Skin Targeting of An Optimized Caffeine Nanostructured Lipid Carrier With Improved Eciency Against Chemotherapy Induced Alopecia

The current investigation was designed to develop and optimize caffeine-loaded nanostructured lipid carriers (NLCs) for topical alopecia treatment. Screening of drug solubility in various excipients was executed. The 2 3 full factorial design was employed for NLCs optimization. Lipid type, surfactant type, and drug concentration were the independent variables. Entrapment eciency (EE), particle size, polydispersity index (PDI) and % drug release were the chosen responses. Physiochemical evaluation, in vitro release, ex-vivo permeation, and stability study were achieved. The solubility of caffeine in stearic acid and glyceryl monostearate (GMS) was 47.11 ± 3.048 and 32.67 ± 2.955 mg/g, respectively. Oleic acid: garlic oil mixture at ratio 1:1 v/v was the oily phase. Tween 80 and Cremophor EL, Transcutol HP, carbonate buffer (pH 10.8 and ionic strength 200Mm) were chosen as a surfactant, co-surfactant, and aqueous phase, respectively. The optimized formula showed particle size, %EE, PDI, zeta potential of 358nm, 72.55 %, 0.912, -24.8, respectively. The % release was 92.9 ± 4.9 % after 4hours. Confocal laser scanning microscopy showed an improved permeation of caffeine-loaded NLCs to the whole skin layers. The histological examination proved the eciency of caffeine NLCs optimized formula on promoting hair growth compared to the market formula. mixture was agitated at 200 rpm and 85ºC for 1hr in a water bath shaker (B S-11, Isothermal shaker, Korea) and then allowed to cool at room temperature for solidication. The miscibility of liquid/solid lipid mixture was examined by placing a smear of the cooled solid mixture on a lter paper. The existence or loss of oil droplets on the lter paper was observed. The absence of oil droplets will be judged as a homogenous miscible mixture. The liquid/solid lipid mixture that showed good miscibility was estimated for caffeine solubility using the above-described method for solid lipid [12]. slidge microtome, cleared xylene, and in paran 56 air for 24h. The obtained tissue sections were collected on glass slides, deparanized and the spreading of the dye was noticed using a confocal laser microscope. Skin pieces were inspected using normal and UV light (excitation at 405 nm, 458nm, 488 nm, 514nm, 543nm), and emission spectra were recorded at 633 nm. The CLSM photographs in bright-eld were overlaid to observe RB distribution [22]. reach 2.146 and 1.413 at pH 7.4 and 7.8, respectively. The same result was clearly observed with carbonate buffer at pH 10.8, whereas the drug solubility decreased from 3.320 % at 50mM ioinc strength to 0.801 at 200mM. The results revealed that increasing the buffer ionic strength statistically decreased the drug solubility, especially if it is accompanied by an increase in the pH of the buffer system. The lower solubility in buffers with high ionic strength was related to increasing the concentrations of dissolved ionic salts that in turn could decrease the solubility of the weak electrolytes. Moreover, these results could be due to weak basic properties of the drug that showed low solubility at high pH and high solubility at lower pH. At acidic pH, the drug has high solubility due to its ionization, whereas, at basic pH affecting the NLCs formulation were optimized using a factorial design program. The FTIR study of pure caffeine and excipients conrmed that there was no remarkable chemical interaction between them. DSC studies conrm the transformation of the crystalline nature of the drug into amorphous nature. The TEM analysis showed the NLCs were formulated and having a spherical shape. The results of zeta potential predicted good particle stability because the repulsive forces prevent aggregation over time. The results of the present study suggest caffeine NLCs as promising carriers for topical delivery of drugs applied for alopecia treatment. The NLCs carriers promote a higher growth rate of hair over the currently marketed formula of caffeine.

as particle size, %EE, in-vitro drug release, and drug permeation through the skin were judged through the independent variables for NLCs loaded caffeine.

Pre-formulation studies and excipients screening
Solubility studies Solubility of caffeine was estimated in distinct liquid lipids, namely; oleic acid, labra l M1944, IPM, caproyl 90, amla oil, rosemary oil, garlic oil, caprylic/capric triglyceride, and jojoba oil or oil mixtures; oleic acid: garlic oil (O: G) at ratios (2:1, 1:2 and 1:1% w/w). The solubility in surfactants (Tween 80 and Cremophor EL) and co-surfactants (Transcutol HP, Span 80, and Span 20) was also evaluated. An excess amount of caffeine (150mg) was dissolved in 1gm of each mentioned vehicle put into small stoppered vials. The drug-vehicle mixtures were constantly shaken for 72h at 37˚C±1.0˚C in a mechanical shaker (Shaking water bath, B S-11, Isothermal shaker, Korea). After equilibrium, the samples were centrifuged at 3000rpm for 20min. The supernatant was diluted with methanol and ltered through a 0.45μm membrane lter [9].
Caffeine concentration was measured by the HPLC-UV method after drug separation through Zorbax Eclipse Plus column C18 (150 mm x 4.6 mm, 5 µm particle size). The column temperature was kept at 40°C. The mobile phase consists of (water: methanol at ratio 60:40 % v/v). An injection volume of 20 µl, a ow rate of 1 ml/min, and a wavelength of 275 nm were adopted [10]. Caffeine solubility in distinct solid lipids namely, stearic acid and glyceryl monostearate (GMS) were also performed. One gram of each lipid was rstly melted at 80°C in a10 ml stoppered tube. The drug was gradually added to the melted lipid by increasing 5 mg per time then allowed to stir at 100 rpm via (digital hot plate magnetic stirring, Jenway, 1000, UK.) until the melted lipid failed to dissolve any more drug. Drug solubility in the molten lipid was judged optically within 15min by noticing the absence or presence of drug crystals. The experiment was conducted in triplicate [11].

Screening of lipids miscibility and drug solubility
From the previous solubility studies, the liquid and solid lipids that showed the highest solubility of caffeine were chosen for the miscibility screening test. Stearic acid or GMS and O: G at ratio (1:1) were blended at ratios; 50:50, 60:40, 70:30, 80:20 and 90:10. Each mixture was agitated at 200 rpm and 85ºC for 1hr in a water bath shaker (B S-11, Isothermal shaker, Korea) and then allowed to cool at room temperature for solidi cation. The miscibility of liquid/solid lipid mixture was examined by placing a smear of the cooled solid mixture on a lter paper. The existence or loss of oil droplets on the lter paper was observed. The absence of oil droplets will be judged as a homogenous miscible mixture. The liquid/solid lipid mixture that showed good miscibility was estimated for caffeine solubility using the above-described method for solid lipid [12].
Screening the optimum aqueous phase for preparation NLCs Phosphate buffer at pHs 7.4 and 7.8 and carbonate buffer at pH 10.8 at different ionic strengths were evaluated for drug solubility. Both phosphate buffer at pH 7.4 and 7.8, and carbonate buffer at pH 10.8 with ionic strength (50,100,150, and 200 mM) were prepared. The equilibrium solubility values of caffeine in aqueous and organic phases were measured by the classical saturation shake-ask method in duplicate [13].
Firstly, a saturation state for both solvents was done. An aliquot of 25 ml of aqueous phase (buffer system) was added to 25 ml of n-octanol in a separating ask and allowed to be shaken in a mechanical shaker for 24h. The solvents were allowed to stand enough time for phase separation and used for checking the drug solubility. Partitioning of caffeine between the two phases (organic and aqueous) is carried out by transferring an accurate amount of the drug (20 mg) into a stoppered separating funnel containing equal volumes of both phases. The funnels were subjected to constant shaking at 25°C for 3hr then allowed to stand with rotation from time to time through 180° about its transverse axis to allow enclosed air to rise through the two phases and such rotations were su cient to create partition equilibrium separation. The drug concentration in all buffer systems with their different ionic strength was determined according to the above-mentioned HPLC method. The buffer system showing the lowest drug solubility will be chosen as an optimum solvent for the formulation of caffeine-loaded NLCs.
Optimization of caffeine-NLCs software, State-Ease Inc., Minneapolis, USA). The drug concentration (1.5 and 3%w/w), lipid type (O: G: stearic or (O: G: GMS) at ratio 15:15:70), and emulsi er type (Tween 80/ transcutol HP and Cremophore EL/ transcutol HP) were the independent variables. The EE, particle size, PDI, and drug release were chosen as dependent variables (Table 1). The design suggests 8 formulas and experimental error could be estimated through repeating the experiments three times.

Preparation of caffeine loaded NLCs
Pre-emulsion/ultrasonication technique was the selected method for preparing caffeine-loaded NLCs. The lipid phase (2%w/w) of O: G: stearic acid or O: G: GMS at ratio 15:15:70 was weighed and allowed to melt at 60-70°C above their melting point by 5°C-10°C. An accurate amount of drug equivalent to 30 and 60 mg (1.5 and 3%w/w, respectively) dissolved in 0.5 ml methanol (to ensure complete drug solubility), was added to the hot lipid blend upon stirring (6000 rpm) to become clear. The aqueous phase was prepared by solubilizing of 3% w/v S/CoS mix (Tween 80/Transcutol or Cremophore /Transcutol HP at ratio 3:1) in preheated carbonate buffer solution. The aqueous phase was added portionwise to the hot lipid blend under constant stirring at 6000 rpm for 30-40 min to obtain a primary emulsion. The resulting emulsion was cooled at ambient temperature (25°C) and then subjected to ultra-sonication for 20min, 35kHz as frequency, the amplitude of 100%, and at 37°C to decrease particle size of solid lipids. The aqueous dispersions of NLCs were allowed to stand at ambient temperature 25°C overnight to allow complete separation of NLCs. The aqueous dispersion underwent cooling centrifugation at 19,000 rpm for 1h (high speed refrigerated centrifuge-cooling centrifuge (Remi cooling microfuge, Remi instrument Ltd, Mumbai, India). The separated NLCs were standing for drying at 25°C [14].

Characterization of caffeine loaded NLCs
Fourier transformation infrared spectroscopy (FTIR) FTIR spectroscopic study was achieved (Shimadzu IR/FTIR spectrophotometer (435 U-O4), Japan) and the FTIR spectra were logged in a wavelength region of 4500-400 cm -1 at ambient temperature with a resolution of 4cm -1 . Caffeine, stearic acid, GMS, drug: lipid physical mixture at ratio (1:1), blank NLC, and optimized caffeine-NLC formula were separately mixed with potassium bromide (KBr) at ratio 1:100 then pressed with a hydraulic press into discs for 5 min under continuous pressure of 5 tons [15].

Differential scanning calorimetry (DSC)
Thermograms of pure caffeine, stearic acid, GMS, drug: lipid physical mixture at ratio (1:1), blank NLC, and optimized caffeine loaded NLC formula were assessed using (DSC 60, Shimadzu, Japan). Five milligrams of each sample were placed into aluminum pans as standard (40μL), with a temperature range from 25°C to 400°C at a constant heating rate of 10°C/min. Through thermal analysis, dry nitrogen was the e uent gas, and purged at a constant ow rate (25mL/min) using reference standard [15].
Determination of the percentage of yield, entrapment e ciency (EE), and drug loading (DL) Percentages of yields of NLCs were calculated by weighing the amount of NLCs dry powder and then apply equation 1: Where W1 is the weight of the dry NLCs (g), W2 is the weight of the drug (g), and W3 is the weight of the lipids and surfactants (g).
The % EE was carried out by estimating the free drug amount dissolved in the supernatant after the separation of solid particles. Two milliliters of NLC dispersion was centrifuged at 19,000 rpm for 1hr at 4°C [16]. The supernatant was analyzed for unencapsulated caffeine at 275nm following HPLC assay previously described under the solubility study. Entrapment e ciency (%) was calculated following equation 2.
The % of drug loading (%DL) was calculated by applying equation 3. The drug analyzed in NLCs was related to the total drug amount, lipid, and oil excipients added for each formula [17]. pH determination and electroconductivity study was measured three times at ambient temperature. The electrical conductivity (σ) of the NLCs dispersions was determined using (HANNA digital conductometer instrument, USA) at a constant frequency of 1Hz at ambient temperature. The average reading ±S.D of triplicate measurements was recorded.
Determination of particle size, PDI, and zeta potential The particle size, polydispersity index (PDI), and zeta potential of caffeine loaded NLCs were determined by (Malvern Zetasizer nano series, Nano ZS 3000, Malvern, U.K) with a 50 MV laser. An amount of 1 gm from each formula was dispersed in 100 ml of distilled water under constant shaking. An aliquot of 1ml of this suspension was put into the cell for particle size and PDI analysis which was carried out at a xed angle of 90° and temperature of 25°C. For zeta potential analysis, the diluted samples were exposed to an electric eld (1V). The data were expressed in triplicate ± S.D [18].
In-vitro release study Caffeine release from NLCs was performed using cellulose membrane (Sigma, molecular weight cut off is ranged from 12,000-14.000 Da) pre-soaked overnight at 25°C in phosphate buffer pH 7.4. One milliliter of each formula was put in a glass tube of 2.5 cm diameter and 6 cm length. One end was tightly sealed with the dialysis membrane and the other end was hanged to the shaft of the dissolution apparatus (Hanson dissolution apparatus I, USA). The tubes were put into phosphate buffer solution (100ml), which was constantly stirred at 100rpm and about 37°C. Two-milliliter aliquots of release medium were taken at different intervals (15,30,45, 60, 120 min.) and substituted by a fresh buffer solution [19]. The aliquots were analyzed for the determination of drug concentration using the HPLC method previously described at 275 nm. Analysis was run in triplicate and the mean value ±S.D was calculated.

Ex-vivo permeation and skin retention study
This test was performed for the formulas which exhibited a high percentage release compared to the market formula. This test was done as mentioned in the in vitro release test with the exception of using excised neonate rat skin instead of cellulose membrane. Aliquots of 2ml were taken from receptor partition at time intervals (0.5, 1, 2, 4, 6, 8, 10, 12, 24hr) and substituted with the same volume of fresh buffer. The withdrawn samples were ltered through 0.22µm lters and analyzed for drug concentration using the HPLC method mentioned above.
The cumulative amounts of caffeine permeated throughout the skin per unit surface area (Eq.4) were plotted against time to construct the drug permeation pro les.
At the end of the experiment (after 24hr), the skin was dehydrated with cotton and washed several times with distilled water then cut into small pieces (2×2cm 2 ). These pieces were dipped into methanol (25ml) for caffeine extraction then subjected to ultrasonication for 30 minutes at 40°C, vortexed for 3 minutes, ltered through 0.22µm, and analyzed for drug concentration using HPLC [9].

Permeation data analysis
The cumulative amount of caffeine permeated through the rat skin (Q, μg/cm 2 ) was plotted as a function of time (t, hr) for each formula. The permeation rate at the steady-state ( ux) (Jss, μg/cm 2 /h) and time were calculated from the slope and the intercept of the straight line obtained by plotting the cumulative amount permeated per unit area of skin versus time at steady-state condition, respectively. Permeability coe cient (Kp) was calculated by dividing the ux by initial drug concentration (Co) in the donor portion of the cell following the equation below [9]: Kp = J ss /Co Enhancement ration (Er) was calculated by dividing the Jss of the respective formula by the J ss of the market formula as given below: Er = J ss of formulation/Jss of control

Statistical analysis
The in-vitro release, ex-vivo permeation and skin retention data expressed as mean ± standard deviation were statistically analyzed by utilizing the principles of one way ANOVA test "SPSS version 17.0 for Windows" (SPSS Inc., USA) followed by post HOC Duncan alpha (0.05)(95% con dence interval).
Transmission electron microscopy (TEM) (1:10 %w/v) with double distilled water. One drop of the diluted sample was put onto a copper grid (200-mesh), negatively stained with 1% phosphotungstic acid, and then dried on the carbon-coated grid. After four minutes, a lter paper was used to eliminate extra liquid droplets at the copper grid edge. The sample was dried at 25°C and then observed by TEM [20].
Stability studies for the optimized formula as per ICH guidelines Stability studies of the optimized caffeine-loaded NLC formula were accomplished according to ICH Q1A (R2) guidelines. The optimized formula was kept at temperature; 25±2 °C/ relative humidity; 65% ± 5% for 12 months. The stored formula was periodically analyzed for pH, particle size, PDI, entrapment e ciency, zeta potential, phase separation, color, and clog formation [21].

In-vivo studies
The animal protocol was carried out according to the experimental protocol was approved by the Animal Care Ethics Committee, Faculty of Pharmacy, Cairo University No (88) for 31/12/2018.

Confocal laser scanning microscopy (CLSM)
Rhodamine B (RB); a commonly uorescent dye at a concentration (5µg/ml), was loaded with caffeine NLC optimum formula. Experimental rats were split into two groups (each containing 3 rats at least). Group 1 was treated with caffeine NLC optimum formula, group 2 was treated with caffeine aqueous solution loaded with RB (control). Animals in different groups were provided the amount of drug equivalent to 15 mg of caffeine. The drug application was occurred by rubbing on the dorsal side of rat skin at the square area (4×3 cm 2 ) twice per day for 48h. The rats were anesthetized and sacri ced after drug administration for 48h. The rat skin was cleaned with a sponge and immersed in 1% paraformaldehyde solution for 30min. Three perpendicular cuts (10 µm) of each sample were gotten via slidge microtome, cleared in xylene, and embedded in para n at 56 degrees in a hot air oven for 24h. The obtained tissue sections were collected on glass slides, depara nized and the spreading of the dye was noticed using a confocal laser microscope. Skin pieces were inspected using normal and UV light (excitation at 405 nm, 458nm, 488 nm, 514nm, 543nm), and emission spectra were recorded at 633 nm. The CLSM photographs in bright-eld were overlaid to observe RB distribution [22].

Hair growth evaluation
Twelve to eighteen months-old female rats with an approximate weight of (200-250g), got from the animal house of The National Organization for Drug Control and Research, were used for in-vivo assessment of caffeine-loaded NLC as an anti-hair loss agent. The rats were subjected to nutritive adaptation, under well-ordered environmental conditions; temperature, 25±1 °C; relative humidity, 45±5 %; and 12-h light/ dark cycle) for 1 week prior to the experiment. Alopecia induction of rats was convinced by intraperitoneal (i.p) injection of a single dose of a chemotherapeutic agent of 1.5 mg/kg etoposide once daily for 3 consecutive days [23]. The rats were divided into four groups, each contains 3 animals. A positive control (Group 1) was used, whereas animals were inducted with a chemotherapeutic agent and not received any treatment. Group two were treated with the market formula, group three were treated with optimized caffeine-loaded NLC formula and group four were treated with the blank formula of the optimized NLC. The animals in each group except group four were delivered topically an amount of the drug equivalent to 3mg twice/day for 30 days at a square area of 10×5cm of the dorsal part and rubbed 3-4 times. This process was repeated every morning and evening for 4 weeks. Rat's skin was observed for hair growth by taking photographs by using a D70 model camera (Nikon, Tokyo, Japan) after 1, 7, 14, 21, and 30 days of treatment [24].

Histopathological analysis
During the period of treatment, skin biopsies (pieces) were taken for histological examination on the 10 th and 20 th day of treatment after temporary anesthesia of the rats. On the 30 th day of treatment, the rats were sacri ced and skin specimens were also taken. These skin pieces were settled in 10% formalin solution for 24hr, rinsed with water, and diluted several times with absolute, methyl, and ethyl alcohols for dehydration. They were puri ed in xylene and xed in para n at 56 degrees in a hot air oven for 24hr. Tissue para n blocks of 4 µm thickness were obtained using slidge microtome. The tissue blocks were placed on glass slides, depara nized, stained by hematoxylin and eosin stains, and scanned by the light electric microscope for histopathological inspection [25].
Hence, the extent of drug solubility in lipid excipients greatly affects the drug loading; it was expected to choose oleic acid for NLCs preparation. In this experimental part, the authors not only focused on the drug solubility, but also on how to maximize the therapeutic e ciency of the applied dosage form. This was accomplished through the drug used and the synergistic effect that could be achieved from vehicles having bene ts in hair follicle targeting and hair growth-promoting for alopecia treatment [27]. Accordingly, oleic acid: garlic oil (O: G) mixture at ratio 1:1 v/v was picked as the liquid oil phase for NLCs preparation. Oleic acid acts as a skin permeation enhancer through interaction and modi cation of SC lipids, thus, forms a new type of lipid domain that can decrease the capacity of skin barrier function. Consequently, it will disorganize and increase the lipid packing uidity, decrease the diffusional resistance, and facilitates permeation [28]. Garlic oil has different health-promoting bene ts, especially its effect as antifungal, antimicrobial, and hair growth promoting properties [26].
The solubility of caffeine in Tween 80 and Cremophor EL was 8.87 ± 0.49 and 8.66 ± 0.756 mg/g, respectively ( Table 2). Both surfactants exhibited similar HLB values which may explain the slight difference in their ability to solubilize the drug. Both are nonionic surfactants that are less toxic and used with low concentrations for effective reduction in interfacial tension. The solubility of caffeine in co-surfactant, Transcutol HP, Span 20, and Span 80, was 25.82 ± 0.0832, 14.2 ± 0.448, and 6.25 ± 0.68 mg/g, respectively ( Table 2). The results revealed that Transcutol HP showed the highest solubilization capacity for caffeine and that was attributed to the more hydrophilic properties of Transcutol HP relative to Span 20 or Span 80. Transcutol HP was supposed to raise the interfacial uidity of a surfactant's external surface in the micelles thus enhancing the solubilization and emulsi cation [29].
Upon comparing the solubility level of the drug in various estimated phases, co-surfactants, surfactant, solid and liquid lipid, it was found that the highest solubility of the drug was in the solid lipid (about twice or more drug solubility in liquid lipid or surfactant or cosurfactant). This result was the same informed by Amanda, et al., [30] who found that solubility of uconazole (hydrophilic in nature) was the highest solubility in solid lipid reaching 250 mg/g stearic acid and reaches 45mg/g castor oil.
Miscibility of liquid and solid lipid, surfactant and co-surfactant, and drug solubility in mixtures The NLCs are modulated by the addition of liquid lipids to the solid lipids which make their difference from SLNs. The solid/liquid lipids mixture imparts imperfection or amorphous structures to NLCs matrices. NLCs lipid molecules generate large distances among fatty acid chains, more drug accommodation occurred, and thus increase drug loading during storage [31]. Accordingly, the miscibility test was carried out for determining the ratio paper that ensure immiscibility of the two phases. These results previously reported that a liquid lipid content of 30% w/w or less was optimal, as higher oil concentrations resulted in the generation of immiscible mixtures and the possibility of nanostructuring of NLC decreases [32].
The solubility of caffeine in selected lipid blends containing stearic acid/O: G at ratio 1:1 at ratios 90:10, 80: 20, and 70: 30 was 25, 30, and 35 mg/g, respectively, and was 25, 30, and 40 mg/g, respectively, for lipid blends containing GMS as solid lipid. It was clear that drug solubility in the lipid blends increases with increasing liquid lipid ratio. The increase in the volume of liquid lipid decreases the melting point with keeping the solid-state of the solid lipid, decreases the viscosity of the mixture, and increases the dynamic equilibrium for drug molecules that in turn increases drug solubility and minimizes the drug escaping as well as gives better stability to the blend [33].

Solubility study of the drug in different aqueous media
This experimental part has potential importance for the preparation of water-soluble drugs as solid lipid nanocarriers, which was accepted to increase the percentage of drug-loaded into the NLCs and reduce the amount of drug escape to the aqueous medium. Solubility of caffeine in different buffer systems, pHs, and ionic strengths are presented in table 3. The results showed that caffeine exhibited the lowest solubility in carbonate buffer at higher pH 10.8 and ionic strength 200 mM that reached 8.01%.
Using phosphate buffer systems at pH 7.4 and 7.8 showed that increasing the pH had an obvious effect on decreasing the drug solubility, especially with increasing the ionic strength of the medium (Table 3). At ionic strength 50mM, the drug solubility was nearly the same at both pHs The results revealed that increasing the buffer ionic strength statistically decreased the drug solubility, especially if it is accompanied by an increase in the pH of the buffer system. The lower solubility in buffers with high ionic strength was related to increasing the concentrations of dissolved ionic salts that in turn could decrease the solubility of the weak electrolytes. Moreover, these results could be due to weak basic properties of the drug that showed low solubility at high pH and high solubility at lower pH. At acidic pH, the drug has high solubility due to its ionization, whereas, at basic pH (carbonate buffer at pH 10.8) the drug is in the un-ionized form, consequently the solubility decreased leading to precipitation in a crystalline or amorphous form [34]. These ndings were similar to Mayer, et al. [35] who informed that altering the pH of the buffer system was the best choice for encapsulating doxorubicin (water-soluble drug) inside the liposomes. Accordingly, carbonate buffer at pH 10.8 with ionic strength 200 mM was selected as the optimum aqueous medium for encapsulating caffeine into NLCs.
Optimum excipients for caffeine-NLCs preparation Table 4; illustrates the composition of caffeine-NLCs formulas. The S/CoS mixture at ratio 3:1 was used at a xed concentration of 3% w/v. The lipid blends were used at a xed concentration of 2% w/v whereas, the ratio of solid lipid: liquid lipid was 70:30 and drug concentration was 1.5 and 3% w/v. Carbonate buffer pH 10.8 at ionic strength 200 mM was used as an aqueous phase for the emulsi cation process due to low drug solubility that is expected to increase the %DL and %EE. The formulas were characterized for DSC, FTIR, %Yield, %DL, %EE, particle size, zeta-potential, PDI, pH, electroconductivity, in-vitro release, and ex-vivo permeation.
Characterization of caffeine loaded NLCs DSC Thermal analysis by DSC was used for detecting drug-excipient incompatibilities, melting point, and recrystallization performance of bulk lipids used in the formulation of caffeine loaded-NLCs [36]. Fig. 1 showed DSC thermograms of the components of caffeine-loaded NLC; (a) caffeine, (b) GMS, (c) stearic acid, (d) physical mix at ratio (1:1), (e) blank NLC, and (f) optimum NLC formula C3. Caffeine showed a sharp endothermic peak with a melting point of 235ºC which was similar to that reported by Carmelo et al., [37], Fig. 1(a). DSC thermogram of GMS revealed a sharp endothermic peak at 56.59ºC revealing its melting point ( Fig.1(b)).
These ndings agreed with that of Freitas and Müller, [38] who studied the DSC of GMS that exhibited an endotherm melting point around 60ºC. DSC pro le of stearic acid showed an endothermic peak at 57.93°C, Fig. 1(c) [39]. DSC thermogram of the physical mix, blank NLC, and optimum NLC formula for caffeine, (Fig. 1(d), 1(e), and 1(f) presented a broad endothermic peak at 300ºC, which revealed the absence of caffeine peak. It is known that when the drug endothermic peak does not appear in thermogram of nanopreparations or in the physical mixture, drug converted from crystalline to amorphous phase that was con rmed by its complete solubilization into NLC [40].

FTIR Studies
FTIR spectrum of the drug identi es its nature through clearing the functional groups that are related to the drug structure. Fig. 2(a-F) showed the IR spectra of pure caffeine, GMS, stearic acid, physical mix (1:1), blank NLC, and caffeine-loaded NLC of the optimum formula (C3).  [42]. FTIR spectrum of stearic acid showed the peaks at 2848.86 cm −1 and 2916.37 cm −1 which was due to the stretching vibrations of -CH2 groups and the peak at 1734.8 cm −1 related to a carbonyl group of stearic acid [43], Fig. 2(c).  Fig. 2(f). This result con rms satisfactory solubilization and complete entrapment of drug into lipid phase that evidenced by the low intensity, slight shift, broadening, and disappearance of the drug peaks in the formula [44].

%Yield, %EE, and %DL
The % yield of caffeine-loaded NLCs (C1-C8) was in the range of 90-98%, these high yield values revealed an optimum method for NLCs preparation. The %DL was in the range of 5.99-22.36, and the %EE was in the range of 38.91-72.55 (Table 4). Further investigation for the effects of several factors on the %EE of caffeine-loaded NLCs was carried out. The data was analyzed using 2 3 full factorial designs at two levels and three variables strategies.
Data analysis according to the factorial design program was presented in Fig. 3 (a, b, and c). It was found that the lipid type, S/CoS type, and drug concentration had a signi cant effect on the %EE of caffeine (P<0.005).
With respect to lipid type, the results revealed that the NLCs formulas containing stearic acid showed a high entrapment e ciency (Fig. 3a). The entrapment e ciency of NLCs was improved upon increasing the fatty acid carbon chain length which led to higher hydrophobicity and thus increasing drugs residence within its core matrix [45]. encapsulate the hydrophilic drugs at low drug concentration (1.5%). The %EE of formula C3 prepared with cremophore EL as a surfactant (72.55%± 0.12) was more than %EE of formula C1 prepared with Tween 80 (60.13%± 1.34). The same results were observed with C5 and C7 (Table 4).
Cremophore EL is identi ed as oxylated triglycerides of ricinoleic acid, this structure decreases the interfacial tension between the oil and aqueous phases during the emulsi cation, consequently, improving drug entrapment [46].
Referring to drug concentration (Fig. 3(c)), the results proved that increasing the drug concentration from 1.5 to 3% decreasing %EE, (P<0.005), especially upon using cremophor EL/Transcutol HP as S/CoS type. The formula C3 has 22.36 and 72.55 %DL and %EE, respectively, that was decreased to 7.58 and 48.96, respectively for the formula (C4) when drug concentration increased from 1.5% to 3% w/v. The same results were also observed upon comparing the results of C7 to C8 (Table 4). These results were contributed to the fact that increasing the amount of the drug resulted in an extra porous structure in the nanostructured lipid particles. The drug readily leaked out to the external medium through the cavities and channels lled with the drug. Moreover, upon raising drug concentration in the nanoformulations, the osmotic pressure difference between the inner and outer phases was altered, as a result of some injuries to the formed quasi-emulsion droplets, which in turn caused the drug to leak from the inner phase [47].
Particle size analysis Table 4 presented the results of particle size and PDI of caffeine-loaded NLCs formulas (C1-C8). The particle size and PDI ranges were 240-772 nm and 0.541-0.958, respectively, revealing the existence of the drug in the nanosize range (<1000nm). Figure 4 presents the data analysis according to the factorial design software. It was found that the lipid type, S/CoS type, and drug concentration had a signi cant effect on particle size (P<0.005).
Concerning the lipid type, NLCs formulas prepared using GMS possessed the smaller particle size and PDI compared to those prepared using stearic acid ( Fig. 4 (a) and Table 4). The addition of GMS could decrease the particle size as it has self-emulsifying property due to its structure which has saturated carbon chain length. This structure enables it to capture more drugs and have a uniform particle size in the nanosize range [48]. These ndings were agreed with Gardouh, S. G. et al., [49] who found that lipid hydrophilicity and self-emulsifying properties affected by the lipid crystal's appearance (and hence the surface area) had an indirect effect on the nal size of NLC dispersions.
The results presented in (Fig. 4(b) and Table 4) showed the important effect of surfactants on the NLCs particle size. The formulas, which contain Cremophore EL as surfactant exhibited particle size smaller than those prepared with Tween 80 at the two levels of drug concentrations 1.5% and 3%w/v. Cremophore EL has more emulsi cation e cacy than Tween 80 that ensures particles uniformity. Comparing C1 to C5, the particle size was 585 and 400um, respectively, C2 and C6 was 772 and 696 nm, respectively), C3 and C7 was 358 and 240nm, respectively, and (C4 and C6 was 624 and 498nm, respectively. This result was attributed to molecular structure differences between both surfactants whereas; Cremophore EL had a slightly lower degree of ethoxylation and unsaturation. At the level of pharmaceutical industries, they have been dedicated much attention for the application of Cremophor EL is more than Tween 80, due to its higher emulsifying capacity and its capability to encapsulate, solubilize, and protect both lipophilic and hydrophilic compounds [50]. Concerning the drug concentration, the results showed that increasing drug concentration from 1.5% to 3% w/v had an obvious effect on increasing NLCs particle size (Fig. 4(c), regardless of the lipid or S/CoS types (see C1, C3, C5, and C7 in comparison to C2, C4, C6, and C8 in Table 4). These results may be due to the concept of crystallization theory which is in uenced by increasing the drug concentration [51]. Higher drug concentration, higher super-saturation rate leading to higher diffusion-controlled growth and faster crystal growth as a result the agglomeration process was higher, resulting in larger particle size. These ndings agreed with Dingler, A. et al., [52] who reported that the higher NLCs particle size was attributed to higher drug concentration.
From the analysis of the factorial design, it was found that only the lipid type had a signi cant effect on PDI (P<0.005), Fig. 5 (a). The formulas C1, C2, C3, and C4 prepared with stearic acid exhibited PDI of 0.839, 0.829, 0.912, and 0.958, respectively, which were higher than that exhibited by the formula prepared with GMS C5, C6, C7, and C8; 0.61, 0.541, 0.556, and 0.665, respectively. These results were attributed to the higher particle size of the formulas prepared with stearic acid more than those prepared with GMS as solid lipid. Large particles form clusters or aggregates which in turn lead to higher PDI and broad size distribution [53]. In addition, the e cacy of GMS to act as an emulsi er, provide stability to NLCs dispersions, and ensure nanoparticles homogeneity.

Zeta potential measurements
All NLCs formulas showed higher negative zeta potential (ZP) values > -19 mv (Table 4), which revealed better stability of the colloidal dispersed nanoparticles. This stability was related to electrostatic repulsion that prevents the nanoparticles from aggregation. The higher values of ZP (±30 or more) indicate higher nanoparticle stability [54]. Table 4 and Fig. 6, formula C3 showed a slightly lower ZP of -24.8± 0.70, than formula C4 -29.1± 1.78. This may be attributed to increasing the dug concentration, decreasing the encapsulation e ciency and drug loading and at lower loading, there is a low a nity between the drug and lipid matrix, consequently, more negative charged particles (ionized drug) are released in the external medium and the drug can be retained by the electrostatic interactions [55].

pH & Electroconductivity
Since NLCs formulas were developed for topical application, the measurement of pH is an essential process to avoid skin damage or severe irritation.
The pH of the prepared caffeine NLCs formulas was measured and was found to be in the range of (5-5.97) that mimic the pH of the cuticle as shown in Table 4. Electrical conductivity is utilized for the evaluation of the electro-conductive behavior of the compounds. The conductivity values of the NLCs formulas were high and ranged from 105.36 to 197.33 µS/cm, Table 4. These elevated values are due to high water content which enables ions to move more freely. It was found that conductivity raised with increasing the drug concentration and diminished with increase % encapsulation e ciency. The formula C2 (197.33 ± 9.29 µS/cm) increased by 0.8 fold than the formula C1 (158.13±7.88 µS/cm). These ndings were attributed to the dependence of conductivity on the concentration, mobility, and valence states of the ionized species in a solution. Increasing the amount of the drug led to a higher concentration of free and unbounded drug ions in water, which nally resulted in higher conductivity [56].

In-vitro drug release studies
The release pattern of caffeine-loaded NLCs formulas compared to the market formula was investigated over 2hr and presented in Fig. 7.
The results revealed that all tested formulas allowed for high % drug release during the rst 0.25min. The market formula released about 47.8%, whereas, NLCs exhibited % drug release in the range of 35.09-55.88. At that time, the drug release was expressed as rapid release in which the drug was released from the exterior shell of NLCs, then release occurred from the inner core. This fast release could be due to the increase in the chemical potential gradient, which resulted in escaping of the encapsulated active ingredients gradually from NLCs into the external phase. At this stage, the release rate would be much higher as less energy was required to overcome the phase barrier for it to be released to the continuous phase [57]. From the analysis of the factorial design (Fig. 8 a, b, and c)  Regarding the drug concentration, upon increasing the drug concentration from 1.5 to 3%, the % drug release slightly increased. After 1h, the % of drug release from formula C2 was greater than C1(85.2±1.72>67.4±2.20) and C4 than C3 (89.24±3.04>72.8±3.18), respectively (P***<0.001). The faster release is achieved by higher drug concentration when the adsorption of the drug particles which are weakly bound to the nanocarriers rather than to the drug incorporated in polymer nanoparticles [59].

Ex-vivo skin permeation studies
Permeation studies were done on the optimized formula, which had the smallest particle size and reasonable % EE. The formulas C3 and C7 showed particle size 358.6 ±1.45, 240.4± 22.87 nm, respectively, and % EE 72.55%±0.12, 67.20%±5.98. The % of drug release for C3 and C7 was 84.28±3.18 and 100.75±0.77 after 2hr, respectively. Consequently, a permeation study was performed on these optimized formulas in comparison to their market formulas. Permeation pro les present the cumulative amounts of caffeine permeated from NLCs formulas and market per unit time for 24hr are presented in Fig. 9, Table 5.
The formula C3 showed the highest amount permeated drug through 24hr. The tested formula could be arranged discerningly as follows: C3 > C7 > market; the cumulative amount permeated (µg/cm 2 ) was as follows: 150.832> 87.2804>55.514, respectively, Table 5. The amount of drug permeated from formulas C3 and C7 present 2.7 and 1.6 fold of market, respectively. These results were due to the mechanism of oleic acid as a chemical permeation enhancer in the NLCs formulas, Its absorption by stratum corneum reduces lipid temperature changes and so, improves the permeation of the drug. Additionally, it decreases the viscosity of lipids of the super cial layer [60]. Cremophor EL and Transcutol HP as S/CoS mixture in NLCs formulation had an e cient effect on drug permeation, which in turn allow better permeation pro le [9]. Cremophor EL increases the epithelial permeation through disturbing connections in the epithelial wall, enabling drug transportation via the membrane. So it e caciously improved topical permeation of caffeine compared to market missing Cremophor EL [61].
At the end of the permeation experiment, the e cacy of caffeine NLCs formulas to be accumulated in skin layers was estimated. The amount of accumulated caffeine in rat skin after subjecting to NLCs formulas (C3 and C7) and the market formula for 24hr was determined, Table 5. The amount of caffeine accumulated in the skin following topical delivery of NLCs formulas was clearly high compared to the market formula. The amounts of accumulated caffeine from C3 and C7 were; 4830.4716±41.461 and 4398.359±298.8837 µg/cm 2 after 24hr, respectively, These amount presents about 3.1 and 2.8 times enhancement over the market formula (1547.670±125.3430 µg/cm 2 ). According to Fei,H. et al. [62], NLCs enhance the occlusive effect, lm formation, and close contact with SC of skin and thus increase the amount of entrapped drug into the skin, improving its penetration.

Transmission electron microscopy studies
Morphology of caffeine NLCs optimized formula (C3) scanned by TEM revealed that it was successfully formulated with a smooth surface, distinct integrity, and spherical shape particles, (Fig. 10). It also showed the sizes ranges (300-500 nm), without obvious aggregation [63].
Stability studies of the optimized formula (C3) as per ICH guidelines Table 6, presents the results of stability studies of the NLCs optimized formula (C3). There was no remarkable alteration in color, pH, particle size, PDI, zeta potential, and % EE. Additionally, it kept its homogeneity without the appearance of any phase separation or clog formation, which proved the good physical stability of the NLCs optimized formula.

Confocal Laser Scanning Microscopy (CLSM)
To evaluate the drug distribution performance when loaded on NLCs within the skin layers (epidermis, dermis, and hypodermis), CLSM was employed.
The CLSM images of perpendicular slices of rat skin gotten after 48hr application of caffeine RB-NLCs (C3) are shown in Fig. 11(a). The NLC formula (C3) loaded with RB showed higher uorescence intensity of RB in the hypodermal layer, whereas the drug was distributed through all layers of the skin, indicating high skin targeting e ciency and highly permeation. These results are consistent with previously reported ndings of Pople and Singh [64]. Control formula loaded with RB in which drug was applied as a suspension in aqueous media showed drug distribution in the epidermal layer only with low permeation e ciency, Fig. 11(b). Accordingly, this experimental part clearly proved the potential e cacy of NLCs nanoformulation as a skin targeting drug delivery system with a small mean particle size of 419 nm. Conjugation of drug loaded-NLCs with RB has no effect on the particle size of the NLCs.
In-vivo study

Morphological examination of rat's dorsal skin in different groups
This experimental part has crucial importance for evaluating the e cacy of different tested formulas on rat hair growth. The effect of caffeine NLCs optimum formula (C3), blank, and market formulas on the hair growth were compared to control (un-treated rats) Fig. 12 (a-d). The photographs (a-d) showed the thick hair of dorsal rats' skin at zero-day before chemotherapeutic induction in different rat groups treated with; (a) the optimum formula C3, (b) blank formula, (c) market formula, and (d) control. Fig.12(e-h) the rat's bald skin without hairs after chemotherapeutic induction with etoposide (1.5 mg/kg), once daily for three consecutive days. Fig. 12(i-l), presents hair growth for rats after 30 days of treatment. Control group (Fig. 12(i) presents minimum hair growth. The rat group treated with the formula C3 exhibited complete and heavy hair growth ( Fig. 12(j)) in comparison to the market formula ( Fig. 12(k) and the blank formula ( Fig. 12(l) that showed incomplete hair growth. Hair regrowth was noticed during the whole treatment period. The effective NLCs permeation and fast drug release con rm well therapeutic e ciency. The improvement of drug activity of NLCs formula was attributed to the presence of oleic acid, that acts as a permeation enhancer due to its good wetting and moisturizing properties [65]. Oleic acid permits drug permeation through skin layers and in addition, it moistens the dry hair. The lower e cacy of the market formula was related to solvent evaporation that not only decreases drug permeation but also yields dry skin. The blank formula that was presented in aqueous media lacks to permeation enhancer.

Histopathological evaluation of the skin specimens
The effect of topical application of caffeine-loaded NLCs formulas on hair growth was studied through the histopathological examination of the skin layers, Fig. 13 (a-d). Fig. 13 (a, b) showed the histopathological examination of the normal skin specimens with a normal histopathological structure of the epidermal layer, underlying dermis, sebaceous gland, and subcutaneous tissue. Also, it showed multiple mature hair follicles with hair shaft and bulb that are characterized by spindle-like dermal papilla (DP) and Y-shaped melanin granules around the DP without ectopic melanin in the bulb.
While Fig. 13 (c, d) showed the histological examination of skin specimens after 3 days of chemotherapeutic induction, a dystrophic change of hair follicles was observed which is characterized by the presence of ectopic melanin granules and swollen DP. These melanin granules are smaller than keratinocyte nuclei and present as clumps in the bulb and hair shaft. The hair follicles showed a catagen-like structure with a ball-like small DP [66] Fig. 14 (a,d,g, and i) presents biopsy specimens of rat skin after 10 days of treatment; (a) the optimum formula C3, (d) blank formula, (g) market formula, and (i) control. The skin of rats in the control group showed reduction and atrophy in the hair follicles (decrease in size and number). The animal group treated with optimum NLCs formula (C3) showed multiple hair follicles of the dermis. The group treated with blank formula showed acanthosis in the epidermis with the appearance of mature hair follicles. The animals treated with market caffeine showed atrophy in most hair follicles. Fig. 14 (b,e,h, and k) presents biopsy specimens of rats' skin after 20 days of topical treatment. The control group (Fig. 14 (k) showed few immature hair follicles of the dermis, and the group treated with optimum NLCs formula (C3) showed multiple numbers of mature and immature hair follicles of the dermis (Fig.14 (b). Skin rats of the group treated with blank formula showed few appearances of immature hair follicles and less mature ones (Fig. 14 (e). The rat group treated with market caffeine showed the appearance of multiple mature hair follicles but less than the group treated with the NLCs optimum formula C3 (Fig. 14 (h).
After 30 days of topical treatment, the animals were sacri ced and the histopathological examination of rats' skin specimens was presented in Fig. 14 (c,f,i, and l). The skin of rats in the control group (Fig. 14 (l) showed focal necrosis and ulceration in the epidermis, acanthosis, and invagination of the epidermis and dermis. The subcutaneous tissue showed the excessive formation of tissue granulation with in ammatory cells in ltration. Fig. 14 (c) is for skin rats of the group treated with NLC formula C3. It showed multiple numbers of mature and immature hair follicles of the dermis. Skin rats of the group treated with the blank of formula C3 showed less appearance of mature hair follicles and less immature ones (Fig. 14 (f). Skin rats of the group treated with market caffeine showed the appearance of multiple mature hair follicles but less than that recognized in the group treated with optimum NLCs formula of caffeine (C3), Fig. 14 (i). The appearance of mature and immature hair follicles for skin rats in different groups can be arranged descendingly as follows; group treated with NLCs optimum formula (C3) > group treated with market formula > group treated with blank formula> group treated with control. The appearance of acanthosis, necrosis and granulation tissue formation with in ammatory cells in ltration for skin rats in different groups can be arranged descendingly as follows; group treated with control > treated with blank formula> treated with market formula > treated with optimum formula (C3).

Conclusion
In the present study, caffeine-loaded NLCs were formulated using a combination of stearic acid or GMS and oleic acid: garlic oil (1:1) as lipid phase.
The pre-emulsion ultrasonication technique was employed. Tween 80 and cremophore EL and Transcutol HP were used as surfactants and cosurfactant, respectively. The factors affecting the NLCs formulation were optimized using a factorial design program. The FTIR study of pure caffeine and excipients con rmed that there was no remarkable chemical interaction between them. DSC studies con rm the transformation of the crystalline nature of the drug into amorphous nature. The TEM analysis showed the NLCs were formulated and having a spherical shape. The results of zeta potential predicted good particle stability because the repulsive forces prevent aggregation over time. The results of the present study suggest caffeine NLCs as promising carriers for topical delivery of drugs applied for alopecia treatment. The NLCs carriers promote a higher growth rate of hair over the currently marketed formula of caffeine.

Declarations
Ethics approval and consent to participate The animal protocol was carried out according to the experimental protocol was approved by the Animal Care Ethics Committee, Faculty of Pharmacy, Cairo University No (88) for 31/12/2018.

Consent for publication
All authors consent for publishing this work.

Availability of data and material
The data sets generated during this work can be available upon request.
Competing interests: The authors declare no competing interests.       Main factor plot showing the effect of (a) lipid type, (b) surfactant type, (c) drug concentration on the particle size of caffeine loaded NLCs In-vitro release pro le of drug loaded NLCs formulas (C1-C8) and their (market formula) in phosphate buffer at pH 7.4 Ex-vivo permeation studies of caffeine NLCs formulas C3 and C7 compared to market formula   (a-d) showed the photographs of dorsal rats skin with thick hair taken at zero days before chemotherapeutic induction of rats in different groups in treated with; (a) the optimum formula C3, (b) blank formula, (c) market formula and (d) control. After chemotherapeutic induction with etoposide (1.5 mg/kg), once daily for three consecutive days, the rats showed bald skin without hairs, (e-h). After 30 days of treatment, (i-l), control group presented minimum hair growth while rat group treated with the formula C3 exhibited completely hair growth more than the market formula and the blank formula Figure 14 (a) Biopsy of skin rats treated with optimum formula (C3) after 10 days of treatment showed multiple follicles in dermis, (b) after 20 days of treatment, it showed mature and immature hair follicles (arrow), (c) after 30 days, it showed increased in the number of mature and immature hair follicles (arrow), (d) blank aminexil showed showed mild acanthosis in the epidermis after 10 days, (e) after 20 days, it showed multiple immature hair follicles and less mature ones (arrow), (f) after 30 days, mild epidermal acanthosis appeared (arrow), (g) market formula (A7) showed atrophy in most hair follicles (arrows) after 10 days of treatment, (h) after 20 days of treatment, it showed multiple hair follicles, (i) multiple mature hair follicles appeared after 30 days of treatment, (j) after 10 days, control group showed reduction and atrophy in the hair follicles (arrow), (k) after 20 days, it showed few immature hair follicles of the dermis, (l) after 30 days, control group showed showed focal necrosis, ulceration in the epidermis, and acanthosis

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.