DESs are usually composed of at least one HBD and one HBA, which presents a deeper depression in the melting point when combined at a suitable stoichiometric ratio. Here we synthesized DESs using OMT and LA by one-step method (Fig. 1A). To distinguish whether the blend is a simple eutectic mixture or a “deep” eutectic solvent, we first established the solid-liquid phase diagram of the binary system. Assuming that OMT and LA are miscible, the activity coefficient approaches unity, so eutectic behavior can be represented as an ideal V-shaped phase diagram [29]. While deep eutectic is exhibiting non-ideal negative deviation. As shown in Fig. 1B, the ideal eutectic point of OMT and LA was about 41 °C with a 0.14 mol fraction of OMT. In terms of appearance, blends with OMT/LA mole ratios of 9:1, 8:2, 7:3, 2:8, and 1:9 were presumed to be partially melted and presented as a mixture of solid and liquid or wet pasty (Fig. 1A). At OMT/LA mole ratios of 6:4, 5:5, 4:6, and 3:7, blends appeared to be in a homogeneous liquid state at room temperature, which implied that such liquid mixtures belong to DES.

DSC thermograms showed that liquid mixtures at these mole ratios began to melt at lower temperatures compared with OMT or LA alone and appeared very significant negative deviation from ideality (Fig. 1B and C; additional file 4). The COSMO-RS model reasonably described the non-ideality of the OMT/LA system, and the eutectic point occurred at the OMT/LA mole ratio of 4:6. Thus, the formation of DESs at wide mole ratios was further confirmed by comparing an experimental curve with an ideal curve, and they were named DES (6:4), DES (5:5), DES (4:6), and DES (3:7) in the following investigation. It is worth noting that DES (6:4), DES (5:5), and DES (4:6) exhibited an evident Tg, suggesting their non-crystalline nature. Among them, DES (4:6) accrued a recrystallization event followed by corresponding to the melting of the crystalline mixture. However, DES (3:7) only showed a single sharp peak with a melting temperature of -9.6 °C. These results suggested that DES aggregation changed gradually from amorphous to crystalline with the increase of the mole fraction of LA.

To estimate the thermal stability of the synthesized DESs, TGA/DTG thermograms were recorded from 30 to 350 °C (Fig. 1D and E). It can be seen that the mass losses of DESs mainly occurred at two stages. The initial mass losses (0.8–1.5%) occurred in the temperature ranges of 50–120 °C, which may be attributed to the evaporation of adsorbed water molecules [12]. The mass losses (89–93%) of the second stage started at near 170 °C, this corresponded to the actual thermal decomposition of DESs. Moreover, the decomposition temperatures of DESs were between OMT and LA (Additional file 4), indicating that DESs had good thermal stability. Peak temperatures for DTG curves showed the maximum rate of mass loss in the temperature ranges of 187–250 °C. The thermal stability of DESs at different mole ratios mainly depended on the composition fraction of LA, because LA was relatively easier to decompose at a high temperature.

Furthermore, DESs were characterized by FTIR, 1H NMR, and 2D NOESY which provided information confirming the intermolecular interactions of DESs. The peaks at 1606 and 2287 cm− 1 were attributed to the C = O vibration and the N+-O− vibration of OMT (Additional file 4). After forming DESs, the N+-O− vibration of OMT disappeared. Meanwhile, a blue shift in C = O stretching vibration was also observed and increased with decreasing proportions of OMT (DES (6:4), 1622 cm− 1; DES (5:5), 1623 cm− 1; DES (4:6), 1626 cm− 1; DES (3:7), 1635 cm− 1). This may be related to the changes in the molecular environment and hydrogen bonding partner. On the other hand, when LA was present alone, the peak at 1697 cm− 1 for the carboxyl group was very sharp; however, with the addition of OMT, the intensity of the peak weakened (Fig. 1F), possibly due to the involvement of OMT in new intermolecular interaction formation. According to our previous report [6], this was confirmed to be due to the presence of charge-assisted hydrogen bonding. In addition, the vibration frequency of LA carboxyl group increased as the proportion of LA increased, confirming the dissociation of carboxylic acid dimeric structure and growth of one hydrogen bond pattern [34].

1H NMR spectra of DESs in DMSO-d6 were shown in Fig. 1G. The peak areas OMT in δ = 5.01 ppm (H5) and LA in δ = 0.85 ppm (H12) were chosen as the benchmark, respectively. We could observe that the values of δ5.01/δ0.85 were approximately equal to 1.32, 1.06, 0.65, and 0.42 for DES (6:4), DES (5:5), DES (4:6), and DES (3:7), which corresponds to the molar ratios of OMT/LA of 6:4, 5:5, 4:6, and 3:7, respectively. Thus, the successful synthesis of DESs containing OMT and LA in a stoichiometric ratio was further confirmed by combining the DSC and 1H NMR analysis. Also, the part proton signals of OMT inside DESs presented obvious changes, with further upfield or downfield chemical displacement. These protons were concentrated on the 5th, 13th, and 15th carbon atoms (δ = 5.02, 3.22, and 2.82 ppm), presumably due to the effect of the electrostatic interaction with LA carboxyl group [35]. The disappearance of the proton signal at the carboxyl group of LA in all DES spectra comparison. While in the spectrum of LA alone, the proton signal was observed in the δ = 11.94 ppm. These results indicated that hydrogen bonding may be established between the N-oxide from OMT and the carboxyl group from LA. In addition, NOESY was used to investigate the intermolecular interactions occurring in DESs. The cross-peaks reveal the spatial correlation between the protons. As indicated in Fig. 1H, multiple H-H cross-peaks were identified in the NOESY spectra of DESs, indicating the presence of multiple NOE between the two components. This supramolecular network was the characteristic of DES. In the case of DES (5:5) and DES (4:6), obvious interactions were observed between the protons (H13 and H15) surrounding the nitroso of OMT and protons (H3) surrounding the carboxyl group of LA. For DES (3:7), this correlation weakened and interactions between the protons (H2) surrounding the carboxyl group of LA increased, which was attributed to the molar excess of LA. Thus, these DESs had different microscopic interactions on a molecular level and might have different properties.

Preparation and characterization of DESs. (A) Preparation process and appearance photos of DESs; (B) Solid-liquid phase diagram of OMT + LA binary mixtures; (C) Differential scanning calorimetry (DSC) thermograms of DESs; (D) Thermogravimetry analysis (TGA) of DESs; (E) Differential thermogravimetry (DTG) curves of DESs; (F) Fourier transform infrared (FTIR) spectra of DESs; (G) Proton nuclear magnetic resonance (1H NMR) spectra (400 MHz, DMSO-d6) of DESs. (H) 2D Nuclear overhauser effect spectroscopy (NOESY) spectra (500 MHz, DMSO-d6) of DESs. The cross-peaks (interactions) between H13 and H15 of OMT and H3 of LA were circled in black

MD simulations (Additional file 1) further presented the supramolecular network formed by hydrogen bonds at different mole ratios of DESs (e.g. O-H (LA)•••N+-O− (OMT), O-H (LA)•••C = O (OMT)). Nevertheless, there was the strongest correlation between the O19 atom at the N-oxide of OMT and the H15 atom at the carboxyl group of LA for all DESs, which means that the heterodimer synthons consisting of O-H (LA)•••N+-O− (OMT) hydrogen bonds play a dominant role in the DES formation. DFT calculations were further applied to confirm the formation mechanism of DESs. The strongest pair-wise interaction for each possible combination (OMT-OMT, LA-LA, and OMT-LA) was considered in the calculations. The results from interaction energy (ΔE) suggested that the hydrogen bonding established between OMT and LA was much stronger than the hydrogen bonding when pure substances were present alone (Fig. 2A). This might be due to the oxygen binds with the nitrogen atom through the coordination bond, and even though reducing the alkalinity of OMT, the lone electron of the oxygen makes it more polarity. As such, N-oxide became a better hydrogen bond acceptor than a carbonyl group. In fact, OMT exists in the hydrate form under natural conditions [36]. When OMT and LA are mixed, OMT preferentially forms hydrogen bonds with LA since the hydroxyl group of LA is a stronger HBD than that of water. The above assumption was further supported by electron density analysis (Additional file 2). The color depth in the molecular surface illustrated the intensity of electrostatic potential (ESP). As shown in Fig. 2B, the negative ESP regions of OMT were mainly located on oxygen atoms, while the positive ESP region of LA was mainly located on the carboxyl hydrogen atom. For the dimer formed by OMT and LA, the positive region of LA was attracted to the negative region of OMT, thereby leading to electron transfer and delocalization [37]. According to the independent gradient model based on Hirshfeld partition (IGMH) analysis, non-covalent interactions between OMT and LA were visualized (Fig. 2C). IGMH can intuitively display the interaction intensity and characteristics between different fragments [38]. It could be seen that the strong electrostatic interaction and van der Waals attraction between OMT and LA played an important role in the binding of the two species.

Density functional theory (DFT) calculations of DESs. (A) Geometries and interaction energies of most probable pairs of interacting molecules in DESs, pure OMT, and pure LA; (B) Molecular electrostatic potential (ESP) on the van der Waals surface (0.001 a.u.) of OMT, LA, and OMT-LA molecular ion pair; (C) Scatter plots between δg vs. Sign (λ2) ρ of OMT-LA molecular ion pair and corresponding color-filled isosurface through the independent gradient model based on Hirshfeld partition (IGMH) analysis

The physicochemical properties of DESs are crucial to their development and application of transdermal formulations. We determined the properties of DESs including viscosity, pH, conductivity, and stability. From the flow curves at a shear rate from 0.1 to 100 s− 1, all synthesized DESs suggested Newtonian behavior (Fig. 3A). The mole ratios of OMT/LA significantly influenced the viscosity of liquid mixtures, and the viscosity was between 1.0 and 2.5 Pa∙s (Fig. 3B). The prepared DESs showed pH values ranging from 7.1 to 10.8 at room temperature (Fig. 3C). With the increase of LA mole ratio, the pH of DESs gradually decreased. This is due to that pH reflects the scale of acidity of a liquid depending on the relative acidity or basicity of materials that are mixed, as well as their stoichiometry [39]. Considering that the water content in the skin increases up to 60% as the depth increases [40], we also evaluated the effect of water addition on the supramolecular structure of DESs by conductivity measurement. Generally, excessive dilution of DES can destroy hydrogen bond interactions between components, allowing charges to migrate more readily through the solution, thus resulting in increased conductivity [16]. Figure 3D shows the conductivity results under different water content. The conductivity substantially increased and then decreased with the increase of the water content, displaying the highest value at the range of 40–60 wt% water. In addition, when adding water into the DESs, we observed the formation of a gel-like substance in DES (4:6) (Additional file 4), which may be attributed to hydrophobic interactions between the tails of LA and LA anion [41]. The increase in viscosity caused a lower conductivity of DES (4:6) + water mixtures, given the diffusion of molecules [42]. That is, the cluster structures of DESs were probably destroyed gradually when the amount of water was high, owing to the hydration. Additionally, recrystallization was observed for DES (6:4) during storage (less than one month). Polarized light microscopy appearing birefringence also confirmed the presence of crystalline substances (Fig. 3E). This behavior might be attributed to the crystal growth of an excess of OMT by absorbing water. In contrast, DES (5:5), DES (4:6), and DES (3:7) were still stable enough to maintain homogeneous liquid for at least one year.

Skin toxicity/irritation of DESs was assessed using two cell models and histopathological examination of skin. The poor stability of DES (6:4) limited its application, only DES (5:5), DES (4:6), and DES (3:7) were selected for further analysis. As shown in Fig. 3F and G, DESs at different mole ratios had no cytotoxicity on HaCaT and HSF at tested concentrations, showing a wide safety range. Furthermore, histopathological examination showed that no obvious loose edema, necrosis, and inflammatory cell infiltration were observed for DES groups in comparison with the control skin (Fig. 3H), which implied good biocompatibility. From these results, DESs formed from OMT and LA with suitable physicochemical properties and skin safety are worth further exploring for application in transdermal delivery systems.

Physicochemical properties and skin safety of DESs. (A) Shear stress-shear rate relationship of DESs; (B) Shear viscosity-shear rate relationship of DESs; (C) The pH values of DESs; (D) Conductivity of DESs under different water contents; (E) Stability of DESs under storage at room temperature; (F) The viability of Human immortalized epidermal cells (HaCaT) after incubation with DESs; (G) The viability of human skin fibroblasts (HSF) after incubation with DESs; (H) Histopathological examination of rat skin treated with DESs for 24 h

Formulation development of sparingly soluble drugs is very challenging. Hence, the solubilization ability of DESs on different drugs was measured and compared with that in water and a commonly used hydrophobic oil solvent IPM. Figure 4A-G presents the solubility results of these drugs including MIN, PIR, TA, QUE, BAI, ADA, and FA. It can be seen that MIN and BAI were more soluble in water than in the oil, while PIR, TA, QUE, ADA, and FA were more soluble in the oil. However, the solubility of these drugs was very low in both the water and oil. In contrast, the solubility of all drugs in the DESs was significantly enhanced. Among them, TA solubility in the DES (4:6) increased by 4348-fold compared with that in the water and increased by 384-fold compared with that in the oil.

The solubilization mechanism of DESs on various drugs was further studied by QSAR. In additional file 3, the input data containing drug structural and topological descriptors were shown, as well as the dependent variable logarithm of solubility (log S). The training (6 drugs) and test (1 drug) set were randomly selected. The independent variables were first selected using correlation analysis. Descriptors with high cross-correlation values (> 0.7) were not considered. Genetic function approximation (GFA) and partial least square (PLS) were then used to carry out the linear regression analysis. The results showed that GFA models containing Tm, ΔSP, and pKa could well describe the changes in log S. The adjusted R2 was similar to the goodness of the model R2, indicating that there was no overfitting. The predicted value of the test drug (MIN) compared with the experimental value was within the applicability domain (Fig. 4H and I). The linear regression equations for DESs were listed as follows:

$$\eqalign}\left( \right)\;:\;\log S &= - 0.1890 - 0.4508\Delta SP \cr &\quad+ 0.1161pKa + 1.5138\, = 0.9983,_ \cr &\quad= 0.9957, = 0.9916,s \cr &\quad= 0.0155,F = 390.45,p < 0.05}$$

(3)

$$\eqalign}\left( \right):\;\;\log \;S\; &= \; - 0.2279\; - \;0.2916\Delta SP\; + \;0.1222pKa\; \cr &\quad+ \;1.4642 = 0.9850,_ \cr &\quad= 0.9625, = 0.5743,s\cr &\quad = 0.0368,F = 43.72,p < 0.05 }$$

(4)

Regrettably, the solubilizing behavior of DES (3:7) on drugs could not be well predicted (p > 0.05), which might be attributed to the influence of more factors. Tm reflects the crystal lattice energy of molecules. It has been previously shown that for drugs with a Tm of more than 200 °C, the crystal lattice has a strong influence on the solubility [43]. SP reflects the interaction energy between molecules. According to the solubility parameter theory introduced by Hildebrand and Scott, mutual solubility would be higher if the SP values of the drug/solvent are closer [44]. pKa reflects the ability to accept electrons or donate electrons, mainly affecting the ionization degree of the drug [45]. Based on the solubility results of different drugs, good molecular compatibility could more efficiently solubilize drugs (e.g. ADA and TA), while the presence of acid groups had little effect on solubility. This was consistent with the results of correlation analysis that there was the strongest relation between log S and ΔSP, followed by Tm and pKa (Additional file 3). In general, our prepared DESs showed good solubilization performance on different drugs, and the established model could be extrapolated and used to predict the solubility of other transdermal drugs.

Solubility of various drugs and quantitative structure-activity relationship (QSAR) analysis. (A) Minoxidil (MIN); (B) Piroxicam (PIR); (C) Triamcinolone acetonide (TA); (D) Quercetin (QUE); (E) Baicalin (BAI); (F) Adapalene (ADA); (G) Ferulic acid (FA); (H and I) Plots of experimental versus predicted solubility based on the genetic function approximation (GFA) model

To clarify the self-assembly behavior of synthesized DESs in non-polar media, we preliminary assessed their intersolubility with various oil phases. Figure 5A shows that DES (5:5), DES (4:6), and DES (3:7) can be dispersed evenly in IPM, castor oil, tea tree oil, and squalene to form an optically transparent solution, especially for DES (4:6) and DES (3:7) which can be dispersed in the oil at any ratio. In contrast, pure OMT created a granular precipitate, while the recrystallization occurred quickly (< 15 min) for LA dissolved in the oil phase. We inferred that DESs composed of hydrophilic OMT and hydrophobic LA probably formed a nanostructure with RM characteristics through the solvent-induced self-assembly effect [46, 47]. Further investigation elucidated the self-assembly mechanisms as follows.

DLS was used to measure the particle size and distribution of the self-assembly. The RM-forming tendency of DESs was confirmed by measuring the CRMC using pyrene as a fluorescent probe. The micromorphology of RM was characterized by TEM. FTIR and conductivity measurements were used to explore the interactions and structural features inside DESs. Figure 5B shows the size and distribution of the self-assembly. Three DESs dispersed in the oil phase exhibited nano-size with 13.65, 7.38, and 10.83 nm, respectively. A clear Tyndall light-scattering effect was observed. The CRMC of DES (5:5), DES (4:6), and DES (3:7) was 126, 10, and 8 mg/g, respectively, verifying the existence of a nano-structure of RM (Fig. 5C). Figure 5D and additional file 5 display the FTIR spectra of DESs dispersed in IPM. Compared with characteristic vibration frequencies of components of DESs, no obvious change was observed, which indicated that basic structural units of DESs still were retained in the oil phase. Moreover, the spectra of DESs dispersed in IPM were also compared with the spectra of DESs dispersed in water. With increasing water content, the carbonyl vibration of OMT gradually red-shifted and the carboxyl vibration of LA disappeared (Additional file 5), confirming the dissociation and hydration of DESs. A previous study has shown a connection between hydrogen bond strength and microenvironment polarity [48]. Hydrogen bond strength increases as local polarity decreases, and the non-polar environment is more conducive to the formation of hydrogen bonding. Conversely, water tends to break the originally established hydrogen-bonds network of DES while reforming new hydrogen bonds with DES components. Taking the nano-dispersion made by DES (4:6) in IPM as an example, the morphology observed by TEM was spherical-like particles with nanometer size (Fig. 5E), which was around 5–10 nm and basically consistent with the DLS result. The dependence of conductivity on the mass fraction of DES (4:6) in IPM was shown in Fig. 5F. The curve could be divided into three stages, similar to the phase transition of the microemulsion [17]. The < 45% of the increase region, 45–59% of the increase region, and > 59% of the decrease region corresponded to DES-in-oil, bicontinuous (BC), and oil-in-DES, respectively. When the mass fraction of DES in IPM was < 59%, OMT tended to gather and became a hydrophilic core, while hydrophobic LA with similar properties to IPM became the shell, the core/shell structure DES-RM subsequently self-assembled to form in the IPM without surfactant/co-surfactant.

Self-assembly mechanism of DESs. (A) Intersolubility of DESs with pharmaceutically acceptable oil phases; (B) Size distribution of the reverse micelles (20% DES-RM); (C) The critical reverse micelle concentration (CRMC) detection of DESs by pyrene radiometric method; (D) FTIR spectra of 20% DESs dispersed in IPM; (E) Transmission electron microscope (TEM) imaging of 20% DES (4:6)-RM; (F) Plot of conductivity as a function of DES (4:6) mass fraction in IPM. The curve was divided into three different subdomains: DES-in-oil (< 45%), bicontinuous (45–59%), and oil-in-DES (> 59%)

The self-assembly process of DESs was also visualized by MD simulation. The aggregation behavior of three DESs in final trajectory frames was presented in Fig. 6. As expected, DESs self-assembled to a nanocluster structure at a low concentration. This nanocluster structure was possibly easier to form as the OMT/LA molar ratio decreased since the molecular area of OMT exposed to the non-polar environment gradually decreased (Fig. 6A), consistent with the lowest CRMC of DES (3:7) of 8 mg/g. To further investigate the driving forces during the self-assembly of DES to RM, a quantitative analysis of hydrogen bond interactions between OMT and LA molecules was carried out. Figure 6B shows the results for RDF, g (r), between specific atoms of OMT and LA. It was obvious that there was a strong correlation between the nitroso of OMT and the carboxyl group of LA at 0.135–0.145 nm, indicating that charge-assisted hydrogen bonding might be the main driving force for self-assembly. Meanwhile, there was also a correlation between the carbonyl group of OMT and the carboxyl group of LA at 0.165–0.175 nm for DES (4:6) and DES (3:7) systems. This was due to the excess LA encircling the OMT carbonyl group. When the proportion of IPM was equal to DESs, a bicontinuous microstructure was formed in the DESs and oil continuous phase with a “worm-like” structure (Fig. 6C). The final snapshot presented the distribution of polar and non-polar domains, in which oil molecules were dispersed in crevices surrounding the hydrophobic regions or, mainly, near LA molecules. Similarly, RDF analysis showed the presence of hydrogen bond interactions between LA and OMT molecules (Fig. 6D).

Through a series of representations and MD simulations, we reasonably speculate the self-assembly mechanism of DESs (Fig. 6E). Firstly, OMT and LA form the complex units that makeup DESs by electrostatic and hydrogen bond interactions. Subsequently, OMT molecules tend to aggregate into clusters in the non-polar environment due to the solvophobic effect. However, the strong molecular interactions among DES components can act as a ‘‘glue” to bond the head groups of OMT and LA together. Hence, these chain aggregates are further stacked through molecular van der Waals forces to form the core/shell structure with an inner hydrophilic core and outer hydrophobic shell. Further, the increase in DES concentration will increase the probability of RM particle collisions and attractions, leading to promote the formation of conductive chains. These chains will be closely connected and fused to each other to form a bicontinuous structure.

Molecular dynamic assembly of DESs. (A) Final trajectory frames and molecular aggregates of 10% DESs; (B) Radial distribution function (RDF) curves for specific atoms among 10% DESs; (C) Final trajectory frames of 50% DESs and representative snapshots of molecular interaction patterns; (D) Radial distribution function (RDF) curves for specific atoms among 50% DESs; (E) Schematic diagram of the self-assembly. (blue: OMT; green: LA; brownish yellow: IPM)

Based on the result of solubility study, we chose TA as a model drug and further investigated the potential of DES-RM for transdermal drug delivery (Fig. 7A). Initially, we evaluated TA solubility in DES (4:6)-RM. TA was dissolved in DES (4:6) and the mixtures were then added to IPM to equilibrate for 24 h. The results showed that the solubility of TA in DES (4:6)-RM was higher than that of traditional pharmaceutically approved solvents such as ethanol, DMSO, and Tween 80 (Fig. 7B). The good solubilizing ability should be due to the molecular interactions between OMT and TA, which facilitated more TA entrapment and adsorption at the cores of RM or the DES/oil interface, as supported by the 1H NMR spectra and MD simulation (Additional file 6). Among them, TA-loaded 10% DES (4:6)-RM reached the concentration of commercial TA formulations, and the loading content and loading efficiency were 1.07 ± 0.04% and 93.00 ± 3.65%, respectively. The stability of RM is important for drug delivery. After being frozen and thawed three times, DES (4:6)-RM at different concentrations showed no drug precipitation and phase separation, and no significant change in the size was observed (Additional file 7), indicating that the systems were stable.

Then, the skin penetration curves of TA from different DES (4:6)-RM formulations containing different concentration of DES were assessed (Fig. 7C). The cumulative penetration amounts from low concentration to high concentration of DES (4:6)-RM (0/total IPM, 10%, 20%, 30%, 40%, 50%, 100%/total DES) were 10.37 ± 1.96, 56.83 ± 3.64, 66.83 ± 2.89, 72.96 ± 2.46, 54.99 ± 8.53, 40.49 ± 6.97, and 33.99 ± 4.07 µg/cm2, respectively. Compared with IPM and DES alone, DES (4:6)-RM significantly improved the skin penetration of TA. The penetration enhancement was probably attributed to the following reasons. First, the high solubility of TA in RM benefited its permeation across the skin in a passive diffusion manner [49]. The increase in the thermodynamic activity might enlarge the concentration gradient, thereby enhancing drug penetration through the skin. Second, RM had good affinity with the skin lipid layer, therefore, TA encapsulated in RM could easily penetrate the skin through diffusion and fusion. However, the cumulative penetration amount of TA increased as the concentration of DES (4:6) increased below 30%, but then decreased as the concentration of DES (4:6) continued to increase. It is worth noting that DES (4:6)-RM below the concentration of 40% exhibited rapid penetration behavior within 12 h and then slowly penetrated up to 24 h, while 50% DES (4:6)-RM exhibited a linear behavior that approached zero-order within 24 h. This indicated that the RM nanodroplet more helped the drug transport compared with the BC structure. After 24 h of in vitro skin penetration, TA retention in the skin was measured (Fig. 7D). The skin retention of 10% DES (4:6)-RM and 50% DES (4:6)-RM was similar and significantly higher than that of IPM and DES alone (P < 0.01). In addition, DES (4:6)-RM significantly enhanced TA penetration and retention compared with TA commercial cream (P < 0.001, Additional file 8).

The impact of DES-RM at different mole ratios of OMT/LA on TA penetration and skin retention was also assessed. To form RM, the concentration of all DESs was set above CRMC. As shown in Fig. 7E, DES (5:5)-RM, DES (4:6)-RM, and DES (3:7)-RM exhibited similar skin penetration curves at a DES concentration of 20%, with cumulative penetration amounts of 62.90 ± 7.27, 66.83 ± 2.89, and 64.74 ± 2.31 µg/cm2. However, from the result of TA retention, DES (4:6)-RM exhibited higher drug retention and was superior to DES (5:5)-RM and DES (3:7)-RM (Fig. 7F). In summary, both transcutaneous penetration and skin retention of TA from DES (4:6)-RM were greatly improved.

In vitro skin permeation studies. (A) Schematic of transdermal delivery of the drug via RM; (B) TA solubility in different medium; (C and D) Effect of DES-RM systems at different concentrations on TA penetration and retention; (E and F) Effect of DES-RM systems at different mole ratios of OMT/LA on TA penetration and retention; (G) Skin penetration depth of coumarin 6 (C6) under incubation with DES-RM systems at different mole ratios of OMT/LA using confocal laser scanning microscopy (CLSM) visualization; (H) Scanning electron microscope (SEM) images of stratum corneum longitudinal section after incubation with DES-RM systems at different mole ratios of OMT/LA. *P < 0.05, **P < 0.01, and ***P < 0.001

Furthermore, DES-RM was loaded with hydrophobic C6 as a fluorescent probe to visualize the penetration depth (Fig. 7G). The fluorescent images revealed that IPM alone only offered relatively low C6 fluorescence intensity and permeation depth (up to 30 μm) in the skin. A large amount of C6 may be embedded in the lipid matrix (0 ~ 10 μm) of SC. In contrast, the skins treated with DES-RM presented a significantly deeper permeation depth (up to 80 μm) with stronger green fluorescence intensity. Such penetration depth was sufficient to reach the active epidermis layer and dermis layer. Consequently, DES-RM delivered the drug more efficiently into the skin in contrast to the oil phase of IPM.

The skin penetration enhancement has three possible mechanisms: (1) lipid extraction and lipid fluidization [6]; (2) interaction with intracellular keratin [50]; (3) increasing partitioning into the skin [49]. These mechanisms are mainly related to the “brick and mortar” structure of SC, which is usually considered the main rate-limiting step to drug absorption through the skin. Hence, we evaluated the effect of DES-RM on the structure of SC by SEM and FTIR. Figure 7H shows the microstructure images of the SC after treatment with DES-RM formulations. Normal skin treated with normal saline was used as a control. After incubation with IPM alone or DES-RM, it was found distinc