Penetration Depth of Propylene Glycol, Sodium Fluorescein and Nile Red into the Skin Using Non-Invasive Two-Photon Excited FLIM

The stratum corneum (SC) forms a strong barrier against topical drug delivery. Therefore, understanding the penetration depth and pathways into the SC is important for the efficiency of drug delivery and cosmetic safety. In this study, TPT-FLIM (two-photon tomography combined with fluorescence lifetime imaging) was applied as a non-invasive optical method for the visualization of skin structure and components to study penetration depths of exemplary substances, like hydrophilic propylene glycol (PG), sodium fluorescein (NaFl) and lipophilic Nile red (NR) into porcine ear skin ex vivo. Non-fluorescent PG was detected indirectly based on the pH-dependent increase in the fluorescence lifetime of SC components. The pH similarity between PG and viable epidermis limited the detection of PG. NaFl reached the viable epidermis, which was also proved by laser scanning microscopy. Tape stripping and confocal Raman micro-spectroscopy were performed additionally to study NaFl, which revealed penetration depths of ≈5 and ≈8 μm, respectively. Lastly, NR did not permeate the SC. We concluded that the amplitude-weighted mean fluorescence lifetime is the most appropriate FLIM parameter to build up penetration profiles. This work is anticipated to provide a non-invasive TPT-FLIM method for studying the penetration of topically applied drugs and cosmetics into the skin.


Introduction
With an area of about 1.5-2 m 2 , the skin covers most of the body and contributes to the mechanical and biochemical defense system [1]. Acting as an effective barrier against pathogen penetration and solar light exposure, it protects the body against temperature changes and water loss [2][3][4]. Due to this function, the skin forms a barrier against topically applied pharmaceuticals and cosmetics, i.e., ointments, creams and drugs. The stratum corneum (SC) is the uppermost layer of the skin consisting of corneocytes embedded in the structurally organized lipid matrix [3,5] and forms the strongest non-homogenous in-depth barrier between body and environment [6,7]. Cosmetics like sunscreens and skin moisturizers act in the SC and should not permeate the skin barrier [8,9], while drugs must first cross the SC to reach the living cells of the viable epidermis [10]. To ensure the permeation through the SC, different methods are used [11] including the addition of penetration enhancers into the formulations [12]. In general, molecules which can pass the SC can usually penetrate into the deeper layers of the skin [13]. Therefore, to inflammatory compound [59], and silver nanoparticles [29]. Nevertheless, this application of TPE-FLIM has so far been rarely used in practice. Therefore, this work aims to explore the efficiency of the TPT-FLIM method for penetration studies on further hydrophilic and lipophilic substances with and without their own fluorescence.

Porcine Skin Samples
In this study, fresh porcine ears were obtained from a local butcher a few hours after slaughter. The ears were cleaned with cold tap water and dried with soft paper tissue. For each experiment, six different ears were used. The hair was carefully removed using scissors so that the SC was not damaged, then the ears were stored in a fridge at a temperature of 4 • C until the next day. Measurements were performed on fresh ears within 48 h of delivery to exclude an influence on the penetration experiments of post-mortem changes [60]. On the measurement day, a 2 cm × 2 cm area was selected and marked on each ear for the application of 20 µL/cm 2 of the examined substance. A finger massage was applied gently in circular motion for 1 min to ensure that the solution was homogeneously distributed over the entire area. After that, the treated and untreated skin samples were incubated for 30 min at a temperature of 32 • C, which mimics the in vivo conditions of human skin surface temperature [61,62]. After the incubation, the rest of the substances was removed from the skin surface using a soft dry tissue, then cleaned with wet tissue 3-5 times. A 1 cm × 1 cm area was excised from the skin for the TPT-FLIM measurements. For tape stripping, a 6 cm × 6 cm sample was treated with 72 µL of the examined solution for 30 min. Before pressing the tapes onto the skin, a small piece of approx. 1 cm × 1 cm was excised for the Raman micro-spectroscopic measurements.

Propylene Glycol (PG)
PG (1,2-propanediol-CH 3 CH(OH)CH 2 OH)) is a colorless viscous liquid and is used as a vehicle for drugs as well as as a hydrophilic penetration enhancer [63]. The action mechanism of PG on the skin has not yet been well understood [64]. PG is thought to act by dissolving keratin through binding at the hydrogen bonding sites. It may also act by interacting with the polar head groups of the intercellular lipids and can thus weaken the barrier function of the SC [10,65]. In this study, PG (SIGMA-Aldrich Chemie GmbH, Steinheim, Germany) was applied undiluted on the skin and incubated for 30 min at 32 • C. The pH value of the PG was 7.29 ± 0.04. PG has no TPE fluorescence at 760 nm excitation.

Sodium Fluorescein (NaFl)
NaFl (C 20 H 10 Na 2 O 5 ) is an orange-red salt powder of very good solubility in water, which has the ability to bind proteins [66]. NaFl was selected because it is often used in dermatology as a marker substance for penetration measurements in vivo and ex vivo [13,67]. The fluorescence lifetime of NaFl is pH dependent and varies between 3770-4110 ps in a pH range of 6.5-8.6 [68]. For TPT-FLIM and LSM measurements, 1 mL of NaFl solution was prepared by diluting 10 µL of NaFl 2% w/v eye-drop solution (Bausch & Lomb GmbH, Berlin, Germany) in 990 µL of phosphate-buffered saline (PBS), whose pH value is 7.4. After TPE by 760 nm a fluorescence lifetime of τ = 4090 ± 50 ps was measured using single-exponential decay. For tape stripping and confocal Raman micro-spectroscopy, a NaFl solution (Fluoreszein SE Thilo ® eyedrops, 1.7 mg/mL, Alcon Pharma GmbH, Freiburg, Germany) was applied to the skin. The treatment time was 30 min at 32 • C for both of the NaFl-solutions.

Nile Red (NR)
NR was chosen due to its strong lipophilicity and its emission-dependence on the polarity of the environment [69]. For example, phospholipids show a red-shifted emission spectrum, while the neutral lipids such as cholesterol and triglycerides show a yellowshifted spectrum [69,70]. Not only the emission but also the fluorescence lifetime of NR varies depending on the polarity of its environment. The fluorescence lifetime of NR was examined in different solvents [70,71]. Taking into consideration a strong occlusion and swelling of the SC induced by topically applied petrolatum [72], which results in the formation of a waterfront and the reduced penetration depth of lipophilic NL [73], paraffin oil was chosen as a solvent. To prepare the NR solution, 1 mg of NR powder (SIGMA-Aldrich Chemie GmbH, Steinheim, Germany) was dissolved in 1 mL of paraffin oil (SIGMA-Aldrich Chemie GmbH, Steinheim, Germany) and mixed using a vortex shaker for 15 min, followed by 15 min in ultrasonic bath under a temperature of 30 • C. At last, the skin was treated with 80 µL of NR for 30 min at 32 • C. The fluorescence lifetime of the solution was τ = 3176 ± 53 ps, determined using a single-exponential decay after a TPE at 760 nm. The solvent used (paraffin oil) is itself non-fluorescent, can be used as a skin moisturizer, and does not permeate the SC, and thus cannot enhance the penetration and reach the viable epidermis [74,75]. The control skin samples were treated with the vehicle of NR, which is paraffin oil, to ensure that both samples were examined under the same conditions.

Two-Photon Tomography in Combination with Fluorescence Lifetime Imaging (TPT-FLIM)
A two-photon tomograph (TPT, Dermainspect, JenLab GmbH, Jena, Germany) equipped with a tunable (710-920 nm) Ti:sapphire laser (Mai Tai XF, Spectra Physics, Milpitas, CA, USA) operated with a 100 fs pulse at a repetition rate of 80 MHz was used for the horizontal imaging of the skin based on TPE. A drop of immersion oil (Immersol™ 518F, Carl Zeiss Jena GmbH, Oberkochen, Germany) was placed between the objective and the objective ring with glass (No.1, 18 mm, VWR, Darmstadt, Germany) along with a drop of water between the glass and the skin to provide refraction-index matching. A bandpass filter (410-680 nm) was used to detect the fluorescence signal after TPE at a wavelength of 760 nm. In addition, the TPT has a 375-385 nm bandpass filter for the detection of the second-harmonic generation (SHG) signal. The lateral and axial resolutions of the TPT-FLIM are approximately 0.5 µm and 1.2-2.0 µm, respectively, with a horizontal scan field of up to 350 µm × 350 µm and a vertical field of up to 200 µm and a numerical aperture of the objective of 1.3 [76]. The acquisition time to record one image was 6.8 s with a scan field of 75 µm × 75 µm. Stack images were acquired at 4 µm increments from the skin surface to a depth of 32 µm. The laser power was adjusted experimentally depending on the fluorophore applied on the skin regarding the imaging depth (Table 1).  0  10  1  5  10  10  18  2  12  15  20  23  10  20  20  30  33  23  28  28  40  40  33  40  40 The TPE-FLIM data were processed and analyzed using SPCImage software version 8.4 (Becker & Hickl, Berlin, Germany). The fluorescence lifetime decay was determined in each pixel of the skin images using a bi-exponential function and the maximum likelihood estimation method. The fast lifetime and amplitude decay components are τ 1 , α 1 ; the slow ones are τ 2 , α 2 . The amplitude-weighted mean fluorescence lifetime τ m was defined as shown in Equation (1): The fluorescence lifetime decay was averaged over the 48 neighboring pixels using a binning value of 3. The TPE-FLIM parameters of each image were exported using SPCImage software and the data were analyzed using Microsoft Excel 2019 to create parameter-depth charts. The utilized TPT-FLIM was described in detail by our group [29,50].

Confocal Laser Scanning Microscopy
A confocal laser scanning microscope (VivaScope ® 1500, Multilaser, MAVIG, Munich, Germany) was used in fluorescent mode for one-photon excitation. The system is equipped with three laser diodes (488, 685 and 785 nm) and an objective with a numerical aperture of 0.9. In this study, the 488 nm wavelength was chosen to investigate the skin in the fluorescent mode. A drop of immersion oil (Crodamol STS, Croda Inc., Snaith, UK) was applied between the skin and the ring glass (adhesive window with crosshair, Lucid Vision Labs GmbH, Ilsfeld, Germany). Ultrasonic gel (Aquasonic 100, Parker laboratories Inc., Fairfield, CT, USA) was placed between the objective and the objective ring glass to maintain the optical contact. The utilized LSM system was described previously by our group [77,78].
Stack images (500 µm × 500 µm) were made at 1.5 µm increments from the skin surface down to approximately 80 µm depth. The laser power was fixed at 5 mW for the treated and untreated skin, aiming at comparing the fluorescence intensities. Using ImageJ 1.53K software (Wayne Rasband, National Institute of Health, Bethesda, MD, USA), an area of 188 µm × 188 µm was selected on a fixed position of the images from the sample surface down to approximately 80 µm. The mean gray values of the selected areas were measured and then averaged at the corresponding depth.

Tape Stripping (TS) and UV/VIS Spectroscopy Measurements
The TS method was used to study the penetration of NaFl into porcine ear skin. A 72 µL quantity of the NaFl was applied onto a 6 cm × 6 cm area of each skin sample for 30 min, then the rest of the solution was removed. Adhesive tapes (Tesa ® , No. 5529, Beiersdorf AG, Hamburg, Germany) were pressed onto the skin using a rubber roller weight of 746 g and rolled 5 times without external pressure; then the tape was removed with one swift movement. Following the method described by Jacobi et al. [79], approx. 84% of the SC was removed after 30 tapes had been stripped from the same area.
The tape strips were cut to a size of 1.9 cm × 3.4 cm and the fluorescein on them was extracted by solving each of them in 6.46 mL ethanol (Uvasol ® Ethanol 99.9%, Merck KGaA, Darmstadt, Germany) using an ultrasonic bath, then purified by centrifuge (MR 1812, Jouan GmbH, Unterhaching, Germany). The extract was decanted in a UV cuvette using a pipette and the absorbance was measured by a spectroscope (Lambda 650S, PerkinElmer, Frankfurt/Main, Germany) at 480 nm for the detection of NaFl.

Confocal Raman Micro-Spectroscopy (CRM)
Raman spectra were recorded using a skin-composition analyzer (Model 3510 SCA, River D International B.V., Rotterdam, The Netherlands) and CRM measurements were performed on the same 6 porcine ear-skin samples which had been used for the UV/VIS spectroscopy measurements. The fingerprint Raman spectra (400-2000 cm −1 ) were recorded from the skin surface down to a depth of 40 µm with a step size of 2 µm. The acquisition time was 5 s/spectrum, the excitation wavelength was 785 nm, and the maximal power at the surface was set to 20 mW. For each skin sample, 10 different points were measured. The semiquantitative concentration profiles of NaFl in the SC were determined for each measurement point using the unconstrained multiple least square fit method (available in the SkinTools software developed by RiverD International B.V.) [80] and averaged for further analysis. The utilized CRM was described previously by our group [27].

Results
TPT was used to determine the thickness of the SC based on the appearance of cells attributed to the stratum granulosum. On average, this was 16.0 ± 3.3 µm, which is consistent with data from the literature [3,75,81]. Thus, 16 µm was considered in the following results as the SC thickness.

Penetration of PG
After applying 40 µL of PG on 2 cm × 2 cm skin sample, it was incubated for 30 min at a temperature of 32 • C. TPT-FLIM measurements were done starting from the skin surface down to a depth of 32 µm. PG has a pH = 7.30 ± 0.04, which is about 1 more than the skin surface of porcine ear [82]. Using pseudocolor scale of τ m = 1000-2000 ps, a comparison between the TPE-FLIM images for untreated and PG-treated skin is shown in Figure 1 for different depths.  Table 1.
Generally, an increase in τm in the whole image down to a depth of 12 μm in SC was observed after the treatment with PG. In Figure 1f, τm was about 334 ± 65 ps longer in the corneocytes, while in the areas between them (Figure 1f Table 1. Generally, an increase in τ m in the whole image down to a depth of 12 µm in SC was observed after the treatment with PG. In Figure 1f, τ m was about 334 ± 65 ps longer in the corneocytes, while in the areas between them (Figure 1f, arrows) the increase in τ m was about 280 ± 55 ps. Those areas were no more visible in the deeper SC, therefore it is assumed that those areas refer to the extracellular area separating the corneocytes. At 12 µm depth (Figure 1h), a small increase from τ m = 1134 ± 87 ps to τ m = 1278 ± 107 ps was measured.
Compared to the untreated skin in Figure 1i-l, τ m of the viable epidermis was not affected after the treatment with PG (Figure 1m-p) showing a non-significant ∆τ m of only 14-40 ps.
TPE-FLIM parameters of both samples are presented in Table 2. τ 1 , τ 2 , and τ m increased in the SC after the treatment with PG, while the relative amplitudes were not affected since no additional fluorophore was applied on the skin. At 16 µm depth, the viable epidermis starts and the differences in TPE-FLIM parameters disappear. Furthermore, the penetration profile of PG in porcine ear skin using the average of τ m for the six ears is shown in Figure 2. The penetration profile exhibited the same results shown in Figure 1, where PG was only detectable in the SC.  Table 2. τ1, τ2, and τm increased in the SC after the treatment with PG, while the relative amplitudes were not affected since no additional fluorophore was applied on the skin. At 16 μm depth, the viable epidermis starts and the differences in TPE-FLIM parameters disappear. Furthermore, the penetration profile of PG in porcine ear skin using the average of τm for the six ears is shown in Figure 2. The penetration profile exhibited the same results shown in Figure 1, where PG was only detectable in the SC.

Penetration of NaFl Using TPE-FLIM and LSM
The fluorescence lifetime of the NaFl solution was measured by a single-exponential decay to be τ = 4090 ± 50 ps. The TPE-FLIM parameters of the untreated and the NaFltreated skin were determined after a treatment time of 30 min and a TPE by 760 nm. Figure  3 shows the distribution of NaFl in the skin after treatment, where the τm of the SC increases on average from τm = 1522 ± 89 ps in the untreated sample (Figure 3a-d)  PG-treated Untreated

Penetration of NaFl Using TPE-FLIM and LSM
The fluorescence lifetime of the NaFl solution was measured by a single-exponential decay to be τ = 4090 ± 50 ps. The TPE-FLIM parameters of the untreated and the NaFl- The white arrows refer to the extracellular areas between the keratinocytes. τm was measured after an excitation with 760 nm using a TPT and is shown by a pseudocolor scale of 500-3000 ps. Scale bar: 20 μm. Acquisition time: 6.8 s. Excitation power is depth-dependent and shown in Table 1.
Furthermore, τm increased in the viable epidermis from τm = 1667 ± 69 ps in the untreated (Figure 3i-l) to τm = 2079 ± 42 ps in the NaFl-treated sample (Figure 3m-p). The TPE-FLIM parameters are shown in Table 3 for both skin samples. The fast and slow fluorescence lifetime components τ1 and τ2 had increased down to a depth of 16 μm, while τm and the relative amplitude α2 indicated an increase even in the viable epidermis till at least 32 μm depth. The fluorescence lifetime of the NaFl solution amounts to τ = 4090 ± 50 ps and is close to the τ2 of the viable epidermis of the untreated skin, which varied in a range of τ2 = (3553 ± 131)-(3956 ± 181) ps. Thus, τ1 and τ2 may not show a noticeable increment, but with their relative amplitudes α1 and α2 will, because they also represent  (m-p) The viable epidermis of the NaFl-treated skin; (o,p) The white arrows refer to the extracellular areas between the keratinocytes. τ m was measured after an excitation with 760 nm using a TPT and is shown by a pseudocolor scale of 500-3000 ps. Scale bar: 20 µm. Acquisition time: 6.8 s. Excitation power is depth-dependent and shown in Table 1.
Furthermore, τ m increased in the viable epidermis from τ m = 1667 ± 69 ps in the untreated (Figure 3i-l) to τ m = 2079 ± 42 ps in the NaFl-treated sample (Figure 3m-p). The TPE-FLIM parameters are shown in Table 3 for both skin samples. The fast and slow fluorescence lifetime components τ 1 and τ 2 had increased down to a depth of 16 µm, while τ m Pharmaceutics 2022, 14, 1790 9 of 19 and the relative amplitude α 2 indicated an increase even in the viable epidermis till at least 32 µm depth. The fluorescence lifetime of the NaFl solution amounts to τ = 4090 ± 50 ps and is close to the τ 2 of the viable epidermis of the untreated skin, which varied in a range of τ 2 = (3553 ± 131)-(3956 ± 181) ps. Thus, τ 1 and τ 2 may not show a noticeable increment, but with their relative amplitudes α 1 and α 2 will, because they also represent the amount of the contributing fluorophores. This increase is therefore also detectable by τ m because it is amplitude-weighted lifetime. Table 3. TPE-FLIM parameters of untreated and NaFl-treated skin. Mean ± SEM. N = 6.

Untreated Skin
NaFl-Treated Skin  Figure 4 presents the different penetration profiles of NaFl based on τ m and τ 2 and it shows also that the influence of NaFl on the viable epidermis using τ 2 as a representative TPE-FLIM parameter is not detectable. the amount of the contributing fluorophores. This increase is therefore also detectable by τm because it is amplitude-weighted lifetime.  Figure 4 presents the different penetration profiles of NaFl based on τm and τ2 and it shows also that the influence of NaFl on the viable epidermis using τ2 as a representative TPE-FLIM parameter is not detectable. To ensure that the penetration depth indicated by τm is correctly determined, we treated three skin samples of different porcine ears with NaFl and measured the fluorescence intensity of the untreated and treated samples via LSM. The treatment time was 5 min and 30 min at 32 °C. Figure 5a shows that the fluorescence intensity in the viable epidermis of the 5-and 30-min treated skin was higher compared to the untreated skin. On the other hand, the 30-min treated sample showed a higher intensity than the 5-min treated sample indicating that more NaFl reached the viable epidermis after a longer treatment time. This result was confirmed by the relative amplitude α2 (Figure 5b), which reflects the relative amount of the contributing fluorophores in τ2, which includes NaFl because it has a long fluorescence lifetime. In the viable epidermis from 16 μm to 32 μm  To ensure that the penetration depth indicated by τ m is correctly determined, we treated three skin samples of different porcine ears with NaFl and measured the fluorescence intensity of the untreated and treated samples via LSM. The treatment time was 5 min and 30 min at 32 • C. Figure 5a shows that the fluorescence intensity in the viable epidermis of the 5-and 30-min treated skin was higher compared to the untreated skin. On the other hand, the 30-min treated sample showed a higher intensity than the 5-min treated sample indicating that more NaFl reached the viable epidermis after a longer treatment time. This result was confirmed by the relative amplitude α 2 (Figure 5b), which reflects the relative amount of the contributing fluorophores in τ 2 , which includes NaFl because it has a long fluorescence lifetime. In the viable epidermis from 16 µm to 32 µm depth, α 2 was on average 5.2 ± 0.2% higher in the 30-min treated than in the 5-min treated sample.

Penetration of NaFl Using CRM and TS
Further experiments were performed to study the penetration depth of NaFl TS and CRM methods. The thickness of the SC in the examined skin samples was 2 The Raman spectrum of the NaFl is presented in Figure 6a and the corresponding tration profiles of NaFl in the skin are shown in Figure 6b. Results obtained using show the exponential decay: at 2 μm depth, the concentration of NaFl decreased 100% to 58 ± 10% and to 21 ± 12% at 4 μm depth. At 8 μm, no signal was detected.  Table 1. SC-thickness is 16.0 ± 3.3 µm. N = 3.

Penetration of NaFl Using CRM and TS
Further experiments were performed to study the penetration depth of NaFl using TS and CRM methods. The thickness of the SC in the examined skin samples was 20 µm. The Raman spectrum of the NaFl is presented in Figure 6a and the corresponding penetration profiles of NaFl in the skin are shown in Figure 6b. Results obtained using CRM show the exponential decay: at 2 µm depth, the concentration of NaFl decreased from 100% to 58 ± 10% and to 21 ± 12% at 4 µm depth. At 8 µm, no signal was detected.
Regarding TS and the UV/VIS spectroscopy measurements, the absorbance o tracted NaFl was measured at 460 nm and the recorded value of the first TS was ered as 100%, so the absorbance from the next TS was calculated as a correspond centage value. Within the first 2 μm, the absorbance fell from 100% to 27 ± 10% μm to 3 ± 7%. At 4.5 μm depth, no more absorbance was measured. As shown i 6b, the value of the penetration-depth profile obtained by CRM is higher than by

Penetration of NR
In the last part of this study, the penetration of the lipophilic NR in the s investigated. The prepared NR solution had a fluorescence lifetime τ = 3176 ± 53 p lated by single exponential decay after TPE at 760 nm. Regarding TS and the UV/VIS spectroscopy measurements, the absorbance of the extracted NaFl was measured at 460 nm and the recorded value of the first TS was considered as 100%, so the absorbance from the next TS was calculated as a corresponding percentage value. Within the first 2 µm, the absorbance fell from 100% to 27 ± 10% and at 4 µm to 3 ± 7%. At 4.5 µm depth, no more absorbance was measured. As shown in Figure 6b, the value of the penetration-depth profile obtained by CRM is higher than by TS.

Penetration of NR
In the last part of this study, the penetration of the lipophilic NR in the skin was investigated. The prepared NR solution had a fluorescence lifetime τ = 3176 ± 53 ps calculated by single exponential decay after TPE at 760 nm.  Table 1.
NR was found to penetrate only in the SC without being able to reach the viable epidermis. In the first 4 μm (Figure 7a,b,e-f), NR was distributed in the SC, inducing an average increment of about Δτm = 1963 ± 92 ps. Following this, the presence of NR was limited in specific areas of SC (Figure 7g, arrows), where τm = 1926 ± 29 ps, while the rest of the same image of treated skin showed τm = 1232 ± 35 ps.
The stars in Figure 7f refer to the background, as they are located on the black areas of the image. In Figure 7g-h, those areas are colored blue, because NR appears, thereby referring to the skin surface due to the furrows. Thus, stars' locations were not considered as having been penetrated by NR at the corresponding depth.
The TPE-FLIM parameters are shown in Table 4. Significant differences in α1, α2, and τm were detected down to a depth of 12 μm, but only to an 8 μm depth in τ1 and τ2. The untreated skin had a fluorescence lifetime τ2 = 3155 ± 192 ps at 12 μm depth, which is similar to the fluorescence lifetime of NR solution with τ = 3176 ± 53 ps. Therefore, the presence of NR at this depth did not induce changes in τ2. However, the relative amplitude α2 increased at 12 μm from 32.4 ± 0.9% to 49.0 ± 1.6%. The viable epidermis started at around 16 μm, where all of the TPE-FLIM parameters were comparable in both skin samples.   Table 1. NR was found to penetrate only in the SC without being able to reach the viable epidermis. In the first 4 µm (Figure 7a,b,e,f), NR was distributed in the SC, inducing an average increment of about ∆τ m = 1963 ± 92 ps. Following this, the presence of NR was limited in specific areas of SC (Figure 7g, arrows), where τ m = 1926 ± 29 ps, while the rest of the same image of treated skin showed τ m = 1232 ± 35 ps.
The stars in Figure 7f refer to the background, as they are located on the black areas of the image. In Figure 7g,h, those areas are colored blue, because NR appears, thereby referring to the skin surface due to the furrows. Thus, stars' locations were not considered as having been penetrated by NR at the corresponding depth.
The TPE-FLIM parameters are shown in Table 4. Significant differences in α 1 , α 2 , and τ m were detected down to a depth of 12 µm, but only to an 8 µm depth in τ 1 and τ 2 . The untreated skin had a fluorescence lifetime τ 2 = 3155 ± 192 ps at 12 µm depth, which is similar to the fluorescence lifetime of NR solution with τ = 3176 ± 53 ps. Therefore, the presence of NR at this depth did not induce changes in τ 2 . However, the relative amplitude α 2 increased at 12 µm from 32.4 ± 0.9% to 49.0 ± 1.6%. The viable epidermis started at around 16 µm, where all of the TPE-FLIM parameters were comparable in both skin samples.
The mean fluorescence lifetime τ m was chosen to create the penetration profile of NR in the skin and Figure 8 shows that τ m of the treated and untreated samples overlaps in the viable epidermis and hence NR did not permeate the SC.  The mean fluorescence lifetime τm was chosen to create the penetration profile of NR in the skin and Figure 8 shows that τm of the treated and untreated samples overlaps in the viable epidermis and hence NR did not permeate the SC.

Discussion
The penetration depth of three different substances-PG, NaFl and NR-into porcine ear skin ex vivo was evaluated non-invasively using TPE-FLIM. It was possible to study the penetration of a non-fluorescent substance like PG taking advantage of the dependence of TPE-FLIM on the environmental parameter, which in this study was the pH value.
PG shows no fluorescence at 760 nm, but its pH value is about 1 higher than the SC [83,84]. Thus, the pH-dependence of the fluorescence lifetime [35][36][37]85] was taken into account, aiming at proving the presence of PG indirectly through the influence of its pH value on the fluorescence lifetime of the endogenous fluorophores. Therefore, the FLIMparameters τ1, τ2, and τm increased, which was shown in Table 2. Based on the TPE-FLIM images in Figure 1, the increase in τm was detected in the whole space of the SC images, thus PG affected the corneocytes and the extracellular region. This leads to the conclusion that PG penetrated transcellularly, passing the intra-and the intercellular penetration

Discussion
The penetration depth of three different substances-PG, NaFl and NR-into porcine ear skin ex vivo was evaluated non-invasively using TPE-FLIM. It was possible to study the penetration of a non-fluorescent substance like PG taking advantage of the dependence of TPE-FLIM on the environmental parameter, which in this study was the pH value.
PG shows no fluorescence at 760 nm, but its pH value is about 1 higher than the SC [83,84]. Thus, the pH-dependence of the fluorescence lifetime [35][36][37]85] was taken into account, aiming at proving the presence of PG indirectly through the influence of its pH value on the fluorescence lifetime of the endogenous fluorophores. Therefore, the FLIMparameters τ 1 , τ 2 , and τ m increased, which was shown in Table 2. Based on the TPE-FLIM images in Figure 1, the increase in τ m was detected in the whole space of the SC images, thus PG affected the corneocytes and the extracellular region. This leads to the conclusion that PG penetrated transcellularly, passing the intra-and the intercellular penetration pathways in the SC, which is typical for lipophilic substances [1,65,86]. Nevertheless, PG could not be detected anymore in the viable epidermis, which starts at around 16 µm depth. Mujica Ascencio et al. [63] showed by using multivariate analysis of Raman spectra that PG can permeate the SC and reach the depth of max. 22.0 µm in the skin, where the thickness of SC was ≈18 µm. This penetration depth could not be precisely detected using TPE-FLIM because PG and the viable epidermis share the same pH value of pH ≈ 7.4 [82] and thus cannot induce further changes in its fluorescence lifetime, which can be explained by the limited sensitivity of TPE-FLIM method to the non-fluorescent PG.
For fluorescent dyes like NaFl and NR with τ= 4090 ± 50 ps and τ= 3176 ± 53 ps, respectively, τ m was more suitable than τ 1 and τ 2 for studying the penetration depth, due to the similar values of τ 2 in the viable epidermis and the fluorescence lifetime of both examined substances. In this case, the lifetime components of the untreated skin will overlap with the components of the treated skin and the presence of the exogenous fluorophore is detectable only when the relative amplitude is taken into consideration because this amplitude is sensitive to the relative amount of the fluorophores. The application of TPE-FLIM to study the penetration of zinc oxide nanoparticles was presented by Roberts et al. [57,87] and showed the limitation of TPE-FLIM due to the overlapping of the τ 1 and τ 2 of zinc oxide with the autofluorescence lifetime of the endogenous fluorophores. As a solution, they used appropriate emission-bandpass filters to exclude the emission and lifetime of specific endogenous fluorophores. In this study, we show that this limitation can be overcome without a bandpass filter by taking the relative amplitudes into account, because they reflect the proportion of τ 1 and τ 2 . Thus, τ m can distinguish between two samples that have similar fast and slow lifetime components but in different fractions.
NaFl reached the viable epidermis and diffused into it. This result was presented by the TPE-FLIM parameters shown in Table 3, excluding τ 1 and τ 2 for the reason mentioned above. The penetration depth was confirmed by fluorescence-intensity analysis using LSM, where the gray values of the intensity images were considered as fluorescence intensity. The fluorescence intensity was measured after 30 min of treatment with NaFl and was, e.g., at 40.5 µm depth I 30 min-treated =6.8 ± 0.5 a.u., while in the untreated skin I untreated = 3.9 ± 0.3 a.u. which refers to the presence of NaFl in the stratum spinosum. Furthermore, the same measurement was repeated on each skin sample after 5 min of treatment and the fluorescence intensity was I 5 min-treated =5.2 ± 0.7a.u., which is lower compared to 30 min of treatment but still higher than in the untreated skin. Consequently, NaFl is able to penetrate at least 40.5 µm into the epidermis within only 5 min, and more NaFl diffuses into the viable epidermis after 30 min. As the relative amplitude α 2 reflects the relative amount of the contributing fluorophores in τ 2 , the parameter α 2 was measured in the three samples. Again, more NaFl was detected in the viable epidermis after 30 min than after 5 min which is confirmed by LSM data.
The CRM data showed a smaller penetration depth of approximately 8 µm. However, with only approximately 4 µm, the penetration depth using TS and UV/VIS spectroscopy was the smallest compared to the other methods. According to these CRM and TS results, it can be concluded that NaFl does not permeate the SC, while it was detectable in the stratum spinosum down to at least 32 µm using TPE-FLIM, and until at least 40 µm using LSM. The differences in the results for TS, CRM, TPE-FLIM, and LSM could be attributed to differences in the sensitivity of each method. According to O'goshi and Serup [66], NaFl can hardly permeate through the SC. Further information about the penetration depth of NaFl applied topically onto the skin could not be found in the literature. However, in this study, we demonstrated deeper penetration into the viable epidermis using TPT-FLIM and LSM.
The hydrophilic NaFl penetrated via the transcellular pathway in the SC, as all points on the TPE-FLIM images of the SC (Figure 3e-h) indicate an increase in τ m . In the viable epidermis, the structure of the keratinocytes was more apparent, where the lifetime of the intercellular region between the keratinocytes had increased more with τ m = 2169 ± 20 ps (Figure 3o-p, arrows) compared to the keratinocytes with τ m = 1918 ± 21 ps. This leads to the assumption that NaFl crossed the lipophilic barrier of the SC transcellularly and intercellularly and diffused into the hydrophilic viable epidermis, with a higher presence in the intercellular space of the viable epidermis. The intercellular penetration pathway of NaFl in the human viable epidermis was shown previously by Roberts et al. [87] who injected NaFl into the skin and investigated it using fluorescence-intensity images after TPE at 920 nm.
Based on the analysis of τ m , NR was found not to permeate the SC. It is known that the fluorescence lifetime of NR depends on the polarity of its environment [70] as well as on the kind of lipids it is binding to. For example, cholesterol and phospholipids induce a longer lifetime of 4200 ps, while triglycerides induce a shorter lifetime of approx. 3000 ps [70,88,89]. The lipid content in the viable epidermis is small compared to the SC [90], but even if NR was present in the viable epidermis, it should have bound to those lipids and therefore induce a longer τ m compared to the untreated skin. Thus, we conclude that NR did not reach the viable epidermis. In addition, the viable epidermis being a hydrophilic layer [91] with a water content of 60-70% [92] should prevent NR from penetration into the viable epidermis. Comparing to the literature, the penetration depth of NR solved in PG [93] as well as of NR gel [94] was examined using LSM. Both studies showed the same results: that NR does not reach the viable epidermis.
TPE-FLIM images of the NR-treated SC showed at 8 µm depth (Figure 7g, arrows) areas with longer τ m = 1926 ± 92 ps compared to the rest of the image with τ m = 1232 ± 35 ps. It is assumed that those areas refer to the lipid matrix in the intercellular penetration pathway, because the intercellular lipids should bind the NR molecules as shown by Talreja et al. [95] due to the strong lipophilicity of NR, thereby inducing a longer τ m as we observed in this study. On this basis it can be summarized that NR penetrates into the SC through the lipophilic intercellular pathway, but does not permeate the SC.

Conclusions
The outcome of this study demonstrates that TPE-FLIM can be a powerful non-invasive method for skin-penetration studies. We have shown that TPE-FLIM is not limited to fluorescent dyes, but can also indirectly detect non-fluorescent substances depending on their influence on the molecular environment of the skin's endogenous fluorophores. However, this influence may disappear if skin parameters, e.g., the pH differences between the SC and the viable epidermis, change. Furthermore, the study showed how the choice of the TPE-FLIM parameters can affect the evaluation of the penetration depth of fluorescent dyes, i.e., if the examined substance has a fluorescence lifetime, which is similar to one of the fluorescence-lifetime components (τ 1 or τ 2 ) and therefore the amplitude-weighted τ m is proposed as a representative parameter for the penetration profile in the skin. Based on this study, we found that both PG and NR cannot permeate the SC and may penetrate at least 12 µm into the SC (SC thickness is 16.0 ± 3.3 µm), while NaFl reaches the viable epidermis down to a depth of at least 40 µm. In contrast, TS and CRM detected NaFl only in the superficial SC depth, and the observed differences are explained by the different sensitivity of the applied methods. PG and NaFl penetrate transcellularly as they are hydrophilic, and NR intercellularly due to its lipophilic property.  Institutional Review Board Statement: Freshly slaughtered pig ear skin without additional treatment was provided by a local butcher. Ethical approval was obtained from the Veterinary Board Dahme-Spreewald, Germany.

Informed Consent Statement: No applicable.
Data Availability Statement: The data are available from the corresponding author upon reasonable request.