Endogenous ω-3 Fatty Acid Production by fat-1 Transgene and Topically Applied Docosahexaenoic Acid Protect against UVB-induced Mouse Skin Carcinogenesis

The present study was intended to explore the effects of endogenously produced ω-3 polyunsaturated fatty acids (PUFAs) on ultraviolet B (UVB)-induced skin inflammation and photocarcinogenesis using hairless fat-1 transgenic mice harboring ω-3 desaturase gene capable of converting ω-6 to ω-3 PUFAs. Upon exposure to UVB irradiation, fat-1 transgenic mice exhibited a significantly reduced epidermal hyperplasia, oxidative skin damage, and photocarcinogenesis as compared to wild type mice. The transcription factor, Nrf2 is a master regulator of anti-inflammatory and antioxidant gene expression. While the protein expression of Nrf2 was markedly enhanced, the level of its mRNA transcript was barely changed in the fat-1 transgenic mouse skin. Topical application of docosahexaenoic acid (DHA), a representative ω-3 PUFA, in wild type hairless mice induced expression of the Nrf2 target protein, heme oxygenase-1 in the skin and protected against UVB-induced oxidative stress, inflammation and papillomagenesis. Furthermore, transient overexpression of fat-1 gene in mouse epidermal JB6 cells resulted in the enhanced accumulation of Nrf2 protein. Likewise, DHA treated to JB6 cells inhibited Nrf2 ubiquitination and stabilized it. Taken together, our results indicate that functional fat-1 and topically applied DHA potentiate cellular defense against UVB-induced skin inflammation and photocarcinogenesis through elevated activation of Nrf2 and upregulation of cytoprotective gene expression.

. UVB-induced inflammation was ameliorated in fat-1 transgenic and DHA-treated mouse skin. (A) fat-1 +/− haired mice (a) were crossed with SKH-1 hairless mice (b) to generate fat-1 −/− and fat-1 +/− progenies. Hairless fat-1 −/− and fat-1 +/− mice during (c-e) and after (f) first hair cycle. After an initial hair cycle, the skin became eventually hairless, suitable for subsequent studies undertaken. (B) Both hairless fat-1 −/− and fat-1 +/− littermates born to the same mother were maintained on a diet (10% safflower oil) high in ω-6 and scarce in ω-3 PUFAs. After the dietary regimen, the PUFA profiles of mouse skin were analyzed by gas chromatography. Significantly elevated levels of ω-3 vs. ω-6 fatty acids were observed in skin tissues of hairless fat-1 +/− mice as compared to those of hairless micelacking the fat-1 gene. Data are expressed as means ± SE. ***p < 0.001 versus WT mice. Newly developed female hairless fat-1 as well as WT mice (n = 5 per each treatment group) were irradiated on their backs with UVB light (180 mJ/cm 2 ) for 2 h. (C) Skin thickness in hairless fat-1 transgenic and WT mice was visualized by H&E staining at 2 h following UVB exposure. Formalin-fixed and paraffin-embedded tissues from UVB-irradiated mice were also immunostained for of antioxidant enzymes play a central role in counteracting excessive ROS accumulation, thereby maintaining the cellular redox balance. Examples are heme oxygenase-1 (HO-1) and NAD(P)H:quinone oxidoreductase 1 (NQO1). The proximal promoter regions of genes encoding HO-1 and NQO1 harbor a consensus sequence known as antioxidant response element (ARE) which is a preferred binding site of nuclear factor-erythroid related factor-2 (Nrf2) 15 .
In the present study, hairless fat-1 transgenic mice were generated by cross-breeding of male fat-1 +/− mice with female SKH-1 hairless mice. By utilizing these hairless fat-1 transgenic mice maintained on ω-6 PUFA containing diet, we investigated the effect of endogenously formed ω-3 PUFAs on the development of UVB-induced tumors as well as oxidative stress and inflammation in the skin.

Results
Generation of hairless fat-1 mice and their skin PUFA profile. To avoid the inconvenience of frequent hair removal in determining whether an increased ω-3 PUFA tissue accumulation in fat-1 mice is protective against skin carcinogenesis induced by repeated UVB irradiation, we generated hairless fat-1 mice by crossing fat-1 +/− haired and SKH-1 hairless mice (Fig. 1A). Thus, the fat-1 transgenic mice of the C57BL6 strain were subsequently backcrossed five times to SKH-1 hairless mice. The fat-1 +/− animals were segregated for the hairless phenotype. The resulting mice had the predominant genetic background of Skh:hr-1 hairless (>96.875%). All the experiments were performed by comparing mice of the same genetic background with or without carrying the fat-1 transgene. The hairless mutation showed normal development of the first hair coat in the first hair cycle. Starting at the 2 week of age, they lost their hair coat rapidly 24 . At weaning, they were completely hairless. Both hairless fat-1 −/− and fat-1 +/− littermates born to the same mother were maintained on an identical diet containing ω-6 PUFAs, but deficient in ω-3 ones. Analysis of the total lipids extracted from skin tissues of hairless fat-1 +/− mice showed a PUFA profile distinct from that of hairless fat-1 −/− (Fig. 1B and Table 1). There were significantly lower levels of ω-6 PUFAs and much higher proportions of ω-3 PUFAs in the skin of hairless fat-1 +/− mice compared to those fat-1 −/− mouse skin. The ratios of the ω-6 (18:2n-6, 18:3n-6, 20:3n-6 and 20:4n-6) to the ω-3 PUFAs (18:3n-3, 20:3n-3, 20:5n-3 and 22:6n-3) in skin tissues of hairless fat-1 −/− and fat-1 +/− mice were 15.48:1 and 3.77:1, respectively. This indicates that the transgene is functionally active and transmittable. UVB-induced acute skin inflammation was ameliorated in fat-1 transgenic and DHA-treated mice. The thickness of the non-keratinized epidermis was decreased in the skin of short-term UVB-irradiated hairless fat-1 mice compared with wild type (WT) hairless littermates (Fig. 1C, upper). Persistently activated STAT3 plays an important role in photocarcinogenesis through upregulation of genes involved in tumor-associated inflammation, anti-apoptosis and proliferation 25 . One of the essential events in activation of STAT3 signaling is phosphorylation at its tyrosine 705 residue (Tyr 705 ). As shown in Fig. 1C (lower), WT hairless mice irradiated with UVB exhibited markedly increased expression of P-STAT3 (Tyr 705 ) which appeared as brown color staining. A significantly reduced proportion of phosphorylated STAT3-positive cells was observed in fat-1 transgenic mouse skin irradiated with UVB for 2 h. fat-1 mouse skin also showed markedly decreased nuclear translocation of P-STAT3 after UVB irradiation compared with WT mouse skin (Fig. 1D). Topical application of DHA also ameliorated UVB-induced epidermal hyperplasia (Fig. 1E, upper) and attenuated UVB-induced expression of P-STAT3 (Tyr 705 ) in WT hairless mice (Fig. 1E, lower). UVB-induced skin tumor development was suppressed in hairless fat-1 mice compared with that in WT mice. After repeated UVB irradiation thrice a week for 23 weeks, representative photographs of mice from different groups were taken. Figure 2A shows that the tumor burden in the hairless fat-1 mice was much lower than that in the hairless WT mice. The tumor multiplicity between the two experimental groups was significantly different (Fig. 2B). In the hairless fat-1 mice, the average number of tumors per mouse was 7 compared with 16.5 in the WT littermates.
UVB-induced expression of COX-2 and phosphorylation of STAT3 were diminished in the papillomas of hairless fat-1 mice compared with those in the WT mice. Inappropriate upregulation of COX-2 induces malignant changes in epidermal keratinocytes 26 and prolongs the survival of malignant or transformed cells with metastatic potential 27 . STAT3 is a well known transcription factor involved in the regulation of phospho-STAT3 (Tyr 705 ), and counterstained with hematoxylin. Magnifications, ×100. Scale bar is 100 µm. (D) Nuclear extracts from different groups were subjected to SDS-PAGE and immunoblotted to detect phosphorylated forms of STAT3 (Tyr 705 ). **p < 0.01 versus UVB-irradiated WT mice. (E) Dorsal skin of female HR-1 hairless mice (n = 5 per treatment group) were treated topically with DHA (10 µmol) 30 min before UVB (180 mJ/cm 2 ) irradiation. Mice were sacrificed after 2 h of UVB irradiation. Irradiated skin tissue sections were stained with H&E to examine inflammatory changes resulting in increased skin thickness. Control animals were treated with acetone alone and left unirradiated. The sections of skin tissues were also subjected to immunohistochemical analysis of phosphorylated of STAT3 at Tyr 705 . Magnifications, ×200. Scale bar is 100 µm.
Scientific RepoRts | 7: 11658 | DOI:10.1038/s41598-017-11443-2 COX-2 expression 13 . A significant decrease in expression of COX-2 at both transcriptional ( Fig. 2C) and translational ( Fig. 2D) levels was noted in the papillomas of the hairless fat-1 mice. Concomitant with downregulation of COX-2, the level of P-STAT3 (Tyr 705 ) was also markedly decreased in fat-1 mouse skin irradiated with UVB ( Fig. 2D). The above findings were also verified by hematoxylin and eosin (H&E) staining and immunohistochemical analysis. The thickening of epidermal layer and inflammatory cell infiltration in WT mouse skin caused by long-term UVB exposure were reduced in fat-1 skin (Fig. 2E, upper). Likewise, expression of COX-2 ( Fig. 2E, middle) and P-STAT3 (Fig. 2E, lower) in the skin of fat-1 mice was substantially abrogated.
DHA inhibited UVB-induced papillomagenesis in hairless mouse skin. fat-1 mice have a substantially enhanced proportion of ω-3 PUFAs in which an increase in the DHA production is most prominent (Table 1). So, we examined the effect of repeated topical application of DHA on UVB-induced skin carcinogenesis in normal hairless mice. Figure 3A shows that UVB-induced skin papilloma formation was diminished by DHA pretreatment. Repeated topical application of DHA prior to UVB exposure significantly reduced the number of skin tumors per mouse (Fig. 3B). The inflammatory skin damage, as evidenced by H&E staining of skin papillomas, was ameliorated in the DHA-pretreated mice (Fig. 3C, upper). DHA treatment also significantly dampened the phosphorylation of STAT3 at the Tyr 705 in the skin papillomas (Fig. 3C, lower). UVB-induced oxidative skin damage was attenuated in fat-1 transgenic and DHA-treated mice. Multiple lines of evidence suggest that UVB-induced generation of ROS is responsible for skin hyperplasia 28 as well as lipid peroxidation 29 . 4-Hydroxy-2-nonenal (4-HNE) and malondialdehyde (MDA) are representative lipid peroxidation products 30,31 . There was lesser 4-HNE-induced protein modification in the hairless fat-1 mouse skin as compared to WT one upon acute UVB irradiation (Fig. 4A). Topical application of DHA also attenuated UVB-induced accumulation of 4-HNE-and MDA-modified proteins in mouse epidermis (Fig. 4B). Moreover, repeated exposure to UVB up to 23 weeks increased the 4-HNE-modified protein expression, which was blunted in the skin of DHA-pretreated ( Fig. 4C, upper) mice. Likewise, repeated topical application of DHA prior to UVB irradiation reduced the UVB-induced accumulation of MDA-modified protein in mouse skin papillomas (Fig. 4C, lower).

The Nrf2-mediated induction of cytoprotective gene expression was elevated in the skin of fat-1 and DHA-treated mice.
To investigate whether the decreased number of papillomas formed by UVB irradiation is attributable to augmentation of cellular defense capacity that counteracts excessive ROS accumulation, we analyzed the antioxidant gene expression in the skin of fat-1 and DHA-treated mice. We found a robust enhancement in the expression of two representative cytoprotective enzymes HO-1 and NQO1, at both transcriptional and translational levels, in the fat-1 mice fed diet containing 10% safflower oil rich in ω-6 PUFAs ( Fig. 5A and B). Nrf2 plays a fundamental role in transcriptional activation of genes encoding HO-1 and NQO1. Elevated accumulation of Nrf2 was observed in the skin of fat-1 mice ( Fig. 5B and C). However, no significant change in the epidermal level of Nrf2 mRNA was found in the fat-1 mice compared with their respective age-matched skin of WT mice (Fig. 5A). Consistently, the stabilization of Nrf2 protein and upregulation of its target protein, HO-1 were observed in the hairless fat-1 mouse skin at the 23 rd week, as measured by Western blot analysis (Fig. 5D). In addition, repeated application of DHA thrice a week for 23 weeks induced the HO-1 expression in mouse skin (Fig. 5E).

UVB-induced inflammation and oxidative stress were less severe in mouse embryonic fibroblasts (MEFs) from fat-1 mice than those from WT animals.
To further confirm the in vivo findings, MEFs were isolated from both WT and fat-1 transgenic mice. A significantly lower ω-6/ω-3 fatty acid ratio was observed in fat-1 MEFs as compared to WT MEFs ( Fig. 6A and Table 2). Reduced levels of ω-6 PUFAs, especially linoleic acid (LA) and arachidonic acid (AA), were found in the fat-1 MEFs as compared to the WT  MEFs. However, the reduction of ω-6 PUFA levels in the fat-1 MEFs was not as dramatic as increases in ω-3 fatty acids. To elucidate the protective role of the fat-1 transgene in UVB-induced oxidative insults, MEFs were incubated with two representative ω-6 PUFAs, LA and AA, followed by UVB irradiation. As illustrated in Fig. 6B, UVB-irradiated MEFs showed an enormous increase in the ROS accumulation as compared to control MEFs. The ROS generation was further enhanced in the fat-1 deficient MEFs incubated with ω-6 PUFAs, LA and AA. However, the decreased ω-6/ω-3 fatty acid ratio blocked UVB-induced ROS overproduction in the fat-1 MEFs.
The fat-1 MEFs exhibited upregulation of the two representative antioxidant enzymes, HO-1 and NQO1, at both transcriptional and translational levels, to a greater extent than WT MEFs upon treatment with LA plus AA (Fig. 6C). Further, siRNA knock down of nrf2 gene abrogated the expression of HO-1 and NQO1 in fat-1 MEFs (Fig. 6D). This finding indicates that Nrf2 is essential for the induction of cytoprotective gene expression in fat-1 MEFs.
DHA increased the stability of Nrf2 protein in mouse epidermal JB6 cells. In another experiment, mouse epidermal JB6 cells were transfected with either pCMV-HA-control vector or pCMV-HA-fat-1 vector.
Overexpression of fat-1 gene increased the Nrf2-mediated expression of cytoprotective enzymes (Fig. 7A). In contrast to the elevated Nrf2 protein expression, JB6 cells transiently transfected with pCMV-HA-fat-1 vector exhibited no significant alteration in the steady state level of its mRNA transcript (Fig. 7A). We also examined the effect of DHA on the nuclear localization of Nrf2 in cultured mouse epidermal JB6 cells. As shown in Fig. 7B, DHA increased the accumulation of Nrf2 in the nucleus, which was confirmed by immunocytochemical analysis.
To determine whether DHA stabilizes Nrf2 protein without inducing de novo synthesis, we monitored the degradation of basal and DHA-induced Nrf2 protein in JB6 cells after inhibition of protein synthesis by cycloheximide.
As illustrated in Fig. 7C, Nrf2 protein in untreated control cells rapidly degraded after addition of cycloheximide, whereas Nrf2 in the DHA-treated JB6 cells underwent degradation more slowly. Many labile regulatory proteins, including signal-activated transcription factors, are commonly degraded by the 26S proteasomes. Treatment of JB6 cells with the proteasome inhibitor, MG-132 significantly increased the level of Nrf2 (Fig. 7D). Moreover, most proteins subjected to proteasomal degradation are marked by prior, covalent ligation of ubiquitin molecules   to the ε-amino group of specific lysine residues. A ubiquitination assay revealed that the appearance of ubiquitinated Nrf2 was decreased with a concomitant increase in Nrf2 accumulation in cells treated with DHA (Fig. 7E).

Discussion
Mounting evidence from epidemiological and laboratory-based studies suggest that long chain ω-3 PUFAs possess a broad range of health beneficial properties 16 and exert protective effects against oxidative stress-induced inflammation 32 , aging 33 , cancer 34   The present study was intended to assess the chemopreventive effects of endogenously produced ω-3 PUFAs against photocarcinogenesis using fat-1 transgenic mice without the need for exogenous supply. Homozygous mutations in hairless gene cause a permanent hair loss, referred to as alopecia in both humans and mice 37 . The hairless mice have been used historically in studying experimentally induced skin toxicity and carcinogenesis 38 . To avoid the tedious procedure of repeated removal of hair for long-term UVB irradiation, we have generated hairless fat-1transgenic mice by cross-breeding of male fat-1 +/− transgenic mice with female SKH-1 hairless mice. We noticed that an inflammatory response to UVB irradiation was significantly alleviated in the hairless fat-1 +/− mice compared with that in the WT mice, which is associated with their relative resistance to UVB-induced papilloma formation.
COX-2 has been identified as a molecular link between inflammation and tumor promotion 39 . Abnormally elevated expression of COX-2 was observed in mouse skin papillomas formed after chronic exposure to UVB 40,41 . In the present study, the irradiation of hairless mice with UVB (180 mJ/cm 2 ) for 23 weeks resulted in a marked increase in COX-2 expression. Hairless fat-1 +/− mice capable of spontaneously producing ω-3 PUFAs exhibited significantly reduced expression of COX-2 in their dorsal skin compared with WT mice, and were protected from photocarcinogenesis. These findings indicate that not only topically applied, but also systemically administered ω-3 PUFAs could be efficient for chemoprevention of photocarcinogenesis.
A critical role for STAT3 in promoting proliferation, metastasis and immune evasion of tumor cells has been shown in various pathophysiologic conditions 42 . Kim et al. reported that constitutive activation of STAT3 enhanced UVB-induced skin carcinogenesis 25 . Likewise, STAT3 activation promoted keratinocyte survival and proliferation in response to UVB radiation 43 . Thus, inhibition of aberrantly overactivated STAT3 signaling is considered as a pragmatic approach for achieving the prevention of skin cancer. In our study, the repeated UVB irradiation caused Tyr 705 phosphorylation of STAT3 in the skin papillomas, and this was attenuated in fat-1 transgenic mice and DHA-treated WT mice.
ROS-mediated oxidative stress is capable of fostering the development of various pathological conditions mainly by overwhelming skin homeostatic antioxidant defenses. Body protects against oxidative stress and maintains the optimal intracellular redox environment through activation of a battery of antioxidant enzymes 15 . In line with this notion, mice expressing wild type HO-1 exhibited the lower multiplicity of skin papillomas than did HO-1 knockout animals 44 . Likewise, NQO1-null mice were susceptible to experimentally induced skin tumorigenesis to a greater extent than wild type mice 45,46 . Nrf2 plays a fundamental role in regulating expression of many anti-oxidative and cytoprotective proteins including HO-1 and NQO1. Nrf2 knockout mice were more prone to develop chemically induced skin tumors 47 . In our present study, one of the notable changes accompanying a lower degree of oxidative stress as well as inflammation and the reduction of papilloma formation in the skin tissues of fat-1 mice challenged with UVB irradiation was the Nrf2-mediated induction of cytoprotective gene expression.
Since fat-1 mice exhibited pronounced constitutive induction of HO-1 and NQO1 as well as accumulation of Nrf2 in their skin, we explored the mechanistic basis for the anti-oxidative effects elicited by endogenously produced ω-3 PUFAs by utilizing the fat-1 MEFs. The inhibition of HO-1 and NQO1 expression in the Nrf2 siRNA-transfected fat-1 MEFs suggests that the expression of these anti-oxidative enzymes was induced mainly through upregulation of Nrf2 signaling in the fat-1 mice. Furthermore, DHA enhanced nuclear translocation of Nrf2 through stabilization of this transcription factor by inhibiting ubiquitination and 26S proteasomal degradation in the JB6 cells.
In conclusion, the present study demonstrates that constitutive ω-3 fatty acid production in the fat-1 mice and topical application of DHA in wild type mice protect against UVB-induced inflammatory and oxidative skin damage and papillomagenesis through suppression of STAT3 activation and upregulation of Nrf2-mediated  Preparation and maintenance of MEFs. Male fat-1 +/− and female WT mice were paired and the pregnancies were monitored. Embryos were obtained at the day 13.5 after paring and used to prepare fibroblasts after removing head, heart and legs. The tails of the embryos were used to confirm the fat-1 +/− genotype by PCR, and the embryo bodies were minced into small pieces and cultured in high glucose Dulbecco's modified Eagle's medium (DMEM; Gibco BRL, Grand Island, NY) supplemented with 10% fetal bovine serum (FBS; Gibco BRL, Grand Island, NY) and kept at 37 °C with 5% CO 2 .
Analysis of PUFA composition. The levels of representative ω-6 and ω-3 PUFAs were determined by using gas chromatography as described previously 48  Histological and immunohistochemical analyses of paraffin-embedded skin sections. Sections of harvested mouse skin were washed with phosphate-buffered saline (PBS) and fixed with 10% buffered formalin and embedded in paraffin. Each section (4 μm) was stained with H&E. The H&E stained sections were examined under light microscope to detect the presence of lesions. Immunohistochemistry (IHC) was performed to detect the protein expression. The sections were cut on silanized glass slides, deparaffinized three times with xylene and rehydrated through graded alcohol bath. For the detection of respective protein expression, slides were incubated with antibodies raised against P-STAT3 (Cell Signaling Technology, Beverly, MA), COX-2 (Cayman Chemical Co., Ann Arbor, MI), proteins modified with 4-HNE and MDA (JaICA, Nikken SEIL Co. Ltd., Shizuoka, Japan) and Nrf2 (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), and then developed using horseradish peroxidase-conjugated secondary antibodies (rabbit or mouse) (Dako, Glostrup, Denmark). The peroxidase binding sites were detected by staining with 3,3′-diaminobenzidine tetrahydrochloride. Finally, counterstaining was performed using Mayer's hematoxylin.
Tissue lysis and protein extraction.  (1:100) of primary anti-Nrf2 antibody was made in PBST with 5% BSA, and cells were incubated overnight at 4 °C. After three washing steps with PBST, the cells were incubated with a fluoresceinisothiocyanate-conjugated goat anti-mouse IgG secondary antibody in PBST with 5% BSA for 1 h at room temperature. Cells were also stained with propodium iodide and rinsed with PBST. Stained cells were analyzed under a confocal microscope (Leica Microsystems, Heidelberg, Germany) and photographed.
Immunoprecipitation. JB6 cells were treated with DHA and/or AA for 6 h and lysed in 250 mM sucrose, 50 mM Tris-HCl (pH 8.0), 25 mM KCl, 5 mM MgCl 2 , 1 mM EDTA, 2 μM NaF, 2 μM sodium orthovanadate, 1 mM PMSF and 10 mM N-ethylmaleimide. Total protein (500 μg) was subjected to immunoprecipitation by shaking with Nrf2 primary antibody at 4 °C for 24 h followed by the addition of 40 μL of 25% protein G-agarose bead slurry (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) and additional shaking for 2 h at 4 °C. After centrifugation at 10,000 g for 1 min, immunoprecipitated beads were collected by discarding the supernatant and washed with cell lysis buffer. After final wash, immunoprecipitate was resuspended in 50 μL of 2X SDS electrophoresis sample buffer and boiled for 5 min. Forty five-μL of supernatant from each sample was loaded on SDS-PAGE. The ubiquitinated Nrf2 was visualized by antibody against ubiquitin (Life Technologies, Carlsbad, CA).

Statistical analysis.
Except for the data on the tumor multiplicity expressed as the mean ± standard deviation (SD), all other values were expressed as the mean ± standard error (SE) of at least three independent experiments. Statistical significance was determined by Student's t-test and a p-value of less than 0.05 was considered to be statistically significant.
Data availability. All data generated or analysed during this study are included in this published article and its Supplementary Information files.