High‐energy visible light at ambient doses and intensities induces oxidative stress of skin—Protective effects of the antioxidant and Nrf2 inducer Licochalcone A in vitro and in vivo

Abstract Background Solar radiation causes skin damage through the generation of reactive oxygen species (ROS). While UV filters effectively reduce UV‐induced ROS, they cannot prevent VIS‐induced (400‐760 nm) oxidative stress. Therefore, potent antioxidants are needed as additives to sunscreen products. Methods We investigated VIS‐induced ROS formation and the photoprotective effects of the Nrf2 inducer Licochalcone A (LicA). Results Visible spectrum of 400‐500 nm dose‐dependently induced ROS in cultured human fibroblasts at doses equivalent to 1 hour of sunshine on a sunny summer day (150 J/cm2). A pretreatment for 24 hours with 1 µmol/L LicA reduced ROS formation to the level of unirradiated cells while UV filters alone were ineffective, even at SPF50+. In vivo, topical treatment with a LicA‐containing SPF50 + formulation significantly prevented the depletion of intradermal carotenoids by VIS irradiation while SPF50 + control did not protect. Conclusion LicA may be a useful additive antioxidant for sunscreens.

on the effects of infrared (IR) light have resulted in conflicting effects and conclusions [2][3][4][5] ; physiological doses are mostly deemed non-hazardous 6,7 and even may be beneficial. 8 The intermediate visible spectrum (VIS 400-760 nm) was long regarded harmless although more than 50% of solar radiation reaching the surface of the earth is visible light with the global spectral irradiance at a latitude of 45°N at solar noon peaking at around 500 nm. 9 Kielbassa et al, 10 on the other hand, had reported oxidative DNA damage induced by short-wave VIS (400-450 nm) in Chinese hamster cells already in 1997.
In contrast to dermatology, the clinical relevance of high-energy visible light (HEVIS) induced ROS effects on ocular tissues, in particular the retina, has been acknowledged much earlier, prompting questions regarding potentially harmful effects on sun-exposed skin. 11,12 Recently, UV/VIS (385-405 nm) was reported to induce delayed CPD formation in vivo. 13 In contrast to other wavelengths of VIS, HEVIS exposure leads to a significant decrease in viability of different skin cell lines and more pronounced shrinkage of the extracellular matrix (ECM). [14][15][16] Like UVA, HEVIS appears to exert its effects mainly through the generation of ROS, accounting for a substantial part of the amount generated by natural midday sunlight in human skin. 17,18 Blue light photon ROS production efficacy corresponds to 25% of UVA in human keratinocyte mitochondria. 19 ROS detoxification in skin is achieved by low molecular weight antioxidants, such as Vitamins C and E, and carotenoids like β-carotene, as well as by enzymes and antioxidant proteins, many under the control of nuclear factor erythroid 2-related factor 2 (Nrf2), the master regulator of cellular redox signaling and antioxidant defenses. 20,21 Carotenoids are photoprotective, lipophilic plant-derived pigments with highest concentration in the superficial stratum corneum (SC) and aggregation at skin surface. 22,23 Carotenoids mainly exhibit absorbance maximum at wavelengths in the range of visible light 21 and are rapidly degraded by blue light radiation in human skin ex vivo 24 and in vivo, indicating ROS formation. 25 However, endogenous restoration takes up to 24 hours. 25 Furthermore, blue light exposure can activate Nrf2 as a protective endogenous response, in human epidermal cells. 26,27 Loss of Nrf2 has negative consequences for skin homeostasis, repair, and disease, suggesting that further activation is beneficial also for augmented skin photoprotection. 20,28,29 Hence, HEVIS leads to a ROS-induced imbalance between protective and aggressive factors, resulting in tissue damage, permanent pigment darkening, photodermatoses, 30-33 melasma, 34,35 and skin aging. 19,36 Accordingly, HEVIS, but not red light, induces hyperpigmentation, in particular in subjects with more pigmented skin. [37][38][39][40] Furthermore, VIS (400-700 nm) and UVA1 (340-400 nm) synergistically induce skin pigmentation and erythema. 41,42 Based on these effects, photoprotection against VIS/HEVIS is increasingly advocated. [40][41][42][43][44][45][46][47] Diffey and Osterwalder 48 even postulated that the labelled sunscreen sun protection factor (SPFs) may overestimate protection in natural sunlight due to the greater spectral output in the visible region compared with UV solar simulation, contributing 17% to an erythemal reaction. Adding VIS-absorbing mineral filters to sunscreens significantly improves protection against the development of VIS-induced hyperpigmentations. 49,50 However, inclusion of such mineral pigments, for example, iron oxide and non-nano-titanium dioxide, results in tinted formulations, 45,49,50 limiting their broad application. Recent evidence suggests that plant-derived antioxidants can protect VIS-exposed skin from ROS-induced oxidative stress, 51-53 but their effects on Nrf2 remain to be established.
A number of botanical ingredients including bixin, salidroside, tanshinones, caffeic acid, ferulic acid, quercetin, rutin, and the algae-derived mycosporine-like amino acids (MAAs) shinorine and porphyra-334 were shown to mitigate UV-induced cell damage via upregulation of Nrf2, 20,28,29,[54][55][56][57] but little is known about their protective effects against VIS. Licochalcone A (LicA) extracted from the roots of Glycyrrhiza inflata was identified as very potent antioxidant, inhibiting of UV-induced ROS generation, and activator of Nrf2 in primary human fibroblasts. [58][59][60] LicA stimulated the Nrf2/ARE signaling pathway by factor 9 at 2 µmol/L concentration. 61 Here, we present data on the effect of VIS on cutaneous oxidative stress levels at doses and intensities representing one hour of sun exposure in summer in Central Europe. Furthermore, we present results that show the protective effect of LicA on VIS-induced oxidative stress and Nrf2 induction in vitro, and as protectant against SC carotenoid degradation in vivo.

| Active ingredients
Licorice extract from the roots of Glycyrrhiza inflata contained 21% LicA and was purchased from Beijing Gingko. For cell culture experiments, a solution of the LicA-rich licorice extract in DMSO was prepared and diluted with DMEM, and the final DMSO concentration in culture was 0.1%. The sunscreen (in vivo SPF 50+/ in vitro UVA-PF 40) applied in the in vitro and in vivo studies was an oil in water emulsion containing 0.025% licorice extract, corresponding to 0.005% LicA (for ingredients according to INCI see Appendix S1).

| Light sources
Various light sources with different filters were used to irradiate cells in vitro and skin in vivo. All doses are given as physical, not erythemally weighted doses, since the erythema inducing potential of the various spectra is quite different. For detailed information, see Appendix S1, Table S1 and Figure S1, S2, S3, and S4. In order to compare the spectral output of the light sources with the ambient sunlight, the solar spectrum in Hamburg, Germany, was measured on a sunny and a cloudy summer day on top of a building free of any shadowing. Every few minutes, the radiometer (Spectro 320D, Instrument Systems) recorded a spectrum in the range of 280-1700 nm with 1 nm steps.

| Absorption spectra of sunscreen formulations used in the studies
The absorption spectra of sunscreen formulations ( Figure 1) were determined following the methodology described in the ISO24443 for the determination of UVA protection. An UV/VIS Spectrometer Lambda 650 S (PerkinElmer) equipped with 150 mm integrating sphere was used for the measurement.

| Cell culture
Primary human dermal fibroblasts of Caucasian donors, phototypes I to III, were isolated from skin biopsies derived from plastic surgery in healthy donors as described elsewhere. 62 Cells were cultured in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal calf serum (FCS; PAA), penicillin/streptomycin 50 µg/mL, and 1% L-glutamine (all from Invitrogen) at 37°C and 7% CO 2 . Details of the Nrf2 activation assay are provided as Appendix S1.

| Irradiation experiments
In order to assess the potency and wavelength dependency of VIS irradiation to induce ROS formation in vitro, cultured primary human fibroblasts were washed with PBS and irradiated with VIS at wavelengths of ≥400 nm, ≥450 nm, ≥500 nm, and ≥585 nm, respectively.
The overall ROS levels after irritation were measured applying a modified DCF H2DCFDA (Life Technologies GmbH) method, which is not selective for single ROS species, 63   Committee of Charite, Berlin (EA1/228/17). Skin carotenoids were measured noninvasively on three test areas on the inner forearm as baseline value using resonance Raman spectroscopy with an excitation wavelength at 488 nm. 22,25,64 On two of the areas, 2 mg/cm 2 of the sunscreens was applied, respectively, whereas the third area served as untreated positive control. The total test area was immediately irradiated with VIS (Skintrek ® PT3, blueVIS mode: 410-600 nm, maximum at 440 nm, 100 J/cm 2 ) for 42 minutes. Immediately after completion of irradiation, the skin carotenoids were determined again as described above. For more details, see Appendix S1.

| Statistics
The in vitro data were analyzed using the t test function of Microsoft Excel. A P-value <.05 was regarded as statistically significant. For the statistical analysis of the in vivo study the nonparametric Wilcoxon test, SPSS Statistics 19 was used. A P-value <.05 was regarded as statistically significant.

| In vitro studies
The conditions for our in vitro experiments were based on measurements of solar spectral radiation from extraterrestic satellite data 1 and incident radiation measurements in Hamburg, Germany, on a sunny and a cloudy summer day. They showed that a dose of 150 J/cm 2 VIS, as used in the in vitro experiments, can be acquired within 1 hour (Table 1) (Table 1). With water-filtered IRA, we could not induce any ROS, even with 600 J/ cm 2 in cultured fibroblasts under tightly controlled temperature conditions. The irradiation with 600 J/cm 2 corresponds to almost 7 hours of sun exposure at noontime and clear skies, and we even applied the doses at irradiance five times higher than the real sun under clear sky.
Treatment of the cell cultures with sulforaphane and LicA revealed an approximately five times more potent Nrf2 activation by LicA (see Figure S8, Appendix S1). Hence, instead of sulforaphane, LicA served as high control in all further experiments with additional antioxidants. TA B L E 1 Solar irradiances in Hamburg in summer on a cloudy and a sunny day. Averaged data calculated from various spectra measured between 10 am and 2 pm on the days indicated F I G U R E 2 Reactive oxygen species formation in cultured human fibroblasts depending on wavelength. Cultured cells were exposed to 150 J/cm 2 of visible light >400 nm, >450 nm, >500 nm, and >585 nm, respectively. The Oriel 1600 W Solar Simulator filtered for VIS irradiation was used for the experiments. Results of unirradiated cells were set to 100%; n = 9; mean ± SD. Significant differences were marked (*P < .05, ***P < .  Figure S9, Appendix S1). In cultures exposed to UV (2.5 J/cm 2 ), 2 µmol/L LicA reduced ROS generation to the level of unirradiated cells ( Figure 4A). When cell cultures were protected only by a PMMA plate covered with the sunscreen SPF 50+/UVA-PF 40 with LicA ( Figure 5A), this had no effect on VIS-induced ROS formation ( Figure 5B). Hence, neither UV filters nor LicA had any VIS-filtering effect. However, when 2 µmol/L LicA was added to the culture medium, ROS levels were reduced below those obtained with unirradiated cells.
Various other antioxidants were also tested but did not provide any Nrf2 induction and protection from ROS formation at comparable concentrations (see Figure S6 and Figure S7, Appendix S1).

| In vivo study
The basic carotenoid levels before irradiation and treatment in the test areas (control, sunscreen without and with LicA) were F I G U R E 3 Formation of ROS in cultured human fibroblasts induced by solar simulated UV, VIS, and water-filtered IRA. Cells were irradiated with various doses of UV and VIS and a fixed dose of water-filtered IRA. Results of unirradiated cells were set to 100%; n = 7 for UV and VIS; n = 8 for wIRA; mean ± SD. Significant differences were marked (*P < .05, **P < .01, ***P < .001). An Oriel 1600 W Solar Simulator filtered for UV irradiation, another Oriel 1600 W Solar Simulator filtered for VIS irradiation, and a Hydrosun wIRA 505 for IRA irradiation were used F I G U R E 4 Reactive oxygen species formation in irradiated cultured human fibroblasts after pretreatment with LicA. After incubation with LicA at various concentrations, cells were exposed either to (A) UV (2.5 J/cm 2 ) or (B) VIS >400 nm, 150 J/cm 2 . An Oriel 1600 W Solar Simulator filtered for UV irradiation and another Oriel 1600 W Solar Simulator filtered for VIS irradiation were used. Gray bars show oxidative stress levels without irradiation, and black bars represent the additional oxidative stress induced by irradiation. Results of unirriadiated and untreated cells were set to 100%; n = 7; mean ± SD. Significant differences were marked (*P < .05, **P < .01, ***P < . The relative changes in skin carotenoids to initial values were significantly different between both sunscreen formulations and between the sunscreen with LicA and the control area (P < .05; Figure 5C).
Skin temperature showed only a modest increase by 1°C at average during irradiation. respectively. Significant differences were marked (***P < .001). An Oriel 1600 W Solar Simulator filtered for VIS irradiation was used. C, Mean carotenoid levels after irradiation relative to initial values measured in vivo in the skin. Prior to irradiation with blue light (100 J/ cm 2 ), for the spectrum of the Skintrek ® PT3 filtered for UV irradiation see Figure S3 in Appendix S1, skin areas were either left untreated (white bar), or protected by a sunscreen (SPF 50+, UVA-PF 40) containg only UV filters (gray bars) or the same sunscreen containing additionally LicA (black bar). Carotenoids in skin were measured in vivo by resonance Raman spectroscopy (n = 10; mean ± SD; *P < .05)

| D ISCUSS I ON
The development of sunscreens initially focused on protection from UV as the main cause of skin photodamage. Recent findings on the negative effects of UV/Visible radiation on the skin including delayed CPD formation 12,13,18 underline an important role of antioxidants in photoprotection. Vitamin E has been applied in sunscreens for decades, and a recent study provided evidence that it effectively inhibits DNA damage by UVA1-induced photosensitization reactions 66 Furthermore, ROS formation in retinal pigmented epithelial cells upon exposure to blue light and tobacco smoke toxins was significantly reduced by incubation with Vitamin E. 67 However, these effects occured at much higher concentration (100 and 10 µmol/L, resp.) than analyzed in our studies (1-2 µmol/L, Figure   S6 and S7, Appendix S1) and as corroborated by our experiments with Vitamin E at high dosages (Appendix S1, Figure S9) A growing body of evidence suggests complex effects of blue light on skin cells. [14][15][16]19 Blue light at different wavelengths induces varying degrees of oxidative stress, which may involve carbonylated proteins acting as photosensitizers. 76 83 Such findings further support recent recommendations to extend sun protection "beyond UV radiation". 36,71 However, our experiments show that the relevance of nonthermal IR at intensities representing natural solar radiation is neglectible with respect to ROS formation in cultured human fibroblasts, as corroborated by others. 3,6 Efficient photoprotection including HEVIS is essential for lightskinned individuals as well as individuals with darker skin complexion who respond to VIS radiation with a more intense and sustained pigmentation than to UVA1. 37  The contribution of Nrf2 activation to the in vivo protection from HEVIS-induced oxidative stress remains to be elucidated in further studies.

ACK N OWLED G EM ENT
We kindly thank Dr Rolf Binder (CCR GmbH, Berlin, Germany) for his support in preparing the manuscript.

CO N FLI C T S O F I NTE R E S T
TM, KE, FR, MT, AB, and LK are employees of Beiersdorf AG, and none of the other authors has a conflict of interest to declare.