The photoprotective properties of α-tocopherol phosphate against long-wave UVA1 (385 nm) radiation in keratinocytes in vitro

UVA1 radiation (340–400 nm), especially longwave UVA1 (> 370 nm), is often ignored when assessing sun protection due to its low sunburning potential, but it generates reactive oxygen species (ROS) and is poorly attenuated by sunscreens. This study aimed to investigate if α-tocopherol phosphate, (α-TP) a promising new antioxidant, could protect against long-wave UVA1 induced cell death and scavenge UVA1 induced ROS in a skin cell model. HaCaT keratinocyte cell viability (24 h) was assessed with Alamar Blue and Neutral Red assays. The metabolism of α-TP into α-T, assessed using mass spectrometry, and the compound's radical scavenging efficacy, assessed by the dichlorodihydrofluorescein (H2DCFDA) ROS detection assay, was monitored in HaCaTs. The mechanism of α-TP ROS scavenging was determined using non-cell based DPPH and ORAC assays. In HaCaT keratinocytes, irradiated with 226 J/cm2 UVA1 in low-serum (2%, starved) cell culture medium, pretreatment with 80 µM α-TP significantly enhanced cell survival (88%, Alamar Blue) compared to control, whereas α-T pre-treatment had no effect survival (70%, Alamar Blue). Pre-treatment of cells with 100 μM α-TP or 100 μM α-T before 57 J/cm2 UVA1 also significantly reduced ROS generation over 2 h (24.1% and 23.9% respectively) compared to the control and resulted in α-TP bioconversion into α-T. As α-TP displayed weak antioxidant activity in the cell-free assays thus its photoprotection was assigned to its bioconversion to α-T by cellular phosphatases. Through this mechanism α-TP prevented long-wave UVA1 induced cell death and scavenged UVA1 induced ROS in skin cells when added to the starved cell culture medium before UVA1 exposure by bioconversion into α-T.

. The emission spectrum of the 385 nm UVA1 source (dotted line) and the absorbance spectra test compounds (400 µM) in methanol (solid lines). α-TP and α-T show absorbance peaks in the UVB region at 288 nm and 291 nm, respectively. This excludes any photoprotection by optical filtration.

Scientific Reports
| (2021) 11:22400 | https://doi.org/10.1038/s41598-021-01299-y www.nature.com/scientificreports/ suggested that although there were some initial signs of chemical degradation, the changes in the absorbance spectra were small and hence α-T showed good photostability under the conditions used in the experiment. The maximum absorbance for α-TP showed no significant change across the data set (p > 0.05), but again there was some slight shifts in the absorbance spectra, which increased the area under the UVA spectral range by almost 50% (Fig. 2 and Table S2). The increase in area under the curve was due to a small red shift and again probably indicated initial signs of chemical degradation, but like α-T there was no drop in UV absorbance and the changes in the absorbance spectra were small and hence α-TP was considered to have good photostability under the conditions used in the experiment.
Photoprotective compound cell tolerability. α-TP was well tolerated in HaCaT keratinocytes and the LD 50 was comparable using both the Alamar Blue and Neutral Red assays (1676 and 1070 µM, respectively), which suggested that there was no reduction of the resazurin by the test compounds (Fig. 3). α-T was also well tolerated by HaCaT keratinocytes dissolved in 0.5% ethanol in the cell culture medium with no statistical reduction in cell viability after the application of test concentrations 0.61-10,000 μM ( Fig. S2 in the supporting information).
Oxygen radical absorbance capacity (ORAC) assay. α-T was not assessed in the ORAC assay because it is hydrophobic and insoluble in the reaction buffer. Trolox was the reference control because it is a water-soluble vitamin E derivative. The concentrations of α-TP and l-ascorbic acid (L-AA) equivalent to 100 µM Trolox were 7599 and 107.02 µM, respectively (Fig. 7). The order of activity was Trolox > L-AA > α-TP.
Alpha tocopherol phosphate metabolism. The α-TP dissolved in cell culture media applied to the HaCaT cells passed into the cells gradually over the 24 h experiment (Fig. 8). The free α-TP was not depleted in the cell culture media over time demonstrating that a strong concentration gradient is required to drive the compound into the cells. The cell culture media used in the experiments did not contain α-T and hence the appearance of this compound in the cell culture media and the cell lysates demonstrated that α-TP was metabolised into α-T by the HaCATs. The concentrations of α-TP were significantly higher in the cell culture media and the cell lysates compared to α-T across all time points (p < 0.05, ANOVA, Fig. 8). The α-TP concentrations within the cell lysates increased sevenfold over the experiment. In contrast, α-T was not detectable in the cell lysate initially and only increased to 0.7 µg/mL after 24 h, but this only represented cytoplasmic concentration that is readily available as an antioxidant rather than any presence in cell membranes, because an organic extraction process was not used to determine the membrane associated concentrations in this work.

Discussion
Exposure to UVA1 has been linked to DNA damage that results in formation of CPDs 4 , MMP-1 gene overexpression that may result in photoageing 4 , release of nitric oxide (NO) that may be beneficial for hypertension 22 , and skin opsin-3 upregulation that causes pigmentation 23 . Unlike sunscreens, topical antioxidants have been shown to reduce oxidative damage from ROS induced by UVR without the potential to interfere with vitamin D synthesis. α-T is one of the classical topical antioxidants that has been used to inhibit UVR-induced oxidative damage 24 , it has also been shown to inhibit dark CPD in model systems 6,25 . Such CPD are formed after UVR exposure by a delayed reaction between DNA and reactive compounds, currently believed to be excited carbonyls 25,26 . Dark CPD have been reported in human skin in vivo after 385 nm 4 and SSR 27 , but their biological significance is unknown. Numerous limitations of α-T hinder its effectiveness in cosmeceutical products. Adding a phosphate ester to form a provitamin is a strategy that has been shown to overcome these limitations. α-TP has shown important  (Left) DPPH radical scavenging ability of α-TP and α-T. l-ascorbic acid (L-AA) was used as a positive control. Aliquots (12.5 μL) of antioxidants at increasing concentrations (0.006-50 mM) were added to wells containing 187.5 μL DPPH (100 μM). IC 50 values were determined by Graphpad Prism non-linear curve fitting (Gaddum/Schild EC50 fitting). Data represent mean ± SD of three separate experiments. (Right) Ability of α-TP (0-6750 µM) to quench the ROO˙ radical (APPH) using the ORAC assay. Trolox and l-ascorbic acid (0-50 µM) were used as positive controls. Fluorescence degradation over 30 min was assessed with AUC as the readout. Analysis was by linear regression (Trolox: y = 0.2978 x + 1.544, p = 0.0027; l-ascorbic acid: y = 0.2885 x + 0.4473, p = 0.0001; α-TP: y = 0.0042 x + 0.5935, p < 0.0001). 7599 and 107.02 µM of α-TP and l-ascorbic acid respectively were equivalent to 100 µM Trolox. SD were too small to be displayed (n = 3). Ability of α-TP (0-6750 µM) to quench the ROO˙ radical (APPH) using the ORAC assay. Trolox and l-ascorbic acid (0-50 µM) were used as positive controls. Fluorescence degradation over 30 min was assessed with AUC as the readout. Analysis was by linear regression (Trolox: y = 0.2978 x + 1.544, p = 0.0027; l-ascorbic acid: y = 0.2885 x + 0.4473, p = 0.0001; α-TP: y = 0.0042 x + 0.5935, p < 0.0001). 7599 and 107.02 µM of α-TP and l-ascorbic acid respectively were equivalent to 100 µM Trolox. SD were too small to be displayed (n = 3).

Scientific Reports
| (2021) 11:22400 | https://doi.org/10.1038/s41598-021-01299-y www.nature.com/scientificreports/ UVA protective properties in the skin, but previous work did not establish if these benefits were due to its antioxidant properties or its ability to act as a sunscreen, as it forms liposomes (with different optical properties) when applied to skin 13 . Furthermore, previous work with α-TP used UVR sources with UVB content that overlapped with the absorption spectrum of α-TP (generated by the compound's two aromatic rings) 12 . However, to the best of our knowledge no studies to date have investigated the protection of α-TP against spectrally pure UVA-induced skin damage by its antioxidant properties 12 .
The longwave UVA source was selected for a number of reasons. Firstly, given the lack of overlap between the absorption spectra of α-TP and α-T with the UVA1 source it can be concluded that neither compound was capable of absorption/sunscreening effects. Secondly, this is a spectral range that has not been well studied and new LED technology now makes this possible. Thirdly, this is a region known to readily induce oxidative stress with limited induction of other types of damage. Finally, long-wave UVA1 is poorly protected by sunscreens, which focus on preventing the more erythrogenic and shorter wavelengths (< 370 nm). Improved sunscreen protection at shorter wavelengths will result in more exposure in the 385 nm region. As such, there is a need to discover strategies that prevent damage from the longer wavelengths of UVR (370-400 nm).
The UVR absorbance of α-TP slightly increased with increasing SSR dose, most likely due to small changes in molecular aggregation, whereas α-T absorbance did not change, thus both did not showing any significant signs of photodegradation. The SSR dose in this study was 51 J/cm 2 , which corresponds to 3.4 h of UK summer sun, previous reports that showed α-T photodegradation used a much higher dose of 198 kJ/m 2 (19.8 J/cm 2 ) using SSR (290-500 nm) and resulted in a subsequent loss of the compound's antioxidant activity 28 .
α-TP reduced HaCaT keratinocyte viability more than α-T (LD 50 was 1.1-1.8 mM vs > 4.70 mM, respectively) suggesting the phosphate group enhanced the molecule's ability to enter the cells and cause cytotoxicity. This has also been shown in oral mucosal cells in a study that hypothesised that toxicity was caused by disruption of intracellular enzymes 29 . Previous studies in other cells lines support selective cell toxicity by α-TP, but α-T is generally well tolerated 30 . However, the cytotoxicity of TP strongly depends on its isomeric form with the γ-form showing better growth inhibitory activity than δ-form (IC 50 was 30 vs. 55 µM, respectively) in human colorectal cancer cells (HCT116 and HT29 cells). In the same study, the α-form was the least inhibitory compared to other isomers (IC 50 > 100 µM) 31 .
Interestingly, pretreatment of human HaCaT keratinocytes with α-TP in starved low serum media (2% FBS) for 24 h exhibited superior protection against long-wave UVA1 (385 nm) induced cell death compared to α-T. A similar effect has been reported by Nakayama and colleagues who found that incubation of skin cells with 2% FBS significantly depleted endogenous antioxidant α-T by 50% after 24 h incubation. They also showed that treatment with 0.5% α-TP dissolved in 2% DMEM not only inhibited depletion of endogenous α-T, but also increased its level by two-fold compared to untreated cells. In the same study, UVB irradiation was found to significantly reduce endogenous α-T in cultured skin by 50% compared to unirradiated skin. This reduction was inhibited by pre-UVB treatment with α-TP for 3 h which caused a two-fold rise in α-T compared to irradiated cells treated with vehicle control 12 . This may be due to the release of α-T from the provitamin E within keratinocytes by their phosphatase enzymes, which prolongs the protection effect even after 24 h. Another possible reason is that the incorporation of α-TP is essential to prevent the propagation of radical species 32 .
It is well known that ROS are significantly increased in cells during and following UVA irradiation 33,34 . As UVA-related biological effects are largely mediated by ROS, their elimination is essential for protection against UVA damage. In this work, pre-irradiation treatment with either α-TP or α-T result in reduction in UVA1induced ROS generation in HaCaTs to a similar extent (by ~ 24.1% and 23.9%, respectively) compared to the   24 . The difference in the degree of ROS suppression with our findings using the dichloro-dihydro-fluorescein (DCFDA) assay, may be due to the differences in UVA1 source (365 nm vs. 385 nm) and UVA1 dose (8 J/cm 2 vs. 57 J/cm 2 ), the concentration of tested compound (2 mg/ mL = 4.6 mM vs. 100 µM), the concentration of H 2 DCFDA (5 µM vs. 20 µM), and the level of supplements in growth medium (serum-free vs. low-serum medium). Studies were performed with DPPH and ORAC assays without cells to determine if α-TP afforded innate antioxidant activity or was the outcome of its potential bioconversion to α-T. As free radical scavenging by antioxidants can proceed through single electron transfer (SET) or hydrogen atom transfer (HAT), two assays (DPPH and ORAC respectively) were used to identify the mechanisms 35 . α-TP exhibited poor DPPH activity compared to ascorbic acid and α-T. The α-T IC 50 of 19 µM was lower than previously reported at 40.6 μM, but within normal inter-laboratory variability 36 . The ascorbic acid IC 50 of 24 µM was slightly lower than previously reported, i.e., 34 µM 37 , 54.6 µM 38 , 49.5 µM 36 , but confirmed that assay worked well. This poor activity suggested that α-TP had very little antioxidant capacity per se via SET whereas α-T and ascorbic acid were potent antioxidants via this mechanism. α-TP and water-soluble ascorbic acid were compared to Trolox in the ORAC assay. Trolox activity was almost equivalent to ascorbic acid, but was 75-fold more effective than α-TP. Our findings agree well with those of Rezk et al. (2004) using the Trolox equivalent antioxidant capacity (TEAC) assay, in which they determined the TEAC for Trolox and α-TP (1.0, and 0.01, respectively), i.e., Trolox was 100-fold more potent than α-TP 39 . Weak α-TP activity in two different assays indicated that bioconversion to α-T was probably responsible for the UVR protective in a biological model.
The metabolism studies were designed to investigate the availability of α-TP and α-T in the cell culture media and cell cytoplasm over time after the dosing of the cells with α-TP. No extraction of the agents from the cell membrane was performed as it was expected that the metabolism studies would explain the α-TP intracellular ROS savaging observed after UVA1 exposure despite its weak in vitro activity. The µg/mL detection of α-T in both the cell culture media and the cell lysate after α-TP, whilst not being present in the control experiments that only applied cell culture media, demonstrated that the cells did metabolise the α-TP into α-T. Previous work in THP-1 cells has shown that α-TP does metabolise into α-T and that the concentration of α-T found in the cells was around 3% of the initial α-TP added 40 . In the HaCaT cells only 0.4% of the applied α-TP converted into α-T and passed into the cell cytoplasm. In the current study LC-MS was employed to specifically detect cytoplasmic α-T with no lipid extraction method being applied, whereas in the previous work radiochemical detection was employed with a lipid extraction method. It is reasonable to suggest that lipid extraction coupled with the radiochemical method would account for the membrane associated α-T as well was that in the cytoplasm, and this provides an explanation for the differences in the reported values. The metabolism studies also suggested that the α-TP penetrated the cells more quickly and to a greater extent compared to α-T. This is in accordance with our own previously published work, which demonstrated that α-TP penetrates the skin more readily than α-T, because the latter is trapped within the lipid regions of the cell membranes 15 . This work was able to demonstrate this effect as it did not mimic the published methods to study tocopherol metabolism as they also extracted the compounds from the cell membranes, which can overestimate the α-T available to quench ROS within the cell.
A limitation of this work is that it was performed in immortalised HaCat keratinocytes rather than primary keratinocyte cell lines or more complex skin models. It can be problematic to study oxidative stress in complex systems due the tissue isolation or growth processes inducing oxidative stress. The use of HaCaTs has been established as valid in photobiological research [41][42][43][44] , and the aim of this study was to determine the relative antioxidant effects of the test compounds at the cellular level given our previous work had demonstrated that α-TP reaches the epidermis after topical administration. Also, we have found in our previous studies a good correlation between the HaCaT model and human in vivo studies 45 .
In conclusion, α-TP displayed weak antioxidant activity when assessed by chemical assays. In HaCaT keratinocytes it was well tolerated, and had superior photoprotection properties compared to α-T. Pre-irradiation treatment of HaCaT keratinocytes with α-TP resulted in a similar reduction in ROS generation as α-T, but α-TP penetrates more effectively into the cells and the skin compared to α-T. The biological activity of α-TP was found to be a consequence of its bioconversion to α-T by endogenous cellular phosphatases, which increase the intercellular level of α-T even after UVA1 exposure ends. That α-TP proved to be a photostable provitamin E form that has greater protection against cell viability reduction and comparable reduction in level of ROS generated by UVA1 exposure compared to the parent molecule and thus it is an excellent candidate to include in sun-protection products to reduce the impact of UVA1-induced skin damage.
The 385 nm spectrum was measured at a distance of 40 cm and the SSR spectrum was measured at a distance of 0 cm. Irradiation distances were based on the output of the source being measured and were selected to be within the dynamic range of the spectroradiometer. Irradiances were routinely measured with hand-held radiometers. This was a Loctite UVA/Vis radiometer (Loctite, Henkel Ltd, UK) for the 385 nm source. A typical irradiance of 88.5 mW/cm 2 with irradiation time of 11 min gave a dose of 56.5 J/cm 2 . The solar simulator spectrum was measured using a Solar Light PMA 2100 radiometer (Solar Light, Glenside, Pennsylvania). The solar simulator was calibrated using spectroradiometric readings such that an irradiance of 1100 μW/cm 2 for 18 s gave 2-3 standard erythema dose (SED).
Absorption spectra of photoprotective compounds. α-TP and α-T were prepared at a concentration of 400 µM in methanol. Their UVR absorbance spectra were determined with a Perkin Elmer Lambda 2 UV/VIS Spectrometer (Perkin Elmer & Co GmbH, Oberlingen, Germany) between wavelengths 250-420 nm, to assess for possible sunscreening effects.
Photostability. SSR was used to test photodegradation of 1 mM (0.05% w/v) α-TP dissolved in Tris buffer pH 7.4 and an equivalent molar concentration of α-T dissolved in ethanol using increasing doses of SSR (10-50 SED or 17.8-88.9 J/cm 2 ). Protected samples were used as dark controls (wrapped with aluminium foil). Absorbance of the dark controls A 0 and samples (A T ) was measured between 280-400 nm (n = 3) after each exposure using a FLUOstar-Omega microplate reader (BMG LABTECH Offenburg, Germany). Percent degradation was calculated from the ratio of the change in absorbances (A T − A 0 ) after exposure and the absorbance of dark control (A 0 ) at 288 nm for α-TP and 291 nm for α-T, after normalising by subtracting the blank absorbance.
Cell culture. The immortalised human HaCaT keratinocyte cell line was purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA). This cell line has two p53 spontaneously transformed point mutations. HaCaT cells were cultured in DMEM supplemented with 10% FBS, 5% penicillin-streptomycin, and 5% glutamine, maintained in a humidified incubator at 37 °C with 95% air and 5% CO 2 . Cells were cultured to around 80% confluence in 75 cm 2 plastic flasks (Corning, New York, USA). Cells were plated into plastic bottom black wall 96 well plate (Perkin Elmer) and left to reach a confluence of 70-80% before being used for experiments.
Cell tolerance of photoprotective compounds. The cytotoxic potential of α-TP and α-T was evaluated using the Alamar Blue 16 (Fisher Scientific, Loughborough, UK) and Neutral Red 17,18 (Sigma-Aldrich, Dorset, UK) viability assays in HaCaT keratinocytes. Alamar Blue is a metabolic function (redox) indicator, which measures viable cells' ability to reduce resazurin, to fluorescent resorufin. The Neutral Red assay assesses the ability of viable cells to incorporate the cationic dye into their lysosomes. Two assays with different mechanisms of action were employed to strengthen conclusions. The cells were treated for 24 h with α-TP (0.0001-10 mM) in prewarmed cell culture media at 37 °C (cell culture media was used as a control) or α-T (0.0001-5 mM) in 0.5% ethanol mixed with prewarmed full media at 37 °C (0.5% ethanol in cell culture media was used as a control). The test compounds were aspirated from each well and the culture media containing 1:10 www.nature.com/scientificreports/ 80 µM of α-TP (1.5% tris buffer in medium was used as a control) or α-T (0.5% ethanol in medium was used as a control), and then incubated at 37 °C with 5% CO 2 for 24 h. After aspiration of treatment solutions, the cells were washed twice with warmed PBS (100 μL/well), then fresh PBS (100 μL/well) was added to each well and the cells were irradiated with UVA1 without the plate lid. Each well was exposed individually for up to 44 min to obtain a maximum UVA1 dose of 226 J/cm 2 . Cells were kept on a cooling platform to keep them at around 37 °C. Unirradiated controls (covered with foil) were kept in the same conditions as the longest exposure time to ensure any differences observed were due to the UVA1 exposure rather than any confounding factors. The cells were washed, and the cell viability was assessed using both the Alamar Blue and Neutral Red assays 24 h post-irradiation.  Alpha tocopherol phosphate metabolism studies. HaCaT keratinocytes were plated (9 × 10 5 cells per well), in triplicate, into a 6 well plate using 1 mL of media. α-TP 6 mM stock solution was made in 0.1 M tris buffer, diluted in DMEM, and (1 mL, 500 μM) was applied to groups of triplicate cells that were incubated either for 2 h, 6 h, 18 h or 24 h (final conc. 250 μM, eq 133 µg per mL). Samples were taken from the cell culture media and the cells were harvested. Cells were then lysed through three freeze-thaw cycles and mechanical syringe homogenisation to release their contents. Lysates were centrifuged to remove cell debris, transferred to a fresh Eppendorf tube and snap frozen at − 80 °C. When ready for analysis, the lysates and cell culture media samples were thawed and the internal standard tocopherol acetate (50 µg/mL), dissolved in acetonitrile, was added. The acetonitrile precipitated the protein in the samples, and they were centrifuged. The supernatants were passed through a solid phase extraction plate (Oasis HLB, Waters, UK) preconditioned using phosphoric acid and methanol. The analytes adsorbed to the solid phase extraction plate were washed with 20% methanol 80% water and eluted into a 96 well plate using 200 µL of isopropyl alcohol. A series of calibration standards were subjected to the same processes as the cell samples to quantify αT and αTP. The LC-MS analysis used a Waters Aquity ultra high-performance liquid chromatography and Xevo TQ MS quadrapole mass spectrometer (Waters, UK). A Waters Aquity BEH C18 1.7 µm, 2.1 mm × 50 mm column (Waters, UK) was employed for the analysis with a mobile phase comprising 90% isopropyl alcohol: 10% water containing 0.1% formic acid. The mass spectrometer settings were optimised for each compounds, but each used a source temperature of 150 °C, a desolvation temperature of 500 °C, a desolvation gas flow of 600 L/h and a capillary flow of 3.9 kV. α-TP was detected using negative ESI using MRM transition of 509. 22

Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.