3.2. Experimental Design: CER-NP-Loaded Liposome Optimization
The optimization of CER-NP-loaded liposomes was carried out based on RSM in this study. The intercellular matrix in the
stratum corneum consists of different ceramides, including CER-NP, fatty acids, and cholesterol, in a specific range. Through the RSM, CER-NP and other lipid substituents are used in harmony. The responses for the design were particle size and PDI value of liposomes. During the optimization process, the concentration of CER-NP was 2.19% of the total lipid amount, which is the value of lipid replacement therapy [
35]. The oleic acid and cholesterol were in more than CER-NP molar concentrations to improve the release and diffusive properties of CER-NP. At this point, the RSM could have assisted the formulation optimization by specifying the lipid substituent amounts and maintaining the quality of CER-NP-loaded liposome.
Non-ionic surfactants such as Tween 80 and sodium cholate are widely used to form liposomes. However, destroying the orthorhombic lateral packing of the lipid barrier is observed due to applying a cosmetic formulation that includes ionic and non-ionic surfactants [
36]. However, phosphatidylcholine needs surfactants to form a vesicle structure, even if this could be cholesterol. In this study, it is a priority to use cholesterol, which occurs in the skin’s extracellular matrix, as a surface-active agent in encapsulating the essential amount of CER-NP.
In the pre-formulation process, CER-NP and cholesterol-containing liposome formulations do not have a fine particle size, and the PDI value indicates the issue of the low solubility of CER-NP and membrane fluidity derived from cholesterol [
37]. In the same manner, the presence of oleic acid changes the particle size and PDI value response in the pre-formulation study. However, the CER-NP-loaded liposomes should contain CER-NP, fatty acid, and cholesterol at an optimum ratio. With this design, the particle size response was minimized to accumulate the CER-NP-loaded liposome on the
stratum corneum. The stability of CER-NP-loaded liposomes is typically associated with PDI values and zeta potential. The goal of this study was to minimize and reach a PDI value of less than 0.3 and a zeta potential of −15 < z < +15 mV [
38].
3.3. Determination of Responses: Particle Size and PDI Value
In the optimization process, there are various models, such as quadratic, cubic, and factorial, to fit the response data in an efficient model. In our case, it was the most efficient model, quadratic for both particle size and PDI value based on the Design of Expert. The equation defining the quadratic model is provided in Equation (2).
Equation (2): Quadratic model equation for particle size and PDI value responses.
The predicted value is defined as Y in the equation. A and B are independent variables. β
1 and β
2 define linear coefficients; β
12 represents interaction coefficients; β
11 and β
22 define quadratic coefficients. The formulation inputs about material attributes are listed in
Table 3. During the optimization process, the formulations listed in
Table 3 were prepared and examined to take the responses.
The second-order quadratic equation was used when minimizing the particle size. The predicted particle size was calculated via Equation (3), which is provided below.
Equation (3): Particle size responses equation based on quadratic model.
The statistical analysis of variances is provided in
Table 4 for particle size. The particle size responses showed a good F-value (26.91) and low
p-value (<0.0001), which means the equation of the model was significant. The lack-of-fit value was insignificant (
p = 0.9384), which means that the model fitting is fine. The adjusted and predicted R
2 values are 0.9247 and 0.8966, respectively. The values show that the experimental and predicted values are correlated.
The second-order quadratic equation provided below was used when minimizing the PDI value.
Equation (4): PDI value responses equation based on quadratic model.
The statistical analysis of PDI value responses is provided in
Table 5. The PDI value responses show a good F-value (16.88) and low
p-value (0.0001), indicating a significant model equation. The lack-of-fit value was not significant
(p = 0.6827), which indicates fitting the model is fine. The adjusted and predicted R
2 values are 0.8827 and 0.7587, respectively. The values show that the experimental and predicted values are correlated.
The desirability factor, which is crucial in optimization methods including multi-response parameters, was found to be 0.9387 in this optimization method. The value proves that the experimental value would meet the target of the predicted value.
According to the statistical analysis, CER-NP (factor A) is not significantly effective on both the particle size and PDI value (
p = 0.7028 and 0.6571). However, cholesterol (factor B) and oleic acid (factor C) were significantly effective on the particle size and PDI value (
p < 0.05). The one-factor analysis shows the individual effect of each factor on particle size and PDI values. The one-factor analysis is provided in
Figure 2.
Figure 2a,d shows the insignificant effect of CER-NP on particle size and PDI values.
Figure 2b,e explains the point that the cholesterol increase raised the particle size and PDI value. The increase in oleic acid reduced the particle size and PDI value, as shown in
Figure 2c,f.
Three-dimensional response surface graphs show the effects of the relations of ingredients.
Figure 3a shows the effect of relations between oleic acid and cholesterol; meanwhile, CER-NP amount is maximized.
Figure 3c indicates that oleic acid has a crucial effect on particle size in maximized cholesterol with any levels of CER-NP. The maximum amount of oleic acid can keep the particle size under control in many levels of cholesterol and CER-NP based on
Figure 3e.
Figure 3b shows that the PDI value was controlled by cholesterol in maximized CER-NP with any levels of oleic acid. However, the CER-NP and oleic acid may have an optimum ratio to obtain a minimum PDI value in the maximum level of cholesterol, according to
Figure 3d. When oleic acid is maximized, an increasing amount of cholesterol can reduce the PDI value at a maximum level of CER-NP in
Figure 3f.
The idea about raising particle size by high cholesterol amounts in the low level of oleic acid is supported by studies on liposomes containing cholesterol [
39]. It is reported that an overrated amount of sterol causes an increase in the mean diameter of the liposome [
40]. The increase in particle size might be related to interactions between lipid chains close to the head group of phosphatidylcholine and membrane stretching in the presence of sterols. Therefore, by increasing the cholesterol ratio, more than adequate cholesterol molecules may diffuse the phosphatidylcholine bilayer membrane. For this reason, the membrane fluidity is increased, and the close packing on the phosphatidylcholine bilayer membrane is prevented.
In addition, the study reports that the ceramide/stearic acid/cholesterol-based liposomal membrane shows that the disarrangement of molecules and phase softening occurred because of the increase in cholesterol concentration [
41]. In our study, the oleic acid provides advantages on this point via increasing surface activity on liposomes. In this way, close packing on the phosphatidylcholine bilayer membrane could have been promoted. In the preformulation study, the impact of oleic acid on particle size, PDI value, and zeta potential was realized. The effect of cholesterol on particle size may be solved in the presence of oleic acid in liposomes. Kurniawan J. et al. showed that the oleic acid causes electrostatic repulsion on the dipalmitoylphosphatidylcholine membrane at neutral pH apart from low acidic pH [
42]. Therefore, we thought that oleic acid could have improved the particle size by causing electrostatic repulsion and increasing the surface activity on the phosphatidylcholine bilayer membrane [
42,
43].
On the other hand, the high amount of oleic acid in a low level of cholesterol with maximum CER-NP can cause a rise in the PDI value. The rigidity derived from cholesterol provides stability to liposome structure, which is why cholesterol has a crucial role in PDI value. However, the cholesterol is not enough to obtain a fine PDI value without oleic acid based on preformulation studies (with oleic acid PS: 136.1, PDI: 0.248; without oleic acid, PS: 494.7, PDI: 0.706). Therefore, we thought that the PDI value brings up to the mark as the oleic acid/cholesterol ratio optimized.
Evaluation of Zeta Potential
Zeta potential responses were in the range between (−)1.85 and (−)10.3 mV that showed no significant difference statistically. Therefore, zeta potential has been descoped on responses in this study. Even so, oleic acid-containing liposomes have been negatively charged rather than the absence of oleic acid according to pre-formulation studies (with oleic acid ZP: −10.7 ± 1.11 mV; without oleic acid ZP: −1.30 ± 0.47 mV) (unpublished data).
3.4. Selection of the Optimized CER-NP-Loaded Liposomes
The optimized liposome formulation was selected based on minimizing the particle size and PDI value by the Design of Expert. The software suggested that the concentrations of CER-NP, oleic acid, and cholesterol are 2.4, 3.76, and 5%, respectively. In this hypothetical case, the particle size and PDI value would be 132.6 nm and 0.278, respectively.
The experimental particle size and PDI data of optimized liposomes, which performed at least three replications, were 136.6 ± 4.05 d.nm and 0.248 ± 0.012, respectively. As a result, the experimental responses were in close agreement with the predicted responses. The DLS result of the optimum CER-NP-loaded liposome is provided in
Figure 4.
3.4.1. Encapsulation Efficiency and CER-NP Content of the Optimized CER-NP-Loaded Liposomes
The optimum liposomes were evaluated due to their encapsulation efficiency. Considering the physicochemical structural similarity between ceramide subtypes and phosphatidylcholine, it might be possible to diffuse to CER-NP into the phosphatidylcholine bilayer membrane. Therefore, the selected CER-NP-loaded liposome results were provided in context. As a result, the CER-NP content was determined to be 97.54%. The encapsulation efficiency was 93.84 ± 0.87%. The high encapsulation efficiency could result in the increased lipophilicity of CER-NP.
3.4.2. Confocal Laser Scanning Microscope Imagining of Optimized CER-NP-Loaded Liposomes
The optimized liposomes prepared with various molar concentrations of NBD-CER instead of CER-NP have been observed in CLSM. The fluorescent nitro group of NBD-CER is inevitably shown in a vesicle structure. The CLSM images are provided in
Figure 5. The NBD-Cer amount was evaluated based on CLSM figures.
Figure 5A contained the highest molar concentration of NBD-Cer.
Figure 5B–D may contain less localized fluorescent areas rather than
Figure 5A [
44].
The localization of ceramides in liposomes is crucial in terms of the release characteristics of liposomes. This study focused on two preparation processes of ceramide-loaded liposomes, showing that ceramide is mainly encapsulated in a liposome bilayer unless there is another polymeric phase [
45]. The CLSM imagining shows that the NBD-CER was successfully encapsulated in vesicles. In addition, we presume that the CER-NP was localized in the phosphatidylcholine layer apart from the aqueous inside of vesicles due to its lipophilicity. However, it is crucial to specify the NBD-Cer molar concentrations used instead of CER-NP in liposomes to obtain fine CLSM imagining. In the literature, the NBD-CER containing liposome was imagined under CLSM at various concentrations. Based on the authors’ view, the NBD-CER was localized on the bilamellar membrane of the liposome [
30]. This consideration supports the idea that CER-NP and other lipid substituents might be composed in harmony within the phosphatidylcholine bilamellar membrane.
3.4.3. ATR-IR Analysis of Optimized CER-NP-Loaded Liposomes
The NBD-CER has a specific 1,2,5 oxadiazole group on its chemical structure. In addition, the oxadiazole ring is bonded to 4-nitroaniline and becomes a nitrobenzoxadiazole (-NBD) functional group as fluorescent functionality. The chemical structure and functional group of NBD-CER are provided in
Figure 6.
The liquid phase of optimized liposomes containing NBD-CER was analyzed via ATR-IR spectroscopy in comparison with the liquid dispersion of empty liposomes based on transmittance (%). The spectral data are provided in
Figure 7. According to the results, the C-N stretch was shown up in 1224.80 cm
−1 as amide III in liquid disperse liposome with NBD-CER (red). The specific regions were seen in only liquid disperse liposomes with NBD-CER (red), not in liquid disperse empty liposomes (blue) as compared to the spectral data.
In addition, there are other characteristic peaks of liposomes in ATR-IR. The PO
2 symmetric stretching region coming from phosphatidylcholine is shown at 1082.07 cm
−1 in lyophilized liposomes with CER-NP (black) [
46,
47,
48]. This region gives an idea about the microenvironment of PO
2− groups in liposomes. The sharp peak on 1737.86 cm
−1 on the phosphatidylcholine spectrum is derived from the non-hydrogen bonded C=O group. However, in aqueous media, water may diffuse into the side of the carbonyl group on the lipid bilayer and make hydrogen bond interaction on the carbonyl group area [
49,
50,
51,
52]. Therefore, the sharp peak may split in aqueous dispersion because of the lipid–water interface on the side of the carbonyl group [
47,
53]. The CH
2 symmetric stretch and CH
2 asymmetric stretch were shown at 2854.65 cm
−1 and 2924.09 cm
−1 in liquid-dispersed liposomes with NBD-Cer (red). In many studies, the CH
2 vibration at 2852–2855 cm
−1 is considered a sign of a liquid crystalline phase that leads to liposome formation [
54,
55,
56]. These are derived from the characteristics of lipid acyl chain packing and fluidity on liposomes [
53,
57]. In lyophilized liposomes with CER-NP (black), the CH
2 vibration region shows a lower frequency at 2852.72 cm
−1. This indicates a more regular and ordered lipid organization, which is like orthorhombic lateral packing of skin lipids in a healthy stratum corneum. Le Deygen I.M. et al. reported the specific peaks of liposomes in ATR-IR analysis [
47]. Stretching vibrations of the CH
2 group are symmetric and asymmetric, stretching into 2850 cm
−1 and 2919 cm
−1 bands, which are like our findings.
3.4.4. FE-SEM Imagining of Optimized CER-NP-Loaded Liposomes
The lyophilized CER-NP-loaded liposomes dispersed in mannitol as cryoprotectant are shown as specific liposome images [
58]. The spherical feature of liposomes is observed in FE-SEM images (
Figure 8). The lyophilized CER-NP-loaded liposomes smaller than 500 nm were indicated with blue arrows on the images.
Figure 8A,B shows the close view of liposomes (Mag: 25,000×).
Figure 8C,D shows the distribution of liposomes in a wide perspective on a small scale (Mag: 10,000×). In electron microscopy, it is well known that high voltage can degrade the samples. The FE-SEM principle minimizes the voltage applied on the sample surface. In this way, the liposomes are observed under low kilovolts without any degradation and no need for sample covering. The surface morphology of liposomes was observed in 3D and real time in FE-SEM. The morphology of liposomes was hexagonal, similar to those observed in the literature [
59]. The small-size liposomes are mostly dispersed in mannitol. The particle sizes and FE-SEM images are also comparable to those reported in the literature [
59,
60].
3.4.5. Physical Stability Studies
The physical stability results of optimized CER-NP-loaded liposomes are provided in
Table 6. The stability was evaluated for 3 months based on particle size and PDI value, which are crucial characteristics indicating physical liposome stability. During the 30 days, the optimum CER-NP-loaded liposomes were observed in monophasic features at 25 °C temperature and 60% relative humidity. However, the liposomes could not keep the monophasic feature under accelerated stability conditions even for 30 days. The DLS results are provided in
Table 6.
Based on the DLS result, the formulation shows a good stability property for the first 30 days period at 25 °C temperature and 60% relative humidity. The main point might be the zeta potential provided by oleic acid. However, the fine zeta potential is not all issues. The cholesterol in optimized ratio might have an impact on PDI values during the 30 days, but it is well-known that dispersed systems are not suitable for long-term stability without viscosity-increasing excipients. The phospholipids’ transition temperature (Tc) has a pivotal role in the stability of the liposomal system [
61]. In our study, Lipoid S100 has an approximate Tc value with the temperature condition of the long-term stability test [
62]. The particle size and PDI values increased comprehensibly under the 25 °C temperature conditions. However, the particle size and PDI value increased dramatically under the 40 °C temperature conditions. This state might explain why the liposomal system is dramatically unstable under accelerated stability test conditions.
3.4.6. In Vitro CER-NP Release from Optimized Liposomes
It is well-known that liposomes provide a sustained drug release via lipid bilayer. Therefore, the in vitro CER-NP release from liposomes is crucial to observe the CER-NP release rate as a function of time. In vitro release of CER-NP from optimized liposomes was evaluated based on the cumulative released percentage of CER-NP from liposomes. As was seen in
Figure 9, more than 50% of CER-NP was released from the optimum liposome until the 6th hour. Almost the entire amount of the CER-NP (89%) was released at the end of the 8 h. The release rate was calculated at 0.54 ± 0.013 (k, µg.cm
−2·h
−1).
In general, non-linear regression models are commonly preferred to express the release profile of liposomes. According to the results, the cumulative release may fit the Korsmeyer–Peppas kinetic model. The Korsmeyer–Peppas equation (M
t/M
∞ = kt
n) shows an exponent value between 0.5 < n < 1.0, which means the formulation expresses a non-Fickian diffusion profile [
63,
64]. The n value is close to 1.0, and the drug release rate is independent of time. In
Figure 10, the regression coefficient of the Korsmeyer–Peppas model was over 1.0, which means the release profile is interpreted as close to the super case II transport drug release mechanism.
In the literature, there are many cases in which the drug release from liposomes fitted the Korsmeyer–Peppas kinetic model [
63,
65,
66]. The CER-NP release from optimized liposomes showed a similar release pattern as reported in the literature.
3.4.7. Cytotoxicity Analysis of Optimized CER-NP-Loaded Liposomes
Even the CER-NP, oleic acid, and cholesterol are naturally bio-transformed through the skin surface; the concentration of skin lipids in the formulation can cause skin irritation. To work the safety assessment, the liposomes without CER-NP and the optimized CER-NP-loaded liposomes were applied to the keratinocyte cell line in comparison with SLS as positive control, which is acceptable as cytotoxic on human keratinocytes within each concentration used in this study [
67,
68].
As a result, the relative light unit on 100% concentration of the optimized CER-NP-loaded liposomes was 6.89 times higher than those of the highest concentration of SLS. The results are provided in
Figure 11. This signifies that the optimized CER-NP-loaded liposomes could be considered relatively safe for lipid replacement therapy on AD via topical administration.
In addition, the optimum CER-NP-loaded liposomes had just 0.63 times lower RLU than those without CER-NP. This means there may be no concern about the CER-NP content used in this study.