Photo-Oxidation Products of Skin Surface Squalene Mediate Metabolic and Inflammatory Responses to Solar UV in Human Keratinocytes

The study aimed to identify endogenous lipid mediators of metabolic and inflammatory responses of human keratinocytes to solar UV irradiation. Physiologically relevant doses of solar simulated UVA+UVB were applied to human skin surface lipids (SSL) or to primary cultures of normal human epidermal keratinocytes (NHEK). The decay of photo-sensitive lipid-soluble components, alpha-tocopherol, squalene (Sq), and cholesterol in SSL was analysed and products of squalene photo-oxidation (SqPx) were quantitatively isolated from irradiated SSL. When administered directly to NHEK, low-dose solar UVA+UVB induced time-dependent inflammatory and metabolic responses. To mimic UVA+UVB action, NHEK were exposed to intact or photo-oxidised SSL, Sq or SqPx, 4-hydroxy-2-nonenal (4-HNE), and the product of tryptophan photo-oxidation 6-formylindolo[3,2-b]carbazole (FICZ). FICZ activated exclusively metabolic responses characteristic for UV, i.e. the aryl hydrocarbon receptor (AhR) machinery and downstream CYP1A1/CYP1B1 gene expression, while 4-HNE slightly stimulated inflammatory UV markers IL-6, COX-2, and iNOS genes. On contrast, SqPx induced the majority of metabolic and inflammatory responses characteristic for UVA+UVB, acting via AhR, EGFR, and G-protein-coupled arachidonic acid receptor (G2A). Conclusions/Significance Our findings indicate that Sq could be a primary sensor of solar UV irradiation in human SSL, and products of its photo-oxidation mediate/induce metabolic and inflammatory responses of keratinocytes to UVA+UVB, which could be relevant for skin inflammation in the sun-exposed oily skin.


Introduction
Numerous mechanisms have been evolved in human skin to sense environmental stimuli and to mount adaptive responses in order to maintain homeostasis in the skin and protect the entire organism. The ultraviolet component of solar light consists of UVB and UVA parts, which differentially penetrate the skin barrier and thus affect prevalently epidermal (UVB) or dermal (UVA) skin cells and corresponding extracellular structures/ molecules [1]. Skin surface lipids (SSL) play an important role for essential human skin functions such as mechanical and chemical barrier, thermoregulation, and photo-protection [1][2][3]. Laying on the surface of the skin, SSL are also exposed to the highest doses of UVA+UVB and they form the first-line defence against its potential danger. Major photo-protective components of SSL are a-tocopherol and squalene (2,6,10,15,19,23,-hexamethyl-2,6,10,14,18,22-tetracosahexaene, Sq), both working as sacrificing antioxidants, since they block photo-induced lipid peroxidation in cellular and acellular skin components, either by chain-breaking mechanism (a-tocopherol) [1], [4], [5] or by quenching singlet oxygen (Sq) [6]. Both are continuously produced by skin surface-open sebaceous glands to maintain their physiologically essential levels and substitute photo-destructed molecules [1][2][3]. Sq is the most abundant oxidisable component of SSL, its concentrations in adult human skin reaching up to 20%, while its levels are negligible in other organs [2], [7]. Upon action of environmental oxidants and microbial residents of the skin, fast oxidative degradation of Sq occurs giving rise to a wide spectrum of by-products, such as monohydroperoxides, epoxides, and aldehydes ( Fig. 7A) [1], [2], [8]. Physiological doses of UVA oxidise Sq at much higher rates than UVB [1], [2]. The role of Sq and its oxidation products (SqPx) in skin photo-protection [9] and in the induction of inflammatory responses of keratinocytes in the context of acne pathogenesis [10] has been evaluated and recently reviewed in [2], [3].
The search of endogenous extracellular sensors of UV mediating its effects on skin cells has received growing interest in the last years. Products of tryptophan photo-oxidation, 6formylindolo [3,2-b]carbazole (FICZ) in particular, were extensively studied and shown to mimic UVB in the induction of aryl hydrocarbon receptor (AhR)-controlled metabolic cascade in hepatocytes and HaCaT [11][12][13] , and of melanogenesis in melanocytes [14]. AhR is a cytosol-associated and ligand-activated receptor with transcription factor functions. Upon stimulation by a ligand, AhR liberates from its chaperon heat shock protein 90 (Hsp90), co-chaperon p23, and XAP-2 and moves to the nucleus, where it binds to its specific nuclear co-partner Arnt to acquire full binding capacity to the promoter of target genes, first of all CYP1A1 and CYP1B1, to start the metabolic program [12], [13]. Over-expression of CYP1A1 and CYP1B1 enzymes has been detected in UVB exposed human skin [15]. As shown recently, the physiological role of AhR is not limited to the control of xenobiotic metabolism, but it extends to numerous cell functions such as breakdown of endogenous metabolites, proliferation, cell-to-cell contacts, immune and inflammatory responses, melanogenesis, and circadian rhythm [16], [17]. Among the endogenous AhR ligands known so far [18], [19], there are the products of tryptophan photo-oxidation [13], prostaglandins [20], bilirubin/ biliverdin [21], and kinurenic acid [22], all of them present in the skin cells/cellular membranes.
Another UV-sensitive receptor is the epidermal growth factor receptor (EGFR), located on the cellular membrane of keratinocytes, which is a key regulator of numerous essential processes underlying skin development, homeostasis, stress responses, and repair [23][24][25][26]. 4-Hydroxy-2-nonenal (4-HNE), the final product of membrane-bound arachidonic acid photo-oxidation, appears to be a signalling molecule of UVB-induced EGFR nuclear translocation in keratinocytes [27][28][29], where the receptor physically or functionally interacts with other transcription factors with DNA-binding activity, including the signal transducer and activator of transcription 3 (STAT3). These events lead to the upregulation of distinct genes controlling cell proliferation, DNA repair, as well as COX-2 and iNOS [23]. Moreover, the cytoplasmic components of the EGFR system, the EGFR-ERK1/2 or EGFRphosphoinositol-3-phosphate (PI3P)-protein kinase B (Akt1) pathways, are involved in the regulated expression of inflammatory chemokines in normal human keratinocytes [23], [25], [30], including CXCL8/IL-8, a chemokine sensitive to UV irradiation and reasonably involved in UV-induced granulocyte recruitment to the irradiated skin sites as well as in solar hyperkeratosis [25].
The major goal of the present study was the search for a firstline endogenous extracellular UV sensor(s) located in the outmost skin surface lipids (SSL). Here, we evaluated the role of Sq, a major photo-oxidisable component of SSL, in sensing solar UVA+UVB and transmitting their signal to underlying skin keratinocytes. We showed for the first time that photo-oxidised SqPx directly applied to primary human epidermal keratinocytes (NHEK) mimics the majority of UVA+UVB effects towards primary human epidermal keratinocytes (NHEK), by stimulating the AhR-controlled metabolic pathway and affecting EGFR-and G2A-connected inflammatory responses.

Ethics Statement
All experiments with human material (skin biopsies and skin surface lipids) were carried out in accord with Helsinki Declaration, the protocols were approved by Ethical Committee of Istituto Dermopatico dell'Immacolata, Rome, and the healthy adult donors signed the informed consent.
Skin surface lipid collection and analysis SSL were obtained from sebum of 10 healthy adult male volunteers by non-invasive cup extraction method, by applying 10 ml diethyl ether on the upper dorsal skin with a glass cylinder with open ends (5.3 cm 2 ), allowing contact for 2 min. Then, the freshly collected SSL extracts were pooled, filtered, and evaporat- ed under nitrogen flux [7]. SSL collection was performed during autumn-winter period to avoid effects of recent sun exposure. The SSL were derivatized at 80uC for 30 min by the addition of N,Obis-(trimethylsylil)-trifluoroacetamide with 1% trimethylchlorosilane and analyzed by gas chromatography-mass spectrometry (GC-MS) on a SHIMADZU GC-17A/QP-5050, equipped with a J&W DB-1 capillary column (25 m60,2 mm60,33 mm). The following GC-MS parameters were used: 1 ml/min (helium) flux; split ratio, 1:10; injector temperature 250uC; detector temperature 270uC; temperature gradient: 100uC to 270uC in 33 min, peak detection method: selected ion monitoring technique [8]. All solvents, reagents, and standards of a-tocopherol, cholesterol, and squalene for GC-MS analysis were from Sigma-Aldrich (Milan, Italy).

Squalene isolation from skin surface lipids
Measured aliquots of SSL samples were separated by thin layer chromatography (TLC) with benzene/hexane (70/30) elution system, with commercial naïve and irradiated Sq (Sigma-Aldrich) as standards. The Sq fraction was extracted quantitatively with chloroform/methanol (2/1) from the silica spots of the SSL lane (Fig. S1), thoroughly filtered, evaporated and diluted in minimal quantity of methanol. To verify the purity of isolated Sq, the fraction was subjected to GC-MS analysis, and the typical spectrum ( Fig. S2) showed more than 97% purity.

Irradiation of cells and skin surface lipids with solar simulated UVA and UVB
NHEK monolayer was exposed to low-dose UVA+UVB irradiation (time of irradiation 30 s, distance from cells 30 cm, dose UVA 1.0 J/cm 2 +UVB 0.1 J/cm 2 ) produced by Solar Simulator (Dermalight Vario with filter A2, Dr. Hoehnle AG,  UV Technology, Planegg, Germany) with emission spectrum from 280 nm and emission peak at 375 nm. The light effluence rate on the cell monolayer was 40 mW/cm 2 .
In the in vivo experiments, dorsum skin of 10 healthy donors was exposed to the same UVA+UVB lamp (the distance 80 cm, irradiation time 30 min, dose UVA 30.0 J/cm 2 +UVB 3.0 J/cm 2 ) and SSL were immediately collected and analyzed.
In the ex vivo experiments, dried-under-nitrogen SSL extracts (0,95 mg) were placed to glass Petri dishes (diameter 2 cm), and irradiated by the same UVA+UVB source (the distance from the bottom of dishes was 30 cm, irradiation time range 1-20 min, and dose range from UVA 1.0 J/cm 2 +UVB 0.1 J/cm 2 to UVA 40.0 J/cm 2 +UVB 4.0 J/cm 2 ). Immediately after irradiation, SSL extracts were dissolved in chloroform/methanol (2:1) with the addition of butylated hydroxyanisole to stop lipid peroxidation, and analyzed quantitatively for a-tocopherol, cholesterol, and squalene levels. Results of triplicate experiments were expressed as percent amount of a-tocopherol, squalene, or cholesterol in SSL as compared to non-irradiated controls. The same irradiation procedure was applied to commercial Sq, to use it as a reference standard of irradiated squalene for TLC and in the experiments with NHEK.
In the experiments with 4-HNE (Cayman Chem, Ann Arbor, MI), reported to mediate UVB signaling to EGFR pathway in keratinocytes [27], [35], a 25 mM 4-HNE solution in DMSO or vehicle alone were added to NHEK. After 0.5-4 h of incubation, the conditioned medium was collected for further cytokine assay, while cells were processed for total lysate preparation and RNA isolation.  To elucidate possible SSL effects as sensors/mediators of solar UV signals to skin cells, NHEK cultures were exposed to the lipids normalized by the content of Sq, or SqPx isolated from ex vivo irradiated SSL, or SqPx-in vivo isolated from in vivo irradiated SSL (final concentrations of Sq, PxSq, or SqPx-in vivo were 1.0 mg/mL).
The effects of tert-butylperoxide and photo-oxidized linoleic acid (both from Sigma-Aldrich, Milan, Italy) were compared to the effects of photo-oxidized SSL and photo-oxidized Sq. Tertbutylperoxide was added to NHEK at two concentrations (0.1 mg/mL and 0.5 mg/mL). Linoleic acid was irradiated by UVA+UVB using the same protocol as for SSL, and the products of its photo-oxidation were added to NHEK at final concentrations 0.1 mg/mL and 0.4 mg/mL, the latter corresponds to the physiological ratio of squalene:linolenic acid % concentrations in SSL equal to 10:1 [2], [7]. The concentration of photo-oxidized linoleic acid 0.4 mg/mL corresponds to its 2 mM. The same molar concentration of PxSq was used in all experiments.
Transfection with AhR-specific small interference RNA AhR was knocked-down by using a pool of four small interfering RNAs (siRNAs) provided by Euroclone (Milan, Italy) (ONTAR-GETplus SMARTpool, L-004990-00-0005). In parallel, a pool of four nontargeting siRNAs was used as negative control (ON-TARGET plus siCONTROL, D-001810-10-05). NHEK were incubated with a mixture of 50 nM SiRNA and 4 mL/mL INTERFERIN TM transfection reagent (Polyplus Transfection, Euroclone), according to the manufacturer's instructions. After 48 h incubation in starvation medium, SiRNA-transfected cells were exposed to FICZ for the indicated time interval. Cells were then lysed for protein extraction, or for RNA extraction.

Preparation of cell extracts and immunoblotting
Total non-phosphorylated and phosphorylated EGFR, ERK1/ 2, Akt1, and phosphorylated p65 subunit of NFkB were investigated in total cell lysates as previously described [36]. Anti-EGFR antibodies were from Santa Cruz, whereas anti-ERK, anti-Akt1, and anti-p65 NFkB antibodies were from Cell Signaling Technology (Beverly, MA).
NHEK were differentially lysed to obtain nucleic extracts according to [37]. The antibodies used for the detection of AhR, Arnt, non-phosphorylated and phosphorylated EGFR were from Santa Cruz Biotechnology, CA.

RNA isolation and quantitative real-time-PCR assay
Total RNA was isolated using the GenElute Mammalian Total RNA Kit (Sigma-Aldrich, Milan, Italy) and was reverse-transcribed using the iScript cDNA Synthesis Kit (Bio-Rad, Hercules, CA). cDNA was amplified with IQ SYBR green Supermix (Bio-Rad, Hercules, CA), using the MiniOpticon Real-Time PCR Detection System (Bio-Rad, Hercules, CA). Two housekeeping genes, ribosomal 18S and beta-actin were chosen as reference, and fold changes were calculated [38]. The primer sets were synthesized by Eurofins MWG Operon

Assays for inflammatory cytokine production by NHEK
The pro-inflammatory cytokines TNFa, IL-6, and IL-8 were measured in cell supernatants at 24 h post-irradiation or SqPx exposure, using BD OptEIA Elisa kits from BD Biosciences (San Diego, CA) [36].

Statistics
All measurements were done in triplicate, and data of at least three independent experiments were statistically evaluated. Statis-tical evaluation was carried out with the software package for Windows XP. Results were expressed as the mean 6SD. To evaluate the difference between experimental groups, the twotailed Student's t-test was applied and P values,0.05 were considered to be significant.

Physiological doses of solar simulated UVA+UVB induce inflammatory responses time dependently in NHEK
Exposure of NHEK to UVA+UVB led to time-dependent bellshaped induction of genes encoding the pro-inflammatory cytokines TNFa, IL-8, and IL-1b and delayed induction of IL-6 ( Fig. 1A). Maximal although slight expression of TNFa and IL-1b was reached at 3 h, while the other pro-inflammatory cytokines were strongly and maximally expressed at 6 h. The protein level of TNFa did not change, whereas those of IL-8 and IL-6 were significantly increased in NHEK supernatants at 24 h postirradiation (Fig. 1B). At 6 h, maximal expression of COX-2 mRNA was observed (Fig. 1C). The mRNA expression of the two membrane receptors TLR4 and IL-1R1 was first temporarily suppressed and then, gradually increased up to 12-24 h (Fig. 1D). Physiological doses of solar simulated UVA+UVB activate AhR system and downstream cytochrome P450 isoforms CYP1A1 and CYP1B1 UVA+UVB irradiation stimulated AhR and Arnt nuclear translocation ( Figs. 2A and 2B), an early event preceding the induction of CYP1A1 and CYP1B1 genes, which reached maximal levels 12h post-irradiation (Fig. 2C). AhR and Arnt nuclear translocation was significantly and dose-dependently triggered also by the product of tryptophan photo-oxidation FICZ (Figs. 3A and  3B). At the same time, total AhR protein level were not affected by FICZ (Figs. 3C and 3D), implying that the aromatic compound is an AhR ligand, which did not induce its de novo synthesis, but rather re-distributed the cytoplasmic receptor between nucleus and cytoplasm. Accordingly, both CYP1A1 and CYP1B1 were also strongly, dose-dependently up-regulated by FICZ, with maximal effect already achieved at 4 h (Figs. 3E and 3F). In addition, FICZ slightly but statistically significantly induced delayed AhR expression at 6 h (Figs. 3E and 3F). Upon AhR protein level knock-down by specific siRNA (Fig. 4A), both spontaneous and FICZassociated AhR were dramatically down-regulated as expected (Fig. 4B), and correspondingly also CYP1A1 levels were remarkably reduced (Fig. 4C). Of note, constitutive IL-8 was not affected by AhR silencing or by the presence of FICZ (Fig. 4D). Similarly, FICZ did not perturb the expression of other pro-inflammatory genes, including TNFa, IL-6, and COX-2, the induction of which is characteristic of NHEK response to UV [33].
Physiological doses of solar-simulated UVA+UVB stimulate signal transduction through EGFR phosphorylation, downstream cytoplasmic signal transduction, and EGFR translocation to the nucleus NHEK exposure to UV led to a rapid (5 min) and transient EGFR phosphorylation, while downstream ERK phosphorylation gradually increased up to 20 min post-irradiation when examined in total cell lysates (Figs. 5A and 5B). At 20 min post-irradiation, nuclear levels of both non-phosphorylated and phosphorylated EGFR were significantly increased (Figs. 5C and 5D). Conversely, exposure to FICZ did inhibit both EGFR and ERK in a dosedependent fashion (Figs. 5E and 5F). Moreover, 1 mM FICZ significantly decreased nuclear levels of both non-phosphorylated and phosphorylated EGFR (Figs. 5G and 5H).

Exogenous 4-HNE dysregulates EGFR-connected signal transduction pathways time dependently, induces COX-2 expression but it only slightly affects pro-inflammatory cytokines and iNOS
We observed that a non-toxic dose of 4-HNE (25 mM), known to stimulate EGFR nuclear translocation and to mimic UVB effects on primary human keratinocyte cultures [29], inhibited downstream ERK and affected Akt1 phosphorylation in a biphasic fashion, with transient inhibition at 0.5 and 1 h, but stimulation at 2 h (Figs. 6A and 6B). With regard to inflammatory gene response, exposure to 4-HNE induced rapid and significant COX-2 expression, slight and delayed iNOX (Fig. 6C), and durable although slight IL-6 expression (Fig. 6D). None of the other EGFR-connected pro-inflammatory chemokines, IL-8 or MCP-1 or IP-10 (data not shown), were affected by 4-HNE. Protein expression of IL-6, IL-8, and MCP-1 corresponded to their mRNA behaviour (Fig. 6E).

Squalene derivatives isolated from photo-oxidised skin surface lipids but not intact lipids/squalene induce UVsimilar mRNA pattern of pro-inflammatory cytokines and receptors
Freshly collected SSL were exposed ex vivo to solar simulated low dose UVA+UVB irradiation for 5-20 min, and the content of UV-sensitive (a-tocopherol and Sq) and UV-insensitive (cholesterol) substances was monitored. In vivo skin irradiation with UVA+UVB (30+3 J/cm 2 ) also resulted in the degradation of photo-oxidisable components of SSL (Fig. 7B-7D).
Intact SSL, SSL irradiated ex vivo for 20 min (oxSSL), Sq isolated from intact SSL and from oxSSL (SqPx), and from in vivo irradiated SSL (SqPx-in vivo), as well as commercial Sq photoirradiated for 20 min (Fig. S1), were normalised by Sq or SqPx content (1 mg/mL) and added to NHEK cultures. The purity of SqPx levels isolated from ex vivo or in vivo irradiated SSL was .99% (Fig. S2), and the levels of Sq or SqPx in SSL or oxSSL, respectively, ranged from 10 to 15%. Fig. 8A shows that co-incubation of NHEK with intact SSL for 3 h practically did not affect gene expression of pro-inflammatory cytokines (TNFa, IL-8, and MCP-1) and only slightly induced IL-6. In addition, SSL did not exert any effect on COX-2 (Fig. 8B) and IL-1R1 mRNA, while it slightly down-regulated G2A, TLR4 and its downstream co-activator MYD88 (Fig. 8C). Sq isolated from intact SSL exhibited similar mode of action. On the other hand, oxSSL Figure 8. Effects of skin surface lipids on gene expression in NHEK. NHEK cultures were treated with skin surface lipids normalised by the content of squalene (Sq): intact freshly prepared skin surface lipids (SSL, with final Sq content 1 mg/mL), Sq isolated from the SSL (final content 1 mg/mL), SSL exposed to UVA+UVB for 20 min (oxSSL with final content of oxidised Sq (SqPx) 1 mg/mL), or PxSq isolated from oxSSL (final content of SqPx 1 mg/mL) or SqPx-in vivo isolated from the in vivo photo-exposed skin surface lipids (final content of SqPx-in vivo 1 mg/mL). The panels present fold of mRNA changes for proinflammatory cytokines (A), COX-2 (B), and membrane receptors (C). *P,0.05 and 1 P,0.01 versus control treated with the vehicle for the lipids (0.1% methanol). doi:10.1371/journal.pone.0044472.g008 up-regulated TNFa, IL-6, and IL-8, but it did not change MCP-1 (Fig. 8A) or COX-2 (Fig. 8B), IL-1R1, and MYD88 expression, while it slightly inhibited G2A and TLR4 (Fig. 8C). SqPx strongly up-regulated TNFa, IL-6, IL-8, IL-1R1, G2A, and TLR4 ( Fig. 8A and 8C), although it down-regulated COX-2 and MYD88 ( Fig. 8B  and 8C). SqPx-in vivo up-regulated TNFa, IL-6, IL-8, while downregulating COX-2 ( Fig. 8A and 8B). The behaviour of commercial Sq upon UVA+UVB exposure was similar to SqPx (Table 1). It could be, however, noticed that photo-oxidized commercial Sq was less effective than Sq isolated from SSL exposed to the same dose of UVA+UVB. This could be explained by the fact that some lipids in SSL were co-oxidized with Sq thus accelerating its photooxidation. Also, trace amounts of transition metals, such as Fe and Cu, in SSL could catalyze photo-oxidation of Sq. Therefore, either more SSL squalene was photo-oxidized or deeper oxidation of SSL Sq (different oxidation metabolites) occurred as compared to pure squalene.
To evaluate whether hydroperoxides of photo-oxidized linoleic acid, which is a minor oxidizable component of human SSL [2], [7], could possess solar UV-mimicking effects similar to PxSq, photo-oxidized commercial linoleic acid was added to NHEK in physiologically relevant concentrations (0.1 and 0.4 mg/mL). Contrary to UV and PxSq, photo-oxidized linoleic acid inhibited TNFa, IL-6, IL-8, and IL1R1. Stimulation of IL-8 expression occurred upon NHEK exposure to 0.4 mg/mL of photo-oxidized linoleic acid (Table 2). Tert-butyl hydroperoxide, a synthetic organic hydroperoxide traditionally used to model UVA effects on skin cells, at millimolar concentrations induced TNFa, IL-8, and IL1R1 while suppressed IL-6 ( Table 2).
Skin surface lipids profoundly inhibit EGFR-ERK axis and activate AhR system, but exclusively SqPx up-regulates cytochrome P450 isoforms (CYP1A1 and CYP1B1) All forms of skin surface lipids studied activated AhR translocation to the nucleus and the effect changed in the range: SqPx.SSL.oxSSL = Sq (Figs. 9A and 9B). Surprisingly, only SqPx and SqPx-in vivo up-regulated CYP1A1 and CYP1B1 genes, downstream targets for AhR (Fig. 9D). Both photo-oxidized linoleic acid and tert-butyl hydroperoxide significantly downregulated CYP1A1 and CYP1B1 genes ( Table 2). Several recent publications [34] have explained the initial transient inhibition of AhR nuclear translocation and CYP1 expression by early transient activation of NFkB. Indeed, p65 phosphorylation was significantly induced only by oxSSL (Fig. 9C), while it did not stimulate AhR (Fig. 9B) or CYP1 gene expression (Fig. 9D). The SSLs were also effective inhibitors of EGFR and ERK phosphorylation, SqPx being the strongest (Figs. 9A and 9C).

Discussion
Mechanisms of solar UV irradiation effects towards human skin have been the field of intense research mostly targeting UVassociated carcinogenesis and premature skin ageing [1], [5], [15], [38]. Based on the different capability of UVB and UVA light to penetrate in the epidermal and dermal layers, most of the studies  Table 1. Induction of inflammatory and metabolic genes (mRNA, fold of induction) by photo-oxidized commercial squalene and squalene from photo-oxidized human skin surface lipids (SSL). Commercial squalene or SSL were exposed to UVA+UVB irradiation (dose UVA = 20 J/cm 2 and UVB = 2 J/cm 2 ). Photo-exposed SSL were subjected to TLC and squalene containing spot was collected as per Materials and Methods. Squalene samples or 0.1% methanol (a vehicle for squalene) were added to NHEK cultures (final concentration -1 mg/mL) for 3 h. Then, cells were thoroughly washed and RNA was isolated. have been focused on the endogenous photo-receptors localised in the viable cellular compartments of the skin, such as DNA, urocanic acid, melanin precursors, porphyrins, tryptophan, dihydropyridins, etc. [2], [14], [39]. These cell-located molecules selectively absorb UV light within a distinct wavelength range and, in the presence of oxygen, initiate oxidative photo-reactions affecting cell signalling and enzymatic activities, which result in characteristic UV responses of the skin cells and UV-induced changes of extracellular skin matrix. Lipid skin surface mantel represents the first-line protective barrier between solar UV and viable human skin structures: it directly and very effectively absorbs UVB and UVA rays [2], [5], inhibits free radical-driven lipid peroxidation (a-tocopherol and coenzyme Q 10 ) [1][2][3][4], and scavenges reactive oxygen species such as singlet oxygen (squalene) [6]. Here, we confirmed that at least two UV-sensitive SSL constituents, a-tocopherol and Sq, were dose-dependently degraded during exposure to solar-simulated UVA+UVB (Fig. 7).
The products of SSL photo-oxidation, such as Sq monohydroperoxides and epoxides (collectively, SqPx), aldehydes (Fig. 7A), and 9-HODA may easily diffuse in the skin depth reaching viable keratinocytes and melanocytes beneath stratum corneum [1], [2], thus exerting their UV-like biological effects. Indeed, SqPx was shown to induce melanocyte dendricity and melanogenesis [3] and IL-6 release from HaCaT [10]. Nuclear translocation of UVB-induced EGFR in primary human keratinocytes was mediated by 4-HNE [27], [35], and a UVB-like cytokine pattern was observed upon keratinocyte exposure to 9-HODA through UVB-like expression of G2A receptor [31]. Comparing data on direct UV-associated ( Fig. 1) and photo-oxidised oxSSL-and SqPx-induced effects (Fig. 8) related to the expression of pro-inflammatory cytokines (TNFa, IL-6, IL-8, and IL-1b) or keratinocyte membrane receptors important for cytokine production (IL-1R1 and TLR4), a striking similarity could be noticed. At the same time, the TLR4 coactivator MYD88, essential for downstream gene transcription [25], [40] was down-regulated by the lipids. Also, in accord to published data on G2A up-regulation by UVB [31] , SqPx exerted similar action (Fig. 8C). Of importance, naive SSL collected from non-photoexposed human skin, as well as Sq isolated from these lipids were practically inactive, while SqPx alone induced the highest gene up-regulation as compared to the photo-oxidised lipid mixture oxSSL containing the same content of SqPx. We hypothesised that the complex mixture may also contain components such as linoleic acid with a mode of action opposite to SqPx. Characteristically, up-regulation of the secondary cytokine IL-6 was slower upon irradiation than upon SqPx action (temporary down-regulation at 3 h in the latter case and a 4-fold up-regulation in the former), which suggests the necessity of UVinduced oxidative signalling molecules to start IL-6 transcription [12]. With regard to induction of COX-2, another marker of UV response (Fig. 1C), exclusively SqPx exhibited opposite inhibiting action (Fig. 8B). COX-2 [11], [23], [28], [29] as well as iNOS [23] expression is controlled by EGFR activation and nuclear translocation. In a good agreement with these data, solarsimulated UV induced fast (maximum at 5 min) EGFR phosphorylation (Figs. 5A and 5B), followed by downstream ERK phosphorylation (gradual increase until 20 min) and nuclear translocation of both EGFR and P-EGFR ( Fig. 5C and 5D). Logically, time-dependent COX-2 induction was observed. We hypothesised that inhibitory effects of SqPx and SqPx-in vivo towards COX-2 could be associated with inhibited phosphorylation (Fig. 9C), followed by inhibited nuclear translocation of EGFR. On contrast, very fast COX-2 and iNOS induction was associated with NHEK exposure to non-toxic dosage of 4-HNE ( Fig. 6C), able to mediate UVB-induced EGFR phosphorylation and translocation to the nucleus [28], [35]. Usually, iNOS upregulation becomes detectable at 6 h post UV [41]. However, in the presence of the pre-formed redox mediator (4-HNE), a statistically significant difference was seen at 4 h. At the same time, Figure 9. Effects of intact and UV-exposed skin surface lipids on signal transduction and AhR-CYP1 axis in NHEK. NHEK cultures were exposed to skin surface lipids normalised by the content of squalene (Sq): intact freshly prepared skin surface lipids (SSL, with final Sq content 1 mg/mL), Sq isolated from the SSL (final content 1 mg/ mL), SSL exposed to UVA+UVB for 20 min (oxSSL with final content of oxidised Sq (PxSq) 1 mg/mL), or SqPx isolated from oxSSL (final content of SqPx 1 mg/mL), or SqPx-in vivo isolated from the in vivo photoexposed skin surface lipids (final content of SqPx-in vivo 1 mg/mL). After 20 min of co-incubation, differential lysis was carried out, cytoplasmic and nuclear proteins extracted, and subjected to electrophoresis (A). Western blots were quantified by densitometry: (B) nuclear levels of AhR and (C) cytoplasmic levels of phosphorylated EGFR (P-EGFR), ERK (P-ERK), and the NFkB constituent p65 (P-p65). Transcripts of CYP1A1 and CYP1B1 were determined by RT-PCR at 3 h co-incubation (D). *P,0.05 and 1 P,0.01 versus control treated with the vehicle for the lipids (0.1% methanol). doi:10.1371/journal.pone.0044472.g009 4-HNE did not completely mimic UV in terms of inflammatory responses in NHEK, since it did not practically induce cytokines (Figs. 6D and 6E) while dramatically, although temporarily, inhibited ERK and Akt1 phosphorylation downstream EGFR (Figs. 6A and 6B).
The AhR has been suggested to be a light sensor in keratinocytes and melanocytes, since its transcriptional function and downstream CYP1 enzymes are activated by both visible and UV light, and mediated by endogenous high affinity AhR ligand FICZ, the tryptophan photo-product [11], [13], [14], [34]. Accordingly, in our hands solar-simulated UV time-dependently stimulated AhR machine (20 min, Figs. 2A and 2B), followed by CYP1A1 and CYP1B1 over-expression reaching maximum at 12 h (Fig. 2C). Similar events were induced by FICZ (Fig. 3) although maximal CYP1 expression was reached earlier (4 h). Experiments with AhR silencing showed that CYP1A1 expression was AhR-dependent (Figs. 4A and 4B), however, 1 mM FICZ slightly but significantly up-regulated AhR transcript at 6 h (Fig. 3D), but down-regulated this gene at 24 h (Fig. 4B), possibly due to negative feed-back mechanisms active in NHEK at this last time-point. Of utmost importance, we showed for the first time, that all the types of skin surface lipids studied induced nuclear translocation of AhR, although SqPx was the best, and oxSSL were the worst inducers (Figs. 9A and 9B). SqPx (approx. 1.5 mM) induced a much higher activation of AhR than FICZ (1 mM), therefore it can be regarded as a high affinity ligand for the receptor, as dioxin [13]. AhR activation resulted in an enhanced CYP1A1/CYP1B1 expression only in the presence of SqPx (Fig. 9D). The strange co-incidence that both FICZ and SqPx also inhibited EGFR phosphorylation could be a hint on the possible involvement of EGFR in the mechanism of CYP1A1/ CYP1B1 induction. However, in our recent publication [36], we reported that resveratrol, being a strong inducer of AhR-CYP1 axis, did not suppress spontaneous or pro-inflammatory stimuliinduced EGFR phosphorylation. On these grounds, we concluded that EGFR pathway was not directly involved in the stimulation of classical AhR-CYP metabolic mechanism activated by both FICZ and SqPx. The induction of distinct inflammatory responses by SqPx could be EGFR-dependent in accord with previous publications [25], [30]. Repressed AhR and CYP1 in the presence of oxSSL corresponded to significantly activated NFkB (Fig. 9), probably induced by cholesterol oxidation products present in oxSSL [42][43][44]. Oxidative stress is known to induce NFkB pathway, whilst repressing AhR translocation and CYP1 transcription [34].
In general, a wide variety of lipid oxidation products such as 4-HNE [35], [42], cholesterol peroxides [44], enzymatically generated lipid hydroperoxides [46], and many others [43] have been considered endogenous signalling molecules for gene transcription. On these grounds, their roles in cell physiology and different human pathologies have been discussed. For the first time, we obtained convincing data that so far unidentified products of Sq photo-oxidation could also be regarded as lipid signalling molecules interacting with several transduction pathways in human keratinocytes. Of particular interest, Sq is present exclusively in the skin of humans, water apes, whales, otters, and other water mammals [47]. Its physiological role has been mainly discussed in terms of singlet oxygen scavenging, and thus photoprotection for the hairless human skin [2]. Our present data seem to open a new protective way of Sq action: the photo-oxidized metabolites of Sq are signalling molecules for keratinocytes, to mount adaptive/protective metabolic, proliferative, and inflammatory responses to the UV component of solar light. However, upon chronic exposure to UV or/and in the case of deregulated molecular pathways in the aged or pathological skin, photooxidized Sq may cause persistent inflammation and overstimulated skin metabolism. These processes may well be further exaggerated in the oily skin, overproducing sebum Sq. Collectively, SqPx may be regarded as a novel molecular target for the prevention and treatment of UV-associated skin defects/pathologies, such as solar hyperkeratosis, acute and chronic inflammation, acne eruption, premature aging, and epidermal tumours. The exact mechanism and cellular specificity of SqPx action should be further elucidated. Figure S1 Thin layer chromatogram of the lipid fractions of SSL, performed with elution system as described in Materials and Methods. The squalene fraction is identified, and the width of the band isolated and extracted for experiments framed in the rectangle. (TIF) Figure S2 GC-MS total ion chromatogram (elution conditions as described in Materials and Methods section, scan mode) of squalene isolated from irradiated SSL, confirming the purity of the extracted band. SSL sample was derivatised by N,O-bis-(trimethylsylil)-trifluoroacetamide with 1% trimethylchlorosilane. (TIF)