2.1.

Sample Preparation

Pig ear skin samples from animals sacrificed the previous day were obtained from the Dermatological Clinic of the Charité, Berlin, Germany. The ears were washed and the hair was removed using a razorblade or scalpel. Samples from the ear were cut out. On these skin samples, the PS was applied topically in a vanishing crème at an amount of $\sim 20$ to $40\mathrm{mg}/\mathrm{c}{\mathrm{m}}^2$ and slightly massaged for $\sim 1\min$. After 60 min, the samples were washed with water to remove residual (not absorbed) crème from the skin surface.

As PS pheophorbide-a (pheo), obtained from spirulina Arthrospira platensis, is used. Physiogel^® hypoallergen vanishing crème was obtained from Stiefel Laboratorium GmbH, Germany. Pheophorbide-a was suspended in Physiogel^® (0.1% by weight) to yield a molar concentration of $1.7·{10}^{-3}\mathrm{M}$.

2.2.

Luminescence Detection Setup

Singlet oxygen luminescence was detected using photon counting as described in Ref. 7. The excitation source comprises a frequency-doubled diode pumped $\mathrm{Nd}{3}^{+}\text{-}\mathrm{YAG}$ laser and a customized dye laser. For all spatially resolved measurements, the excitation was done at 666 nm, in the lowest-energy absorption band of pheo. The light intensity for excitation was maintained at 11 mW for the setup as in Fig. 1(a) and at 2 mW for the setup as in Fig. 1(b), measured with a LabMax/J 10MT- 10 kHz (Coherent, Germany). The repetition rate was 12.2 kHz and the pulse length 10 ns.

Fig. 1

(a) Schematic of the luminescence acquisition setup and the geometry of the flat fiber-tip. (b) Schematic of the optics used for high-resolution scanning.

The NIR luminescence was detected with a photomultiplier, H-10330-45 (Hamamatsu Photonics, Germany), sensitive from 950 to 1400 nm with a nominal quantum efficiency of 2% at 1270 nm. An $f=0.6$ lens transfers collimated light onto the effective area of the anode (1.6 mm diameter). The rise time of 900 ps and the transit time spread of 300 ps may be neglected in the ${}^1{\mathrm{O}}_2$ luminescence decay time domain. The NIR ${}^1{\mathrm{O}}_2$ luminescence is discriminated using an interference filter with a central wavelength at 1270 nm. ${}^1{\mathrm{O}}_2$ emission is verified by comparing to measurements with filters at 1200 and 1300 nm. The high sensitivity of the setup allows the determination of rise and decay times with an accuracy and reproducibility of 0.1 μs in many biological samples.⁷ Larger error margins are given for absolute values due to sample variance.

For the measurements reported here, the ${}^1{\mathrm{O}}_2$ luminescence was detected through an optical fiber. Two approaches were made. One comprises a tri-furcated fiber, with a flat fiber-tip [Fig. 1(a)]. The other comprises the fiber for detection perpendicular to the excitation; excitation and detection are focused on the sample and separated by a dichroic mirror.

The flat fiber-tip comprises separate fibers for excitation and detection of fluorescence and ${}^1{\mathrm{O}}_2$ luminescence, each connected to one leg of the tri-furcated fiber. The fibers of all three legs are arranged side by side in the fiber-tip. The flat design without further optics results in a tip that is easy to handle. The signal maximum is found at a short distance of a few millimeters above the sample surface. This can be expected for this probe geometry and scattering samples such as skin as described in Ref. 32. The signals obtained with this setup result from averaging over a certain area of the sample, but offer only low spatial resolution.

The PS fluorescence was detected using a fiber spectrometer, C10083CA (Hamamatsu). The excitation light was discriminated using a cut-on filter with 550 nm cut-on wavelength for measurements with 532 nm excitation wavelength and one with 700 nm cut-on wavelength for measurements with 666 nm excitation wavelength.

The second setup, as in Fig. 1(b), comprises a single core fiber for the ${}^1{\mathrm{O}}_2$ detection; excitation light is delivered perpendicular to the ${}^1{\mathrm{O}}_2$ detection. A dichroic mirror separates the wavelengths. The lenses allow focusing the laser spot on the sample; the spot is as narrow as 0.2 mm in diameter. Thus, higher resolution scanning is possible with a resolution limit determined by the laser spot and the scattering properties of the investigated sample.

2.3.

Singlet Oxygen Kinetics

The time-dependency, ${}^1{\mathrm{O}}_2(t)$, of the amount of PS-generated ${}^1{\mathrm{O}}_2$ in a homogeneous environment is simplified as described by Eq. (1):

On the microsecond time scale, the PS molecules can be assumed to reach the triplet state instantaneously following a nanosecond laser pulse excitation. The fitting constant C accounts for the number of initially excited PS molecules, while ${\tau }_{\mathrm{\Delta }}$ and ${\tau }_T$ represent the ${}^1{\mathrm{O}}_2$ and PS triplet state decay times, respectively. The sign of the pre-exponential factor can change depending on the relative values of the two decay times; a signal rise is determined mainly by the faster process and the decay by the slower one. Hence, the firm assignment of PS triplet and ${}^1{\mathrm{O}}_2$ decay times to the fitting parameters of the luminescence signal requires additional experiments.¹⁰ In vitro the PS triplet lifetime can be determined directly by flash photolysis³³ or indirectly by the addition of quenchers in variable concentrations followed by the monitoring of the changes in ${}^1{\mathrm{O}}_2$ decay time. Since these methods cannot be applied here, the samples were exposed to pure ${\mathrm{N}}_2$ atmosphere and the change in kinetics under readmission of ${\mathrm{O}}_2$ was evaluated. Considering the high ${}^1{\mathrm{O}}_2$ quantum yields, the PS triplet decay time is mainly determined by the efficiency of the energy transfer to ${\mathrm{O}}_2$. Thus, reducing the ${\mathrm{O}}_2$ concentration leads to an increased PS triplet decay time while the ${}^1{\mathrm{O}}_2$ decay time remains constant, allowing a quantitative differentiation between the two processes and assigning of ${\tau }_{\mathrm{\Delta }}$ and ${\tau }_T$ to the fitted parameters.³⁴ For skin samples, as investigated previously, the shorter time reflects the triplet decay, while the longer time can be attributed to ${}^1{\mathrm{O}}_2$ decay.⁸ It should be noted that Eq. (1) is strictly applicable only for homogenous environments. However, the model is widely accepted and used here for a first evaluation of the data. Due to the signal artifact from the skin samples at times shortly after the laser pulse, the first 1.6 μs were excluded from the fitting procedure.

3.1.

Fiber Influence on Detection of Singlet Oxygen Luminescence Kinetics

It is known that some common glass types may exhibit luminescence in the NIR, e.g., when glass cuvettes are used. Since the glass emission causes artifacts in time-resolved ${}^1{\mathrm{O}}_2$ luminescence measurements, we evaluate the influence of the light transmission through the optical fiber to exclude the possibility of signal distortion by the influence of the fiber optics. For this experiment, the setup as in Fig. 1(a) is used; the fiber length is 2 m.

To investigate the influence of the fiber on the ${}^1{\mathrm{O}}_2$ luminescence kinetics, we use the PS in solution—pheo in ethanol. The kinetics of this system are described by Eq. (1) exactly. The solution of pheo was put in a cuvette and measured directly with the setup described previously.⁷ We compare that measurement with the measurements using the fiber. Therefore, the fiber-tip was dipped in the cuvette to obtain the luminescence signal via the fiber. For the determination of errors, 100 identical measurements were performed, fitted, and the standard deviation of the parameters was determined. The ${\chi }^2$-test, respectively, the reduced ${\chi }^2$ value, is used to quantify the goodness-of-fit. The luminescence signals obtained from measurements of pheo in ethanol in a cuvette without using the optical fiber can be fitted with a reduced ${\chi }^2$ value of 1.00, proving a perfect congruence of theory and experiment.

The measurements shown in Fig. 2 were fitted with $0.18\pm 0.02\mu \text{s}$ (cuvette) and $0.17\pm 0.02\mu \text{s}$ (fiber) triplet decay time and $13.8\pm 0.1\mu \text{s}$ (cuvette) and $13.6\pm 0.1\mu \text{s}$ (fiber) for the ${}^1{\mathrm{O}}_2$ decay time. The reduced ${\chi }^2$ value obtained for the measurement using the fiber is 0.99, also proving a very good correlation of theory and experimental result. Considering the fitted parameters and the errors, a possible influence of the fiber on the kinetics of the ${}^1{\mathrm{O}}_2$ luminescence signal is below the measurement error.

Fig. 2

Singlet oxygen luminescence signal photosensitized generated by pheophorbide-a in ethanol. Dark blue dots: measurement in cuvette without optical fiber; light red squares: measurement through the optical fiber with the setup as shown in Fig. 1(a). Solid black line: two exponential fit of the data. The fitted parameters are $0.18\pm 0.02\mu \text{s}$ (cuvette) and $0.17\pm 0.02\mu \text{s}$ (fiber) triplet decay time and $13.8\pm 0.1\mu \text{s}$ (cuvette) and $13.6\pm 0.1\mu \text{s}$ (fiber). The reduced ${\chi }^2$ values are 1.00 (cuvette) and 0.99 (fiber).

3.2.

Singlet Oxygen Luminescence Detection in Skin

Samples of pig ear skin were investigated using the setup as in Fig. 1(a). The sample of pig ear skin was prepared as shown in Fig. 3(a): the upper half of the sample was prepared with PS-crème according to the procedure described in Sec. 2.1; the lower half was delimited using a fixed tooth pick to prevent massaging the PS-crème in the part without PS and thus avoid the penetration of the PS in the skin. The investigated area on the sample was marked with black dots to allow a correlation of the photo and the luminescence measurement. The fiber-tip was fixed above the sample at the distance of the signal maximum. The sample was moved pixel by pixel to scan the area. The fluorescence spectrum was integrated for 100 ms per pixel and the ${}^1{\mathrm{O}}_2$ luminescence for 10 s.

Fig. 3

(a) Photo of the sample; the black dots mark an area of $\sim 11.2$ by $6.5\mathrm{m}{\mathrm{m}}^2$. (b) Fluorescence spectrum of pheophorbide-a in ethanol. (c) Normalized pheophorbide-a fluorescence intensity at 720 nm in skin on the area between the black dots in (a). Excitation was at 666 nm, with 11 mW and 100 ms integration time per pixel. (d) Typical skin fluorescence spectra acquired with 532-nm excitation at a location with pheophorbide-a (black curve) and a location without (red curve). For the better display of the fluorescence spectra in (b) and (d), 532 nm was chosen as excitation wavelength and a 550-nm long-pass filter was used for discrimination of the excitation light. The scanning was done using excitation light at 666 nm, as this is the relevant excitation wavelength for this photosensitizer in, e.g., photodynamic therapy applications.

The characteristic fluorescence spectrum of pheo in solution is shown in Fig. 3(b). This can be clearly identified in the skin sample, on the area that was treated with pheo in crème [Fig. 3(c) and spectra in Fig. 3(d)]. For better display of the fluorescence spectra in Figs. 3(b) and 3(d), 532 nm was chosen as the excitation wavelength and a 550-nm long-pass filter was used for discrimination of the excitation light. The skin itself shows everywhere some broad background luminescence, which does not hinder the identification of the pheo fluorescence.

In Fig. 3(c), the fluorescence intensity map of the sample is displayed. The 720-nm emission of pheo is quantified by integrating the fluorescence spectrum from 710 to 730 nm. The effect of background luminescence of the skin was neglected. The black dots used for position marking appear nonfluorescent. The fluorescence intensity map shows that the pheo emission can be clearly attributed only to half of the sample that was prepared with PS-crème, while the untreated area does not fluoresce beyond the background fluorescence.

Figure 4(a) shows the ${}^1{\mathrm{O}}_2$ luminescence intensity distribution across the sample, which is congruent with the PS fluorescence intensity distribution of the sample shown in Fig. 3(c). The ${}^1{\mathrm{O}}_2$ luminescence was acquired time-resolved for every pixel. Figure 4(b) shows the ${}^1{\mathrm{O}}_2$ luminescence kinetics for a $5\times 5\text{pixel}$, which is a detail of Fig. 4(a). The central pixel is fitted [Fig. 4(c)] with the biexponential model [Eq. (1)], yielding a rise time of $0.4\pm 0.2\mu \text{s}$ and a decay time of $13.6\pm 0.5\mu \text{s}$. Despite differences in the measurement techniques, the results obtained via scanning match our previously published kinetics where a rise time changing from $0.3\pm 0.2$ to $0.9\pm 0.2\mu \text{s}$ and a decay time changing from $12.5\pm 0.5$ to $19.2\pm 0.5\mu \text{s}$ with illumination of $20\mathrm{J}/\mathrm{c}{\mathrm{m}}^2$ at 666 nm was found.⁸ It should be noted that those kinetics were obtained by averaging over large and not precisely controlled areas but offered the ability to resolve the change of kinetics with illumination. Currently, the scanning procedure does not allow precise tracking of the change of the ${}^1{\mathrm{O}}_2$ kinetics with illumination, because during scanning, it cannot be avoided that the whole sample is illuminated, mainly due to scattering of the sample itself. This illumination may cause changes of the kinetics already before a certain pixel is measured.

Fig. 4

Time-resolved singlet oxygen luminescence signals from pig ear skin, acquired through a fiber, with 11 mW excitation power at 666 nm and 10 s integration time per pixel. (a) Normalized singlet oxygen luminescence intensity on the area between the black dots in Fig. 3(a). (b) Grid of the singlet oxygen luminescence kinetics of the $5\times 5\text{pixel}$ area marked in (a). Axes and scales were omitted for presentation (each $x$ axis shows 35 μs full-scale, the $y$ axis 400 counts full-scale). (c) Singlet oxygen luminescence kinetics of the central pixel in (b), fitted according to the biexponential model. The signal rise time is 0.4 μs, and the decay time is 13.6 μs.

3.3.

Spatially Resolved Singlet Oxygen Kinetics in Skin

A sample of pig ear skin prepared with pheo as described in Sec. 2.1 was scanned at high resolution using the focusing setup as in Fig. 1(b). A scanned area of $\sim 4\mathrm{mm}$ by 4 mm is shown in Fig. 5. The intensity distribution [Fig. 5(a)] clearly shows several distinct maxima of the ${}^1{\mathrm{O}}_2$ luminescence intensity. A correlation with a photo [Figs. 5(b) and 5(c)] of the scanned area suggests that the ${}^1{\mathrm{O}}_2$ luminescence intensity maxima correlate with the macrostructure of the skin. Likely, the topography of the skin allows a higher accumulation of PS at certain locations. Structures like hair follicles and furrows seem to pose accumulation sites for the PS and show higher ${}^1{\mathrm{O}}_2$ luminescence intensity.

Fig. 5

(a) Normalized singlet oxygen luminescence intensity map. The size of the scanned area is $\sim 4\times 4\mathrm{m}{\mathrm{m}}^2$. (b) Photo of the scanned area. The black dots on the photo are markings to correlate the scanning area with the location on the photo. (c) Overlay of (a) and (b).

Even though it could be assumed that the follicular orifice poses a location for the accumulation of the PS-crème, the luminescence kinetics allows negating this assumption. The ${}^1{\mathrm{O}}_2$ luminescence kinetics differ from the kinetics that would be expected for the PS in the crème. The luminescence kinetics in the skin sample show a maximum of the ${}^1{\mathrm{O}}_2$ decay time of 19 μs at the presumed location of follicles. Previously reported kinetics of ${}^1{\mathrm{O}}_2$ luminescence in crème yield ${}^1{\mathrm{O}}_2$ decay times of $>23\mu \text{s}$.³⁵ Taking into account the short diffusion range in the order of 100 nm, this leads to the conclusion that ${}^1{\mathrm{O}}_2$ is interacting with the skin. A significant accumulation of the PS-crème within the comparably large follicular orifice would show a much longer decay time—similar to crème. Cryo-biopsies of pig ear skin samples prepared with pheo support this hypothesis. The samples show an accumulation of pheo in the hair follicle as well as in the stratum corneum [Fig. 6(b)].

Fig. 6

(a) Singlet oxygen decay times in microsecond of the scanned area in Fig. 5. (b) Fluorescence microscopy image of a cryo-biopsy of pig ear skin prepared with pheophorbide-a in crème. The fluorescence of pheophorbide-a is shown in red and originates from within the stratum corneum as well as the hair follicle.

Present results still do not allow to clearly differentiate whether the higher ${}^1{\mathrm{O}}_2$ luminescence at the hair follicles is merely an effect of the much higher surface area due to the geometry of the hair follicle or whether the hair follicle provides a preferred accumulation site for pheo due to different cell populations within the hair follicle.³⁶ A quantitative analysis of the ${}^1{\mathrm{O}}_2$ luminescence intensity, which could help elucidate this effect, is technically difficult due to the scattering processes within the skin. As well, a correlation with the increased surface area at the hair follicle will be practically impossible since the geometry of the specific hair follicle is not known.