Safety Assessment by Multiphoton Fluorescence/Second Harmonic Generation/Hyper-Rayleigh Scattering Tomography of ZnO Nanoparticles Used in Cosmetic ProductsDarvin M.E.a · König K.b · Kellner-Hoefer M.b · Breunig H.G.b · Werncke W.a · Meinke M.C.a · Patzelt A.a · Sterry W.a · Lademann J.a
aCenter of Experimental and Cutaneous Physiology (CCP), Department of Dermatology, Venereology and Allergology, Charité-Universitätsmedizin Berlin, Berlin, and bJenLab GmbH, Jena, Germany
Zinc oxide nanoparticles (ZnO NPs) are commonly used as UV filters in commercial sunscreen products. Their penetration into the skin is intensively discussed in the literature. In the present in vivo study, penetration of ZnO NPs (30 nm in size) into human skin was investigated by multiphoton tomography. Based on the non-linear effects of a second harmonic generation and hyper-Rayleigh scattering, the distribution of ZnO NPs in the horny layers of the epidermis, as well as the furrows, wrinkles and orifice of the hair follicles was analyzed. This method permitted distinguishing between the particulate and dissolved forms of Zn. A detection limit of 0.08 fg/µm3 was estimated. Taking advantage of this sensitivity, it was clearly shown that ZnO NPs penetrate only into the outermost layers of stratum corneum, furrows and into the orifices of the hair follicles and do not reach the viable epidermis.
Copyright © 2012 S. Karger AG, Basel
The whole spectrum of sun radiation can provoke skin injury, which is mostly associated with the action of free radicals on living cells [1,2,3,4]. The UVB-UVA ranges of the spectra are responsible for the formation of free radicals in human skin  and, thus, the amount of UV light reaching the skin should be reduced. The nanoparticles (NPs) whose size is less than 200 nm have higher efficiency in scattering UV light than microparticles. Modern sunscreens contain photostable broadband UV filters in the form of the nano-structured metal oxides titanium dioxide (TiO2) and zinc oxide (ZnO), providing protection against UV radiation [5,6]. Additionally to prevent a metal oxide-induced whitening of the skin and, thus, providing a cosmetically desired appearance of the skin, ZnO and TiO2 are used in the form of nano-sized particles ranging in size from 20 to 200 nm . The action of modern sunscreens is based on the absorption of UV photons by filter substances, reflection and scattering of photons by NPs such as TiO2 and ZnO, and neutralization of free radicals by antioxidants [8,9,10,11,12].
The introduction of NPs in cosmetic products used for sun protection is controversially discussed in the literature [6,13]. On the one hand, NPs offer new perspectives for drug delivery, since it could be demonstrated that they penetrate more efficiently into the hair follicles of human skin and can be stored here one magnitude longer than nonparticulate substances . On the other hand, there is a suspicion that sunscreen NPs can pass through the skin barrier, accumulating in the human organism with harmful consequences [15,16,17]. The most probable pathway for the delivery of NPs through the stratum corneum (SC) was proposed to be the lipid layers between the corneocytes . The main cytotoxicity of ZnO NPs has been attributed to the generation of free radicals (mainly via hydroxyl radicals formed through oxidation), which was shown in vitro in cell experiments [19,20].
In the past, several in vivo human studies have been performed on TiO2. These studies were able to demonstrate that TiO2 NPs at a size of ≥30 nm do not penetrate the intact skin barrier and do not reach viable layers of the epidermis [21,22,23]. However, if the skin is damaged by interaction with chemical irritants or mechanical procedures, various amounts of TiO2 NPs can be detected in the living tissue [6,15,24,25]. The detection of TiO2 NPs in biological tissue is relatively easy when sensitive detection methods are applied because TiO2 is not a natural compound of the human organism. Based on the results of these studies, a risk assessment was undertaken, as a result of which TiO2 was accepted by the European Union Committee on Cosmetics and Non-Food Products for use in cosmetic products as a nontoxic compound entailing no adverse effects on the skin in the year 2000 .
Taking into consideration the results obtained with TiO2, it can be expected that ZnO NPs possess similar penetration properties, i.e. they do not pass through the intact skin barrier if the size of particles is similar. The investigations concerning the safety of ZnO NPs are more complicated compared to those of TiO2. Traces of ZnO are also present in the human organism in dissolved form . Consequently, low concentrations of dissolved ZnO, i.e. Zn2+ ions, should be harmless, whereas the toxicity of ZnO NPs cannot be completely assessed. So far, it has not been possible to distinguish between ZnO traces in nanoparticulate form from those in a dissolved form in human tissue. Attempts to detect differences in ZnO concentrations before and after sunscreen application in the living skin by fluorescence lifetime imaging measurements of ZnO have been performed [27,28]. The extremely short fluorescence lifetime of ZnO (around 100 ps)  was measured at levels near the detection limit of fluorescence lifetime imaging measurements.
Many studies conducted in vitro on human and porcine skin showed the limited penetration of TiO2 and ZnO NPs to the upper layers of the SC and the absence of penetration into the living skin [30,31,32]. Penetration into the hair follicles was also reported [22,33].
Dai et al.  showed that ZnO could be reliably detected by the measurement of photoluminescence induced by multiphoton absorption of ZnO under excitation by a femtosecond near-infrared laser. The first in vivo observations that ZnO does not penetrate through the intact skin barrier were reported by Zvyagin et al. , who analyzed the penetration of ZnO (26–30 nm mean size) into the human skin by multiphoton microscopy. The ‘principal photoluminescence’ of ZnO at 385 nm, which originated at an excitation wavelength of 320 nm (740 nm two-photon excitation), was found to be in the range of 380 ± 10 nm. The detection limit of this method was not evaluated by the authors and the possibility to detect the small concentrations of ZnO in human tissue was not determined. Lin et al.  showed that there was no penetration of ZnO NPs ranging from 10 to 50 nm in diameter into viable human skin in vivo using time-correlated single-photon counting, with the detection limit being limited by two-photon excited skin autofluorescence.
ZnO NPs are able to generate coherent as well as incoherent radiation at half the excitation wavelength, known as second harmonic generation (SHG) and hyper-Rayleigh scattering (HRS), respectively, which allows us to distinguish between the natural components of the skin and ZnO NPs. Both types of scattering have been reported to occur for ZnO NPs at considerable intensities [36,37]. They can be identified and distinguished from autofluorescence because of three basic properties: (i) SHG as well as HRS signals are generated instantaneously, i.e. without any temporal delay with respect to the excitation pulse. (ii) SHG/HRS signals are generated just at half the wavelength of the excitation pulse, i.e. without any spectral red shift from this position, the latter being typical of autofluorescence. (iii) The spectral width of the signal approximately corresponds to the width of the excitation pulse. Usually it will be much narrower than for autofluorescence. As in human skin, only collagen, which is located below 50 µm depth, is able to generate SHG [38,39], the observation of SHG/HRS is highly specific for ZnO NPs, at least in the upper layers of the skin. The high fluorescence background signal of the SC [40,41], as well as the presence of dissolved Zn in human tissue , do not influence the SHG/HRS measurements of ZnO NPs.
Materials and Methods
The experiments were performed on the forearms of 6 healthy female volunteers, aged between 26 and 45 years. All volunteers had skin type II according to the Fitzpatrick classification . All volunteers participating in the study had normal skin without visible abnormalities, such as extremely dry or fatty skin, wounds, skin ‘defects’ or skin diseases. Approval for these measurements had been obtained from the Ethics Committee of the Charité-Universitätsmedizin, Berlin.
A basic sunscreen formulation not containing organic UV filters was used in the experiments. Three different concentrations of ZnO NPs exhibiting a main size of 30 nm (Zinkoxid NDM, Symrise AG) were incorporated into the formulation (weight concentrations of 8, 0.8, and 0.008%). Two ZnO NP concentrations (0.08 and 8%) were chosen to simulate real-life conditions by the application of sunscreens (usual concentration of NPs in sunscreens is around 1–3% and normally does not exceed 5%). The smallest concentration of ZnO NPs (0.008%) was chosen to determine the detection limit of the measuring method used.
The investigations were carried out with a two-photon tomograph (JenLab GmbH, Jena, Germany) equipped with a femtosecond titanium sapphire laser (Mai Tai XF; Spectra Physics, USA). The laser that was spectrally tunable between 710 and 920 nm generated 100-fs pulses at a repetition rate of 80 MHz. For studying the cellular structure in different depths of the skin by autofluorescence [44,45,46] and by SHG/HRS detection, two detection schemes – spectrally resolved broadband detection with a CCD camera and two-channel detection with two photo multipliers – were applied. Employing a fiber-based spectrometer equipped with a CCD camera (BTC112E from B&W Tek, Inc.) with a spectral coverage of 300–1,050 nm, the spectral profile generated by ZnO formulations applied to the skin was analyzed. Under the conditions of broadband detection, the radiation collected from different depths was averaged over slices of 250 × 250 µm2. SHG/HRS signal detection of ZnO applied to the human skin was optimized by spectrally shifting the laser wavelength. As a result, an excitation wavelength of 760 nm was chosen for further investigations. For multiphoton tomography with a lateral and axial resolution of 0.5 and 2 µm, respectively, two-channel detection was used . To distinguish between the spectrally broad two-photon fluorescence (also called autofluorescence) of the skin and the spectrally narrow SHG/HRS signals originating from ZnO, a broadband filter transmitting between 300 and 700 nm and an interference filter (transmission at 380 nm with a spectral width of 14 nm, full width at half maximum), respectively, were used in front of the photo multipliers.
During measurements, the laser focus was moved from the skin surface into deeper parts of the skin in 5-µm increments. The maximum measuring depth into the skin was 200 µm, on account of the limited penetration depth of the laser radiation.
The cellular skin structure and the number of ZnO NPs applied were detected by repositioning the laser focus, i.e., at different skin depths. In in vitro measurements, the laser power usually does not exceed 5 mW in the focus, which was sufficient for detection of SHG/HRS signals of ZnO. For in vivo measurements in the skin, the power was increased, not exceeding 15 mW in the focus.
The detection limit for ZnO was estimated from the signal-to-noise ratio obtained for the 0.008% ZnO formulation that was applied to the surface of the skin.
A dose of 2 mg/cm2 of the three formulations, containing different concentrations of ZnO NPs, was applied to three marked skin areas (size: 4 × 4 cm2) on the forearms of the volunteers. The formulations were rubbed into the skin for the duration of 1 min, which is close to real-life conditions when applying sunscreens on the beach. The position of the skin areas treated with the different formulations was randomized. Then the formulations were left on the skin surface for 60 min for passive penetration. This relatively long penetration time was necessary to simulate the actual conditions on the beach, where the sunscreen is in contact with the skin for many hours. Additionally, it was ensured that the formulation completely penetrated into the skin and that no supernatant remained. Two-photon measurements were initialized 60 min after application of the formulations. Before the measurements started, the skin surface was wiped with filter paper.
At the stage of preinvestigations in in vitro measurements, porcine ear skin was pretreated with 8% ZnO-containing formulation.
Spectrally resolved broadband detection offers the possibility of distinguishing between the SHG/HRS signal and fluorescence generated by the ZnO NPs after two-photon excitation. In figure 1, a spectrum of the topically applied formulation with 8% of ZnO on the porcine skin in vitro, obtained at 760 nm excitation, is presented. A prominent peak with a bandwidth comparable to the width of the excitation radiation is observed at 380 nm – half of the excitation wavelength. By spectral tuning of the excitation wavelength, the SHG/HRS signal is spectrally shifted too, always remaining at a position of half of the excitation wavelength (data not shown). Our observations clearly indicate that the signals measured from ZnO originate from SHG/HRS, whereas two-photon excited fluorescence is not observed.
|Fig. 1. SHG/HRS signal generated at 380 nm by ZnO NPs (8%) applied to porcine ear skin in vitro under two-photon excitation at 760 nm.|
In figure 2a, an SHG/HRS image of the ZnO NPs recorded at 760 nm excitation after topical application of the 2 mg/cm2 formulation with 0.008% of ZnO NPs to the human skin surface in vivo is presented. The signal of the ZnO NPs could be clearly detected. The image shows a cloudy structure on the skin surface. Occasionally, aggregates can be recognized, which appear as small intensely luminous spots (see white spots in fig. 2).
|Fig. 2. In vivo SHG/HRS images (white spots) of the distribution of cosmetic formulations containing 0.008% ZnO NPs (a) and 8% ZnO NPs (b) applied topically to human skin (2 mg/cm2). Two-photon excitation wavelength 760 nm; ZnO-induced SHG/HRS signals measured at 380 nm. Nebular structure due to ZnO NPs as well as several high-intensity spots due to aggregates of ZnO NPs are observed.|
As scattering from NPs with a signal-to-noise ratio of about 10 is observed, a detection limit lower than 0.08 fg/µm3 can be estimated for NPs in the upper layers of the skin. In figure 2b, the distribution of 2 mg/cm2 of the formulation containing 8% ZnO on the skin surface is presented. It can be observed that the skin is totally covered with the NPs, which results in a high SHG/RHS signal. Sixty minutes after application, the NPs were wiped off from the skin surface with filter paper. After the ZnO NP-containing formulation (8%) had been wiped off, mainly aggregates of the NP remained on the skin surface, which are visible as small bright white spots in the images. Similar results were observed for all volunteers participating in the study.
Images of autofluorescence showing the cellular structure (broadband detection) and of the SHG/HRS signal demonstrating the distribution of ZnO NPs (narrowband detection) on the skin (8%) are compared in figure 3. Using the broadband filter, autofluorescence images of the skin, which were obtained from different depths, are presented in the left column of figure 3. In the right column, the SHG/HRS images of ZnO NPs are shown. Without treatment with ZnO NPs, any SHG/HRS signal is lacking. Using broadband detection, the different skin layers can be well recognized because of their characteristic cellular structure when the laser focus is moved into the tissue. After topical application and subsequent wiping off of the ZnO (8%)-containing formulation, the penetration of NPs was investigated. SHG/HRS signals, which strongly decreased with depth, were observed in the lipid layers of the first and second corneocyte layers. Moving the focus deeper into the living tissue they completely vanish, but remain visible in the furrows and wrinkles of the skin and in the orifices of hair follicles (fig. 3, 4) up to a depth of 120 µm. The same penetration profiles of ZnO NPs were observed for all volunteers participating in the study.
|Fig. 3. Depth-dependent in vivo images of two-photon-induced skin autofluorescence (left) and of the SHG/HRS signal of ZnO NPs (right). Human skin is pretreated for 60 min with a cosmetic formulation containing 8% ZnO NPs before the measurements were performed. Obviously ZnO NPs are located in the furrows and wrinkles only and do not penetrate into the living skin.|
|Fig. 4. Penetration of ZnO NPs into the orifice of a hair follicle (depth 30 µm). In vivo images of two-photon-induced skin autofluorescence (a) and of the SHG/HRS signal of ZnO NPs (b). The latter indicate the presence of ZnO NPs only inside the orifice of the hair follicle, but not inside the living skin.|
The autofluorescence signals of the SC and of the applied ZnO NPs are located in the same optical range, making selective detection rather difficult even if temporal resolution of fluorescence decay is applied by time-correlated single photon counting [27,48]. To solve this problem it should be taken into consideration that various particles (crystals) including ZnO NPs possess a distinct structure, allowing an SHG/HRS signal to be generated at half the wavelength of the excitation radiation [36,49]. The SHG/HRS signal is produced only by the ZnO NPs, but not by the dissolved Zn available in the cells. Therefore, the SHG/HRS measurements can be utilized as a sensitive tool for the detection of ZnO NPs. To some extent, these nonlinear optical signals can be distinguished also from autofluorescence of the tissue by time-correlated single photon counting because of their instantaneous temporal response to the excitation pulses. Even more efficiently they can be selected spectrally at half of the wavelength of the excitation pulse, using the corresponding narrowband bandpass optical filter. In this case, SHG/HRS can be easily distinguished from the broadband autofluorescence signal originating from the tissue.
Our experimental results demonstrate strong SHG/HRS generation of ZnO NPs, whereas two-photon-induced fluorescence can be neglected (fig. 1). These findings are in contrast to previous measurements reported by Zvyagin et al. . According to our measurements, these authors erroneously assigned a corresponding radiation observed in the range of half the excitation wavelength to two-photon-induced fluorescence of the ZnO NPs, which was probably due to the fact that their experiments lacked spectrally resolved measurements. Fluorescence spectral profiles of ZnO NPs have a relatively bright prominent maximum red shift with respect to the electronic transition . In a recent paper, Song et al.  reported a detection limit of about 200–250 NPs of ZnO obtained under their experimental conditions optimized for fluorescence detection, which corresponded to a concentration of 5–7 fg/µm3. In contrast, our experiments show that a carefully optimized detection of SHG/HRS allows a detection limit of 0.08 fg/µm3 for ZnO NPs to be achieved. This high sensitivity is sufficient to determine the ZnO NPs in the different layers of the epidermis and to estimate the maximum amount of penetrating ZnO that could still reach the living epidermis.
As demonstrated in figure 3, the ZnO NPs (8%) are located on the skin surface but only to a low extent, as well as around the first and second layers of corneocytes. On account of desquamation, the cellular structure of these two upper cell layers is loosened, so that part of the NPs can enter these layers. When the NP-containing formulation was wiped off 60 min after application, mainly aggregates remained on the skin surface as seen in figure 3. It is well known that ZnO NPs form aggregates in a formulation, if they are incorporated at a concentration of several percent [51,52]. This is in agreement with published results, according to which particles and aggregates more than 100 nm in size stick efficiently to the skin surface as they deposit well in the furrows and wrinkles. Such aggregates do not pass through the intact skin barrier. The aim of the investigations was to check whether the nonaggregated NPs, which had been in contact with the skin for 1 h, would penetrate into the living epidermis. The nonaggregated ZnO NPs occur in contrast to the prominent light spots of the aggregates as a light diffuse nebula.
Moving the focus deeper into the skin, the SHG/HRS signal of the ZnO NPs disappeared extremely quickly in the SC. In deeper parts of the SC and in the living tissue, no traces of ZnO NPs could be found within our detection limit. Only in the furrows and wrinkles and in the orifices of hair follicles could an SHG/HRS signal of the ZnO NPs be detected. The furrows and wrinkles had a maximum depth of approximately 80 µm on the skin areas under investigation, i.e. in a range considerably smaller than the range accessible to our measurements. No ZnO NPs could be detected outside the furrows and wrinkles. Thus, it is clear that no ZnO NPs penetrate into the living tissue from the bottom of the furrows and wrinkles. A similar situation was observed for the hair follicles. Here, the ZnO NPs penetrated up to approximately 120 µm into the orifices of hair follicles. Again, no SHG/HRS signal of the ZnO NPs was found in the area outside the hair follicles in living tissue. In figure 2a, it is demonstrated that also traces of the ZnO NPs can be detected on the skin surface by multiphoton tomography. The detection limit of 0.08 fg/µm3 was calculated, which corresponded to the detection of 0.03 fg ZnO NPs in the focus of the measuring system.
These results are important for tape stripping-based investigations, because not all NPs could be removed by tapes from the furrows, wrinkles and orifices of hair follicles at a depth of 120 µm, and residual NPs could be mistakenly interpreted as being in the living skin.
If amounts less than 0.08 fg/µm3 of the applied NPs pass through the skin barrier, the human organism can be expected to dissolve these small amounts, so that an accumulation of harmful concentrations in the human body can be excluded.
The results obtained in the present study demonstrate that ZnO NPs at a size of approximately 30 nm possess similar penetration properties as described in the literature for TiO2 [22,53]. These NPs do not pass the skin barrier, signifying that they do not present any danger. These results, however, are only valid in the case of an intact skin barrier, which corresponds to the application instructions for sunscreens.
Multiphoton tomography is a well-suited method for the measurement of penetration of different formulations into the skin, including NPs. Penetration of ZnO NPs, which are widely used in cosmetic sunscreen products, can be monitored because of their ability to generate SHG and HRS signals. Based on these types of nonlinear optical scattering, the detection limit for ZnO NPs applied to the human skin surface at a concentration lower than 0.08 fg/µm3 was obtained. This is the first in vivo study on human skin showing such a detection limit for ZnO NPs.
It was shown that topically applied ZnO NPs (30 nm in size) stay on the skin surface and do not penetrate into the viable layers of the epidermis. Penetration was observed only into the first layers of the SC. ZnO NPs were also detected in the furrows, wrinkles and orifices of hair follicles of the skin, where they remain and do not leave the horny barrier.
Our results conclusively demonstrate that – in view of penetration – application of sunscreens containing ZnO NPs at the size used in the study is harmless to human skin.
We would like to thank the Beiersdorf Co., Hamburg, for the preparation of the sunscreen samples utilized in the present study. Furthermore, we thank Prof. Michael Roberts from the School of Pharmacy and Medical Sciences and Sansom Institute for Health Research, University of South Australia, for a stimulating and helpful discussion and the Foundation ‘Skin Physiology’ of the Donor Association for German Science and Humanities for financial support.
Dr. Maxim E. Darvin, PhD, Center of Experimental and Cutaneous Physiology (CCP)
Department of Dermatology, Venereology and Allergology
Charitéplatz 1, DE–10117 Berlin (Germany)
Tel. +49 30 450 518 208, E-Mail firstname.lastname@example.org
Received: December 29, 2011
Accepted after revision: April 17, 2012
Published online: May 30, 2012
Number of Print Pages : 8
Number of Figures : 4, Number of Tables : 0, Number of References : 53
Skin Pharmacology and Physiology (Journal of Pharmacological and Biophysical Research)
Vol. 25, No. 4, Year 2012 (Cover Date: June 2012)
Journal Editor: Lademann J. (Berlin)
ISSN: 1660-5527 (Print), eISSN: 1660-5535 (Online)
For additional information: http://www.karger.com/SPP