Improvement of Imiquimod Solubilization and Skin Retention via TPGS Micelles: Exploiting the Co-Solubilizing Effect of Oleic Acid

Imiquimod (IMQ) is an immunostimulant drug approved for the topical treatment of actinic keratosis, external genital-perianal warts as well as superficial basal cell carcinoma that is used off-label for the treatment of different forms of skin cancers, including some malignant melanocytic proliferations such as lentigo maligna, atypical nevi and other in situ melanoma-related diseases. Imiquimod skin delivery has proven to be a real challenge due to its very low water-solubility and reduced skin penetration capacity. The aim of the work was to improve the drug solubility and skin retention using micelles of d-α-tocopheryl polyethylene glycol 1000 succinate (TPGS), a water-soluble derivative of vitamin E, co-encapsulating various lipophilic compounds with the potential ability to improve imiquimod affinity for the micellar core, and thus its loading into the nanocarrier. The formulations were characterized in terms of particle size, zeta potential and stability over time and micelles performance on the skin was evaluated through the quantification of imiquimod retention in the skin layers and the visualization of a micelle-loaded fluorescent dye by two-photon microscopy. The results showed that imiquimod solubility strictly depends on the nature and concentration of the co-encapsulated compounds. The micellar formulation based on TPGS and oleic acid was identified as the most interesting in terms of both drug solubility (which was increased from few µg/mL to 1154.01 ± 112.78 µg/mL) and micellar stability (which was evaluated up to 6 months from micelles preparation). The delivery efficiency after the application of this formulation alone or incorporated in hydrogels showed to be 42- and 25-folds higher than the one of the commercial creams.


Introduction
Actinic keratosis (AK) is a precancerous lesion caused by excessive exposure to light whose appearance is related to several mechanisms such as oxidative stress, immunosuppression, inflammation, altered proliferation and dysregulation of cell growth, impaired apoptosis, mutagenesis and papilloma virus (HPV) [1]. As actinic keratosis is the most common precursor of cutaneous invasive squamous cell carcinomas, its treatment appears to be imperative. In this sense, imiquimod (IMQ), a synthetic drug belonging to the class of imidazoquinolones, has widely demonstrated its usefulness thanks to its immunostimulant activity mediated by the binding to the toll-like receptors 7 and 8. These receptors play a pivotal role in controlling innate immune system response as they induce the secretion of pro-inflammatory cytokines such as interferons (IFN-α, IFN-γ), tumor necrosis factor-alpha from PMC ISOCHEM (Vert-Le-Petit, France). Oleic acid was purchased from Alfa Aesar (Karlsruhe, Germany) whereas isostearic acid was a kind gift from Biochim (Milan, Italy). Linoleic acid and linolenic acid were purchased respectively from Alfa Aesar (Karlsruhe, Germany) and Sigma-Aldrich (Saint Louis, MO, USA). Plurol ® Oleique (mixture of polyglyceryl-3 esters of oleic acid with a predominant diester fraction) and Peceol™ type 40 were a kind gift from Gattefossé (Saint-Priest, France). Span ® 80 was purchased from Bregaglio (Biassono, Italy). Natrosol™ 250 M (MW 720 kDa) was a kind gift of Ashland (Wilmington, DE, USA) whereas xanthan gum was purchased from Sigma-Aldrich (St. Louis, MO, USA). Sodium carboxymethylcellulose E466 (MW 25 kDa) was purchased from A.C.E.F. S.p.A. (Fiorenzuola d'Arda, Italy) whereas Carbopol ® 934NF (MW 300 kDa) was purchased from Lubrizol (Wickliffe, OH, USA). Sodium hyaluronate (HA; MW 1000 kDa) was a kind gift from IBSA Farmaceutici Srl (Lodi, Italy) while PVA Gohsenol™ EG-30P (MW 98 kDa) was purchased from Nippon Synthetic Chemical Industry Co. Ltd. (Osaka, Japan). Albumin from bovine serum and nile red (NR; 9-(diethylamino)-5H-benzo[a]phenoxazin-5one; MW 318.4 g/mol) were purchased from Sigma Aldrich (St. Louis, MO, USA). High purity water was used (Arium ® comfort, Sartorius, Goettingen, Germany); all solvents were of analytical grade.

Imiquimod Quantification Method
IMQ quantification was performed by HPLC (Flexar, Perkin Elmer, Waltham, MA, USA) equipped with a reverse-phase C 18 column (Kinetex C18 2.6 µm, 100 Å, 75 × 4.6 mm, Phenomenex, Torrance, CA, USA) and a C 18 guard column (SecurityGuard Widepore C18, Phenomenex, Torrance, CA, USA). The mobile phase, composed of methanol:acetonitrile:water:triethylamine (180:270:530: 20), was pumped at a flow rate of 0.5 mL/min. In these conditions, the retention time of the drug was about 4 min. Imiquimod detection was carried out through UV or fluorescence detection. In particular, imiquimod solubility was mostly assessed by UV absorbance (λ 242 nm; injection volume: 10 µL), whereas fluorescence measurement (λ exc 260 nm, λ em 340 nm, injection volume: 1 µL) was used in order to evaluate imiquimod accumulation and permeation through tissues. Calibration curves in the intervals 0.5-15 µg/mL for UV and 0.05-2.5 µg/mL for fluorescence were built starting from a stock solution prepared by dissolving approximately 2 mg of imiquimod in 20 mL of 0.1 M HCl. The HPLC methods were previously validated for sensitivity, precision and accuracy [28].

Fatty Acids Quantification Method
Fatty acids quantification was performed by HPLC (Flexar, Perkin Elmer, Waltham, MA, USA) equipped with a reverse-phase C 18 column (Nova-Pack C18, 150 × 3.9 mm, Waters, Milford, MA, USA) for oleic acid and with a reverse-phase C 8 column (Aeris WIDE-PORE XB, 150 × 4.60 mm, Phenomenex, Torrance, CA, USA) for linoleic and linolenic acid. In both cases, the columns were thermostated at 65 • C and the flow rate was 1.6 mL/min. For oleic acid, the injection volume was 50 µL and the mobile phase was composed of acetonitrile:water with trifluoroacetic acid (TFA) 0.1% in ratio 65:35 (v/v), whereas linoleic and linolenic acids quantification was carried out by using acetonitrile:water with TFA 0.1% in ratio 60:40 (v/v) and an injection volume of 100 µL. Absorbance of oleic acid, whose retention time in these conditions was about 12 min, was measured at 210 nm. The same wavelength was used for quantifying both linoleic and linolenic acid, whose retention times were about 2.7 min and 2.1 min respectively. Details on calibration curves, RSD%, RE% and LOQ values are reported in Table S1.  20, 40, 100, 200 mM) in high purity water to obtain formulations T20, T40, T100 and T200. TPGS micelles were then saturated with oleic acid (O), linoleic acid (L), linolenic acid (LN) and isostearic acid (I) or with oleic acid esters, namely Plurol ® Oleique (PO), Peceol™ (P) or Span 80 ® (SP) (see Table 1 for details on the structures). Briefly, the micellar solutions were added with the above-mentioned excipients at 4:1 volumetric ratio and the obtained mixtures were subjected to 100 revolutions (made by hand) to ensure complete saturation of the aqueous phase. After saturation, the oily phase was removed and a 2-folds filtration (regenerated cellulose, Minisart RC 0.2 µm, Sartorius, Gottingen, Germany) of the aqueous phase was performed in order to get a clear micellar solution. The obtained formulations (composition and details are shown in Table 2) were then saturated with imiquimod (see Section 2.5). In order to deepen the effect of different preparation methods on IMQ solubility, other processes for micelles development were tested using only TPGS and oleic acid. In this case, a mixture of methanol and oleic acid in ratio 1:1 was left under magnetic stirring until a clear solution was obtained. Then, 4 mL of this solution were injected into 8 mL of T20 and left under magnetic stirring for 30 min. Afterwards, methanol was evaporated by stirring the system in a water bath at 50 • C for approximately 2 h. Finally, the system was filtered 2 times (regenerated cellulose, Minisart RC 0.2 µm, Sartorius, Gottingen, Germany) to eliminate the excess of oily phase. Even after filtration, the solution remained opalescent (see DLS analysis, Table 3 for details). The obtained formulation (composition and details are shown in Table 2) was then saturated with imiquimod (see Section 2.5).  A total of 2 mL of oleic acid was saturated with imiquimod (approximately IMQ conc.: 74 mg/mL). After filtration, the saturated solution was added in a 1:4 ratio to T20, as in the method 1 (Section 2.4.1).

Imiquimod Loading in the Polymeric Micelles
In order to load (IMQ), an excess of the drug was added to the blank micellar formulations prepared (a list is reported in Table 2). Mixtures were left under magnetic stirring at room temperature for 48 h. Then, the suspension was centrifuged at 12,500 rpm for 15 min and the supernatant was sampled and analyzed by HPLC-UV or HPLC-fluorescence for imiquimod solubility determination, after 1:100 dilution in 0.01 M HCl.

Micelles Characterization and Stability
Size, polydispersity index (PDI) and surface charge of blank and IMQ-loaded micelles were measured by Dynamic Light Scattering (DLS) using Zetasizer Nano-ZSP (Malvern Instruments, Malvern, UK). Measurements were carried out at 25 • C after 10-folds samples dilution in high purity water. Surface charge determination was performed by 40-folds samples dilution in 0.5 mM KCl.
Stability over time (storage at room temperature) was also determined. Imiquimod concentration in TO20 and TI20 micelles was evaluated after 3 and 6 months from the preparation, using the HPLC method previously described and by diluting the samples 1:100 in 0.01 M HCl. pH of TO20 and TI20 micelles was measured after 0, 100 and 200 days from micelles preparation. In case of TO20 micelles, in order to further describe polymeric micelles behavior over time, size was measured after 100 days from samples preparation. Stability over time of TLN20 and TL20 was evaluated visually, as the formation of a white precipitate occurred after 14 days from micelles preparation.

Hydrogels Preparation
Hydrogels were prepared by hydration of the polymers in high purity water. In case of xanthan gum (XG), hydroxyethyl cellulose (HEC), sodium hyaluronate (HA) and Carbopol ® (CP), the polymer was added to high purity water and left under magnetic stirring at room temperature until complete hydration. Carbopol ® solution was then added with 2 M NaOH until the formation of the hydrogel was observed. Sodium carboxymethylcellulose (CMC) hydrogel was prepared by intermittent vortexing after dispersing the polymer in high purity water. Polyvinyl alcohol (PVA) gel formulation was obtained by dispersion of the polymer in water overnight at room temperature followed by heating at 90 • C until complete hydration. After hydration, all hydrogels were left to rest overnight. When necessary, in order to remove bubbles, hydrogels were centrifuged at 3000 rpm for 10 min. The same procedures described above were followed to prepare micelles-based hydrogels. In this case, hydrogels were obtained by hydration of the polymers in both the blank TO20 and IMQ-loaded TO20 micelles. pH of the formulations was measured for water-based as well as IMQ-loaded TO 20-based hydrogels.

Evaluation of Micelles Diffusion in Water-Based Hydrogels
A preliminary experiment was performed in order to describe the ability of TO20 micelles to diffuse across the different hydrogels. The experimental procedure was carried out as a preliminary screening to select the most promising hydrogels for further testing. With this aim, a saturated solution of nile red in blank TO20 micelles was prepared. Briefly, 10 µL of a 10 mg/mL nile red solution in DMSO (conc. 10 mg/mL) were added to 1.5 mL of blank TO20, which was then centrifuged at 13000 rpm for 5 min to precipitate the excess of nile red.
In a 3 mL plastic tube, 2.5 g of hydrogels were gently transferred, taking care to avoid bubbles formation. Next, 200 µL of nile red-loaded micelles were deposed on the surface of each hydrogel ( Figure S1). Diffusion of the micelles through the hydrogels was visually evaluated at predetermined time points (1 h, 2 h, 24 h).

Skin Accumulation and Permeation Experiments
Skin accumulation and permeation experiments were performed by using porcine tissues obtained from a local slaughterhouse (Macello Annoni Spa, Madonna dei Prati, Busseto, Parma, Italy). The skin was excised from the outer part of pig ears (breed, Large White and Landrance; weight, 145-190 kg; age, 10-11 months; sex, male and female) within 3 h from animal death, and separated from the underlying cartilage utilizing a scalp. Afterwards, the tissue was frozen at −20 • C and used within 3 months. Defrosted skin was mounted on glass Franz-type diffusion cells (DISA, Milano, Italy; 0.6 cm 2 surface area) with the stratum corneum facing the donor compartment. To guarantee sink conditions, the receptor compartment was filled with 1% w/v bovine serum albumin solution in PBS pH 7.4 (IMQ solubility: 143 ± 3 µg/mL) [37]. Formulations tested were TO20, TO20-2, TO100 (see Table 2 for composition), micellar-based hydrogels HA and XG, and the commercial formulation Imunocare ® (composition: isostearic acid, benzyl alcohol, cetyl alcohol, stearyl alcohol, white soft paraffin, polysorbate 60, sorbitan stearate, glycerol, methyl hydroxybenzoate, propyl hydroxybenzoate, xanthan gum and purified water; IMQ concentration: 50 mg/g) as control. All the donors were applied at infinite dose (200 mg/cm 2 , occluded) for 6 h. At the end of the experiment, the receptor solution was sampled, the formulation was removed from the donor compartment and the skin surface was rinsed with water to clear away the last traces of micelles. The tissue was then delicately dried and tape-stripped twice (Scotch Booktape #845, 3 M Co., St. Paul, MN, USA) to remove the most external layers of stratum corneum and eliminate any traces of imiquimod-loaded formulation, thus avoiding potential contamination. Consequently, skin samples were heated (hairdryer for 60 s) and epidermis was separated from dermis using a spatula. This separation procedure could slightly affect skin distribution, increasing drug levels in the deeper skin layers. Nonetheless, the short heat application time should limit this phenomenon; additionally, all tissues were treated equally, and any possible increase in drug diffusion to the dermis would be comparable for all samples. IMQ extraction from tissues was carried out by treating tissues with 1 mL solution of oleic acid:methanol (1:3) for the epidermis and 1 mL solution of PEG 400:methanol: 1 M HCl (1:2:2) for the dermis. Samples were left under these conditions overnight at room temperature, centrifuged at 12,500 rpm for 10 min and analyzed by HPLC-fluorescence. In order to evaluate IMQ permeation, 1 mL of the receptor was added of 50 µL of 70% v/v perchloric acid to precipitate albumin, and centrifuged (12,000 rpm, 15 min). IMQ quantification was performed by HPLC-fluorescence. The extraction procedure from epidermis and dermis and the recovery from the receptor solution were previously validated [15]. Permeation and accumulation experiments were also performed using nile red-loaded TO20 micelles (prepared as in Section 2.8.) to visualize the skin distribution of the fluorescent probe by two-photon microscopy.

Statistical Analysis
All data presented in text, figures and tables are reported as mean value ± SD. The significance of the differences between the results was assessed using Student's t-test. Differences were considered statistically significant when p < 0.05.

Two-Photon Microscopy Method
Porcine skin samples were analyzed with a two-photon microscope Nikon A1R MP+ Upright equipped with a femtosecond pulsed laser Coherent Chameleon Discovery (~100 fs pulse duration with 80 MHz repetition rate, tunable excitation range 660-1320 nm). A 25× water dipping objective with numerical aperture (NA) 1.1 and 2-mm working distance was employed for focusing the excitation beam and for collecting the two-photon excited fluorescence (TPEF) and the second harmonic generation (SHG) signals. TPEF/SHG signal was directed by a dichroic mirror to a series of four non-descanned detectors (three high sensitivity GaAsP photomultiplier tubes and a multi-alkali detector) allowing fast image acquisition. The four detectors were preceded by optical filters allowing the simultaneous acquisition of four separated channels: blue channel (415-485 nm), green channel (506-593 nm), red channel (604-679 nm) and far-red channel (698-750 nm). Imaging overlay of the four channels and processing was performed by the operation software of the microscope. Additionally, a fifth photomultiplier GaAsP detector, connected to the microscope through an optical fiber and preceded by a dispersive element, was used to record the spectral profile of the TPEF/SHG signal (wavelength range 430 to 650 nm with a bandwidth of 10 nm).
For microscope observations, the full-thickness skin samples were placed in a dedicated plexiglass holder and saline solution was used to dip the objective and to avoid dehydration. The samples were left at room temperature under these conditions for 20 min before measurements, to allow complete thawing. Z stack acquisitions of the porcine skin samples were taken from the stratum corneum to the dermis (150-350 µm deep, with steps between 0.8-1.5 µm depending on the experiment). Images were acquired with a typical field of view of 500 × 500 µm (with 1024 × 1024 pixels definition) and an average acquisition time per image between 2 and 8 s in order to achieve a suitable signal-to-noise ratio.
Different excitation wavelengths were used, between 850 and 1100 nm. Tissue autofluorescence and nile red emission fall in the green and red channel (the latter is much more intense and red-shifted independently of the excitation wavelength). Detector gains for the red and green channels were set at equal values for the collection of images of the same sample, except where explicitly reported. When exciting below 970 nm, the SHG signal falls in the blue channel: the gain of this detector is usually set at higher values compared to red and green channels, in order to better appreciate the SHG signal. No significant photobleaching has been observed in our experiments under the conditions employed to acquire both images and emission spectra.

Results
TPGS, a water-soluble vitamin E derivative (Table 1), has demonstrated the ability to remarkably increase water-solubility of various hydrophobic compounds as a result of micelles formation in solutions [38,39]. Additionally, some authors also report a permeation enhancing effect of TPGS itself [20,24,40]. Finally, micelles could also promote drug uptake into the skin due to their nanometric size, as reported for retinoic acid [41], hydrocortisone [42] and econazole [43]. In this paper, TPGS-based micelles were evaluated to increase imiquimod solubility and promote its uptake into epidermis and dermis.

Imiquimod Solubilisation
Imiquimod is a small molecule (molecular weight of 240.3 g/mol) with a very low solubility in water (0.6 µg/mL [44]; 2.4 µg/mL [45]). As the drug is a weak base (pka = 7.3), its water-solubility could be slightly improved by protonation at acidic pH [44]. We evaluated the effect of different formulations on imiquimod solubility. The results obtained are summarized in Table 2.
IMQ solubility was firstly evaluated in solutions containing 20 mM TPGS. This concentration is well above the CMC, that was previously measured and resulted approximately 0.1 mM [23]. This polymer previously demonstrated to significantly increase the solubility of hydrophobic compounds such as dexamethasone [46], cyclosporine [23], econazole [47], taxanes [48], minoxidil [40] and simvastatin [49]. Furthermore, TPGS addition to tetramethylguanidinium tetradecanoate micelles was able to increase imiquimod encapsulation efficiency and drug loading [50]. Even though, IMQ solubility in 20 mM TPGS was lower than 20 µg/mL. This disappointing result is probably linked to the low affinity of im-iquimod for the hydrophobic micellar core, made of tocopherol, together with the small size of the hydrophobic block (approximately 400 Da), that has been correlated with a low efficiency of encapsulation [51][52][53].

Effect of Co-Encapsulation of Fatty Acids and Oleic Acid Esters
In order to increase drug loading, we evaluated the possibility of co-encapsulate a lipophilic compound that could, at least in principle, enter the micellar core and increase its affinity for the drug. This kind of strategy has already been used for enhancing drug encapsulation in other nanocarriers such as niosomes [25] and lipid-based particles [26,27]. As fatty acids have previously demonstrated good imiquimod solubilization properties [15,44], we studied their co-encapsulation into TPGS micelles. Compounds evaluated, liquid at room temperature, were isostearic (C18:0), oleic (C18:1), linoleic (C18:2) and linolenic (C18:3) acids; their structure and relevant physicochemical properties are reported in Table 1. Fatty acids were added to a 20 mM TPGS solution at the saturation and the excess was removed by filtration to obtain a limpid micellar formulation, afterward loaded with IMQ. IMQ solubility values are reported in Table 2 and Figure 1, together with the amount of fatty acid loaded, quantified by HPLC. Figure 1. Dark bars indicate IMQ solubility (mean ± SD, Y axis on the left hand) in TPGS 20 mM solutions saturated with oleic (TO20), linolenic (TLN20), linoleic (TL20) and isostearic acid (TI20). Light bars represent the content (mean ± SD, n ≥ 3) of fatty acids (Y axis on the right hand). In case of isostearic acid, quantification was not performed due to the absence of a detection method (isostearic acid is not an UV-absorber). The table reports the % of the drug in the formulations after 3 and 6 months from preparation (storage at room temperature) as well as the value of the pH after 6 months. The exact composition of vehicles, with values of drug solubility and fatty acids content is reported in Table 2. (**** p < 0.0001, *** p < 0.001, ** p < 0.01).
The solubility value of IMQ in TPGS/fatty acids mixtures was, in all cases, higher than in TPGS alone, supporting a synergistic solubilizing effect produced by the incorporation of fatty acids in the micellar core: in the absence of TPGS, oleic acid water solubility resulted lower than 60 µg/mL (in agreement with literature data [54]), while in the presence of TPGS, its solubility was higher than 3 mg/mL, testifying its loading in the micelles. Features such as molecular weight, chain length, unsaturation [55], pK a [56] and log P octanol/water can affect fatty acids ability to accumulate in the micellar core and impact on micelles internal structure. For instance, unsaturated fatty acids, in consideration of their ability to reduce lipidic molecules packing in model membranes [57], could also increase the fluidity of micellar hydrophobic core facilitating the incorporation of the drug.
From Figure 1 it is evident that linolenic acid has the highest affinity for micelles core, while oleic acid has the lowest (quantification of isostearic acid was not possible, as it is not UV-absorber). However, the amount of fatty acid loaded was not proportional to the imiquimod solubility. In fact, the highest solubility (1154.08 ± 112.78 µg/mL) was reached with oleic acid, while linolenic, linoleic and isostearic acid gave an IMQ solubility respectively of 912.59 ± 27.50 µg/mL, 731.77 ± 52.25 µg/mL and 393.24 ± 13.34 µg/mL (Figure 1).
It is well known that one of the main problems of micellar formulations is represented by the low stability which is affected by several parameters including the cohesion of the hydrophobic core [53]. In case of TL20 and TLN20, drug precipitation occurred after some weeks. The reason is linked to the expulsion of linolenic and linoleic acid from the TPGS micelles. In fact, fatty acid concentration decreased from 7898.09 ± 180.30 to 280.53 ± 7.99 µg/mL (TLN20) and from 3998.00 ± 140.60 to 258.33 ± 22.25 µg/mL (TL20). In case of TO20 and TI20 micelles, on the contrary, no precipitation occurred and IMQ concentration remained stable up to 6 months from preparation ( Figure 1). Additionally, the pH remained substantially stable as well as micelles size (see Section 3.2). The difference in stability between the different fatty acids can be due to their different 3D structure, lipophilicity or pKa (Table 1). Another reason could be linked to the lower chemical stability of linoleic and linolenic acid, caused by the number of unsaturation, which increases their tendency to oxidate.
As the presence of oleic acid demonstrated a positive impact on IMQ solubilization and a good stability over time, we also evaluated the possible solubilizing effect of lipophilic surfactants composed of oleic acid esterified with different polar heads, namely glycerol (Peceol™, formulation TP20), sorbitan (Span ® 80, formulation TSP20) or a bigger and more hydrophilic one (Plurol ® Oleique, formulation TPO20) (see structures in Table 1). This could help in defining the influence of hydrophilic groups features on IMQ solubility. In all cases, IMQ solubility was below 200 µg/mL ( Table 2), suggesting that the presence of a free carboxylic group is essential either for entering the micellar core or for solubilizing the drug. Indeed, from one side the higher MW and hindrance could have prevented the penetration of the hydrophobic chain in the micellar core, and, on the other side, IMQ solubility could be related to an interaction between the positively charged drug (pKa = 7.3) and the fatty acid, with a beneficial influence on IMQ accumulation in the hydrophobic micellar core [58,59].

Effect of TPGS Concentration
When IMQ was added to 40, 100 and 200 mM TPGS, an increased solubility was obtained ( Figure 2 panel A), however very limited and not linearly correlated with TPGS concentration. As oleic acid demonstrated to be the best co-solubilizing agent among all the compounds tested, solutions of TPGS at concentration of 40 and 100 mM were saturated with oleic acid. It was not possible to obtain micellar formulations from 200 mM TPGS saturated with oleic acid, due to the formation of a gel-like system [37] that prevented the separation of the oily phase.
The result (Figure 2, panel A) highlights an increase in solubility roughly proportional to TPGS concentration: in this case TPGS role in drug solubilization appears to be pivotal. By analyzing also oleic acid content, it is clear (Figure 2 panel B) that a linear correlation exists between IMQ solubility and oleic acid concentration, even if the 10-folds increase in oleic acid registered moving from TPGS 20 mM to TPGS 100 mM, reflects in only 2.5 increase in IMQ solubility. The slope of the line indicates that approximately 10 molecules of oleic acid are necessary for the solubilization of 1 molecule of (IMQ), however, the Y-axis intercept suggests that also other phenomena are involved. and without oleic acid (T). Panel (B) reports the correlation between oleic acid concentration (mM; mean ± SD) and IMQ solubility (mM; mean ± SD) in micelles with TPGS at concentration of 20 (produced by using both method 1 and 2), 40 and 100 mM (produced by using method 1).

Effect of Preparation Method
The important role of oleic acid content on IMQ solubilization is also evident if considering a different method used to saturate the aqueous phase with oleic acid (Section 2.4.2). In this case, the obtained 20 mM TPGS micellar formulation (TO20-2, Figure 2 panel B) loads a higher oleic acid amount and, as a consequence, a higher amount of IMQ (1659.34 ± 109.87 µg/mL). However, the oleic acid is not completely solubilized in micelles TO20-2 as testified by a slight opalescence and by the results of the DLS analysis (see loaded TO20-2 in Table 3).
A further preparation method was also considered (Section 2.4.3). In this case, IMQ was first dissolved at the saturation (approximately 75 mg/mL [15]) in oleic acid and the obtained solution was used to saturate the 20 mM TPGS micelles. This method, evaluated to assess the direct loading of a possible oleic acid-IMQ ion pair, brought to a very low drug solubility (304.76 ± 4.06 µg/mL).

Size Determination
As the size of nano-systems has been reported to play a key role in drug delivery to skin deepest layers, determination of micelles dimension by DLS is considered fundamental to better describe their behavior when applied to skin [17,60].
TPGS micelles size has been previously determined both by DLS and by X-ray scattering [47] and resulted approximately 12 nm. Images from literature [21,61] showed a spherical shape. The results obtained here (T20, T40, T100) are in reasonable agreement with this value (Table 3). Apparently the PDI decreases as the polymer concentration increases. No variation of micelles size was registered before and after loading with the drug and, in case of T40 and T100, the PDI decreased probably due to improved stability and cohesion of micelles caused by the development of hydrophobic interactions between the core and the lipophilic drug incorporated [53]. The addition of oleic acid did not substantially change micelles size.
Multimodal particle size distribution was obtained after addition of surfactants such as Peceol™, Plurol ® Oleique and Span ® 80 to the 20 mM TPGS solution in case of both blank and IMQ-loaded micelles (Table S3).
The evaluation of micelles size has been used also to further confirm TO20 and TI20 micelles stability over time which demonstrated to have, respectively, dimensions of 14.90 ± 0.13 nm and 13.40 ± 0.16 nm after 3 months from preparation. In case of TO20 (further investigated in the manuscript), zeta potential was also measured. Blank micelles had slightly negative zeta potential (−5.82 ± 0.44 mV), in agreement with pegylated nanocarriers.

Loading of Polymeric Micelles in Hydrogels
Rheological properties of a formulation play a very important role in dermal delivery, as they influence the retention time on skin surface and even impact on patients' acceptability [62]. In order to preserve the micellar structure, a careful evaluation of the thickening agent should be done. In particular, after incorporation, the behavior of micelles, with specific attention to their diffusion capacity, should be investigated as the nature of polymeric chains as well as their disposition in the hydrogel matrix could prevent micelles release from the system [63]. Polymers for hydrogels preparation were selected considering different characteristics in terms of molecular weight and degree of ionization (neutral or negatively charged polymers). Both natural and synthetic polymers were evaluated. The concentration has been determined empirically on a case-by-case basis, to reach the adequate viscosity for cutaneous application.
The visual evaluation of the diffusion of nile red-loaded micelles through blank hydrogels has been used as marker of the ability of micelles to be easily released from the formulation (Figure 3). At the deposition, nile red-loaded micelles immediately diffuse in HA hydrogel (see also Figure S1) in which completed diffusional process was observed within few hours. On the contrary, PVA hydrogel was not penetrated by nile red-loaded micelles even after 24 h from deposition. HEC, CMC and CP were characterized by a minimal (if any) micelles diffusion after 24 h. A slightly better result was found for Xanthan gum (see the thickness of the colored layer below the black mark at 24 h in Figure 3, Column 1).
These results were also confirmed when polymers were hydrated in blank and scloaded micelles (TO20), Figure 3, Column 2. In fact, in case of HEC, CMC, PVA and CP, the formulation became opalescent or whitish, probably as a result of micelles disruption with consequent release of the oleic acid and the drug. HEC hydrogel showed also a marked phase separation. On the contrary, HA and XG gels preserved their characteristics and were selected to be used for further experiments. Thus, XG and HA hydrogels were identified as the most interesting because of their higher stability and the possibility of micelles diffusion.

Imiquimod Skin Accumulation and Permeation
In order to determine the potentialities of micellar formulations for imiquimod skin delivery, the drug accumulation in full-thickness porcine skin was evaluated after 6 h from the application of different formulations. While IMQ was recovered from epidermis and dermis, no transdermal permeation was observed in the receptor compartment, in which albumin presence ensured sink conditions. The lack of permeation can be attributed to the slow permeation rate and the limited application time (6 h), selected as a reasonable time of formulation persistence on the skin. Formulations tested and accumulation values are listed in Table 4.

Imiquimod Skin Delivery from Micelles
The data obtained in the present work were compared to the one previously obtained starting from a saturated solution of Transcutol ® [15]. IMQ concentration in this vehicle is very similar to TO20 micelles (namely 1.11 ± 0.07 µg/mL), it has the same thermodynamic activity (equal to 1, they are both at the saturation) and, being Transcutol ® a hydrophilic compound, IMQ should have a similar tendence to partition in the SC. However, as illustrated in Figure 4, drug accumulation into the skin from TO20 micelles is 10-fold higher compared to Transcutol ® saturated solution. In particular, the micellar formulation is able to increase penetration depth as dermal levels resulted 25-fold higher for TO20 than for Transcutol ® saturated solution.  Table S4. By increasing TPGS concentration (and thus IMQ solubility) the skin accumulation did not change substantially, and we even found for TO100 a (not-statistically significant) reduction of IMQ skin retention. This could be ascribed to the slightly higher viscosity of the sample, probably due to micelle-micelle interactions [64,65] that limited their freedom of diffusion and thus their interaction with the SC. Additionally, the result supports the substantial absence of penetration enhancing activity of TPGS for imiquimod skin delivery, as recently reported also for rapamycin by Quartier et al. [61].
The formulation TO20-2 gave rise to a significant increase of IMQ accumulation in epidermis, probably attributable to the presence of small oleic acid droplets (Table 4 and Section 3.1.3) and higher IMQ concentration (Table 2). This formulation, however, is not able to promote a deeper IMQ penetration (see dermis concentration), presumably because oleic acid droplets remained trapped in the SC [15].
Based on these results, we can hypothesize that the presence of small micelles (approximately 15 nm in diameter) free to diffuse can significantly enhance IMQ penetration into the skin. The good performance of nanometric micelles was also reported by other authors [16,[66][67][68] and was ascribed to the improved accumulation of the nanosystems in the hair follicles as well as to the affinity of the polymers used for both the stratum corneum and the cellular membrane, which results in a strong change in the membranes fluidity, thus promoting the passive uptake.
In order to visualize the behavior of this formulation on skin tissue, TO20 micelles loaded with nile red were prepared and applied for 6 h on full-thickness porcine skin. As a reference, a blank experiment was also done, by using saline solution as donor.
Two-photon microscopy demonstrated to be a very powerful tool for the study of skin structure and fluorescent probe distribution: details on the technique and selection of the optimal excitation wavelength are presented in Appendix A. Figure 5 illustrates the volume rendering of a porcine skin sample where the different skin layers are clearly visible as well as nile red accumulation in SC and viable epidermis; the green signal is referred to the collagen fibers present in the dermis (SHG signal). A more complete and detailed view is available in Supplementary Movie S1: the probe clearly marks all the cellular elements of the skin, i.e., epidermis and also vascular and corpuscular elements in the dermis. From this sample, at an 80 µm depth ( Figure 5B), the spectral profile of the emitted/generated signal was also recorded and compared with the emission spectrum collected from nile red-loaded TO20 micelles ( Figure 5C) with an Edinburgh FLS-1000 fluorimeter. Emission of nile red is strongly sensitive to the polarity of the environment [69]. The emission of nile red in porcine skin is shifted towards shorter wavelengths compared to the emission of nile red in TO20 micelles. The shift of the emission spectrum of nile red indicates a change of environment polarity, suggesting that the fluorescent probe has been released from the micelles. Overall, these images support the capability of TO20 micelles to efficiently deliver the probe to the skin, integrating the data obtained with imiquimod. A specific skin section with a hair follicle, presented in Figure 6, was analyzed and compared to a control sample (skin treated with saline solution). In this case nile red accumulation in the hair shaft and its penetration into the hair follicle were demonstrated (see also Figure A3). Details on the technique used and further images are presented in Appendix A.

Imiquimod Skin Delivery from Semisolid Formulations
When micelles were included in a gel, essential for skin application [61], drug delivery was slightly reduced, in particular to the dermis (Table 4). However, this reduction is not statistically significant, and the result obtained support the selection of XG and HA as thickening agents capable to preserve micelles diffusion. A similar result was obtained by Lapteva et al. [70] who used mPEG-hexPLA micelles embedded in a carboxymethyl cellulose gel.
Finally, the results can be compared with the commercial cream. The drug level found in epidermis and dermis are comparable to the formulation prepared in the present work, however, the delivery efficiency of Imunocare ® (IMQ concentration: 50 mg/g) is extremely low (Figure 4, panel B), and 42-fold lower than TO20. Thus, the micellar formulation is able to produce similar epidermis and dermis accumulation with a 40-fold lower concentration, with possible advantages in the therapeutic outcome and reduced systemic side effects in case of damaged skin. These side effects, although rare, are fever, vertigo, myalgia and, anecdotal reports of distant inflammatory mucosal reactions [71]. Additionally, IMQ is characterized by important local side effects. Some of them (irritation and erythema) are an extension of the pharmacologic effect, while others (skin infection, skin hypo and hyper pigmentation, alopecia) are unrelated. The dramatic reduction of the concentration of the vehicle could in principle have an effect also on some of these manifestations. Additionally, the presence of TPGS, that can deliver vitamin E to the epidermis, can mitigate some local reactions, as indicated in a recent paper [72]. Indeed, TPGS can be hydrolyzed to vitamin E [23], thanks to the presence of esterase activity in the stratum corneum.

Conclusions
In the present paper, TPGS-based micelles co-loaded with oleic acid were prepared, characterized and exploited for solubilizing and delivering imiquimod to the skin. A linear correlation was found between imiquimod solubility and oleic acid amount in the micelles, underlining the relevant role of this fatty acid in imiquimod formulation. These micelles, prepared by a very easy procedure, were stable at least for 6 months and, applied ex vivo to porcine skin, gave rise to a relevant imiquimod skin deposition, with a delivery efficiency 40-fold higher than the commercial cream. A careful selection of the thickening agents permitted to obtain a gelified formulation preserving the skin delivery performance of the micelles.
In consideration of these findings, it is possible to affirm that the formulation TO20 is a promising vehicle for imiquimod delivery to the skin and, hopefully, for reducing side effects, due to the lower drug concentration in the vehicle and the presence of TPGS.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/pharmaceutics13091476/s1, Table S1: Range of calibration curves, RSD% (relative standard deviation %) and ER% (relative error %) of the fatty acid considered; Table S2: Hydrogels composition, with their codification, appearance and pH values. Table S3: Size (mean ± SD), intensity and PDI of blank and IMQ-loaded micelles of TP20, TPS20 and TPO20. Table S4: p values obtained from t-test analysis for epidermis, dermis and total amount of the drug in the skin for all formulations tested; Figure S1: Hydrogels before and after deposition of 200 µL of Nile-Red loaded micellar solution. Supplementary Movie S1: 3D reconstruction of porcine skin after 6 h treatment with nile-red loaded TO20 micelles (excitation wavelength: 1070 nm); Supplementary Movie S2: 3D reconstruction of untreated porcine skin (excitation wavelength: 850 nm); Supplementary Movie S3: 3D reconstruction of hair follicle after 6 h treatment with nile-red loaded TO20 micelles (excitation wavelength: 850 nm).

Appendix A
Two-photon microscopy images of blank skin samples and selection of the excitation wavelength. Figure A1 reports volume renderings of untreated porcine skin samples reconstructed from the Z-stacks when excited at 850 nm. In this condition, the dermis is blue, due to the second harmonic generation (SHG) signal typical of collagen fibers [73], while stratum corneum and viable epidermis are characterized by a marked green autofluorescence. The same appearance has been detected for control samples treated with saline solution and blank TO20 formulation.
Longer excitation wavelengths were then evaluated: Figure A2 refers to a skin sample (80 µm deep) where, due to the inclination of the tissue, SC, viable epidermis and dermis are visible on the same plane. The two panels refer to excitation at 900 nm (panel A) and 1100 nm (panel B). From the image it is evident that at 900 nm, green autofluorescence is still present in SC and epidermis while, when the tissue is excited at 1100 nm, autofluorescence is not present in case of viable epidermis and only a very limited red signal is present on the SC. This last excitation wavelength was then used to obtain the images reported in Figures 5 and 6 of the main text. Figure A1. Volume renderings of untreated porcine skin samples reconstructed from the Z-stacks (Z-step: 1 µm) acquired with excitation wavelength of 850 nm. XY (A) and XZ (C) views of a hair follicle (total depth: 191 µm); XY (B) and YZ (D) views of a flat skin region (total depth: 209 µm). A more complete and detailed view of the flat skin region 3D rendering is available in Supplementary Movie S2. The blue signal is due to the second harmonic generation (SHG) of the collagen present in the dermis. The green signal is due to SC and epidermis autofluorescence when irradiated at 850 nm. Two-photon microscopy images (excitation at 850 nm) of skin samples treated with nile red-loaded TO20 micelles Despite the spectral overlap between tissue autofluorescence and nile red emission, the latter is much more intense and also images collected with excitation at 850 nm can help in describing nile red penetration into the skin from TO20 micelles. Figure A3 confirms the data previously shown (see Figures 5 and 6) regarding nile red retention in stratum corneum and epidermis. Panel B of the same figure, obtained with red channel saturated using LUTs, highlights the deep penetration of the probe in the hair follicle.   Figure A4, represents a sample of micelles-treated porcine skin at different depths, where images are collected exciting the sample at 850 nm. It is possible to visualize the presence of skin furrows whose width is reduced by proceeding towards the deeper layers and nile red accumulation in the epithelial cells. Furthermore, at 50 µm depth, collagen fibers from the dermis become clearly visible. The circular structure of these fibers has to be referred to the dermal papillae, ridge-like structures extended on the surface area of the dermal-junction in the papillary dermis.