Sinapic acid, cholesterol, L-a phosphatidylcholine (soya lecithin), and lactose monohydrate were supplied by Sigma-Aldrich (St. Louis, MO). Tween 80 (polyoxyethylene sorbitan monooleate) of HLB = 15 and sodium deoxycholate (HLB = 16); cellophane membrane with a (12,000 kDa) molecular weight cut-off; methanol; HPLC-grade acetonitrile; and absolute ethanol were purchased from Fisher Scientific United Kingdom Ltd. (Loughborough Leicestershire, United Kingdom).

Thin-film hydration was used to create SA-loaded transferosomes, with slight modifications identified by preliminary studies (Gamal et al., 2020). Briefly, absolute ethanol was used to dissolve SA (10 mg), surfactant, cholesterol, lactose monohydrate (1,250 mg), and phosphatidylcholine. After that, the mixture was sonicated for 10 min and then moved into a round flask (100 mL). The absolute ethanol was evaporated in a vacuum employing a rotary evaporator (Stuart, United Kingdom) for 1 h at 40°C. Consequently, the obtained dried film was hydrated via 10 mL of an isotonic phosphate buffer (pH 7.4) for 1 h at 60°C. To transform the large vesicles into smaller ones, they were subjected to sonication for 20 min. After that, the developed protransferosomes were separated at 20,000 rpm for 3 h at 4°C in a cooling centrifuge (Hettich Zentrifugen, Germany), and then they were washed using isotonic phosphate buffer. The produced transferosomes were kept in tightly closed amber tubes at 4°C for further research.

User-defined response-surface design was employed using Design-Expert software (version 10.0.0.3, United States) for the optimization procedure (Table 1). Twelve transferosomal formulations were developed (Table 2) using a ratio (w/w) of cholesterol to phospholipid (X1), surfactant type (X2), and the ratio of phospholipid to surfactants (X3) as independent variables. Consequently, the tested dependent variables were particle size (Y1), entrapment efficiency (Y2), and release percentage (Y3). Dependent and independent variables were selected according to literature review and preliminary studies.

The statistical significance obtained from the investigated data of each response was estimated using the ANOVA test at p-value < 0.05. The predicted and adjusted coefficients of determination (R2) were used to determine if the chosen model was suitable for the investigated data. The signal-to-noise ratio was determined using adequate precision to confirm that the applied model may well navigate the model’s design spaces (Das et al., 2015). The optimization process also depended on using the highest desirability index (Gamal et al., 2020). The relationship between different factors and each response was explained according to the polynomial equations (Equations 1–3; Table 3).

The optimized SA-loaded transferosomes were chosen based on the following criteria: maximizing the in vitro released SA, maximizing the EE%, and minimizing the mean particle size. Consequently, the selected optimized formula was developed using thin-film hydration, as previously described, and subjected to further characterization studies. As a result, the optimized transferosomes were prepared and analyzed for vesicle size, PDI, EE%, and % of SA released, as previously mentioned.

The optimal SA-protransferosomes were dried via freeze-dryer EF03 (Edwards High Vacuum, Ltd.) to develop protransferosomes powder. Then, the lyophilized optimal SA protransferosomes were applied to the following characterization studies.

The in vivo performance of the optimal formulation was examined utilizing in vivo pharmacokinetic examination. In addition, the bioavailability of SA was studied in a rabbit model following nebulization of the re-dispersed optimal lyophilized SA protransferosomes and oral administration of the SA suspension.

A total of 12 adult albino rabbits weighing 1.35–1.60 kg were distributed into two groups. Group 1 received 10 mg of the SA suspension orally with the help of a gauge needle (Günther et al., 2004). In comparison, group 2 received a suspension of the optimized lyophilized protransferosomes (10 mg of SA) through nebulization after anesthesia using diethyl ether (Gamal et al., 2020; Pastor et al., 2015). The protocol developed for this in vivo experiment was certified through the Animal Ethics Committee of Nahda University. (Regd. No. NUB-032-022). Blood samples were taken using EDTA tubes at various intervals for 24 h; then, they were centrifuged for 10 min at 5,000 rpm, and then plasma separation and HPLC-UV analysis were performed.

The plasma samples were analyzed using Dionex Ultimate 3000 UHPLC, supplied with an auto-sampler, a diode array detector (Massachusetts, United States), and a quaternary solvent delivery pump. The C18 reverse-phase column ZORBAX Eclipse Plus^® (5 µm, 250 × 4.6 mm) (California, United States) was used for the development and quantitation. The internal standard was ciprofloxacin, while the mobile phase was water–acetonitrile (2:98% v/v) at a 1.0 mL/min flow rate. The test temperature was set to 25°C, and UV detection was carried out at 320 nm. The output signal was checked and processed using Chromeleon software. Different pharmacokinetic parameters, such as MRT, AUC, t_1/2, T_max, and C_max, were computed via the PKSolver program. The statistical data analysis utilized SPSS 16.0 software (Chicago, IL).

The user-defined design was a proper statistical design-of-experiments (DOE) technique, and it was used to perform the optimization process. DOE was utilized for this study to estimate the model parameters because it lowered the variance of the constraints estimator due to its efficient norm reduction of the covariance matrix. In addition, it was suggested that applying a DOE design would lead to fewer samples being needed, as compared to other designs, while not impacting the estimated error and variance, which was strongly desired (Ali et al., 2021).

Twelve transferosomes were developed corresponding to the user-defined experimental design, and then they were characterized for numerous responses. In addition, the interactions among the examined independent and dependent variables were demonstrated mathematically using polynomial equations (Equations 1–3; Table 3).

The quadratic approach was observed as the best-fitting model for the Y1, Y2, and Y3 variables. All the obtained models were significant at a p-value of 0.05. The positive coefficient for the impact of each factor on the measured responses was considered synergistic, while the antagonistic influence was determined by a negative coefficient (Salem et al., 2019a). All the responses indicated good agreement between the predicted R2 of the determination and the adjusted R2.

Given that the applied model’s signal-to-noise ratio had a sufficient degree of precision, that is, more significant than 4, it could be used to explore the design space (Table 3; El-Menshawe et al., 2018). A response-surface analysis was performed to explain how certain independent variables had affected the observed responses. It was clear from the resulting equations (Table 3) that the intercept had a positive influence on each of the three responses. Figure 1 shows how the independent variables affected the measured responses.

The ANOVA test results showed that each of the tested factors significantly affected the vesicle size at a p-value < 0.05, except X2. The size of the transferosomes varied from 141.7 ± 9.95 to 500.90 ± 9.27 nm (Table 2). The examined factors (X1 and X3) had an antagonistic influence on the transferosome mean size, excluding X2, which had a synergistic impact. The increase in the concentration of phospholipid or cholesterol resulted in a significant increase in the transferosome size, while the increase in the surfactant concentration significantly decreased the transferosome size (Khan et al., 2020).

Our results agreed with those of Moawad et al. (2017) as we found that the Tween 80 (HLB = 15) samples showed smaller transferosomes, as compared to the sodium deoxycholate samples (HLB = 16). Despite the higher HLB of sodium deoxycholate, the samples showed larger vesicular sizes, explained by their anionic character. This resulted in a negative charge and caused a powerful repulsion force to develop among the lamellae of the vesicles. Consequently, the aqueous core size increased, producing larger vesicles, as previously mentioned (Moawad et al., 2017). The homogeneity of the vesicular dispersions was assessed using the PDI. A system with a PDI of 0 had mono-dispersed vesicles, whereas a system with a PDI of 1 had high poly-dispersed vesicles (Song et al., 2022). The obtained PDI of the SA-loaded transferosomes ranged from 0.103 ± 0.06 to 1.00 ± 0.25, indicating that some transferosomes had a balanced size distribution and homogeneity (Table 2).

The zeta potential (ZP) is regarded as an essential measure of the physical stability of nano-formulations. The high repelling forces between the vesicles and the large ZP on the surfaces (positive or negative) minimized the aggregation of the vesicles and provided more sustained vesicular stability. The transferosome stability resulted from the decreased interfacial tension caused by using a high surfactant concentration. In addition, the prepared transferosomes had a negative charge ranging from −18.1 to −42.1 mV (Table 2). This high negative charge caused steric stabilization and electrostatic repulsion (Salem et al., 2019c). Nanoparticles have been demonstrated to have a stronger affinity for biological membranes due to the predominant electrostatic interactions. The uptake of strongly negatively charged nanoparticles was preferred by the cells despite the presence of extensive negative charges in the biological membranes due to the predominance of the non-specific interaction mechanisms, as well as the cluster arrangement of the nanoparticles (Salem et al., 2019c).

The ANOVA results indicated that each of the tested factors significantly affected the EE% at a p-value < 0.05, except X2. As shown in Table 2, the entrapment efficiency of the transferosomes varied from 76.28% ± 2.94% to 89.53% ± 3.61%. The high SA entrapment tendency may have been caused by the hydrophobicity of SA, which made it insoluble in water (Shakeel et al., 2016a).

The investigated variables (X1 and X3) revealed synergistic interactions, as compared to variable X2, which displayed a significant negative interaction. It was evident that SA entrapment had been increased by the higher ratio of phospholipids to surfactants. These results could have been related to the liquid membrane rupture caused by the high surfactant concentration, which weakened the membrane entrapping the SA (López et al., 1998). Consequently, the increase in surfactant concentration showed a decline in EE%.

Regardless of the kind of surfactant, the percentage of entrapped SA increased with higher cholesterol-to-phospholipid ratio. The transferosomal membrane became highly rigid, and the spaces between the phospholipid and surfactant molecules were filled with cholesterol. This reduced the drug leakage from the vesicles through the cholesterol interaction with the membrane core during the annealing process. Accordingly, the capacity of the transferosomal bilayer membrane to entrap hydrophobic drugs was increased; this might also have been due to the phospholipid bilayer conversion to the solid-phase bilayer as a result of the inclusion of cholesterol (Gyanewali et al., 2021; Khan et al., 2021).

The ANOVA results showed that each tested factor significantly affected the SA release at p-value < 0.05. The examined factors X1, X2, and X3 had a significant antagonistic impact on the release percentage.

As shown in Figure 2, the examined release varied from 19.74% ± 2.26% to 47.47% ± 1.19%. The increase in the ratio of phospholipid to cholesterol resulted in a significant decrease in the SA released. In contrast, the increase in surfactant concentration significantly increased SA release, owing to the added rigidity and the hydrophobicity of the prepared transferosomes caused by the cholesterol (Abd-Elbary et al., 2008). Using Tween 80 as a surfactant significantly increased SA release, as compared to use of sodium deoxycholate. These findings agreed with those of both Moawad et al. (2017) and Jain et al. (2003).

The optimization process for the prepared transferosomes aimed to identify the levels of the independent variables required to develop the transferosomes with optimal characteristics. The numerical optimization of the Design-Expert^® program was utilized to select an optimal formulation from 12 prepared transferosomal formulations in compliance with the DOE design. The following optimization criteria were used: (i) achievement of maximum percentage of in vitro release; (ii) maximum percentage of entrapment efficiency; (iii) minimal size of transferosomes. Following the numerical optimization, the formulation composed of Tween 80 as a surfactant (85:15% w/w as a ratio of phospholipid to Tween 80) and 1:2 w/w as a ratio of cholesterol to phospholipid was selected as the optimized SA-loaded transferosomes. The developed optimized formulation showed a desirability index of 0.678, an in vitro release of 28.03% ± 2.3%, an EE of 89.47% ± 3.01%, and a mean vesicle size of 232.13 ± 3.30. The observed study and predicted model achieved an acceptable level of agreement, proving the validity of the model.

The re-dispersed lyophilized protransferosomes had a vesicle size of 270.2 ± 5.1 nm and a drug content of 98.11% ± 3.6%. The in vitro percentage of the release of SA from the optimal lyophilized transferosome was 32.8% ± 2.1%, as displayed in Figure 3.

Both SA and its protransferosomes demonstrated good inhibition for most of the bacterial and fungal pathogens tested (Table 4). MIC for the antibacterial activity demonstrated enhanced activity of SA protransferosomes compared to the pure SA, moving it much closer to the positive control (Table 4). Interestingly, in the case of E. faecalis and S. mutans, although the pure SA did not show good activity, SA protransferosomes demonstrated comparable effects to ciprofloxacin, the positive control. Additionally, SA protransferosomes demonstrated comparable MIC to nystatin, the positive antifungal control drug, against both C. albicans and A. niger (Table 4) and a more enhanced effect compared to pure SA.

Phenolic compounds such as SA can exhibit varying antibacterial activity against different bacterial pathogens due to differences in cell envelope composition, permeability, and resistance mechanisms. For example, Gram-negative bacteria possess an outer membrane that can hinder the penetration of phenolics, reducing their efficacy, while Gram-positive bacteria lack this barrier, making them more susceptible. Additionally, variations in membrane-bound enzymes, efflux pumps, and intracellular targets influence how each strain responds to phenolic stress. The compound’s activity may also depend on its ability to disrupt membrane integrity, generate oxidative stress, or denature proteins, mechanisms that can vary in effectiveness depending on the bacterial species and strain.

The higher MIC of SA protransferosomes vs. ciprofloxacin reflects its superior potency, while the lower MIC vs. free SA highlighted the benefits of transfersomes in improving the activity compared to the pure SA, due to its ultra-deformability that enhances penetration through biological membranes due to their elasticity. Transferosomes retain SA longer, so they provide prolonged release, which sustains its activity longer. It also provides a shield for SA from enzymatic degradation through encapsulation. These properties regarding transfersomes may be the explanation for its enhanced activity (Aguilar-Toalá et al., 2022; Opatha et al., 2020).

An in vitro antiviral assay of SA against SARS-CoV-2 showed an IC₅₀ value of 2.29 ± 0.065 μg/mL. This value was comparable to our previous report, where IC₅₀ had been found to be 2.69 μg/mL (Orfali et al., 2021). The IC₅₀ of the reference antiviral drug, remdesivir, was 0.036 ± 0.002 μg/mL. Both SA and remdesivir showed significantly low cytotoxicity to the host cells, with a CC₅₀ of 188.33 ± 10.8 and 170.74 μg/mL, respectively.

The newly formulated SA protransferosomes showed significantly enhanced antiviral activity: 50% viral growth inhibition at significantly reduced concentrations (IC₅₀ = 0.016 ± 0.008 μg/mL; the IC₅₀ of the lyophilized SA formulation = 0.0179 ± 0.0089 μg/mL). Hence, this new formulation was more potent than remdesivir. Accordingly, the present nano-formulation improved the cellular delivery of SA.

The enhanced antiviral activity of SA-loaded protransferosomes can be attributed to the unique physicochemical properties of transferosomes, particularly their ultra-deformability. This high flexibility allows transferosomes to traverse biological barriers more effectively than conventional liposomes by squeezing through tight intercellular spaces and facilitating fusion with cellular membranes, thereby improving intracellular delivery and bioavailability of SA (Elsayed et al., 2006; Gupta et al., 2005). Additionally, transferosomes offer a sustained release profile, prolonging the retention and therapeutic effect of SA (Song et al., 2012). The encapsulation also protects SA from enzymatic degradation, preserving its structural integrity and antiviral efficacy (Aguilar-Toalá et al., 2022; Opatha et al., 2020). These combined features likely contribute to the superior antiviral performance observed with SA protransferosomes. Of course, this should be supported by future in vivo studies.

Each of the calculated pharmacokinetic parameters was determined by constructing a plasma–concentration time curve, as presented in Figure 7. The relative bioavailability of the nebulized SA protransferosomes was 6.1-fold of the oral SA. The AUC for the SA protransferosomes was significantly more than the AUC of the oral SA, by a factor of 6.1 (p = 0.001). The increased bioavailability of the SA protransferosomes can be attributed to their ability to bypass the first-pass hepatic processing. Whereas the oral SA had a T_max of 2 h and a C_max of 118.136 μg/mL, the SA protransferosomes had a T_max of 4 h and a C_max of 287.116 μg/mL, as shown in Table 5. Moreover, the SA protransferosomes had an increased MRT and t1/2 with a considerable (p = 0.001) increase and effectively maintained their plasma concentration. The longer T_max, which was correlated with the release date, may have been caused by the prolonged SA release into the alveolar fluid from the protransferosomes before it could be absorbed via the systemic circulation (Gamal et al., 2020; Saimi et al., 2021).