SLN (composed of glyceryl dibehenate) and NLC (including glyceryl dibehenate/caprylic capric triglycerides 90:10) were prepared by high shear homogenization using an Ultra Turrax^® T18 Basic (IKA^®, Germany). Briefly, a hot 2.5% poloxamer 188 solution (90˚C) was added to the melted lipid phase (solid lipid or lipids blend). This mixture was emulsified at 20,000 rpm for 5 min at 90˚C. The resultant emulsion was cooled down in a cold water bath (7˚C) to room temperature in order to obtain nanoparticles.

Particle size, polydispersity index (PI) and zeta potential were determined by photon correlation spectroscopy using a Malvern Zetasizer^® Nano-ZS90 (Malvern Instruments, USA). For size and PI, the samples were diluted with distilled water to a proper mean count rate prior to the measurements, which were performed at a 90˚ scattering angle at 25˚C (n = 5). To measure zeta potential, samples were diluted with deionized water and the measurements were done considering viscosity, refraction index and dielectric constant of water at 25˚C (n = 5). The morphology of the lipid nanoparticles was evaluated by means of scanning electron microscopy (SEM) using a JEOL JSM-25SII Scanning microscope (JEOL Tokyo, Japan). To prepare the sample, nanoparticle aqueous dispersion was left to dry over a coverslip and covered with gold.

SLN and NLC formulations were stored in glass vials at 5˚C, 25˚C and 40˚C. Particle size, PI and zeta potential were measured at 8, 15, 30, 60 and 90 days (n = 3), as previously described.

Occlusive effect of SLN and NLC was determined using a modified in vitro occlusion test [18] . Briefly, bakers were filled with 40 mL of water and covered with Whatman^® filter paper grade 42 (Sigma-Aldrich, USA). A sample of SLN or NLC (300 µL, 10% w/v lipid content) was spread on the filter surface, using water as reference, instead lipid carriers. Bakers were stored at 32˚C to mimic the temperature of skin surface and weighted at 0, 6, 24 and 48 h to calculate water evaporation through the filter paper in terms of water loss. The occlusion factor F was calculated at 6, 24 and 48 h using Equation (1) [18] .

where R = reference water loss and S = sample water loss. An F = 0 means no occlusive effect while an F = 100 means maximum occlusiveness. At the end of the experiment, filter paper was observed by scanning electron microscopy (SEM), using a JSM-25-SII scanning electron microscope (JEOL, Tokyo, Japan) in order to visualize film formation.

Transepidermal water loss (TEWL) is the outward diffusion of water through skin [19] . Skin was excised from pig ears. Skin slides (700 µm thickness) were obtained using an Electric Dermatome (Zimmer^®, USA). Impaired skin was obtained removing the SC by means of tape stripping (20 tapes). Both impaired and intact pig skin samples were mounted on the receptor compartment of a Franz cell filled with phosphate buffer solution pH 7.4 and basal TEWL value was recorded with a Tewameter^® TM 210 (Courage & Khazaka, Germany). A volume of 50 µL (10% w/v) of SLN or NLC dispersion was spread on the skin surface and TEWL was measured 2 h later. The occlusive effect was determined by the reduction of the TEWL value using Equation (2):

where B is the basal TEWL value and T is the TEWL value after lipid carrier treatment. SEM was used to visualize film formation on the skin. Pig skin was fixed, dehydrated and coated with gold previous observation under the microscope (JEOL JSM-25-SII, JEOL, Tokyo, Japan).

Penetration test through the SC of pig skin was carried out using SLN and NLC containing Nile red as a marker. Nile red-loaded SLN and NLC (0.005% w/w) were prepared as described before, incorporating the dye in the melted lipid phase. Skin excised from pig ears was mounted in Franz diffusion cells containing phosphate buffer pH 7.4 as receptor solution, and the surface of the skin was treated with 1 mL sample of Nile red-loaded SLN or NLC dispersion. The Franz diffusion cells were kept in a water bath at 37˚C and, after 2 h contact, the remaining formulation was removed and tape stripping was performed on the skin. Fifteen tapes (Scotch^® packaging tape) of 2 × 2 cm were used, weighting each tape before and after skin application. Nile red was extracted from the tapes, stirring with a volume of methanol during 24 h. Tapes 1 to 5 were extracted individually and subsequent tapes were extracted in clusters (6 - 10 and 11 - 15). The extracts were filtered and Nile red was quantified by spectrophotometry UV/Vis (λ = 555 nm) (Varian Cary^® 50 UV-Vis spectrophotometer, Australia). Penetration distance was determined from tapes weight data, using Equation (3):

where TWf is the tape weight after the stripping, TW₀ is the tape weight before the stripping, SCD is the SC density considered as 1 g cm³ and A is the tape area.

The statistical analysis was performed with p < 0.05 as level of significance (STATGRAPHICS® Centurion XVI).

Although high pressure homogenization (HPH) is the most used technique for lipid nanoparticle preparation [9] , high shear rate homogenization is a feasible technique to prepare lipid nanoparticles at laboratory scale since it is an easy handle technique, that requires low cost equipment compared to HPH, and that allows to obtain par- ticles of nanometric size (~200 nm) [20] . Table 1 shows the results of SLN and NLC characterization.

No differences in particle size between SLN and NLC (p > 0.05) were found. SLN and NLC of similar particle size were desirable so that this variable was not a factor directly influencing the results observed on the skin (occlusive effect and penetration enhancement ability). PI indicates the width of the particle size distribution. Theoretically, monodisperse distributions are described as PI = 0, however, PI < 0.2 is considered as narrow size distribution [16] . PI > 0.2 for the lipid nanoparticles prepared could be due to the preparation method used, giving a wider size distribution compared to HPH [20] . Zeta potential is a parameter used to predict stability of colloidal suspensions throughout storage time. A zeta potential higher than ±30 mV may assure good physical stability, being optimal when zeta potential is close to ±60 mV [21] . It is well known that zeta potential depends on the nature of the particles and the medium composition. For lipid nanoparticles, the molecular arrangement of lipids in the external layer of nanoparticles and its interaction with stabilizers are determinant. For the lipid nanoparticles prepared, poloxamer 188 was used as stabilizer, and due to its non-ionic nature, this molecule does not contribute with additional charges to zeta potential. Furthermore, the lipid that composes SLN is in fact a blend of acylglycerols (Figure 1(a)): glyceryl tribehenate (28% - 32%), glyceryl dibehenate (52% - 54%) and glyceryl monobehenate (12% - 18%) [22] , all of them being glycerol esters of long- chain-length fatty acids (C22) so that they provide neither charge nor polarity that contributes to zeta potential. On the other hand, NLC are made of the same lipid blend but including a certain amount of caprylic/capric triglyceride, which is a diacylglycerol of medium-chain-length fatty acids (Figure 1(b)). In this case, due to the non-esterified hydroxyl group of the glycerol and the length of the fatty acids, this molecule exhibits certain polarity that contributes to zeta potential, which explains the higher value compared to SLN. A similar result was observed by Teeranachaideekul et al. [18] for nanoparticles composed of cetyl palmitate and medium- chain-length triglycerides. The authors explain that it might be due to the accumulation of oil at the surface of NLC. Being the melting point of the solid lipid higher than that of the oil, when preparing NLC, the solid lipid recrystallizes first, holding a portion of the oil within the solid lipid matrix. Subsequently, the excess of oil remains in the outer shell of nanoparticles, then the oil contributes largely to zeta potential. Despite this difference in zeta potential between SLN and NLC, both values are under the desired ±30 mV, which may indicate physical stability issues during storage. Nevertheless, poloxamer 188 stabilizes lipid nanoparticles because of the steric effect it produces which avoids coalescence of the particles [21] . Such steric effect is possible due to the structure and conformation of poloxamer. Poloxamer 188 is a block copolymer of ethylene

¹Glyceryl dibehenate: caprylic capric triglycerides.

oxide and propylene oxide with the formula HO(C₂H₄O)₈₀(C₃H₆O)₂₇(C₂H₄O)₈₀H. The polyoxyethylene segments are hydrophilic and the polyoxypropylene segment is hydrophobic [22] , so this last part of the molecule is located on the lipid nanoparticle surface whereas the long hydrophilic polymer chains are oriented towards the medium. Regarding the morphology of the lipid nanoparticles, Figure 2 shows SEM images for SLN and NLC. Both systems exhibit similar characteristics since they are assembled out of the same main lipid and were prepared under the same conditions. Particles appear as spherical or almost spherical structures of approximately 200 nm, which agrees with the results obtained from the photon correlation spectroscopy technique.

Physical stability of SLN and NLC was assessed measuring particle size, PI and zeta potential to determine whether there are changes or not, depending on the storage temperature and storage time. Results showed no significant changes in either PI or zeta potential through storage time under different storage conditions (data not shown). Conversely, both storage time and temperature have effect on the particle size of the lipid carriers (Figure 3). For SLN, particle size increased at 90 days and 40˚C, whereas NLC size increased at 90 days for both 25 and 40˚C (ANOVA test showed significant differences between 40˚C and the other two temperatures for SLN and between 5˚C and the other two temperatures for NLC). Hence, storage at 5˚C seems to preserve the particle size more effectively for both SLN and NLC up to 90 days.

Results of in vitro occlusive effect (Figure 4) showed that occlusion factor (Equation (1)) took values between 35.9 and 38.9 throughout the test for SLN and NLC. Although no statistical difference was found between SLN and NLC, differences between 6 h and the other two times were found, which might be due to the formation of a film on the filter paper at the beginning of the test. When a nanoparticle suspension is applied onto the skin, lipid nanoparticles tend to fusion forming a film on the skin [23] , so, it is expected that this effect occurs on the filter paper as well. The formation of this nanoparticle film was confirmed by scanning electron microscopy. Figure 5 shows the presence and characteristics of the film formed with the lipid carriers: fibers that compose filter paper (without treatment) are shown in Figure 5(a), whereas Figure 5(b) and Figure 5(c) show these fibers covered by the lipid carriers. This film is thought to be responsible for the occlusive effect observed in the test, as it acts like a barrier that avoids water vapor to escape through the filter paper. As shown, the films formed by both SLN and NLC are alike between each other. It has been demonstrated previously that the degree

of occlusion of these films depends on the particle size [24] , therefore, as the size of SLN and NLC are very close, it is clear that the inclusion of oil for preparing NLC does not have effect either on the film formation or on the occlusive effect.

In order to evaluate the effect of SLN and NLC on the TEWL value, reduction of this parameter as percentage (Equation (2)) is shown in Figure 6. A significant reduction in TEWL of intact skin was observed for SLN and NLC respectively, without statistical difference between these systems (p > 0.05). Such effect is due to the occlusive effect of lipid carriers as result of film formation on the skin. The film formed at the surface of pig skin was observed by SEM. Figure 7(a) shows intact skin without treatment, where the microrelief lines as well

as hairs are clearly seen. On the contrary, in Figure 7(b) and Figure 7(c), which correspond to intact skin treated with SLN and NLC respectively, microrelief lines and some parts of hair remain under the film formed by the carriers. In the case of impaired skin, carriers did not have a significant effect on the TEWL (p > 0.05). Due to stripping of SC the high humidity on the skin surface prevents film formation, since lipid carrier suspensions applied on the skin have to lose their own humidity in order to form the film. Jenning et al. [25] studied water loss from SLN, by applying lipid nanoparticles on a glass surface and measuring water loss. They determined that formulation lost their own water almost completely within 1 h. In the present work, even 2 h after applying formulations, the film was not formed on the stripped-skin surface and therefore, a reduction in TEWL was not detected.

Tape stripping is considered a useful non-invasive technique that when coupled with other techniques, it can provide information about the transport of substances through the SC [11] . In this work, tape stripping was used to determine the amount as well as the penetration depth (Equation (3)) through the SC of Nile red included in both SLN and NLC. Lipid nanoparticles are unable to penetrate through human skin because of their size, but

they can help actives to reach deeper skin layers due to the interaction of nanoparticle lipids with the extra- cellular lipid matrix of the skin. Figure 8 shows the penetration profile of Nile red included in both SLN and NLC, being NLC the system that favors the penetration of a greater amount of Nile red. At a distance of 12.2 ± 1.5 µm (SLN) and 12.6 ± 1.4 µm (NLC), the total amount of Nile red penetrated was 1.7 ± 0.4 µg and 4.7 ± 1.3 µg, respectively. This means that, even if a similar occlusive effect was found for both systems, the amount of Nile red penetrated was 2.7 fold higher for the NLC. This fact suggests that lipid nanocarrier composition affects the ability to enhance penetration of molecules through the SC, due to differences in the interaction of lipids with skin components. It has been reported that NLC have the ability to get into deeper layers of the SC. Pe- netration through pig skin of flufenamic acid (a model drug) included in SLN and NLC, was monitored using tape stripping. Researchers found that NLC showed the deepest skin penetration [12] . In other study, the pene- tration enhancement ability of lipid nanoparticles was related to the solubility parameter (SP) of lipids [26] . SP is useful to predict the release and skin penetration of any molecule by its interaction with skin lipids. Com- ponents with a SP close to that of the skin may have good miscibility with the skin lipids, and thus, can lead to greater penetration. As reported by Jensen et al. [26] , pig skin SP is estimated in 10, thus lipids with a SP value nearby 10 are expected to exhibit a better interaction with skin lipids and so, the penetration is expected to be greater as well. In this work, SP of glyceryl behenate and caprylic/capric triglyceride was calculated upon the Fedors substituents method [27] . SP values are estimated to be 9.34 for glyceryl behenate and 10.36 for caprylic/ capric triglyceride, therefore, the inclusion of this oil may improve the interaction of NLC with skin lipids as well as the penetration enhancement ability through the SC, since triglycerides are more miscible with skin lipids for having a closer value to the skin SP. Moreover, partition of Nile red within NLC contributes to explain its greater penetration. Nile red distribution within NLC was studied by fluorescence spectroscopy [28] . The authors found that the fluid lipid phase of NLC composed of glyceryl behenate and caprylic/capric triglyceride was enriched in Nile red (65% ± 8%). Hence, a better interaction with skin lipids is expected, allowing Nile red to penetrate into deeper SC layers when it is included in NLC.