2.1.

Materials

Tesafilm® kristall-klar (Tesa, $33\mathrm{m}\times 19\mathrm{mm}$, cut to pieces of $30\times 19\mathrm{mm}$) was purchased from Tesa AG, Hamburg, Germany. UHU superglue (UHU, blitzschnell Pipette) was provided by UHU GmbH & Co, KG, Bühl/Baden, Germany. Mayer’s haematoxylin was obtained from Carl Roth $\mathrm{GmbH}+$ Co. KG, Karlsruhe, Germany.

2.2.

Porcine Skin

Fresh pig ears were obtained from Emil Färber GmbH & Co. KG, Zweibrücken, Germany. The ears were excised before brewing and shipped immediately to the lab. After rinsing with water, they were inspected for evident skin abrasions prior to use. Only healthy-looking skin was selected for the experiments.

2.3.

Human Skin

Fresh human skin for the in vitro experiment was obtained from plastic surgery of female Caucasian patients from the department of Plastic and Hand Surgery, Caritaskrankenhaus, Lebach, Germany. The study included thigh skin from a female donor, 48 years old, previously consented. Immediately upon arrival of the skin, the subcutaneous adipose tissue was removed with a scalpel and the skin was cut into pieces of $10\times 20\mathrm{cm}$, wrapped in aluminum foil and stored in polyethylene bags at $-20°\text{C}$ until use. The study was approved by the ethical commission of Saarland, Germany (Aerztekammer des Saarlandes, 204/08).

2.4.

Cross-Sections of Hair Follicle

To get a cross section of a hair follicle, a biopsy of a $20\times 30\text{-}\mathrm{mm}$ sample was cut out of the intact frozen thigh skin using a scalpel and cross-sections with a thickness of 15 μm were cut using a cryomicrotome (MEV Cryostat, Slee, Mainz, Germany). Nuclear staining was performed using Mayer’s haematoxylin. The optical microscope images were taken using a Zeiss AXIO Scope A1 light microscope (Carl Zeiss Microscopy GmbH, Oberkochen, Germany) equipped with a digital camera (AxioCam ERc 5 s). Images were edited and labeled using the software Zen lite 2011 (Carl Zeiss Microscopy GmbH, Oberkochen, Germany).

2.5.

Cyanoacrylate Skin Surface Stripping

Cyanoacrylate skin surface biopsies were taken by applying a drop of superglue on a predetermined and precleansed location of the excised pig ear, immediately followed by application of a tape strip on top.^10,18 After polymerization of the glue (15 min), this tape was quickly removed along with the content of the hair follicle. For analysis the biopsies were fixed upside down on glass slides, leaving the follicles sticking up.

2.6.

Confocal Raman Microscopy

For confocal Raman microscopic evaluation, cross-sections and cyanacrylate biopsies were placed on calcium fluoride slides. Measurements were performed using a Witec alpha $300R+$ (WITec GmbH, Ulm, Germany). The excitation source was a diode laser with an emitting wavelength of 785 nm. Laser power was adjusted to 50 mW before the objective, which was found to be harmless to biological tissue. Microscopic images were obtained with a $10\times$ objective (Epiplan Neofluar N.A. 0.5, Zeiss, Germany). Raman single spectra and maps were recorded with a $50\times$ objective (Epiplan Neofluar N.A. 0.8, Zeiss, Germany). A confocal pinhole of 100 μm rejected signals from out-of-focus regions. Raman spectra were recorded in the range of 400 to $1780{\mathrm{cm}}^{-1}$ with a spectral resolution of $4{\mathrm{cm}}^{-1}$. For single spectra, three measurements over a 10-s integration time were averaged. Raman mapping was performed by collecting data for 10 s every 5 μm in $x$ and $y$ directions. The data were processed using WITec Project Plus software (WITec GmbH, Ulm, Germany). After cosmic ray removal, the spectral baseline was corrected using a polynomic fit. All spectra were normalized to the most intense peak (1430 to $1480{\mathrm{cm}}^{-1}$ representing $\nu (\mathrm{C}─\mathrm{H})$.¹⁹

2.7.

Optical Profilometry

Surface topography measurements were performed with a True Surface Microscopy sensor integrated in a Witec alpha $300R+$ confocal Raman microscope (WITec GmbH, Ulm, Germany). The sensor can resolve an elevation difference of 3 mm, with an accuracy of 1 μm. Data points were recorded every 5 μm in $x$ and $y$ directions, generating a lateral resolution equivalent to Raman mapping.

3.1.

Comparison of Porcine and Human Hair Follicles

Porcine ear skin represents a commonly used model for analysis of drug penetration into hair follicles. However, besides anatomical and physiological similarity one major aspect for the applicability of such a model is the chemical equivalence with the simulated site in the human body. Thus, in order to verify the suitability of porcine follicles as an in vitro substitute for human follicles, a comparison of the composition of major chemical components is performed by confocal Raman spectroscopy.

Optical and dimensional similarity of porcine and human hair follicles and the adhering tissue are depicted as microscopic sections in Fig. 1(a) and 1(b), respectively. The hair (blue cross), follicular epidermis (yellow cross), deeper skin layers (green cross), and the sebaceous gland (red cross) are clearly visible in the individual cross-sections. To evaluate the spectral comparability, Raman spectra were obtained from the four regions of interest marked by the colored crosses. Raman spectra recorded from porcine and human dermis (green), hair (blue), follicular epidermis (yellow) and a sebaceous gland (red) are depicted in Fig. 1(c) and 1(d), respectively (colored figures available online only). The follicular epidermis spectrum is composed of unspecific bands representing protein at around $1003{\mathrm{cm}}^{-1}$ (aromatic amino acids), at $1440{\mathrm{cm}}^{-1}$ $\delta (\mathrm{CH})$ and at $1650{\mathrm{cm}}^{-1}$ (amide I) with keratin as the main component.¹⁹ The dermis spectrum is additionally characterized by two double peaks at 815 to $850{\mathrm{cm}}^{-1}$ and 920 to $940{\mathrm{cm}}^{-1}$. This pattern is representative for collagen, which is localized in the dermis in significantly higher quantities than in other tissue regions.²⁰ Human and porcine hair reveal a prominent peak at $509{\mathrm{cm}}^{-1}$ specific for $S-S$ cross-links, which appear in high content in keratotic tissues like horn, hoof, and hair.^17,21 To evaluate the recorded spectra of human and porcine hair follicle components regarding similarity, a statistically meaningful method is necessary. Therefore, spectral subtraction and calculation of the area under the curve (AUC) of the resulting graphs are performed. As references, three spectra for each component from different positions were subtracted from each other. This generates graphs specifying the internal spectral variability for each individual component. Figure 2(a) depicts these graphs for hair in blue, follicular epidermis in yellow, dermis in green, and sebum in red (colored figures available online only). The confidence interval (CI) of spectral similarity for each individual component is subsequently calculated based on the absolute AUC of these graphs deriving mean and standard deviation. Finally, component spectra of human and porcine follicles are subtracted from each other and AUC of the resulting graph was determined. The calculated AUC of these graphs as well as the confidence interval for spectral similarity are listed in Fig. 2(b). The graph representing the hair states a significantly higher AUC than the reference. However, a direct comparison of the subtracted spectra, as shown in Fig. 2(c) reveals no major discrepancies but high noise. Moreover, the internal variability of the reference measurement for hair is rather small, explaining the detected difference. For follicular epidermis and dermis, no significant difference is exposed. Noteworthy is the high variability in epidermal spectra. This can be explained by the different follicular epidermal layers. To assure robustness of our method, epidermal spectra were taken from different layers. Paying special attention to the sebum as the potential release medium for drug carrier systems, no significant difference in Raman spectra is noted between human and porcine sebum. However, the direct spectral comparison displayed in Fig. 2(d) exhibits a slight heartbeat shaped discrepancy between 1050 and $1090{\mathrm{cm}}^{-1}$. Even though not significantly different, this result indicates a peak shift in the initial spectra from mainly trans conformational $\mathrm{C}─\mathrm{C}$ stretch vibrations in porcine to more random conformations in human sebum. The characteristic peak shift from 1060 to $1080{\mathrm{cm}}^{-1}$ of the $\nu (\mathrm{C}─\mathrm{C})$ skeletal vibrations confirms these findings.²² Other small discrepancies are intensity related and hint at negligible concentration differences of sebum components. Finally, the inter-individual variability of human and porcine skin composition was not taken into account since a vast number of individuals need to be tested to gain valid results. Yet, Raman spectroscopy has proven its value to compare chemical composition of all main hair follicle components and indicates chemical similarity of the different tissues and sebum in particular.

Fig. 1

Microscopic pictures of porcine (a) and human (b) hair follicle cross-sections. Raman spectra obtained from the areas marked by the colored crosses displayed in corresponding colors from porcine (c) and human (d) follicles.

Fig. 2

(a) Spectral subtraction graphs derived from porcine and human Raman spectra of the four main chemical components. (b) Statistical evaluation of the absolute area under the curve (AUC) of the spectral subtraction graphs. Comparison of single Raman spectra and the subtraction graphs for human and porcine hair (c) and sebum (d).

3.2.

Raman Mapping of Follicle Cross-Sections

For visualization of different tissue components staining procedures represent the most common technique. Figure 3(a) shows a porcine hair follicle cross-section after haematoxylin nuclear staining. A clear differentiation between follicular epidermis, dermis, and hair is possible. The location of a sebaceous gland can be assumed (red arrow) but the sebum remains undetectable. Raman mapping allows the discrimination of different components according to their Raman spectra without staining, visualizing the component distribution. An unstained microscopic section of a porcine hair follicle is displayed in Fig. 3(b). Raman spectra are recorded from the area indicated by the red rectangle in the image with a step size of 5 μm. After processing the spectral data with multivariant analysis algorithms to identify the component distribution, a false color Raman image is obtained allocating each picture pixel with a color representing similar Raman spectra [Fig. 3(c)]. In this binary method, all spots with spectra representing the dermis are shown in green, follicular epidermis in yellow, hair in blue, and sebum in red. In this image, each pixel is explicitly assigned to a component.

Fig. 3

(a) Transmittance-light microscopy pictures of a haematoxylin nuclear stained porcine hair follicle cross-section. The red arrow indicates the location of the sebaceous gland. (b) Incident light microscopic picture of an unstained porcine follicle cross-section. (c) and (d) Raman map of the area indicated by the red rectangle in (b). The hair is displayed in blue, the sebum is depicted in red, whereas epidermal and dermal structures are shown in yellow and green, respectively.

Figure 3(d) depicts the same area after further data processing. This approach displays intensity differences and allows displaying multiple components simultaneously in one pixel. This procedure facilitates a more detailed depiction of the component distribution. A close comparison of Fig. 3(c) and 3(d) reveals the advantages of this more sophisticated method. For instance, traces of sebum (red) can be identified on top of the hair (blue), which is not visible by the binary approach (colored figures available online only). The sebum is mainly located in the sebaceous gland and along the hair shaft. Slight traces of sebum remaining on the hair are considered artifacts generated by the cutting procedure. The Raman map successfully allows insight into the follicle's constitution not feasible by conventional microscopy. Raman microscopy enables visualization and characterization of the follicle’s most important components without any labeling or staining. Furthermore, the methodology bares a high potential to track penetration of drugs and drug delivery systems in the follicular tissue.

3.3.

Visualization of Cyanoacrylate Skin Surface Biopsies

Cyanoacrylate skin surface stripping represents the most common analytical technique to analyze the penetration behavior of drug delivery systems into hair follicles. Removing the hair and the follicular cast as a cyanoacrylate biopsy allows extraction and quantification of penetrated drugs. However, the quality and extent of follicle removal cannot be analyzed with this procedure. Microscopy analysis of a follicle biopsy reveals limitations, as conventional light microscopy only gains information from one focal plane at a time. Figure 4(a) and 4(b) displays two light microscopic pictures obtained from two different focal planes of a porcine cyanoacrylate biopsy visualizing this problem. The biopsy tape is turned upside down with the excised hairs pointing upwards. Only fractions of the hairs can be assessed this way. To evaluate the complete follicle, a three-dimensional analytical approach is necessary.

Fig. 4

(a) and (b) depict light microscopy pictures of a cyanoacrylate skin surface biopsy recorded from two different focal planes. (c) Schematic of the basic principle of optical profilometry. (d) Color-coded surface map of the biopsy area corresponding to (a) and (b).

A versatile method for analysis of structured samples is optical profilometry. The basic principle relies on the effect of chromatic aberration. With a set of hyperchromatic lenses good mapping-point capabilities are associated with a high chromatic error. Thus, white light which is focused through this lens assembly is split in its different colors, assigning different focal lengths to each color. While detecting the reflected light through a confocal pinhole, a topography map can be generated as displayed in Fig. 4(c). By applying the aforementioned methodology, the complete cyanoacrylate biopsy with the excised hairs can be visualized [Fig. 4(d)].

In a second step, this method allows overlaying the sample surface topography map with Raman spectroscopic information. After determining the surface structure, Raman spectra can be obtained from the exact sample surface eliminating weak signal intensity caused by out of focus effects. This enables Raman mapping of highly structured specimen. In this study, the technique was utilized to image a hair follicle obtained from a cyanoacrylate surface biopsy. Figure 5(a) shows a microscopic picture of a single hair bulb. In the area indicated by the red rectangle, the surface structure was assessed by optical profilometry. The resulting topography map of the follicle is presented in Fig. 5(b). The individual Raman spectra of the chemical components within a hair bulb are displayed in Fig. 5(c). Overlaying the surface topography map with the spatially resolved Raman spectral information results in a three-dimensional component distribution map of the follicle [Fig. 5(d)]. The false color image displays hair in blue, sebum in red and epidermal structures in yellow, thus matching the color of the component spectra displayed in Fig. 5(c).

Fig. 5

(a) Light microscopy picture of an excised hair follicle. (b) Surface topography map of the area indicated by the red rectangle in (a). (c) Raman spectra of the individual chemical components in the hair follicle. (d) Raman map displaying the component distribution on the excised hair follicle.

The image depicts an intact follicular epidermis shell enclosing the hair root. Traces of sebum can be found outside the shell, indicating the location of the sebaceous gland. An enhanced sebum presence at the bottom might indicate thinner layers of follicular epidermis at the lower hair bulb. The analytical evidence of the extraction of complete hair bulbs by cyanoacrylate skin surface biopsies corroborates the suitability of this technique for quantification of drugs within the hair follicle.

Furthermore, the novel combination of optical profilometry and confocal Raman microscopy proves its value for the evaluation of skin biopsies and indicates its potential for other analytical approaches.