3.1.2, Table S2). High recovery values (>98% and low percent relative standard deviation (% RSD <2%) for all selected levels of SH-17023 demonstrated excellent accuracy of the developed LC-MS method. The precision of the developed LC-MS method, evaluated at 5, 50, and 500 ng/mL of working standard solution via repeatability, intra-day, and inter-day studies, showed a low % RSD (< 2%) in all cases, reflecting the attainment of a high degree of precision for the developed LC-MS method (supplementary file, section 3.1.3, Table S3). The LOD and LOQ for SH-17023 were found to be 0.531 ng/mL and 1.611 ng/mL, respectively (supplementary file, section 3.1.4, Table S4). Very low values of LOD and LOQ of the developed LC-MS method represent its excellent sensitivity for qualitative and quantitative analysis of SH-17023. Robustness is a measure of the ability of the developed LC-MS analytical method to remain unaffected by minor changes in analytical variables. The robustness of the developed LC-MS method examined by deliberate variations in chromatographic conditions reflected a lower % RSD value (<2%) of the area ratio or analyte to internal standard and retention time (supplementary file, section 3.1.5, Table S5), reflecting robustness of the developed method. This provides the reliability of the developed LC-MS method for qualitative and quantitative analysis of SH-17023 under slight variations in analytical conditions. A summary of the complete validation outcomes is provided in Supplementary Table S6. The overall validation results (linearity and range, accuracy, precision, LOD, LOQ, and robustness) as per the ICH guideline reflected the suitability of the developed LC-MS method for accurate quantification of SH-17023.

3.2. Screening of oil, surfactant, co-surfactant, and co-solvent by solubility study

Screening of suitable excipients is necessary for the development of stable NEs with high drug loading capacity. Oil, surfactants, and co-surfactants are major components of NEs. The oil solubilizes the hydrophobic drugs. The surfactants and co-surfactants adsorbed at the oil-water interface decrease the interfacial tension, leading to globule size reduction and formation of NEs (Smail et al., 2021). Sometimes, surfactant alone is not sufficient to reduce interfacial tension and produce stable NEs. Hence, a combination of surfactants and co-surfactants is needed to produce NEs with ultra-low dimensions and prolonged stability (Smail et al., 2021). A saturation solubility study was conducted to find suitable excipients for the development of NEs. The results of the solubility study of SH-17023 in various oils, surfactants, co-surfactants, and co-solvents are shown in Fig. 1 and Table S7 (supplementary file, section 3.2). Out of various screened oils, rose oil showed the highest solubility, whereas Labrasol® and Transcutol®P showed highest solubility among surfactants and co-surfactants/ co-solvents respectively.

Fig. 1.

Solubility study of SH-17023 (a) solubility in various oils, (b) solubility in surfactants, co-surfactants, and co-solvents.

3.3. Construction of pseudo-ternary phase diagrams

Each oil requires a certain HLB value of surfactants for the effective formulation of NEs. Each component of the NE possesses a specific effect on phase behaviour. So, the construction of a pseudo-ternary phase diagram is crucial for identifying the nanoemulsion region (Mahboobian et al., 2020). The pseudoternary phase diagrams with multiple oils, surfactants, co-surfactants, and co-solvents are shown in Fig. 2. Based on the solubility study, rose oil, Capryol® 90, Labrasol®, Labrafil® M 1944 CS, Tween® 80, and Transcutol® P were used in different combinations for the construction of phase diagrams. The single-phase/monophasic regions of the ternary diagrams represent the desired NE region.

Fig. 2.

Pseudo-ternary phase diagram of (a) rose oil and Labrasol®, (b) rose oil and Labrasol®: Labrafil® M 1944 CS (1:1), (c) rose oil and Labrasol®: Labrafil® M 1944 CS: Transcutol® P (1:1:1), (d) rose oil and Labrasol®: Tween® 80 (1:1), (e) rose oil and Labrasol®: Tween® 80: Transcutol® P (1:1:1), (f) Capryol® 90 and Labrasol®, (g) Capryol® 90 and Labrasol®: Labrafil® M 1944 CS (1:1), (h) Capryol® 90 and Labrasol®: Labrafil® M 1944 CS: Transcutol® P (1:1:1), (i) Capryol® 90 and Labrasol®: Tween® 80 (1:1), and (j) Capryol® 90 and Labrasol®: Tween® 80: Transcutol® P (1:1:1). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.4. Formulation and development of SH-17023-loaded NEs

NEs were prepared by a low-energy spontaneous emulsification method due to the advantage of spontaneous nanoglobule formation with gentle stirring without the need for specialized instruments (Dehghani et al., 2024; Ismail et al., 2020). Such a method utilizes the internal energy of the system to form the NEs spontaneously. Furthermore, they produce smaller globules compared to high-energy methods, offer ease of scale-up, and avoid the degradation of thermostable drugs that may occur with high-energy methods (Ismail et al., 2020).

The choice of excipient is important for the development of stable NEs. Non-ionic surfactants and co-surfactants, such as Labrasol® (Da Silva Gonçalves et al., 2024), Tween® 80 (Attia et al., 2024), and Transcutol® P (Kassaee and Mahboobian, 2022; Mahaling et al., 2024; Mahboobian et al., 2017) were chosen for the development of NEs due to their reported high safety profile in ocular formulations. Further, the surfactants with high HLB values (>10) are required to produce o/w type NEs with good physical stability (Attia et al., 2024). The requirement is fulfilled by the use of Tween® 80 and Labrasol®, having the HLB values of 15 and 12, respectively (Attia et al., 2024). The Transcutol® P was used as a cosolvent in the formulation to further enhance the solubility and stability of the formulation.

Labrasol® is a medium-chain triglyceride used mostly as an amphipathic non-ionic surfactant/co-surfactant/emulsifier and permeation enhancer. It improves the solubilization of hydrophobic drugs and enhances stability and homogeneity of an API in dosage form (Da Silva Gonçalves et al., 2024). It has been utilized as a surfactant/co-surfactant/edge activator/penetration enhancer in multiple ophthalmic formulations, such as nanostructured lipid carrier (Liu et al., 2012), microemulsion (Maulvi et al., 2021; Raval et al., 2018), microemulgel/microemulsion-based gel (Dhaval et al., 2020), penetration enhancer vesicles, transfersomes (Ashraf et al., 2018), spanlastic nanovesicles-loaded gel (Yasser et al., 2024), and NEs (Mahboobian et al., 2017) without causing any tissue damage.

Tween® 80/polysorbate 80 is a non-ionic surfactant widely used in various ocular formulations as a surfactant, emulsifier, edge activator/flexibility enhancer without any irritation or toxicity (Said et al., 2024). It has been utilized in many ocular drug delivery systems, such as NLC (Liu et al., 2012), solution, suspension, emulsion, gel (Ubhe et al., 2024), NEs, ocular inserts (Said et al., 2024), Ocusert (Shaikh et al., 2024), nano crystal (Kiran and Niharika, 2024), microemulsion (Zafar et al., 2024), self-nanoemulsifying drug delivery systems (SNEDDS) (Zingale et al., 2024), and solid lipid nanoparticle, without any sign of eye irritation. In the case of NEs, the Tween® 80 allows better stabilization of NEs, forming a protective layer around the oil globule, causing the emulsification, reducing the surface energy, and preventing their aggregation (Zingale et al., 2024).

Transcutol®P has been reported as cosurfactant for many NDDS, including NEs (Kassaee and Mahboobian, 2022; Mahaling et al., 2024; Mahboobian et al., 2017), microemulsions (Gohil et al., 2020; Salimi and Behrouzifar, 2020; Zhang et al., 2023), microemulgels (Dhaval et al., 2020), SNEDDS (Vikash et al., 2023; Zingale et al., 2024), in situ gelling systems (Ahmed et al., 2024), in situ gelling NEs (Kassaee et al., 2023), polymeric nanoparticles (Beirampour et al., 2024), leciplex (Bagul et al., 2024), and cubosomes (Nemr et al., 2024), for improved permeability in ophthalmic applications without any ocular irritation.

To our knowledge, there are no scientific reports on the ocular use of rose oil; therefore, for the first time, the ophthalmic NE of rose oil was developed and characterized for ocular irritation to identify its suitability for topical ocular application.

The pseudoternary phase diagrams with larger monophasic area (Fig. 2a, d, and e) were considered for development of NEs. Preliminary batches were formulated with the use of pseudoternary phase diagrams due to the maximum NE region, which will offer the maximum region of operability/flexibility for the development of an optimum NE. Initially, three preliminary formulations (200 μL oil, 600 μL surfactant or surfactant-cosurfactant or surfactant-cosurfactant-co-solvent mix, and 200 μL water) were prepared with the use of phase diagrams (Fig. 2a, d, and e), and the outcomes (Z_avg, PDI, zeta potential, and drug loading capacity) were analyzed. The formulation with the phase diagram Fig. 2a, having rose oil, Labrasol®, and water demonstrated very low PDI (0.081 ± 0.001), high zeta potential (−35.166 ± 0.903 mV), and drug loading (6.28 ± 1.356%) compared to other formulations with phase diagrams Fig. 2d and e. However, the globule size was found to be relatively larger (101.2 ± 0.512 nm) compared to the other two formulations (13.573 ± 0.532 nm and 45.6 ± 0.236 nm). After loading the drug in the formulation having rose oil, Labrasol®, and water, it was found to develop precipitates after 40 days. Due to the physical instability of the formulation, the remaining phase diagrams (Fig. 2d and e) were considered for the development of NEs. The formulation with phase diagram Fig. 2d, rose oil, Labrasol®, and Tween® 80 demonstrated a globule size of 13.573 ± 0.532 nm, PDI of 0.314 ± 0.012, zeta potential of −7.03 ± 0.884 mV, and loading capacity of 3.43 ± 2.137%. Due to the low loading capacity, this formulation was not considered for further optimization. Then the formulation with phase diagram Fig. 2e having rose oil, Labrasol®: Tween® 80: Transcutol® P (1:1:1), and water was considered for the QbD-based development of NEs due to the smaller globule size (45.6 ± 0.236 nm), relatively high loading capacity (5.44 ± 1.873%), low PDI (0.290 ± 0.003), excellent zeta potential (−29.16 ± 1.161), and prolonged physical stability (absence of phase separation, precipitation, sedimentation, flocculation) after loading of drugs.

3.4.1. Response surface analysis

Response surface analysis allows for understanding the effects of various factors on responses during formulation optimization. In the QbD approach, the desired critical quality attributes (CQAs) are defined, and the influence of each independent variable is investigated. The lower and upper levels of the independent variables cannot be selected at random when the sum of the formulation components is essential to develop a stable formulation (Barot et al., 2012). Among various experimental designs (central composite design, Box–Behnken designs), the D-optimal mixture design is widely used for formulation optimization, specifically when the proportions of the formulation components are responsible for experimental outcomes (Gohil et al., 2020; Shah et al., 2018). In a D-optimal mixture design, the proportion of each formulation component is more essential than its actual amount in the combination, as the sum of components in each mixture must equal 100% v/v. In contrast, other statistical designs rely on the amounts of formulation components to achieve the desired formulation with the required quality attributes (Shah et al., 2018). The experimental runs in a three-component mixture design are represented in an equilateral triangle in which the components’ boundaries are limited in the design space. Hence, the D-optimal mixture is used for improved prediction power with a minimum number of experimental runs (Barot et al., 2012; Fouad et al., 2013).

In our nanoemulsion formulation, the three-component system includes the % of rose oil (X₁), S_mix (Labrasol® + Tween® 80 + Transcutol® P) (X₂), and water (X₃). The CMAs and the desired CQAs of the nanoemulsion are shown in Table 1. The upper and lower levels of CMAs were selected from the pseudo-ternary phase diagrams and pilot studies on NEs. The design expert software suggested 16 batches for the D-optimal mixture design with the inclusion of factorial points. The exact composition of 16 preliminary batches with their CQAs (Z_avg, PDI, and LC) is shown in Table 2. The statistical model fit summary for various CQAs is shown in Table S8. The R² values for each response shown in the model fit summary (Table S8) were found to be higher (>0.99), representing the significance and suitability of the experimental design.

For the response Z_avg (Y₁), based on very high model R² value (0.999), nonsignificant lack of fit value (0.483 i.e., >0.05), low sequential p-value (<0.0001), reasonable agreement between predicted & adjusted R² values, permissible difference (<0.2) between predicted (0.996) and adjusted R² value (0.998), low standard deviation (0.316), and very high signal to noise ratio/ adequate precision (104.648, i.e.,>4.0). With a Predicted Residual Sum of Squares (PRESS) value of 3.74, the special quartic vs. quadratic model was found to be suitable (Table S8).

For PDI (Y₂), the special quartic vs. quadratic model was also found to be best owing to very high model R² value (0.997), nonsignificant lack of fit value (0.341 i.e., >0.05), very low standard deviation (0.006), low sequential p-value (<0.001), acceptable difference (<0.2) between predicted (0.982) and adjusted R² value (0.994), high signal to noise ratio (48.723, i.e.,>4.0), and suitable PRESS value (0.002) (Table S8).

For LC (Y₃), the cubic model was found to be suitable due to very high model R² value (0.998), nonsignificant lack of fit value (0.815 i.e., >0.05), low sequential p-value (<0.009), reasonable agreement between predicted and adjusted R² values, permissible difference (<0.2) between predicted (0.953) and adjusted R² value (0.995), very low standard deviation (0.080), and very high signal to noise ratio (60.041, i.e.,>4.0). With a PRESS value of 0.932, the cubic model was found to be suitable (Table S8).

The ANOVA results and the model fit statistics of the D-optimal mixture design for Z_avg, PDI, and % LC are shown in Table S9, Table S10, and Table S11, respectively. The model p-values for Y₁, Y_2, and Y₃ are found to be <0.0001, which represents the statistical significance and goodness of fit of the used models. Most of the model terms in the ANOVA table showed the p values <0.05, representing their inclusion in the model equation and suitability/adequacy of D-optimal mixture design factor-response analysis of the nanoemulsion. “Lack of Fit Tests” compares pure error with the residual error from design points, and non-significance (p > 0.05) is desired for such statistical models. The obtained lack-of-fit p values are 0.4833 for Y₁ (Table S9), 0.3413 for Y₂ (Table S10), and 0.8155 for Y₃ (Table S11), indicating the suitability of the used models. The adequate precision measures the signal-to-noise ratio. As per the software, a ratio > 4 is desirable. In our case, the signal-to-noise ratio of 104.648 for Y₁, 48.722 for Y₂, and 60.04 for Y₃ indicates the achievement of adequate signal (Table S9, Table S10, and Table S11). Thus, the selected models can be used to navigate the design space. The model terms with very low p value and high F value represented a significantly stronger influence on a specific response.

Based on the experimental outcomes of the preliminary batches, NEs containing rose oil, S_mix (Labrasol® + Tween® 80 + Transcutol® P), and water were used for the response surface analysis and formulation optimization. The oil and the surfactant phase of nanoemulsion should solubilize the maximum quantity of drug without any precipitation/phase separation to achieve the maximum dose in the minimum formulation. The NEs were formulated using non-ionic surfactants, co-surfactants, and a co-solvent. The obtained diagnostic plots for Y₁, Y₂, and Y₃ are shown in Fig. S5, Fig. S6, and Fig. S7, respectively. The linearity of “normal plot of residual,” linearity of “predicted vs. actual plot,” acceptable value (−0.03 to 1.14, for Y₁, −0.19 to 1.35 for Y₂, and  –1.08–3.03 for Y₃) of lambda in “Box-Cox plot,” persistent range of residuals in “externally studentized residual vs. predicted plot,” and random scattering within the control limits of “externally studentized residual vs. run plot” of diagnostic plots for response Y₁ (Fig. S5), Y₂ (Fig. S6), and Y₃ (Fig. S7) demonstrated the goodness of fit of the used mixture design.

Model polynomial equations for the responses Y₁, Y₂, and Y₃ are represented in Eq.16, Eq.17, and Eq.18, respectively.

$$Z_{\mathit{avg}}\left(Y_1\right)=123.55X_1+55.57X_2+83.83X_3-203.84X_1{X}_2-204.01X_1{X}_3-151.35X_2{X}_3+80.11X_1^2{X}_2{X}_3+774.43X_1{X}_2^2{X}_3+21.89X_1{X}_2{X}_3^2$$ (16)

$$\mathit{PDI}\left(Y_2\right)=1.39X_1+0.3794X_2+0.6911X_3-2.30X_1{X}_2-2.53X_1{X}_3-1.32X_2{X}_3-0.6107X_1^2{X}_2{X}_3+5.43X_1{X}_2^2{X}_3-1.79X_1{X}_2{X}_3^2$$ (17)

$$\mathit{LC}\left(Y_3\right)=-44.46X_1+8.27X_2+38.13X_3+97.78X_1{X}_2+30.39X_1{X}_3-65.50X_2{X}_3-53.62X_1{X}_2{X}_3+65.98X_1{X}_2\left(X_1-X_2\right)+104.86X_1{X}_3\left(X_1-X_3\right)+39.87X_2{X}_3\left(X_2-X_3\right)$$ (18)

Then the influence of each independent variable on response (CQAs) was investigated from 2D counter plots and 3D response surface plots (Fig. 3a-d). The obtained globule size of emulsions was found to be within 30.888 to 55.744 nm, representing effective preparation of NEs. The globule size of the NEs was found to increase with an increase in the percentage of rose oil and decreasing concentration of S_mix (Fig. 3a and Fig. 3b). The result is in agreement with the reported results on the effect of oil and surfactant phase on the globule size of emulsion systems (Dhaval et al., 2020; Gohil et al., 2020; Mohapatra et al., 2024).

Fig. 3.

2D Counter plots and 3D response surface plots showing (a & b) effect of independent variables on globule size, (c & d) effect of independent variables on polydispersity, and (e & f) effect of independent variables on loading capacity.

In a recent study, travoprost-loaded NEs demonstrated a significant decrease in the globule size with an increase in the surfactant (Tween® 80) concentration and a significant increase in the globule size with increasing the oil (Lipophile®) concentration (Ismail et al., 2020). The smallest globule size was reported with the highest surfactant and lowest oil concentration (Ismail et al., 2020). A high oil concentration causes resistance to oil disruption during stirring, leading to increased globule size (Ismail et al., 2020). As the oil phase forms the internal dispersed phase of NEs, an increase in its percentage forms bigger globules, leading to an increase in the Z_avg. Surfactants and cosurfactants act as emulsifiers in the NEs, reducing interfacial tension at the oil-water interface and globule size (Ismail et al., 2020). The increase in surfactant concentration causes increased coverage on the NE globule surface, leading to the formation of smaller globules (Ismail et al., 2020). The co-solvents in the S_mix further reduce the interfacial tension. With an increase in the proportion of S_mix, the decrease in globule size is due to the reduction of the interfacial tension and the lowering of free energy between oil and water. At low concentrations of S_mix, the interfacial tension and free energy are very high, leading to an increase in globule size of the NEs. In a study regarding clotrimazole-loaded ophthalmic microemulsions, the globule size was reported to decrease with increasing the concentration of S_mix from 60% to 70%, which is ascribed to the decrease in the surface tension upon increasing the S_mix concentration (Al-mahallawi et al., 2021). In another study, the effect of percentage of oil (isopropyl myristate), surfactant (Tween® 80), and cationic agent (cetyl trimethyl ammonium bromide) was studied on the globule size, zeta potential, and PDI of dorzolamide hydrochloride-loaded NEs using Box–Behnken design. The size of NEs was found to be increased with increasing the volume of oil phase and decreasing with increasing the surfactant concentration (Kassem et al., 2022). The reduced globule size at higher surfactant concentration was also reported for the reduction of the interfacial tension between the oil and aqueous phase (Kassem et al., 2022). The higher surfactant concentration stabilized the small oil globules by forming a steric barrier on their surface and preventing the coalescence of globules (Kassem et al., 2022).

Here, the globule size of the NE was found to be decreased with an increase in the proportion of dispersion medium/continuous phase (water), which can be ascribed to the very good dispersion of oil globules at high water content. However, after a very high concentration (70%) of S_mix, the size of the NEs was found to be increased. Such an outcome is ascribed to possible accumulation of surfactants, which further attract the surfactants from the oil-water interface and allow the flocculation of oil droplets in the NE, resulting in an increase in globule size. For this reason, the globule size was found to be very high even at low percentage of oil (<15%) due to the very high surfactant concentration.

The PDI value represents the homogeneity of the NEs (Kassem et al., 2022). A PDI value below 0.5 indicates a narrow size distribution and a homogeneous nature of NEs. Uniform globules facilitate the effective corneal delivery of low-water-soluble drugs (Al-mahallawi et al., 2021). In all trial runs (16 batches), the PDI values ranged from 0.19 to 0.412, i.e., <0.5, indicating the monodisperse and homogeneous nature of the developed nanoemulsions. The obtained PDI values were observed to be decreased with a decrease in the proportion of rose oil and an increased proportion of S_mix and water (Fig. 3c and d). The results are in accordance with the PDI results of dorzolamide hydrochloride-loaded NEs (Kassem et al., 2022), and a clotrimazole-loaded microemulsion (Al-mahallawi et al., 2021) where the PDI values are reported to be decreased with increasing the Tween® 80 and S_mix concentration, respectively. It was reported that the greater concentration of surfactant would cause the formation of homogeneous globules (Kassem et al., 2022). In contrast, at elevated oil levels, very large and more heterogeneous globules were generated, resulting in increased PDI value. An increase in the proportion of aqueous phase promoted the fine dispersion of globules, resulting in a lowering of PDI values.

Loading capacity is the actual quantity of loaded drug per unit weight of formulation. The LC for all the NEs is found to be within 4.358% to 8.360%. A higher LC is required to load the maximum quantity of drug within a minimum quantity of formulation to achieve maximum therapeutic effect. From the response surface plots, it was observed that LC increases with increasing proportions of oil and S_mix. Since SH-17023 is more soluble in S_mix than in rose oil, the loading capacity increased with an increased proportion of S_mix, followed by rose oil (Fig. 3e and f). In a recent study, regarding a poorly water-soluble drug (luliconazole) loaded NEs, the effect of concentration of oil phase and S_mix was studied on the drug content and globule size (Yang et al., 2022). The drug loading was reported to increase with increasing the percentage of the oil phase, due to improved drug solubility with increasing oil concentration (Yang et al., 2022).

3.4.2. Optimization of NEs

The optimized nanoemulsion with requisite CQAs was obtained using numerical and graphical optimization methods, with constraints such as lowering of globule size, polydispersity index, and higher loading capacity. The desired NE region was identified inside the yellow design space of the overlay plot (Fig. 4). The flag mark represents the composition of the software-suggested optimized NEs. The suggested composition of optimized NE is 23.061% v/v of rose oil (X₁), 66.939% v/v of S_mix (X₂), and 10% v/v of water (X₃), which would show Z_avg, PDI, and LC of 32.941 nm, 0.204, and 7.522%, respectively.

Fig. 4.

Overlay plot representing the optimized NEs with the inset figure showing the photograph of optimized NEs.

3.4.3. Checkpoint analysis and validation of the software prediction

The % prediction error for the desired responses of the checkpoint batch (Table S12) is found to be within the acceptable limit (<5%), which validates the prediction accuracy of the D-optimal statistical mixture design.

3.5. Characterization of optimized NE

The results of the evaluated parameters of the optimized NE are shown in Table 3 and described further below.

Table 3.

Characterization of optimized NE.

Parameters Evaluated	Optimized NEs
Visual characterization	Transparent, pale yellow in colour
pH	5.58 ± 0.216
Viscosity at shear rate 10 s−1 (mPa·s)	124.486 ± 0.669
Refractive index (RI)	1.347 ± 0.0001
% Transmittance	96.531 ± 0.0
Conductivity (μS/cm)	53.166 ± 0.124
Globule size (nm)	32.832 ± 2.125
Polydispersity index (PDI)	0.201 ± 0.003
Zeta potential (mV)	−29.2 ± 0.45
Accelerated stability	Stable with heating-cooling and centrifugation cycle, unstable at freeze-thaw
Kinetic Stability	Stable with a slight increase in globule size & PDI up to 6 months
Drug content (%)	99.585 ± 7.719
Entrapment efficiency (%)	97.091 ± 1.294

3.5.1. Organoleptic properties

The developed NE was observed as clear, transparent, slightly pale yellow in colour, without any sign of precipitation, phase separation, or flocculation. The transparent nature is characteristic of NEs (De Oca-Ávalos et al., 2017).

3.5.2. pH

The pH of the optimized NE was found to be 5.58 ± 0.216 (Table 3). The NEs showed slightly acidic pH, which can be tolerated on the eye without potential irritation. The acceptable pH range of ocular formulations is within 3.5–8.5 (Dhahir et al., 2021; Kassem et al., 2022) or more precisely within 6.5–8.5 (Choradiya and Patil, 2021; Mohapatra et al., 2025). The pH of the optimized NEs is within the acceptable limits of pH for ocular drug delivery, representing their topical applicability potential. Similar pH values (pH 5.5–5.9) were reported for travoprost-loaded ophthalmic NEs (Ismail et al., 2020). In another report, very low pH values (pH 3.80–4.20) were reported for β-caryophyllene-loaded NEs for the treatment of Acanthamoeba keratitis and reported to be close to the acceptable limits (pH 4–8) for ocular application (Porto et al., 2024).

3.5.3. Conductivity

The conductivity of the optimized NE was found to be 53.166 ± 0.124 S/cm (Table 3). The results of conductivity testing represent the type of NEs. Despite the use of a non-ionic surfactant in the optimized NE, they demonstrated electroconductive behavior. A very high electrical conductivity indicates that the optimized NEs are oil-in-water (o/w) and that they dilute readily in water. The conductivity of the diluted sample remained unchanged at 0, 12, and 24 h, indicating the stability of the NEs after dilution. In case of any instability, phase separation occurs, and the conductivity gradually decreases.

3.5.4. Viscosity

The viscosity of the optimized NE was measured to check their consistency and suitability for topical ocular administration (Mohapatra et al., 2025). Very low viscosity leads to spillage and poor ocular retention, whereas overly high viscosity can make topical ocular administration difficult; therefore, NEs should have an optimum viscosity to ensure both adequate retention and ease of installation (Mohapatra et al., 2025). The obtained viscosity of the optimized NEs was found to be 124.486 ± 0.669 mPa·s at a shear rate of 10 s⁻¹ (Table 3), which is within the permissible viscosity limits (15–150 mPa·s) for topical ophthalmic application and maximum corneal retention (Mohapatra et al., 2025). The developed NE demonstrated low viscosity, which is expected for the consistency of NEs.

3.5.5. Percentage transmittance

The optimized ophthalmic NE was found to be very transparent with a % T value of 96.531. A very high % transparency (>95%) represents the potential of the ophthalmic NEs for topical ocular application without any blur effects. Such a formulation would offer improved patient compliance. The blank NE showed a % T value of 98.552%. The relatively lower %T in the case of optimized NE compared to blank NE is due to the successful entrapment of the drug.

3.5.6. Refractive index

The refractive index (RI) represents the transparency of the ocular formulation (Mohapatra et al., 2025). A transparent ocular formulation offers clear visualization after topical application. An RI value of <1.476 is optimal for topical ocular application without causing blurring or vision discomfort (Dhahir et al., 2021; Ismail et al., 2020; Kassem et al., 2022; Singh et al., 2020). The RI value of the developed NE (1.347 ± 0.0001; Table 3) is close to the RI of water (1.333) and tear fluid (1.340–1.360) (Kassem et al., 2022), representing the optical clarity of NEs and their suitability for topical ocular application (Mohapatra et al., 2025). Further, the RI measurements of three independent samples from optimized NEs were found to be very close to each other, indicating the isotropic nature of the novel ophthalmic formulation.

3.5.7. Dilutability

After dilution with water, PBS (pH = 7.4), and simulated tear fluid, the NE did not show any sign of cracking, precipitation, or phase separation, demonstrating dilution stability (Kassem et al., 2022) and water as the external phase, i.e., o/w type. The o/w type NEs have been frequently employed and are ideal for topical ocular administration, as the external phase is water; this makes them easier to dilute with aqueous tear fluid and offers the potential to deliver hydrophobic drugs (Attia et al., 2024; Kassem et al., 2022).

3.5.8. Globule size, polydispersity, and zeta potential

The globule size, polydispersity index, and zeta potential values of the optimized NE were found to be 32.832 ± 2.125 nm, 0.201 ± 0.003, and –− 29.2 ± 0.45 mV, respectively (Table 3, Fig. 5). A very low globule size will offer improved ocular penetration (Fig. 5a). A globule size of <200 nm is reported to be optimal for ophthalmic drug delivery (Ismail et al., 2020; Mohapatra et al., 2025), since they offer improved effective surface area, increased solubility, and improved ocular permeation via receptor-mediated endocytosis (Mohapatra et al., 2025).

Fig. 5.

Vesicle size, morphology, size distribution, and crystallinity of optimized NEs. (a) Vesicle size, (b) correlogram, (c) zeta potential, (d) phase plot, (e & f) HRTEM photomicrographs at 200 nm (with particle diameters shown) and 100 nm scale, (g) globule size distribution, and (h) SAED results of optimized NE.

PDI value varies from 0 to 1, where 0 represents a completely uniform/homogeneous size distribution with monomodal distribution, and 1 represents polydispersity/heterogeneous globule size with multimodal distribution (Durgapal et al., 2022). A higher PDI value represents that the system possesses a high surface free energy, which leads to phase separation and fusion of globules to achieve a low surface free energy state through aggregation (Durgapal et al., 2022). The PDI value <0.3 represents the homogeneous nature of the formulation and narrow size distribution, whereas a value >0.3 indicates broad size distribution and heterogeneity (Mohapatra et al., 2025). The correlogram is shown in Fig. 5b. Smaller globules move faster, and the correlation decays quickly, producing a steep curve. Smooth single-exponential decay represented monodisperse globule size with narrow size distribution. A very low PDI value avoids Ostwald ripening, thereby improving the stability (Mohapatra et al., 2025).

The negative zeta potential at the surface of NEs will provide repulsion among the nanoglobules, thereby improving colloidal stability (Al-mahallawi et al., 2021). The zeta potential value within ±0–10 mV signifies highly unstable, a value between ±10–20 mV signifies relatively stable, a value between ±20–30 mV signifies moderately stable, and values greater than ±30 mV signify high stability of colloidal dispersion (Mohapatra et al., 2021). The developed NEs showed a zeta potential of −29.2 ± 0.45 mV, reflecting a moderately stable nature (Fig. 5c). An inverse phase plot represents the negative zeta potential on the NE globules (Fig. 5d). A higher zeta potential will offer repulsive forces among the emulsion globules, thereby preventing the aggregation and coalescence, and offering prolonged stability (Mohapatra et al., 2025).

3.5.9. HRTEM, Globule size distribution, and SAED

The HRTEM photomicrographs of the optimized NE showed spherical smooth globules (Fig. 5e and f). The morphology of the globules is consistent with previous reports (Zhang et al., 2022). However, the aggregation of globules is due to the evaporation of the external aqueous medium from the TEM grid during drying. The size distribution curve represented a narrow size distribution with average globule size of 32.798 ± 3.012 nm, which is in accordance with the DLS results (Fig. 5g). However, some of the globules were found to be relatively large in size due to the fusion of the droplets on the TEM grid during sample processing. The SAED results revealed a diffused pattern, reflecting the amorphous nature of the NE (Fig. 5h).

3.5.10. Drug content

The percentage drug content of the optimized NE was found to be 99.585 ± 7.719% by the validated LC-MS technique. A drug content of 85–115% (± 15%) is accepted and considered as a good product, whereas a content of 75–125% (± 25%) is not accepted as a good product (Mohapatra et al., 2025). The evaluated drug content of the developed NE is within the accepted limit.

3.5.11. Entrapment efficiency

EE is a critical quality attribute as it reflects the fraction of the drug associated with the dispersed internal phase. The entrapment efficiency of the optimized NE was 97.091 ± 1.294%. A high entrapment efficiency (>90%) is required for ophthalmic NEs to maximize ocular bioavailability, thereby reducing dosing volume and frequency (Mohapatra et al., 2025). A high % EE was also indicated by the absence of SH-17023's characteristic ATR-FTIR peaks in the optimized formulation (see section 13.5.14), indicating effective drug loading in the internal oil phase and the absence of free drug in the external medium of the ophthalmic NEs (Yang et al., 2022; Youssef et al., 2021).

3.5.12. In-vitro drug release and release kinetics

The dialysis bag method has been widely used for the release study of ophthalmic NEs (Ismail et al., 2020; Mahboobian et al., 2020). The release profile of SH-17023 from solution and optimized NE is shown in Fig. 6a as the cumulative percentage of drug release as a function of time (h). The release of SH-17023 from the solution was found to be faster than the optimized NE. The release of the drug from the solution plateaued after 6 h, while the optimized NE continuously released the drug up to 72 h. At 6 h, the cumulative percentage drug release from solution and NE were found to be 98.274 ± 1.104 and 51.798 ± 3.743%, respectively. The optimized NE demonstrated a biphasic release pattern, i.e., initial rapid release followed by sustained release. The initial rapid release might be due to the release of a small quantity of drug localized at the o/w interface. The sustained release profile of SH-17023 from the optimized NE is attributed to the release of the drug from the oil core, which behaves as a drug reservoir to sustain release (Attia et al., 2024; Ismail et al., 2020). The release of hydrophobic drugs from o/w emulsion involves the partitioning/diffusion of drugs from the oil core to the intermediate surfactant layer, followed by release into the aqueous phase, hence shows sustained release behaviour (Youssef et al., 2021; Youssef et al., 2022). A similar sustained release profile was reported for the travoprost-loaded ophthalmic NEs (Ismail et al., 2020), ciprofloxacin NEs (Youssef et al., 2021), moxifloxacin NEs (Youssef et al., 2022), and prednisolone NEs (Attia et al., 2024). Since SH-17023 is highly lipophilic (log P = 5.28), the maximum concentration of the drug is available at the oil core of NEs. This will prevent frequent administration, improve patient compliance, and provide benefits over conventional eye drops.

Fig. 6.

In vitro release study, ex vivo ocular permeability study across goat cornea, and drug-excipient compatibility study by FTIR and XRD. (a) In vitro release of SH-17023 from optimized NE and drug solution, (b) Cumulative amount of SH-17023 permeated across cornea from SH-17023 solution and optimized NE. Values are represented as mean ± SD (n = 3), (c) ATR-FTIR spectra, and (d) XRD diffractograms of formulation components (rose oil, Labrasol®, Tween® 80, Transcutol® P, and SH-17023), physical mixture, placebo formulation (blank NE), and optimized NE for drug-excipient compatibility study.

The release kinetics of the drug from the optimized NE are shown in Table S13. Among various kinetic models, the Korsmeyer-Peppas model showed the highest regression coefficient (R² = 0.808) compared to the other kinetic models. The release pattern agrees with the previous reports on ophthalmic NEs (Youssef et al., 2021; Youssef et al., 2022) and microemulsions (Ma et al., 2022; Shi et al., 2023; Zhang et al., 2023). The release exponent or “n” obtained from the Korsemeyer-Peppas model represents the mechanism of drug release. A value of n < 0.43 represents Fickian diffusion, within 0.43–0.85 represents anomalous diffusion, 0.89 represents Case II transport, whereas a value >0.85 represents super Case II transport (Kassem et al., 2022). The value of n for the optimized NEs was found to be 0.770, representing anomalous diffusion, which is mainly mediated by more than one mechanism (Kassem et al., 2022). Anomalous diffusion was previously reported for ophthalmic NEs (Kassem et al., 2022). The anomalous diffusion of the NEs is influenced by globule size, size distribution, surfactants, and viscosity.

3.5.13. Ex vivo ocular permeation

Corneal permeability impacts on the ocular bioavailability of topically administered ophthalmic drugs. The ex vivo corneal permeation of optimized NE across the goat cornea was compared with the drug solution to investigate permeation enhancement due to the topical NE formulation. The LC-MS chromatograms of SH-17023 from solution and optimized NE at 4, 24, 48, and 72 h are shown in Fig. S8–11. An increase in the area of SH-17023 was observed in the case of optimized NEs compared to the solution, reflecting the enhanced permeation of NEs across the goat cornea. Fig. 6b depicts the cumulative amount of drug permeated per unit area of cornea from the drug solution and optimized NEs. The NEs showed significantly higher (p < 0.05) amount of drug permeation across the cornea compared to the drug solution. Optimum hydrophilicity and lipophilicity are required to achieve ocular permeability and bioavailability (Ismail et al., 2020). In the case of NEs, due to the presence of oil and surfactants/co-surfactants and maintenance of optimum hydrophilic-lipophilic balance, the permeation was improved. A recent study reported improved ocular bioavailability with a highly lipophilic drug, travoprost (log P = 4.6), via NE formulation (Ismail et al., 2020).

Table 4 shows the results of permeability coefficient (K_p), steady-state flux (J_ss), and enhancement ratio (ER) of optimized NE and drug solution. The K_p and J_ss were significantly higher for the optimized NE compared to the drug solution, with an enhancement ratio of 23.815. The enhanced permeation of NE compared to drug solution is attributed to (i) nanometric globule size offering improved effective surface area to cover and interact with cornea (Mahboobian et al., 2020; Zhang et al., 2022), (ii) improved drug solubilization with maintenance of high drug concentration at corneal surface/high concentration gradient, (iii) improved corneal adhesion due to wetting by surfactants/lowering of contact angle/lowering of surface tension (Durgapal et al., 2022), (iv) better interaction of the cornea with oily nanodroplets surrounded by surfactant and the cosurfactant (Cardoso et al., 2022), (v) disruption and fluidization of corneal tight junctions by surfactants/cosurfactants (Mahboobian et al., 2020; Youssef et al., 2021), (vi) permeation enhancer property of S_mix (Durgapal et al., 2022; Youssef et al., 2022), (vii) receptor mediated endocytosis due to size <200 nm (Youssef et al., 2021; Youssef et al., 2022), and (viii) controlled drug release. The significantly higher drug permeation of NEs will eventually lead to improved ocular bioavailability, hence improved therapeutic outcome. The improved ocular permeation and bioavailability of travoprost (Ismail et al., 2020), β-caryophyllene (Porto et al., 2024), prednisolone (Attia et al., 2024), ciprofloxacin (Youssef et al., 2021), moxifloxacin (Youssef et al., 2022), posaconazole (Nasser et al., 2021), besifloxacin (Kassaee and Mahboobian, 2022), isoliquiritigenin (Zhang et al., 2022) via ophthalmic NEs and acyclovir (Mahboobian et al., 2020) via nanoemulgel, and besifloxacin (Kassaee et al., 2023) by thermoresponsive in situ ophthalmic nanoemulgel were reported by multiple research groups.

Table 4.

Ex vivo corneal permeability of SH-17023 solution and optimized NEs.

Formulation	Permeability coefficient (Kp)	Steady-state Flux (Jss)	Enhancement Ratio (ER)
	(h−1.cm−2)	(μg. h−1.cm−2)
Drug solution	0.002 ± 0.0004	1.575 ± 0.488	–
NEs	0.050 ± 0.012	37.51 ± 9.29	23.815

Due to the sustained-release profile of the optimized NE, we continued the permeation study for up to 72 h. However, this ex vivo study has its own limitations regarding corneal viability. We tried to maintain the corneal integrity as much as possible. Before the ex vivo ocular permeation study, the integrity of the goat cornea was confirmed by visually inspecting for tears, cuts, imperfections, and cracks, and by measuring thickness with Vernier calipers. After the permeation experiments, the excess samples from the corneal surface were removed, and the tissue integrity, thickness, and intactness level were reanalyzed. No substantial changes in membrane integrity, including damage such as cuts, tears, edema, imperfections, or cracking, were observed. In the case of a damaged cornea, the free drug should immediately permeate the cornea in bulk due to the loss of corneal tight junctions and integrity. However, in our permeability experiment, we did not observe a sudden, significant enhancement in free drug permeation up to 72 h, indicating that the corneal barrier function and integrity are maintained.

3.5.14. Drug-excipient compatibility

3.5.14.1. ATR-FTIR spectroscopy

FTIR studies are used for investigating the interactions among NE components (Porto et al., 2024). The results of ATR-FTIR spectroscopy of NEs, their components, and a physical mixture of all the components are shown in Fig. 6c. SH-17023 showed a characteristic broad peak at 3360 cm⁻¹ and 3051 cm⁻¹ due to the stretching of the sp² C—H group. The peaks at 2863 cm⁻¹, 2904 cm⁻¹, and 2957 cm⁻¹ are due to the C—H stretch of the alkane. The characteristic peak at 1616 cm⁻¹ is ascribed to N—H bending. The peak at 1576 cm⁻¹ is due to aromatic C C stretching. The peak at 1450 cm⁻¹ is ascribed to the C—H bend of alkanes. The peak at 1419 cm⁻¹ is ascribed to the stretching of the C—C in-ring aromatic group. The peak at 1361 cm⁻¹ is ascribed to the C—H rocking of alkanes. The peaks at 1329 cm⁻¹ and 1253 cm⁻¹ are ascribed to the C—N stretching of aromatic amines.

Rose oil showed a broad peak at 3361 cm⁻¹, which is ascribed to the -OH vibrations of the major components, such as citronellol, nerol, and geraniol. Sharp peaks at 2959 cm⁻¹ and 2926 cm⁻¹ are ascribed to the -CH₃ and -CH₂ asymmetric vibrations, respectively. The peak at 2872 cm⁻¹ is ascribed to the symmetric vibration of -CH₂. A very weak peak at 2728 cm⁻¹ is ascribed to the C—H stretching of the aldehyde group. Weak peaks at 1720 cm⁻¹ and 1665 cm⁻¹ are attributed to the stretching of the C O bond. A characteristic peak at 1450 cm⁻¹ is ascribed to in-plane bending or deformation of the -CH₃ group. A sharp peak at 1374 cm⁻¹ is due to the C-O-H deformation. The peak at 746 cm⁻¹ is due to the out-of-plane bending of C—H (Bardakçi et al., 2009).

Labrasol® showed two characteristic peaks at 2921 cm⁻¹ and 2863 cm⁻¹ due to C—H stretching. A characteristic sharp peak at 1737 cm⁻¹ is ascribed to the stretching of the C—O group. The peak at 1460 cm⁻¹ is ascribed to C—H bending. A well-defined peak at 1106 cm⁻¹ is ascribed to the C—O stretching (Alkholifi et al., 2023; Yasser et al., 2024). Water has a distinct broad vibration at 3344 cm⁻¹ ascribed to stretching of the -OH bond and a sharp vibration at 1636 cm⁻¹ due to bending of the H-O-H bond.

Tween® 80 demonstrated vibrational peaks at 3491.65 cm⁻¹ due to alcoholic -O-H stretching, 2921.47 cm⁻¹ due to -C-H asymmetric stretching, 2856.56 cm⁻¹ due to -C-H symmetric stretching, 1735.35 cm⁻¹ due to esteric -C=O stretching, 1645.73 cm⁻¹ due to amidic C O stretching. The peaks at 1456.64 cm⁻¹ and 1349.44 cm⁻¹ are ascribed to -C-H bending of the alkane group, and 1296.58 cm⁻¹ to 1091.75 cm⁻¹ are ascribed to C—O stretching of esters, alcohols, or ether groups (Mohapatra et al., 2023).

Transcutol® P spectra showed a peak at 3437.78 cm⁻¹ due to O—H stretching, 2975.64 cm⁻¹ due to C—H aliphatic vibration, 2864.75 cm⁻¹ due to methylene stretch, 1454.71 cm⁻¹ due to CH₃ scissoring, and 1349.67 cm⁻¹ due to CH₂ scissoring, 1108.36 cm⁻¹ due to C—O stretching, and 1068.75 cm⁻¹ due to O—H bending (Abdallah et al., 2023; Pervaiz et al., 2026). The ATR-FTIR spectrum of the physical mixture is a superposition of all the characteristic peaks of SH-17023, rose oil, Labrasol®, Tween® 80, Transcutol® P, and water. All the characteristic peaks of SH-17023 were retained in the physical mixture without any shifting or disappearance. This ruled out the possibility of drug-excipient interaction and incompatibility. The blank NE/placebo NE showed the characteristic peaks of rose oil, Labrasol®, Tween® 80, Transcutol® P, and water. Further, ATR-FTIR results are also useful for efficient investigation of encapsulated drug molecules (Yasser et al., 2024). The spectrum of the NE showed the dominating peaks of S_mix due to its higher amount in the formulation. The vibration spectrum of the optimized NE showed the disappearance of all the characteristic peaks of the drug, reflecting successful entrapment within the nanoemulsion. The ATR-FTIR result of the NE is in accordance with previous reports, where ciprofloxacin-loaded ophthalmic NEs (Youssef et al., 2021), osthole-loaded ophthalmic microemulsions (Zhang et al., 2023), and luliconazole-loaded ophthalmic NEs (Yang et al., 2022) did not show the characteristic vibrational peaks of the drug due to solubilization and successful entrapment in the oil core. The overall results of ATR-FTIR demonstrated drug-excipient compatibility and successful entrapment of the drug inside the NE.

3.5.14.2. X-Ray diffraction

The XRD study of NEs allows evaluation of the physical state/crystallinity of the drug, the presence of surface-adsorbed drugs, drug-excipient interactions, and the effective loading of the drug into the oil globules of NEs. A halo XRD diffraction pattern represents an amorphous nature, whereas sharp diffractions indicate crystallinity. Any shifting of the diffraction peaks represents a possible drug-excipient interaction. The XRD results of NEs and their excipients are shown in Fig. 6d. The XRD pattern of rose oil, Labrasol®, Tween® 80, and Transcutol® P showed halo diffraction patterns with very small notches at 2θ of 16.62° and 13.05°, representing an amorphous nature. SH-17023 also displayed a halo diffraction pattern, with very small diffractions at 2θ of 31.85° and 45.41° representing the amorphous state. The physical mixture of all formulation components showed superimposed peaks of all the diffractions, representing compatibility among the drug and excipients. The optimized NE demonstrated similar diffraction to that of blank NE without any small diffractions of SH-17023 (31.85° and 45.41°), representing successful loading of drugs into the core of the emulsion globules, absence of surface adsorbed drugs, and retention of the amorphous state of the drug in the optimized NEs.

3.5.15. Stability studies

The impact of temperature on the stability/thermodynamic stability of NEs was evaluated by heating-cooling and freeze-thaw stability studies, whereas the resistance of NEs towards sedimentation and phase separation was observed by the centrifugation study (Mohapatra et al., 2024). The optimized NE showed no sign of change in clarity, drug precipitation, cracking, creaming, or phase separation at the end of the heating-cooling and centrifugal stability study. However, it showed temporary turbidity during exposure to the freezing cycle and recovered after exposure to room temperature during the thaw cycle. Due to the change in solubility at low temperature (−20 °C) and ice formation of the external aqueous phase, it showed the signs of instability after the freeze-thaw cycle. Similar stability outcomes for NEs loaded with acyclovir were reported (Mahboobian et al., 2020). The NE showed stability under the centrifugation cycle. However, it demonstrated turbidity when stored at −21 °C and 4 °C but recovered clarity when kept at room temperature (25 °C) and 40 °C. The temporary appearance of turbidity was reported to be due to the lower drug solubility at lower temperature and coagulation of oil droplets or interfacial surface crystallization (Mahboobian et al., 2020). Further, no significant change in the Z_avg, PDI, and zeta potential was observed at the end of heating-cooling (Z_avg = 32.036 ± 0.13 nm, PDI = 0.251 ± 0.005, and zeta potential = −28.566 ± 1.167 mV) and centrifugation studies (Z_avg = 31.93 ± 0.024 nm, PDI = 0.25 ± 0.005, and zeta potential = −30.033 ± 0.518 mV). In contrast, the globule size (749.3 ± 2.174) and PDI (0.259 ± 0.019) were increased after the freeze-thaw cycle with a substantial decrease in zeta potential (−12.133 ± 0.659 mV), reflecting possible instability under such storage conditions. The percentage of drug content was 87.865 ± 3.759% and 92.859 ± 10.025% after heating-cooling cycle and freeze-thaw cycle, respectively. The retention of the higher drug content reflects the maintenance of drug integrity under such storage conditions. The ATR-FTIR spectroscopy showed no significant changes in the characteristic peaks of optimized NEs at the end of heating-cooling, freeze-thaw, and centrifugal stability study, representing the maintenance of chemical integrity (Fig. 7a).

Fig. 7.

Accelerated and kinetic stability study. (a) ATR-FTIR of optimized NEs following accelerated stability conditions (heating-cooling, freeze-thaw, and centrifugal cycle), (b & c) ATR-FTIR spectra of optimized NEs and blank NEs up to 6 months, (d, e, & f) Z_avg, PDI, and zeta potential of the NEs up to 6 months. Data were represented as mean ± SD, n = 3. Z_avg: Globule size, PDI: Polydispersity index, ATR-FTIR: Attenuated total reflectance-Fourier transform infrared spectroscopy.

The kinetic stability evaluates long-term stability. The optimized NE demonstrated excellent kinetic stability up to 6 months without any sign of drug precipitation, cracking, creaming, or phase separation. The ATR-FTIR spectroscopy revealed the maintenance of the chemical integrity without any shifting of the characteristic vibrational bands (Fig. 7b). As per the previous drug-excipient compatibility study, in the stability study, the spectral results of optimized NE (Fig. 7b) were found to be very similar to that of the blank NE (Fig. 7c) during the stability study, representing the maintenance of the stability and presence of loaded drugs without any leaching or precipitation. Further the Z_avg, PDI, and zeta potential were found to be maintained up to 6 months (Fig. 7d–f) followed by a slight increase in the globule size (7th month: 43.6 ± 0.236 nm; 8th month 45.963 ± 0.5169 nm) & PDI (7th month: 0.29 ± 0.003; 8th month: 0.28 ± 0.007), and decreased zeta potential (7th month: −21.233 ± 0.825 mV; 8th month −18.533 ± 0.188 mV). A slight increase in the globule size over time is ascribed to the coalescence of the oil globules (Ismail et al., 2020). This leads to a decrease in zeta potential and an increase in PDI value.

3.6. Ocular irritation/tolerability and toxicity studies

The ophthalmic formulations should be non-irritant and well-tolerated. The ocular tolerability of the developed NE was verified by the Draize technique, histopathological observations, slit-lamp biomicroscopy, fundoscopy, SD-OCT, and FA, and ATR-FTIR studies.

3.6.1. Draize technique

The recorded ocular irritation scores (OIS) as per the Draize scoring (Table S14) on C57BL/6 J mice eyes for each treatment are shown in Table S15-S20. The positive control (1 N NaOH) showed severe corneal abrasion, complete corneal opacity, conjunctival redness, conjunctival discharge, and severe chemosis with lids more than half closed. At the end of the study (7 days after a single application of 1 N NaOH by filter paper), the ocular irritation was still observed. The total OIS after 24 h and 7 days were found to be 11 and 7.75, respectively, for the positive control group, indicating severe ocular irritation (Table 5). However, the 0.9% NaCl as a negative control and the left eyes (untreated control) of each experimental group showed no sign of ocular irritation with the OIS of zero (Table 5).

Table 5.

Irritation scores of various treatments after 24 h and 7 days of administration.

Treatments (b.i.d. except where noted)	Corneal opacity	Conjunctiva	Chemosis/swelling	Total ocular irritation score (OIS) after 24 h	Corneal opacity	Conjunctiva	Chemosis/swelling	Total ocular irritation score (OIS) after 7 days
	After 24 h				After 7 days
0.9% NaCl	0	0	0	0	0	0	0	0
1 N NaOH (once)	4	3	4	11	3.25	3	1.5	7.75
Blank NE	0	0	0	0	0	0	0.5	0.5
Optimized NE	0	0	0	0	0	0.50	0.25	0.75

Both the blank NE and optimized NE were non-irritating to the mouse eye (Table 5) with an OIS of zero after 24 h and 0.5 (for blank NEs) and 0.75 (for optimized NEs) after 7 days of repeated treatment, which is negligible and meets the criteria for non-irritant (Mahboobian et al., 2020). The total ocular irritation score for cornea, conjunctiva, and chemosis was found to be within 0–3, representing the non-irritant nature of the optimized NEs (Fardous et al., 2022). Only slight conjunctival redness and chemosis were observed in two of the experimental mice. Apart from that, no ocular surface, systemic, or retinal toxicity was observed after repeated dosing, representing excellent ocular tolerability of the developed NEs. Further, the eyes showed no sign of tearing/allergic reaction or inflammation. Rabbits are widely used for ocular irritation studies, as their eyes are more susceptible to irritation than human eyes (Kassem et al., 2022; Mahboobian et al., 2020). However, in our case, as we have considered the C57BL/6J mice for the L-CNV model, we have evaluated the ocular irritation in those mice. The irritation potential of the developed NEs may be further evaluated in rabbit or other large animal experimental models.

3.6.2. Slit-lamp biomicroscopy, fundoscopy, SD-OCT, and FA

The slit-lamp biomicroscopy results (Fig. S12) of all experimental groups before starting the ocular irritation studies reflected healthy eyes without any sign of ocular abnormalities. The wide beam slit lamp biomicroscopy of the blank NEs and optimized NE-treated mouse eyes after 7 days of twice-daily administration (b.i.d) reflected no signs of inflammation of conjunctiva, cornea, and eyelid, whereas the narrow beam examination revealed the absence of corneal swelling, corneal inflammation, changes in corneal surface smoothness, and clarity of aqueous humor (Fig. S13). The anterior chamber was found to be normal, with unaltered cornea, healthy epithelia, free from surface abrasion and swelling, clear aqueous humor (absence of flare or cells), and intact lens. The biomicroscopy of the NE-treated eye (right eye) was similar to that of the untreated left eye of the same experimental subject and to the negative control/isotonic saline-treated eyes (0.9% NaCl). Further, the results are similar to the slit-lamp biomicroscopy results of day 0 (before starting the irritation studies) (Fig. S12). In contrast, the wide beam slit lamp biomicroscopy of the positive control (1 N NaOH) treated eyes showed a high degree of inflammation of the cornea, conjunctiva, and eyelid, and the narrow beam reflected corneal swelling, inflammation, surface roughness of the cornea, and clouding of aqueous humour (Fig. S13).

Topical application of sodium fluorescein solution evaluates the corneal surface defect using slit-lamp biomicroscopy (Yang et al., 2022). The anterior segments of the mouse eyes under blue light (fluorescein-stained eyes) photomicrographs at the end of the study revealed no sign of edema, congestion, discharge, or abnormalities in the cornea or conjunctiva, reflecting excellent tolerability of the optimized NE (Yang et al., 2022). The anterior segment of NE-treated mice showed a similar structure to that of the untreated left eye of the same experimental mouse, and the negative control (0.9% NaCl) treated mouse eyes. However, the standard irritant NaOH showed higher fluorescein uptake (compared to other treatments) due to corneal surface epithelial erosion and ocular abnormalities, and disruption of corneal barrier function, which reflected severe ocular irritation (Fig. S13).

The fundoscopy, corneal OCT, and retinal OCT results of the experimental mice eyes before starting the ocular irritation studies represented an intact ocular structure without a visible sign of abnormalities (Fig. S14). The fundoscopy results of mice eyes following repeated topical application of optimized NE and blank NE up to 7 days revealed a healthy retina with uniform retinal fibers, normal vasculature (uniformly organized and with expected branching pattern) and healthy optic nerve. It also demonstrated no signs of exudates, retinal hemorrhage, inflammatory lesions, or alterations in pigmentation. The fundoscopy results are comparable to the untreated control eyes of the same experimental mice and isotonic saline-treated eyes, demonstrating the retention of retinal integrity and non-toxic nature of the ophthalmic NEs. In contrast, the standard irritant (1 N NaOH) treated eyes demonstrated clouding of the fundus image, ocular inflammation, corneal redness, and narrowing of the eyes/pupil (Fig. 8). The fundus and retinal vasculature were difficult to visualize due to ocular clouding, vitreous opacity, and inflammatory involvement, reflecting toxicity and irritation.

Fig. 8.

Anterior eye, corneal OCT, fundus, retinal OCT, and fundus FA of 0.9% NaCl (negative control), 1 N NaOH (positive control), blank NE, and optimized NE-treated eyes after 7 days of consecutive administration of treatments. The green arrow on the fundus represents the OCT plane. Left eye of each group is untreated, whereas the right eye is treated with 0.9% NaCl, 1 N NaOH, blank NE or optimized NE. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

The SD-OCT (Fig. 8) of the cornea and retina of the optimized NE and blank NE-treated mouse eyes showed clearly identifiable and well-organized corneal and retinal layers without any abnormalities or destruction of ocular tissue layers. The OCT of the cornea revealed intact corneal epithelium, stroma, and endothelium without structural alteration, disruption, or edema. The retinal OCT reflected preservation of retinal fibers, inner and outer retinal layers, and photoreceptor/RPE layers without any abnormalities. The results are similar to those of the untreated control eyes and saline-treated eyes. In contrast, the corneal OCT results of 1 N NaOH-treated mice eyes showed loss of corneal integrity and disruption of corneal layers, representing ocular toxicity and irritation. Due to clouding of the eyes, ocular inflammation, redness, and an opaque lens, it was difficult to obtain retinal OCT for these mice.

The fundus FA of the mouse eye enables the in vivo examination of retinal vasculature, permitting the clear visualization of vascular perfusion and integrity. The fundus FA of the NE and blank NE-treated eyes revealed healthy vasculature, well-defined and homogeneously perfused retinal blood vessels without any leaky vasculature, neovascularization, capillary dropout, microaneurysm formation, or abnormalities. The integrity of the vasculature was comparable to that of the untreated control contralateral eye and the isotonic saline-treated eyes. In contrast, due to severe clouding and inflammation of the eyes, it was difficult to evaluate the FA of the eyes treated with the standard irritant.

3.6.3. Histopathology

Histopathology of ocular tissues (Fig. 9) following topical application of optimized NE and blank NE revealed intact morphological features of cornea, iris, lens, and retina, suggesting the non-irritancy/safety of the developed NE ophthalmic formulation (Al-mahallawi et al., 2021). The NE-treated eyes showed no signs of corneal swelling, surface erosion, central or limbal neovascularization, hyphema, or cataract. The histopathological reports of optimized NE and blank NE-treated eyes showed similar tissue morphology to that of isotonic saline-treated eyes and the untreated left eye, revealing excellent ocular tolerance.

Fig. 9.

Histopathology of 0.9% NaCl (negative control), 1 N NaOH (positive control), blank NE, and optimized NE-treated mice eyes after 7 days of consecutive topical administration. Scale bar: 200 μm for the complete eye (4× magnification) and 100 μm for various portions of eyes (20× magnification). Left eye of each group is untreated, whereas the right eye is treated with 0.9% NaCl, 1 N NaOH, blank NE or optimized NE.

In contrast, the standard irritant (1 N NaOH) induced significant pathology (Fig. 9). It caused corneal epithelial erosion and coagulative necrosis involving the squamous, polygonal, and basal cell layers, along with destruction of Bowman’s layer. The corneal stroma showed infiltration of polymorphonuclear leukocytes and exhibited both central and limbal neovascularization. Further, the detachment of Descemet’s membrane, endothelial hyperplasia, and associated neovascularization were also observed. The iris exhibited fragmentation, blending with endothelial cells, and damage to deeper structures of the anterior segment. The anterior chamber (aqueous humor) showed hyphema, profuse central neovascularization, and infiltration of acute inflammatory cells. The lens exhibited capsular thickening with subcapsular anterior epithelial hyperplasia, and some lens fiber cells lost their elongated architecture and became enlarged. In addition, focal lens dislocation and capsular rupture were observed, with the lens being displaced towards the cornea. The lens fibers appeared swollen and disrupted, with abnormal retention of lens fiber nuclei, consistent with cataract formation. However, the posterior chamber (vitreous humour), sclera, choroid, and retina were unremarkable in all cases. The overall histopathological outcomes of 1 N NaOH-treated mice eyes demonstrated noticeable ocular irritation and toxicity, in contrast to the NE formulations.

3.6.4. ATR-FTIR study for NE-cornea interaction evaluation and systemic effects

ATR-FTIR spectroscopy allows the investigation of the interaction of various treatments with the corneal tissue (Zhang and Aksan, 2010). Since the major components of the cornea are collagen and water, a change in the IR intensity or shifting of peaks represents protein denaturation and potential ocular interaction (Porto et al., 2024; Zhang and Aksan, 2010). The results of FTIR of the treated cornea of C57BL/6J mice eyes are shown in Fig. 10. A broad peak between 3600 and 3000 cm⁻¹ is ascribed to -OH stretching of water (Porto et al., 2024; Zhang and Aksan, 2010). The peaks at 2925.23 cm⁻¹ and 2855.67 cm⁻¹ are ascribed to the –CH₂ or –CH₃ stretching bands of the lipid membrane. The peaks at 1642.27 cm⁻¹ and 1546.68 cm⁻¹ are ascribed to the C O stretching of collagen amide I and the NH₂ bending of amide II, respectively (Porto et al., 2024). A very weak vibrational peak at 1248.36 cm⁻¹ is ascribed to collagen amide III C—N stretching coupled with N—H bending (Zhang and Aksan, 2010). A significant decrease in the intensity of the amide II peak at 1546.68 cm⁻¹ was noticed in the cornea of 1 N NaOH-treated eyes without any shifting of peaks, reflecting protein denaturation and potential ocular interaction. The NaOH causes alkali burn, leading to denaturation of protein, disruption of collagen, and saponification of lipid, which was noticed by decreased intensity of the FTIR spectrum. In contrast, no significant frequency shift or intensity change of the FTIR spectra was observed in the cornea that received various other treatments (0.9% NaCl, blank NE, and optimized NE), and the spectra are similar to that of the untreated control eye (left eye), reflecting the retention of corneal integrity without any degradation of corneal protein or potential ocular interaction. The overall FTIR spectral results reflected the ocular safety without any significant interaction or deterioration of ocular tissues.

Fig. 10.

ATR-FTIR spectroscopy of the mouse cornea and body weight measurements of the experimental mice for ocular irritation and toxicity studies. (a) ATR-FTIR spectra of the mouse cornea treated with 0.9% NaCl (negative control), 1 N NaOH (positive control), blank NE, and optimized NE after 7 days of consecutive administration of treatments. The light colour indicates the FTIR spectrum of the untreated control eyes (left eye) of mice, whereas the dark colour indicates the FTIR spectrum of various treatments assigned to the right eye of the same experimental mouse. (b-e) Body weight measurements of the experimental mice treated with (b) 0.9% NaCl, (c) 1 N NaOH, (d) blank NE, and (e) optimized NE. M1 and M2 in each group represent male mice, whereas F1 and F2 represent female mice.

The body weight of all the treatment groups is shown in Table S21 and Fig. 10b-e. The steady increase in body weight throughout the ocular irritation study indicated good health in the experimental mice, with no systemic abnormalities. The ocular safety and tolerability of NEs are attributed to the compatibility of the contained excipients (Labrasol®, Tween® 80, Transcutol® P) of NEs with the ocular tissue. The ocular safety profile of these excipients has been previously described in Section 3.4. Based on the irritation studies, the NE formulation showed its potential for ophthalmic use without causing significant irritation or ocular toxicity.

3.7. Therapeutic effect of SH-17023 NE on L-CNV

The murine L-CNV model is widely used to mimic neovascularization in nAMD (Chobisa et al., 2024; Gong et al., 2015). This model uses an argon laser that ruptures a specific point in the interface of Bruch’s membrane-choroid complex, which induces angiogenesis from the choroid region into the subretinal space, mimicking clinical features of human nAMD (Chobisa et al., 2024; Gong et al., 2015). The therapeutic effect of treatments was evaluated on the L-CNV model up to 7 days via fundoscopy, OCT, FA, and ex vivo vasculature staining (Fig. 11 and Fig. 12) (Chobisa et al., 2024; Gong et al., 2015). C57BL/6J mice of both sexes were used in the L-CNV experiment, as pigmented mouse eyes absorb laser energy and respond consistently to laser photocoagulation in the eyes (Gong et al., 2015). The NEs did not cause any ocular abnormality during in vivo therapeutic activity evaluation at a dose of 15 μL/eye (equivalent to 0.75 mg SH-17023/eye), as observed from the fundoscopy, OCT, FA, (Fig. 11a) and ex vivo evaluations of the choroid (Fig. 12a).

Fig. 11.

In vivo therapeutic activity of SH-17023-loaded optimized NE (15 μL/eye, equivalent to 0.75 mg/eye) reduces murine L-CNV. (a) Representative photomicrographs of L-CNV (fundus) on the day of laser (day 0), fundus, OCT, and FA images obtained 7 days post L-CNV of blank NE and SH-17023-loaded optimized NE (15 μL/eye, equivalent to 0.75 mg/eye) treated C57BL/6 J mice eyes. (b) Quantitative results of in situ neovascularization lesion volume from OCT taken after 7 days. Mean ± SD (n = 12 eyes per group) ***p < 0.001 vs. vehicle treatment; unpaired t-test. (c) bar diagram representing lesion severity in terms of lesion grade evaluated from FA scoring of the eyes after 7 days. The topical administration of SH-17023-loaded optimized NE significantly reduced CNV lesion volume (yellow brackets) on OCT and lesion severity (FA). The black line on the fundus (a) represents the OCT plane. OCT scale bars: 100 μm. NEs: Nanoemulsions, L-CNV: Laser-induced Choroidal Neovascularization, OCT: Optical Coherence Tomography, FA: Fluorescein angiography. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 12.

In vivo therapeutic activity of SH-17023-loaded optimized NEs (15 μL/eye, equivalent to 0.75 mg/eye) reduces murine L-CNV. (a & b) Representative photomicrographs of DAPI and lectin (GS-IB4) stained CNV lesion of blank NE and SH-17023-loaded optimized NE treated C57BL/6 J mice eyes. (c) quantitative lesion volume calculated from the Z-stacked images of RPE-choroid lesions stained with vascular marker (GS-IB4). Mean ± SEM (n = 12 flat mounts per group) *p < 0.05 vs. vehicle treatment; unpaired t-test. The topical administration of SH-17023-loaded optimized NEs (0.75 mg/eye b.i.d) significantly reduced CNV lesion volume in GS-IB4 (green) stained choroidal flat mounts. Mean ± SD (n = 6 mice/treatment) **p < 0.01 vs. vehicle treatment; unpaired t-test. Scale bars: 200 μm and 50 μm (flatmount/GS-IB4 stained lesions). NEs: Nanoemulsions, L-CNV: Laser-induced Choroidal Neovascularization, DAPI: 4′,6-diamidino-2-phenylindole, GS-IB4: Griffonia simplicifolia-isolectin B4. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.7.1. Fundoscopy and OCT

Fundoscopy includes visualization of the interior ocular surface, including retina, its vasculature, optic disc, and posterior pole, whereas OCT is a non-invasive ocular imaging technique that captures high-resolution cross-sectional images of retina and L-CNV lesions, which demonstrate the extent of choroidal neovascularization (Chobisa et al., 2024). The lesions of various treatment groups were monitored to study the in vivo CNV progression by fundoscopy and OCT after 7 days of laser challenge. After 7 days of treatment, the optimized NE showed a significant reduction of OCT-derived lesion volume (p < 0.05) compared to the blank NE group (Fig. 11b).

3.7.2. Fluorescein angiography

Fundus FA revealed vascular leakage of the CNV lesions. The representative FA photomicrographs of various groups are shown in Fig. 11a. Qualitatively, the FA leakage of the optimized NE-treated group is smaller (low fluorescence or almost no fluorescence) compared to the blank NEs-treated group (saturated fluorescence). In contrast, the sham group showed no FA leakage due to the absence of a laser burn. In the semi-quantitative score of FA lesion severity, in terms of lesion grades (grade 0–3) the optimized NE-treated group showed a relatively low percentage of grade 2 and grade 3 lesions and a high percentage of grade 0 and grade 1 lesions compared to the blank NE group (Fig. 11c).

3.7.3. Ex vivo immunostaining and lesion volume quantification

The extent of neovascularization was further investigated in ex vivo choroidal flat mounts of CNV lesions using GS-IB4 as a vasculature marker (Fig. 12). This provides a complementary evaluation technique for quantifying choroidal lesion volumes from z-stacked images. Representative photomicrographs of choroidal flat mounts stained with DAPI and isolectin at 5× magnification and at 20× magnification are shown in Fig. 12a and Fig. 12b, respectively. The quantitative lesion volumes for various treatments, calculated from Z-stacked images of lesions stained with a vasculature marker, are shown in Fig. 12c. The lesions in the blank NE groups showed typical L-CNV morphology (Fig. 12a and Fig. 12b). The mean lesion volume of the optimized NE-treated group showed significantly lower lesion volume (p < 0.05) compared to the blank NE group, which is consistent with the results of OCT quantification. Considering the lesion volume quantification via OCT, ex vivo vasculature staining, and semi-quantitative lesion grading using FA, the optimized NE demonstrated a significant reduction in CNV volume of >45%, without ocular irritation or toxicity.

Earlier reports from our lab showed that SH-17023 inhibits FECH and causes PPIX accumulation in Human Retinal Endothelial Cells (HRECs). The compou also demonstrated in vitro anti-angiogenic activity through inhibition of HREC proliferation, migration, and tube formation. Further, it is verified for in vivo therapeutic activity in the L-CNV model via intravitreal injection (Sishtla et al., 2022). However, the intravitreal injections cause significant patient and health system burden and access challenges. By avoiding the challenges of posterior ocular drug delivery, the NE formulation developed here demonstrated a successful therapeutic option that reduces dependence on intravitreal injections, thereby having potential for improving patient compliance and quality of life plus enhancing care in resource-limited settings where intravitreal injection access may be difficult. Compared to the simple eye drops (drug solution), the optimized NEs represented enhanced trans corneal permeation and sustained drug release, representing their suitability for ophthalmic drug delivery for the effective treatment of CNV.

4. Conclusion

Ophthalmic NEs loaded with a novel FECH inhibitor were developed for the treatment of CNV. The NEs were prepared by the spontaneous emulsification method and optimized using QbD-based D-optimal mixture design for globule size, polydispersity, and loading capacity. The formulation demonstrated nanometric globule size, with narrow size distribution, and excellent loading capacity. Further, it demonstrated excellent pharmaceutical properties, sustained drug release, drug-excipient compatibility, and stability. It demonstrated enhanced ex vivo ocular permeability compared to the drug solution. Finally, it demonstrated excellent in vivo ocular tolerability and therapeutic activity in the L-CNV murine model without significant toxicity or irritation. The topical NEs of novel FECH inhibitor (SH-17023) have promise for further development as an adjunct or alternative to anti-VEGF therapies for CNV.

CRediT authorship contribution statement

Debadatta Mohapatra: Writing – review & editing, Writing – original draft, Visualization, Methodology, Investigation, Formal analysis, Conceptualization. Seung-Yong Seo: Writing – review & editing, Resources. Timothy W. Corson: Writing – review & editing, Supervision, Resources, Project administration, Funding acquisition, Formal analysis, Conceptualization.

Funding

This work was supported by grants to T.W.C. from the National Institutes of Health/National Eye Institute (NIH/NEI) [grant numbers R01EY025641, R01EY035159]; the Natural Sciences and Engineering Research Council of Canada (NSERC) [grant number RGPIN-2025-04563]; the Retina Research Foundation; the Ontario Research Fund; and the Canada Foundation for Innovation (CFI).

Declaration of competing interest

DM, SYS, and TWC are inventors on patents related to this work.

Appendix A. Supplementary data

Supplementary material: Supplementary material, including supplementary materials and methods, tables, and figures.

Data availability

Data are available in the supplementary files and the raw data will be made available on request.