Nanostructured lipid carriers for percutaneous administration of alkaloids isolated from Aconitum sinomontanum

Preparation of AAS-NLCs

The stirring rate (rpm), stirring time (min), homogenization pressure (bar), homogenization cycle, and cooling temperature were evaluated to optimize the preparation of the AAS-NLCs. As shown in Figure 1, the particle size and polydispersity index (PDI) were reduced as the stirring rate was slowed or the stirring time was lengthened. As the homogenization pressure increased, the particle size and PDI were reduced but subsequently increased if the pressure continued to increase. Moreover, particle size decreased as the number of homogenization cycles decreased. In addition, a small particle size and low PDI were produced by cooling the formulation to room temperature in an ice bath. It was previously reported that the instability of the preparation system is caused by an increase in the kinetic energy of droplets due to the increase in homogenization pressure and number of homogenization cycles [15]. Therefore, the production parameters were set as follows: stirring rate, 5,000 rpm; stirring time, 10 min; homogenization pressure, 800 bar; cooling temperature, 0°C; and five homogenization cycles were performed.

Characterizations of AAS-NLCs

The particle size and PDI are important characteristics of NLCs that influence the distribution of nanoparticles [16]. AAS-NLCs and AAS-SLNs had PDI values of 0.082 ± 0.009 and 0.073 ± 0.006, respectively and narrow size distributions (160.5 ± 7.9 and 176.6 ± 19.6 nm for AAS-NLCs and AAS-SLNs, respectively). The ZP values of the AAS-NLCs and AAS-SLNs were −24.8 ± 1.6 and −20.3 ± 2.1 mV, respectively, indicating good stability in these systems. In addition, the EE and DL of the NLC-associated LA were 82.18 ± 1.93 and 4.55 ± 0.52%, respectively while the EE and DL of NLC-associated RAN were 81.87 ± 1.52 and 0.36 ± 0.01%, respectively. The EE and DL of the SLN-associated LA were 77.48 ± 2.56 and 3.61 ± 0.75%, respectively, and the EE and DL of SLN-associated RAN were 74.81 ± 1.97 and 0.21 ± 0.01%, respectively.

The morphologies of the AAS-NLCs and AAS-SLNs were observed using transmission electron microscopy (TEM). As shown in Figure 2, most of the particles were spheroidal and uniform in size.

Differential scanning calorimetry (DSC) was performed to investigate the melting and crystallization behavior of the lipid molecules and drugs in the nanoparticle preparations [17]. As shown in Figure 3, the bulk drug had endothermic peaks at 208.6 and 269.0°C, the Precirol ATO 5 displayed a melting point of 61.1°C, and the cryoprotectant (mannitol) exhibited an endothermic peak at 159.3°C. The physical mixture heating curves exhibited three endothermic peaks that corresponded to the melting points of the three components. These included the reduced endothermic peak of the bulk drug at 269.0°C and the moving endothermic peak of Precirol ATO 5 at 51.5°C. The results indicated that the enhanced dissolution of the partial bulk drug in the solid lipids as the temperature rose led to changes in crystalline structure. All the endothermic peaks of AAS disappeared and the peak of Precirol ATO 5 was significantly shifted to the right in the AAS-NLC and AAS-SLN heating curves, indicating an increase in lattice defects in the AAS-NLC and AAS-SLN preparations. The results suggest that the drug was homogeneously dispersed in the nanoparticles in a disordered crystalline state.

X-ray diffraction (XRD) and DSC are complementary analytical methods used in the assessment of crystallinity and polymorphism [18]. Unlike the physical mixture (Figure 4D), the characteristic peaks of AAS (Figure 4A), precirol ATO 5 (Figure 4B), and mannitol (Figure 4C) in the XRD graph of the nanoparticles preparation (Figure 4E, F) were absent or reduced, confirming the formation of disordered structures in the inner cores of the NLC and SLN particles.

In vitro release

The cumulative release percentage profiles of LA and RAN from the AAS-NLCs, AAS-SLNs, and the mixture of AAS and blank NLCs are shown in Figure 5. The aqueous solubility of LA and RAN in the medium was 849.79 ± 36.95 and 582.06 ± 16.27 µg/mL, respectively, and the sink condition was maintained during the release study. Both SLN and NLC dispersions significantly enhanced the in vitro AAS release for 72 h. The results showed the LA and RAN dispersion from SLNs was 94.6 ± 5.2 and 71.8 ± 1.9%, respectively and from the NLCs was 98.5 ± 4.1 and 79.3 ± 2.7%, respectively. In comparison, their physical mixtures showed LA and RAN from SLNs at 58.8 ± 0.7 and 39.2 ± 1.7%, respectively and from NLCs was 58.8 ± 6.9 and 40.0 ± 2.8%, respectively, under the same experimental conditions. This disparity is most likely due to the disordered crystalline state of the drug that exhibited higher solubilization and was selected for nanoparticle formation [9]. In addition, the total drug released from the NLC formulations was slightly greater than that released from the SLN formulations, which was likely caused by the imperfections in NLCs resulting from the incorporation of liquid lipids into solid lipids [19]. Furthermore, AAS-NLCs exhibited an initial fast release of 12.2 ± 0.7 and 17.2 ± 1.4% for LA and RAN, respectively for 0.5 h, which was possibly due to the portion of AAS that was localized at the surface of the nanoparticles.

Cytotoxicity studies

Most lipid nanoparticles formed from glycerides or their derivatives are safe and well tolerated by organisms [20]. The in vitro cytotoxicity of AAS, AAS-SLNs, AAS-NLCs, and the empty vehicles (without AAS, defined as blank NLCs and blank SLNs) was evaluated in HaCaT and CCC-ESF cells. As shown in Figure 6, blank NLCs and SLNs showed no serious cytotoxicity at the tested concentrations (cell survival rates were higher than 80%). The viability of HaCaT and CCC-ESF cells treated with various concentrations of the formulations containing AAS was reduced dose-dependently, and both AAS-SLN and AAS-NLC formulations were significantly (p < 0.05) less cytotoxic than the AAS aqueous suspension (1,000 µg/mL) was. These results may be due to the fact that the free drug in the AAS suspension was released and immediately permeated the cells by passive diffusion while the nanoparticles had to be internalized via endocytosis or phagocytosis and released through the drug release process. AAS loaded in the tested nanoparticles achieved a biocompatibility with skin cells that was superior to that of the AAS aqueous suspension.

Cellular uptake

As shown in Figure 7, HaCaT and CCC-ESF cells showed time-dependent uptake of C6-labeled NLCs and SLNs. Moreover, differences in the fluorescence intensity between the NLC- and SLN-treated cells at the same incubation time were not significant (p > 0.05). The similar uptake of NLCs and SLNs in HaCaT and CCC-ESF cells suggests that the solid lipids in the NLCs and SLNs had a similar effect on their biocompatibility.

Endocytosis pathway

To further characterize the endocytosis pathways through which the NLCs deliver drugs, seven endocytosis inhibitors were tested in HaCaT and CCC-ESF cells. These inhibitors included chlorpromazine (CPZ, cationic amphipathic molecule that inhibits clathrin disassembly) [21], filipin III (FLP, inhibitor of caveolae-mediated endocytosis via combination with cholesterol) [22], genistein (GNT, inhibitor of tyrosine kinases associated with caveolae-mediated endocytosis) [23], cytochalasin D (CCD, which disrupts actin filaments and inhibits actin polymerization) [24], amiloride (DMA, blocker of epithelial Na⁺ channels that prevents non-specific macropinocytosis) [25], sodium azide (NaN₃, a metabolic inhibitor that restrains cellular energy metabolism) [26], and monensin sodium (MS, a Na⁺ ionophore that has been reported to cause a marked increase in intracellular Na⁺ concentration) [27]. As shown in Figure 8, CPZ, GNT, DMA, NaN_3, and MS treatments did not significantly affect the internalization of the C6-labeled NLCs in CCC-ESF cells. In contrast, FLP and CCD significantly inhibited the uptake of NLCs in comparison with the control treatment (p < 0.05) and the inhibitory effect of ELP was stronger than that of CCD was. This suggests that the uptake of C6-labeled NLCs into CCC-ESF cells was mainly mediated by caveolae-mediated endocytosis but macropinocytosis was also involved. In contrast, of the tested endocytosis inhibitors, only FLP significantly reduced the average fluorescence intensity of the HaCaT cells (p < 0.05), showing that caveolae-mediated endocytosis was involved in the internalization of HaCaT cells. These results suggest that the internalization of the drug cargo of NLCs involved cholesterol and lipid rafts in the HaCaT and CCC-ESF cells.

Confocal imaging

The uptake of C6-labeled NLCs and SLNs by the HaCaT and CCC-ESF cells was visualized using LSCM (Figure 9). Hoechst-stained nuclei appeared blue and LysoTracker Red-labeled lysosomes appeared red. The colocalization of C6-labeled NLCs or SLNs (green) and LysoTracker (red) in the periphery of HaCaT and CCC-ESF cells produced yellow fluorescence in the cytochylema. After incubation with C6-labeled nanoparticles, cells showed strong green and weak red fluorescence in the cytosol, which was probably caused by lysosomal escape of C6 from the NLCs, and the LysoTracker lost its fluorescence under neutral conditions in the cytosol [28]. These findings demonstrate that C6-labeled nanoparticles facilitate drug outflow from lysosomes, and that the fluorochrome is primarily distributed in the cytochylema.

In vitro permeation studies

The enhanced transdermal permeation of drugs packaged in the lipid nanoparticles in comparison with other vehicles is mainly due to the increased surface area in which they interface with skin corneocytes, superior skin occlusion characteristics, and more effective hydration of the stratum corneum (SC) [29]. In our studies, the AAS-NLCs enhanced the transdermal penetration of LA and RAN (Figure 10) more effectively than the AAS-SLNs did. Initially, AAS was released rapidly from both SLNs and NLCs but showed sustained release afterward, which was consistent with the in vitro release study. Over the duration of the experiment, NLCs released cumulative amounts of LA and RAN that were greater than the amounts released by the SLNs. The enhanced drug release from the NLC formulations may be attributable to the greater drug encapsulation achieved with the NLC vehicle. In addition, labrasol, which was used in the NLCs, acts as a surfactant, which may loosen or fluidize the lipid bilayers of the SC [19].

Nanoparticle tracking analysis (NTA)

No fluorescent nanoparticles were found in the blank control sample (Figure 11a), demonstrating that flaking skin particles did not influence the experimental outcome. The C6-labeled NLCs and SLNs were each diluted 1,000,000-fold and fluorescent nanoparticles were detected (Figure 11b, c). This ensured the sensitivity of the method for small quantities of nanoparticles such as those that might be transported across the skin. As shown in Figure 11d and e, fluorescent nanoparticles were not detected in the receptor fluid after permeation by C6-labeled NLCs or SLNs. It is also believed that rigid particles above 10 nm do not penetrate intact skin [30]. This finding is consistent with reports on lipid nanoparticles, which are not considered to penetrate the SC [14].

Effect on SC structure

The surface of blank nude mouse skin (without any treatment) was intact, smooth, and showed a tightly compacted SC layer (Figure 12). However, the surface of skin exposed to the NLC or SLN formulations was jagged and the cracks in the SC were larger in size (Figure 12). The lipids and surfactants in the vehicles may erode the SC, and thereby disrupt its lipid arrangement. This disruptive effect on the SC may have been exacerbated by the occlusive effect of the nanoscale lipid particles.

In vivo skin microdialysis

In the in vitro and in vivo RR validation studies, STD1–STD6 were used as perfusates and a linear correlation was obtained between the drug concentrations in the perfusate and the dialysate (LA, [C_p–C_d] = 0.431C_p − 0.022, r² = 0.997 [in vitro]; [C_p–C_d] = 0.345C_p + 0.215, r² = 0.991 [in vivo]; RAN, [C_p–C_d] = 0.418C_p + 0.394, r² = 0.996 [in vitro]; [C_p–C_d] = 0.318C_p + 0.330, r² = 0.987 [in vivo]). This correlation implied that in vitro and in vivo RR measurements obtained via retrodialysis are independent of perfusate concentrations, validating the feasibility of the microdialysis method to detect true unbound extracellular levels of LA and RAN.

In vivo time-concentration profiles of LA and RAN after topical application of the AAS-NLCs and AAS-SLNs on the abdominal skin in rats are depicted in Figure 13, and results for some pharmacokinetic parameters are summarized in Table 3. The peak concentration (C_max) of the AAS-NLC and AAS-SLN formulations were similar at approximately 8 h after administration. The drug concentrations at the sites where NLCs or SLNs were applied first increased and then decreased, which implied similar patterns of transdermal drug absorption for both lipid nanoparticle formulations. Their occlusive effects and ability to hydrate the SC [31] enhance the transdermal drug permeation achieved by NLCs and SLNs. Drug absorption may be enhanced by the formation of a lipid membrane that fuses with the SC and allows the drug to diffuse into the skin. Moreover, drug molecules must be released from nanoparticles before they can be absorbed, which may cause drug levels in the skin to fluctuate. The higher C_max values (LA and RAN, 207.45 ± 29.89 and 46.60 ± 9.56 ng/mL, respectively) were obtained with AAS-NLCs, and the area under the concentration–time curve (AUC_0–t) of LA and RAN was 3.27- and 4.42-fold higher, respectively than that of AAS-SLNs (Table 3). Altering skin physiology and modifying NLC formulations to influence drug partitioning and solubility are potential strategies to enhance transdermal drug penetration. The addition of labrasol as a liquid lipid increased the drug-loading capacity of NLCs and fluidized the lipid bilayers of the SC, which may enhance the permeation effect compared with that of the SLNs [19].

Parameter	AAS-NLCs		AAS-SLNs
	LA	RAN	LA	RAN
Tmax (h)	7.70 ± 0.84	8.00 ± 0.50	7.70 ± 0.45	7.50 ± 0.00
Cmax (ng/mL)	207.45 ± 29.89*	46.60 ± 9.56*	65.93 ± 6.37	10.07 ± 1.93
MRT0–t (h)	6.04 ± 0.03	6.24 ± 0.07	6.11 ± 0.05	6.04 ± 0.07
AUC0–t (ng/mL × h)	1,239.07 ± 176.39*	248.44 ± 50.85*	378.72 ± 56.04	56.16 ± 8.89

Mean ± SD, n = 5, * p < 0.05 vs. AAS-SLNs.

Animals and cell lines

Male Sprague–Dawley rats and nude mice weighing 200 ± 20 and 25 ± 5 g, respectively were used in this study, which was conducted with the approval of the Animal Ethical Committee of the Shanghai University of Traditional Chinese Medicine (permit SYXK [Hu] 2009-0069). All the animals were acclimatized at a temperature of 25 ± 2°C and relative humidity of 70 ± 5% for 1 week with food and water provided ad libitum. The HaCaT cell line was purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA), and the CCC-ESF cell line was obtained from the Chinese Academy of Medical Sciences (Beijing, China).

HPLC analysis

LA and RAN were analyzed using an Agilent HPLC system (HP 1260, Agilent Technologies, Inc., Santa Clara, USA) with an octadecylsilyl column (Inspire™ C18 (Dikma Technologies, Beijing, China), 250 mm × 4.6 mm, 5 µm). The mobile phase consisted of a solution of methanol and 0.01 mol L⁻¹ monosodium phosphate aqueous solution (80:20, v/v) at a flow rate of 1.0 mL min⁻¹. The column temperature was maintained at 30°C and the detector was set at 252 nm. Quantified samples were filtered through a 0.45-µm filter membrane before automatic injection into the HPLC system.

Preparation of AAS

AAS was extracted using alcohol reflux extraction methods. Briefly, powdered AS was soaked in acidified water (pH 1)/90% ethanol (v/v) at a ratio of 1:10 (w/v), and then extracted by reflux with heating for 2 h in a water bath at 100°C. The extract was collected, and then replaced with fresh solvent. This extraction procedure was repeated twice. After filtration and alkalization (pH > 10) with ammonium hydroxide, the extracts were further treated with chloroform and concentrated by vacuum distillation. To prepare the AAS, the residue was dissolved in acetone and recrystallized with diethyl ether, and the obtained extracts were vacuum-dried (relative vacuum, −0.1 MPa; temperature: 60°C) for 12 h. The main components of AAS, which have superior pharmacological effects were LA and RAN (yield, LA 69.47 and RAN 9.16%, w/w, respectively).

Ultra-performance liquid chromatography (UPLC)-tandem mass spectrometry (MS/MS) analysis

The in vivo dialysate samples were analyzed using a Thermo Ultimate 3000 ultra-performance liquid chromatography system coupled to a Thermo TSQ Quantum triple-quadrupole tandem mass spectrometer (Thermo Finnigan, San Jose, CA, USA) equipped with an electrospray ionization source. The Xcalibur 1.2 data analysis system was used. Chromatographic separation was achieved on an octadecylsilyl column [Hypersil GOLD™ C18 (Thermo Finnigan, San Jose, CA, USA), 100 mm × 2.1 mm, 5 µm] maintained at 30°C. The mobile phase consisted of methanol:water:acetic acid (70:30:0.2, v/v/v) with a flow rate of 0.2 mL/min, and the injection volume was 5 µL. The mass spectrometer was operated in the positive mode. The spray voltage was set to 3,500 kV and the capillary temperature was set at 350°C. Nitrogen was used as the sheath (35 psi) and auxiliary gas (10 psi). The precursor-to-production transitions were monitored at m/z 585 → 356 for LA and m/z 601 → 422 for RAN using the selected reaction monitoring mode. Quantified samples were extracted with diethyl ether and dissolved in methanol before automatic injection into the UPLC-MS/MS.

Preparation of AAS-NLCs

The different AAS-NLC formulations were prepared using the hot-melt high-pressure homogenization method [32] with Precirol ATO 5, labrasol, and Cremophor RH40 as the solid lipid, liquid lipid, and surfactant, respectively. Briefly, AAS was dissolved in the solid (Precirol ATO 5) and liquid (labrasol) lipids, which were melted in a water bath at 75°C. The required amount of the aqueous phase (Cremophor RH40 in double-distilled water) was then added to the oleic phase dropwise under high-speed mixing using a disperser (Ultra Turrax T25, IKA, Staufen, Germany). The resulting pre-emulsion was then immediately passed through a high-pressure homogenizer (NS1001 L, GEA, Parma, Italy). To optimize the NLC formulations, a uniform design with a U*7(7⁴) table was used in this study that included the following factors: the amount of lipids (solid and liquid lipids, factor A), the amount of surfactant (factor B), the ratio of solid lipids to liquid lipids (factor C), and the ratio of drugs to lipids (factor D). This scheme produced seven NLC formulations (NLC1–NLC7) that were characterized with regards to particle size, EE, and drug DL. The AAS-SLN preparations used for comparison were prepared using the same method with Precirol ATO5 (9.0%, w/v), Cremophor RH40 (5.5%, w/v), and the same drug content.

Characterization of AAS-NLCs

In vitro cell uptake

In vitro transdermal studies

Rats were anesthetized, euthanized, their abdominal fur was removed with a razor, and then the skin was excised carefully and washed with normal saline. In vitro permeation experiments were conducted using a Franz diffusion cell (Fulansi Electronic Science and Trade Co, Ltd, Tianjin, China) fitted with excised rat skin [37]. Each donor compartment had a diffusion area of 2.0 cm² and each receptor compartment was filled with 12.5 mL of freshly prepared normal saline containing 20% PEG 400 (v/v) to provide sink conditions [38]. The saline in the receptor compartment was maintained at a temperature of 37 ± 0.5°C and stirred at 500 rpm using a magnetic bar. One mL of each formulation was added to the donor compartment, and it was sealed with parafilm. At predetermined time points, 1 mL of the sample was removed from the receptor compartment, and replaced with an equal volume of receptor fluid equilibrated to 37 ± 0.5°C. The permeation studies were performed in triplicate and the collected samples were analyzed using HPLC.

In vitro fluorescence detection-based NTA in receptor fluid

A Franz diffusion cell fitted with excised rat skin was prepared as described earlier. The receptor compartment was filled with purified water containing 0.02% NaN₃ and the donor compartment was filled with 1.0 mL of the C6-labeled NLC suspension, C6-labeled SLN suspension, or 1.0 mL of C6 suspended in distilled water as a blank control sample. The diffusion cell was stirred at 500 rpm while the temperature was maintained at 37 ± 0.5°C. After 12 h, samples were removed from each receptor compartment and measured using a NanoSight NS300 system (Malvern Instruments, Malvern, UK), which was equipped with a scientific sCMOS camera, a 405 nm laser with temperature control, and a 503 nm long-pass filter for fluorescent particle analysis. NanoSight NS300 NTA 3.0 software was used for the data collection and the temperature was maintained at 25°C throughout the experiment.

Effect on SC structure

A nude mouse was anesthetized and immobilized on a mat with the abdomen facing upward, and then the skin in this region was divided into three areas. A flat cylindrical plastic cover with an area of 1 cm² that served as a drug pool was glued above each skin area using cyanoacrylate adhesive. One mL of NLCs or SLNs was applied to the plastic cover, and the untreated skin area was maintained as a control region. After 8 h, the mouse was euthanized, and the abdominal skin was collected and washed with normal saline. The skin samples were fixed in glutaraldehyde for 24 h, rinsed with PBS (pH 7.4), fixed in 1% osmic acid, dehydrated with ethanol, and then dried using CO₂ critical point drying. Finally, the skin was sputtered with platinum and evaluated using a scanning electron microscope (SEM, Quanta FEG250, FEI, Hillsboro, OR, USA).

Microdialysis system

The microdialysis system used was composed of a WZ-50C6 microinfusion pump (Smiths Medical, St. Paul, MN, USA) with a 20-mL plastic syringe and a linear microdialysis probe. The regenerated cellulose (inner diameter, 200 µm; outer diameter, 280 µm; molecular weight cut-off, 13 kDa) was used to prepare Spectra/Por^® microdialysis hollow fibers (Spectrum Laboratories, Inc., Houston, TX, USA). The linear microdialysis probes were made by gluing the fibers to the quartz capillary tubing (Welch Materials, Inc., Shanghai, China) using cyanoacrylate adhesive (Mxbon^® Super Glue, Cartell Chemical Co., Ltd., Chia-Yi Hsien, Taiwan). The inlet tubes of the probes were connected to the microinjection pump using polyethylene tubing [39]. In all experiments, the length of the membrane accessible to dialysis was 20 mm, and the perfusate flow rate was 0.2 mL/h. Cannulas were used as insertion guides and vials were used to collect the dialysate samples.

Standard (STD) solutions of LA and RAN were prepared by dissolving pure LA and RAN in normal saline containing 20% PEG400 (v/v). The concentrations of the STD solutions (STD1–STD6) were 1, 5, 10, 25, 50, and 100 ng/mL, respectively.

In vivo recovery validation in vivo correction

In vivo microdialysis studies

To determine the concentration of the administered drug, a probe was inserted into the dermis of the abdominal skin of a rat anesthetized with intraperitoneal urethane aqueous solution (1.3 g/kg), and the active dialysis window was placed below the site of topical drug administration. The rat was immobilized on a mat with the abdomen facing upward and the fur in this region was manually shaven, taking care not to damage the skin. The ambient temperature was maintained at 25°C. For the in vivo microdialysis sampling following probe implantation, the connective tubing from the probe was secured to the skin with adhesive tape to fix its position. The probe was continuously perfused with 20% PEG 400 in normal saline (v/v) at a flow rate of 0.2 mL/h. The skin was allowed to equilibrate for 1 h before a blank sample was taken, and 1 mL of AAS-NLCs or AAS-SLNs was applied 1.5 h after the start of perfusion. A flat cylindrical plastic cover of about 1 cm in height and 1.5 cm in diameter, with an edge width of 2 mm was glued above the application site using cyanoacrylate adhesive, and the drug was applied to the cover the skin just above the probe. During the experiment, the application site, and the probe were kept at the same level. Dialysate samples were collected in small vials, which were replaced every 30 min. Dialysis sampling was performed for 10 h.

Statistical analysis

Results are presented as mean ± standard deviation (SD). Differences were analyzed using the Student’s t test with statistical program for the social sciences (SPSS) software version 19.0 (IBM Corp., Armonk, NY, USA). P values <0.05 were considered to be statistically significant. The uniform design with a U*7(7⁴) table was optimized using the SPSS software and VB software version 6.0 (Microsoft Corp., Redmond, WA, USA). The cutaneous pharmacokinetic parameters of LA and RAN were calculated using noncompartmental analysis with the DAS 2.0 software (Mathematical Pharmacology Professional Committee of China, Shanghai, China).