2.1.

Preparation and Characterization of Gold Nanoparticles

Ionically stabilized, polar AuNP were prepared according to the Turkevich method.³³ In short, a 70-ml solution of hydrogen tetrachloroaurate (${\mathrm{HAuCl}}_4·3{\mathrm{H}}_2\mathrm{O}$, Sigma-Aldrich Chemie GmbH, Steinheim, Germany) containing about $100\mu \mathrm{g}/\mathrm{ml}$ gold was brought to boiling under magnetic stirring and then reduced by a solution of trisodium citrate dihydrate (${\mathrm{Na}}_3{\mathrm{C}}_6{\mathrm{H}}_5{\mathrm{O}}_7·2{\mathrm{H}}_2\mathrm{O}$, Sigma-Aldrich) containing five-fold the molar concentration of the gold salt. After the gold colloid formation, its temperature was lowered to room temperature. The colloid was then transferred into a suitable glass container and stored in the refrigerator protected from light.

The optical properties of the prepared AuNP were checked using a $\mathrm{UV}/\mathrm{Vis}$ Spectrophotometer (lambda 35, Perkin Elmer, Rodgau-Jürgesheim, Germany) in the range of 400 to 800 nm. The particles were also characterized, as indicated in our previously published work,⁷ in terms of mean particle size and morphology by TEM (JEOL Model JEM 2010, JEOL GmbH, Eching, Germany) and the zeta potential based on the electrophoretic mobility (Zetasizer Nano, Malvern Instruments, Malvern, United Kingdom).

2.2.

Constructing a Skin Equivalent From Human Foreskin Fibroblasts and HaCaT Cells (Keratinocytes)

A reconstructed SE was grown from human foreskin fibroblasts and HaCaT keratinocytes according to the following protocol: Collagen gel (extracted from rat tail in 0.1% acetic acid) containing human fibroblasts was pipetted on the top of a cell culture insert [Greiner BioOne, Frickenhausen, Germany (pore size: 1.0 µm, PET membrane)] in a six-well cell culture plate. The fibroblast-containing gel was cultured for one day at 37°C and 5% ${\mathrm{CO}}_2$ under submersed conditions, applying a total of 2.5 ml Dulbecco’s Modified Eagle’s medium (Life Technologies GmbH, Darmstadt, Germany) with 10% fetal calf serum (FCS) (Life Technologies GmbH, Darmstadt, Germany) per well and insert.

The following day, a volume of 50 µL fibronectin (Life Technologies GmbH, Darmstadt, Germany) solution ($5\mu \mathrm{g}/\mathrm{ml}$) was then added and incubated again for 10 min at 37°C. HaCaT keratinocytes (CLS Cell Lines Service, Eppelheim, Germany; tested for mycoplasma) in 100 µL keratinocyte growth medium-2 (KGM-2 Kit) (Promocell, Heidelberg, Germany) with 5% FCS were subsequently seeded on top of the collagen gels. After 1 h of incubation at 37°C, a total of 2.5 ml KGM-2 medium containing FCS (5%) was added to each insert. While maintaining submersed culture conditions, medium was exchanged on day 3, 4, and 5, thereby gradually lowering the FCS concentration from 5% to 0%. At the end of the submersed culture period (day 7), inserts with the skin equivalents were transferred into six-well cell culture plates. KGM-2 medium without human epidermal growth factor and bovine pituitary extract but supplemented with calcium chloride solution (Sigma-Aldrich, Taufkirchen, Germany), 1.88 mM, was filled into each well up to the level of the insert membrane (air-lift culture). The air-lift culture was performed for up to 13 days, with medium exchange performed every second day. One skin construct was then used for histological examination to check the formation of the skin layers.

The skin construct should then be used within four to five days, during which the culture medium should be changed daily.

2.3.

Skin Penetration Study of AuNP Through Skin Equivalents

2.3.1.

Penetration of AuNP through nonfixed skin equivalents

The culture medium was carefully removed, and specimens were washed with freshly prepared phosphate buffer saline, pH 7.4, three times. Specimens on their PET inserts (polyethylene terephthalate membrane, pore size diameter: 1 µm) were then transferred into a new six-well plate, where skin penetration study was conducted. The receptor (basolateral)/surrounding compartment was filled with 800 μL KRB, touching the tissue from below (Fig. 1). AuNP, with a concentration of $90\mu \mathrm{g}/\mathrm{ml}$, was carefully applied on the top of the skin surface, in the center to avoid leakage to the deeper skin layers (DSL). The plate was then covered with parafilm and plastic cover and maintained at 37°C, 5% ${\mathrm{CO}}_2$ throughout the experiment (2 or 24 h). Following exposure, the skin surface was gently cleaned with a tissue. Collected skin was examined after longitudinal cryo-sectioning. All experiments were done in triplicates.

Fig. 1

Setup and skin model used for testing nanoparticle penetration. The skin model was developed in vitro from HaCaT cells on the top of fibroblasts in collagen. This developed into an open-edged nonsealed specimen with concave surface.

2.4.

Longitudinal Skin Cryo-Sectioning

Skin cross-sections were performed at $-26°\mathrm{C}$ using a cryo-microtome (Slee, Mainz, Germany). Following fast freezing of the skin specimen, an 8-mm punch in the center of the specimen was placed in a perpendicular direction to the cutting blade to avoid dislocation of the particles from outside into the tissue or vice versa, thus limiting sectioning artifacts.²⁸ Skin sections were placed on microscopy slides and mounted by an aqueous mounting medium (FluorSaveTM reagent, Calbiochem, San Diego) and covered with glass cover slips.

2.5.

Multiphoton Microscopy

Imaging was performed using an inverted confocal/two-photon excitation fluorescence microscope (ZEISS LSM 510 META NLO system, Carl Zeiss, Jena, Germany) equipped with a tunable pulsed NIR laser to evoke nonlinear optical excitation processes (Chameleon, Coherent, Dieburg, Germany). The objective used was a water immersion lens 63X ($\mathrm{NA}=1.2$) with a working distance of around 200 µm. A wavelength of 800 nm was used for both excitation of AuNP and scanning the skin samples using 0.485 and 0.647 mW, respectively. The energy was measured in the focal plane of the microscope after the objective using a power meter. The pulse width at the laser output was less than 140 fs, and the repetition frequency was 80 MHz (specified by Coherent). The gain settings were adjusted for each measurement individually. The optical settings, discussed in detail earlier,^26,28 allowed for a clear separation of both signals from AuNP and the tissue with no signal interference among tracks, as shown earlier.²⁸ $Z$-stacks of the skin samples were taken with steps every 1 μm. Each optical scan is composed of $512\times 512$ pixels and a size of $0.14\times 0.14{\mu \mathrm{m}}^2$. The gain settings were adjusted for each measurement individually.

2.6.

Data Analysis

The method of semiquantitation of nanoparticles in skin is based on measuring the frequency or the number of pixels due to the detected gold luminescence signal in the Stratum Corneum (SC) and DSL. A detailed description of the method of analysis for the acquired $z$-stacks has been published earlier,²⁸ and so has a detailed protocol.³⁴ Briefly, the number of pixels of AuNP signals as semiquantitative data for the distribution of AuNP in the SC and viable DSL were determined ($\sum \text{Pixel}$) for each optical layer (1-µm thickness) of a respective $z$-stack of the examined longitudinal skin sections. The overall sum from all optical sections was calculated and biased by the theoretical optical size and the pixel size to obtain the number of AuNP in the different skin layers to compare different experiments according to Eq. (1):

It should be noted here that the summed value developed by data analysis is regarded as a semiquantitative index to compare penetration into the SC and DSL under the same conditions, as well as for different formulations and conditions. Summed values were found more informative than mean values (SD) due to the nonhomogeneous distribution of the particles throughout the diffusion area. This has been explained in detail earlier.⁷ This value does not give the real amount of particles penetrated.

2.7.

Setting up the Experimental Conditions

Nanoparticle penetration experimentation through skin equivalent was standardized in terms of the experimental setup, sterility level, applied dose/area, cryo-sectioning conditions, and skin integrity check. This will be discussed in detail in the following section.

3.1.

Experimental Setup

The Franz diffusion cell, the most common setup for conducting a penetration/permeation experiment, was not used in this case to avoid shear stress on the skin specimen developing from clamping the receptor and donor compartments together. On the contrary, experiments were run in a six-well plate with inserts in which the basolateral compartment was equivalent to the receptor compartment in the Franz diffusion cell, and particle dispersion was applied on the top of the specimen. Open-edged nonsealed specimens in the insert did not result in leakage of the applied red AuNP colloid, due to the concave surface of the SC that holds the applied dispersion (Fig. 1).

It should be noted here that using Franz diffusion cells in skin permeation studies through reconstructed human epidermis is possible, as has already been shown previously.³⁶ Special inserts could be used to avoid the shear stress on the epidermis. However, this is not possible for full-thickness in vitro tissue, due to thickness variations from one specimen to another. In addition, the low mechanical resistance of this skin model is due to the usage of collagen, which is not the case for other skin models (no dermis), and thus Franz diffusion cells cannot be used in this case.

3.2.

Sterility Conditions

Already formed skin tissue should be still kept under strictly sterile conditions before and during the penetration experiment; otherwise, infection may occur, leading to loss of barrier properties, as shown in Fig. 3(a-ii), in which infection has led to loss of specimen’s ability to retain the applied AuNP and their complete leakage outside the specimen.

3.3.

Measurement of the Diffusion Area

Contraction of the skin constructs on growth was not uniform for all skin specimens, resulting in different diffusion areas, as shown in Fig. 3(a-iv). Therefore, the dose was standardized per diffusion area. Prior to the penetration experiment, the skin specimen was imaged, and the diffusion area was estimated from the area of a circle of the same area with the help of the freeware program Image J ( http://rsbweb.nih.gov/ij/) reference to a standard scale bar.

3.4.

Skin Sectioning

Shear stress should be minimized as much as possible while sectioning to avoid artifacts. Many trials were conducted to have good skin cuts. For this, the whole skin punch was completely embedded and surrounded by binding solution, Tissue-Tek® (Sakura Finetek Europe, Zoeterwoude, Netherlands), on the specimen block on cryo-sectioning. This minimizes stress on edges. Moreover, skin punch should be sectioned at a minimum thickness of 50 µm. Thinner sections were not possible without detaching SC and DSL, as shown in Fig. 3(a-v). This does not reflect on the poor quality of the used skin models, since 10-μm sections of control untreated skin samples of the same model were possible to cut, as has been shown previously.²⁶ This is also not due to a deleterious effect of gold nanoparticles on the skin model, since skin samples grown in presence of gold nanoparticles could be cut in >10-μm sections.²⁶ However, this could be due to the increased swelling of the skin tissue after long incubation time with gold nanoparticle dispersion.

The skin integrity check was performed as mentioned in the methodology. For all the test conditions, the barrier function of the SE was not distorted, as shown in Fig. 3(a-iii).

Figure 3(b) shows the optimal procedure followed when performing a particle penetration experiment through nonsealed SE. Table 1 compares these conditions with that of HS experiments performed in our laboratory. The penetration experiment was then conducted under these optimal experimental conditions for 24 h exposure time. Similar to HS experiments,⁷ results showed nonhomogeneous distribution of AuNP all over the diffusion area, as shown in Fig. 4(a). Therefore, analysis was not dependent on one image field, as explained previously by the method of image analysis we developed earlier.²⁸ However, analysis of $z$-stacks showed depth profiles of AuNP having higher values in the upper skin layer and SC, and decreasing toward the DSL, as shown in Fig. 4(b). On comparing the number of particles penetrating into the SC and DSL in HS and SE experiments (Table 1), one could conclude an overpredictability of the SE model. This underlines again the structural deficit of reconstructed skin regarding the barrier quality (density). A seven-fold increase in the number of particles in the SC of SE versus HS was observed. In SE experiments, the particles were not exclusively localized in the SC but penetrated further into the DSL. This is despite the fact that the effective dose of particles topically applied was lower for the SE experiment than in the comparable HS experiment (Table 1). Overprediction was also the case observed for previously reported drug permeation experiments through commercial SE.^13,36,37 This implies a similar behavior of molecules and particles when penetrating the skin, but a lower penetration rate is expected here, due to the particulate nature and the reduced diffusion.³ This presumption is supported by our previously published work.^6,7

Fig. 4

(a) Representative overlaid multiphoton/transmission images showing gold nanoparticles (indicated as white spots) at different optical layers of a $z$-stack of a longitudinal skin section, in which different amounts of gold nanoparticles in the Stratum Corneum and deeper skin layers after 24 h of skin exposure were detected in each layer. A single layer was therefore not descriptive for the overall penetration pattern, due to nonhomogeneous distribution. (b) The weighted number of particles in all optical layers of the $z$-stack was thus determined, as shown above, showing the depth profile for AuNP concentrating more in the SC rather than in DSL.

Table 1

Comparison between skin penetration experiments through human skin and skin equivalent in terms of experimental conditions and penetration results.

Attempts were therefore adopted to counteract the overpredictability of SE experiments for particle penetration. Two variables were tested: exposure time and tissue fixation.

Exposure of SE to AuNP for a shorter time (2 h versus 24 h) resulted in exclusive localization of AuNP in the SC with no penetration into the DSL (Fig. 5). This result is equivalent to the 24 h penetration experiment through HS (Table 1). This could be positively employed as a fast test in the future for testing the penetration of topically applied nanoparticles, especially for cosmetic preparations, and extraplate this to penetration through HS. However, further experiments are required to validate the relationship, especially for other nanoparticles with different materials and surface chemistries.

Fig. 5

Representative overlaid multiphoton/transmission image showing exclusive localization of gold nanoparticles (indicated as white spots) in the Stratum Corneum after 2 h of skin exposure and no penetration into the deeper skin layers.

Tissues constructed in vitro (SE, in this case) could possibly suffer low compactness; i.e., compact organization is not easy to attain for artificially constructed skin. However, the use of fixatives was previously shown to be effective in the structural preservation of SE.³⁹ Therefore, the effect of fixation on increasing the tissue density was examined as a method to limit particle penetration. On the contrary, skin fixation resulted into a two-fold increase in particle localization in the SC and a slight increase in the frequency of particles penetrating into the DSL relative to unfixed skin tissues exposed to AuNP for the same exposure time, 24 h (Fig. 6). A greater enhancement effect was even evident on shorter exposure time, 2 h, in which tissue fixation resulted in induction of penetration into the DSL, versus exclusive localization for unfixed specimens, as well as a 56-fold increase in the number of particles in the SC (Fig. 6). This could be explained by the higher effect of formalin as a severe skin irritant, resulting in diminished barrier function rather than being a skin fixative. Nevertheless, skin fixation was found not to be a suitable method to avoid the exaggerated particle penetration results through SE.

Fig. 6

The effect of skin fixation on the penetration of gold nanoparticles after time exposures of two and 24 h, as extracted from optical layers of $z$-stacks (1 μm step width) imaged by multiphoton microscopy.