7,3′,4′-Trihydroxyisoflavone, a Metabolite of the Soy Isoflavone Daidzein, Suppresses Ultraviolet B-induced Skin Cancer by Targeting Cot and MKK4*

Nonmelanoma skin cancer is one of the most frequently occurring cancers in the United States. Chronic exposure to UVB irradiation is a major cause of this cancer. Daidzein, along with genistein, is a major isoflavone found in soybeans; however, little is known about the chemopreventive effects of daidzein and its metabolites in UVB-induced skin cancer. Here, we found that 7,3′,4′-trihydroxyisoflavone (THIF), a major metabolite of daidzein, effectively inhibits UVB-induced cyclooxygenase 2 (COX-2) expression through the inhibition of NF-κB transcription activity in mouse skin epidermal JB6 P+ cells. In contrast, daidzein had no effect on COX-2 expression levels. Data from Western blot and kinase assays showed that 7,3′,4′-THIF inhibited Cot and MKK4 activity, thereby suppressing UVB-induced phosphorylation of mitogen-activated protein kinases. Pull-down assays indicated that 7,3′,4′-THIF competed with ATP to inhibit Cot or MKK4 activity. Topical application of 7,3′,4′-THIF clearly suppressed the incidence and multiplicity of UVB-induced tumors in hairless mouse skin. Hairless mouse skin results also showed that 7,3′,4′-THIF inhibits Cot or MKK4 kinase activity directly, resulting in suppressed UVB-induced COX-2 expression. A docking study revealed that 7,3′,4′-THIF, but not daidzein, easily docked to the ATP binding site of Cot and MKK4, which is located between the N- and C-lobes of the kinase domain. Collectively, these results provide insight into the biological actions of 7,3′,4′-THIF, a potential skin cancer chemopreventive agent.

Nonmelanoma skin cancer is one of the most frequently occurring cancers in the United States (1). UV irradiation from sunlight is the major etiologic factor in the development of nonmelanoma skin cancers, including squamous cell carcinomas and basal cell carcinomas (2). Among the forms of solar irradiation, UVB (290 -320 nm) exhibits highly mutagenic and carcinogenic effects in animal experiments, as compared with UVA (320 -400 nm) (3). Prolonged exposure to UVB irradiation causes the development of benign epidermal tumors, most of which become skin carcinomas because UVB functions as a complete carcinogen (4). Therefore, targeting UVB-induced molecular and signaling mechanisms might be effective approaches for the chemoprevention of skin cancer.
Cyclooxygenase 2 (COX-2) 4 is an essential enzyme mediating the conversion of arachidonic acid to prostaglandin, the inducible isoform of cyclooxygenase (1). The inflammatory process affects numerous human malignancies, including skin cancer, by promoting epidermal hyperproliferation and hyperplasia through the secretion of various inflammatory factors, such as prostaglandin E2 (2). In human skin, increased COX-2 expression is observed in response to acute UVB irradiation (5). Similarly, in the mouse model, COX-2 is overexpressed in hyperplastic skin, benign tumors, and malignant tumors following chronic UVB irradiation (2). Indeed, celecoxib, a COX-2-selective inhibitor, in dietary or topical treatment in mice following UVB exposure, attenuated the number and volume of tumors (2). Therefore, regulating the expression of COX-2 might be a potential protection strategy against skin cancer.
Cot was initially identified in a human thyroid carcinoma cell line and in moloney murine leukemia virus-induced rat T-cell lymphomas (Tpl2) (6). Stimulation of Cot causes the activation of a p38 MAPK, JNKs, and transcription factors NF-B and nuclear factor of activated T cells (7). Numerous studies have shown that Cot mediates cytokine production, such as TNF-␣, and leads to several inflammatory diseases, such as rheumatoid arthritis. Recently, several studies have revealed that the activation of Cot is observed in tumor cells, such as T cell neoplasia cell lines (7). Cot also positively regulates COX-2 expression by LPS in macrophages (8). However, the involvement of Cot in UVB-induced COX-2 expression in skin has not been reported.
The mitogen-activated protein kinase kinase 4 (MKK4), a dual-specificity kinase, plays a critical role in the SAPK signaling pathway. The SAPK pathways, including p38 MAPK and JNKs, is activated in response to environmental stressors and extracellular stimuli (9). MKK3 and MKK6 are specific upstream kinases of p38, whereas MKK4 activates both p38 and JNKs (10 -12). Phosphorylation of p38 and JNKs mediates the inflammatory response, thereby inducing the eukaryotic transcription factor NF-B activation (13). In mouse skin, UV irradiation activates p38 and JNKs, which contribute to the development of skin cancer (13). Therefore, regulating the upstream molecules of p38 and JNKs is one possible strategy for the chemoprevention of skin carcinogenesis.
Daidzein and genistein are naturally occurring isoflavones in soy foods. In particular, genistein has received attention as a potential anticarcinogenic compound and has been intensely studied, whereas daidzein and its metabolites have been less well studied. Here, we investigated the chemopreventive effects of 7,3Ј,4Ј-trihydroxyisoflavone (THIF), a major metabolite of daidzein, against UVB-induced skin cancer. We report that 7,3Ј,4Ј-THIF, but not daidzein, is an ATP-competitive inhibitor of Cot and MKK4, and subsequently suppresses UVB-induced COX-2 expression in JB6 Pϩ mouse epidermal cells. In a mouse skin tumorigenesis model, 7,3Ј,4Ј-THIF strongly suppressed the incidence, multiplicity, and volume of UVB-induced mouse skin tumors. Consistent with the tumor data, 7,3Ј,4Ј-THIF clearly attenuated UVB-induced COX-2 expression in hairless mouse skin. Furthermore, 7,3Ј,4Ј-THIF directly bound with Cot or MKK4, resulting in the suppression of Cot or MKK4 activity in hairless mouse skin.
Cell Culture-JB6 Pϩ mouse epidermal (JB6 Pϩ) cell lines were cultured in monolayers at 37°C in a 5% CO 2 incubator in MEM containing 5% FBS, 2 mM L-glutamine, and 25 g/ml gentamicin. The cells were stably transfected with an NF-B luciferase reporter plasmid and maintained in MEM supplemented with 5% FBS containing 200 g/ml G418.
UVB Irradiation-UVB irradiation was carried out in a UVB chamber with a transluminator emitting UVB light protons and fitted with a Kodacel K6808 filter (Eastman Kodak, Rochester, NY) that eliminates all wavelengths below 290 nm. This lamp is one of the most frequently used UVB sources for the study of skin carcinogenesis. Irradiation energy was measured using a UVX radiometer (UVX-31) from UVP (Upland, CA).
Luciferase Assay for COX-2 Promoter Activity or NF-B Transcription Activity-COX-2 (14) or NF-B (15,16) luciferase reporter-transfected JB6 Pϩ cells were constructed as described earlier. The cells (8 ϫ 10 3 ), suspended in 100 l of 5% FBS/MEM, were added to each well of a 96-well plate and incubated at 37°C in a humidified atmosphere of 5% CO 2 . When cells reached 80 to 90% confluence, they were starved in 0.1% FBS/MEM for an additional 24 h. The cells were then treated for 1 h with 7,3Ј,4Ј-THIF or daidzein and then exposed to UVB for 24 h. After treatment, cells were disrupted with 100 l of lysis buffer (0.1 M potassium phosphate buffer (pH 7.8), 1% Triton X-100, 1 mM DTT, 2 mM EDTA), and the luciferase activity was measured using a luminometer (Luminoskan Ascent, Thermo Electron).
Western Blot Analysis-After the cells (1.5 ϫ 10 6 ) were cultured in a 10-cm dish for 48 h, they were starved in 0.1% FBS/ MEM for an additional 24 h. They were then treated with 7,3Ј,4Ј-THIF for 1 h before exposure to UVB (4 kJ/m 2 ) and harvested 30 min later. The harvested cells were disrupted, and the supernatant fractions were boiled for 5 min. The protein concentration was determined using a dye-binding protein assay kit according to the manufacturer's protocol. Lysate protein (30 g) was subjected to 10% SDS-PAGE and transferred to a polyvinylidene difluoride membrane. After blotting, the membrane was incubated with the specific primary antibody at 4°C overnight. Protein bands were visualized by a chemiluminescence detection kit after hybridization with a horseradish peroxidase-conjugated secondary antibody.
In Vitro MKK4 and Cot1 Kinase Assays-The in vitro kinase assays were performed in accordance with the instructions provided by Cell Signaling Technology. Briefly, for the MKK4 assay, 40 ng/l active MKK4 recombinant murine protein (55.8% purity and 964 units/mg activity) and 7,3Ј,4Ј-THIF (20, 40, or 60 M) were reacted at 30°C for 10 min. For each reaction, 5 l of 2 ϫ kinase buffer (10 mM MOPS (pH 7.2), 5 mM ␤-glycerol phosphate, 2 mM EGTA, 0.8 mM EDTA, 10 mM MgCl 2 , 0.1 mM DTT), 5 l of 250 M ATP, and 0.2 g/l of the inactive JNK2 were added. The mixtures were incubated at 30°C for 15 min. A 5-l aliquot was removed from the reaction mixture containing 10 l of 2 mg/ml of ATF-2 substrate peptide, 5 l of 2ϫ kinase buffer, and 5 l of 0.16 Ci/l [ 32 P]ATP solution, and incubated at 30°C for 15 min. Then, 20-l aliquots were transferred onto p81 filter paper and washed three times with 1% phosphoric acid for 5 min per wash and once with acetone for 5 min. Radioactive incorporation was determined using a scintillation counter (LS6500; Beckman Coulter). Each experiment was performed three times. For the Cot1 assay, 150 ng/l active Cot1 recombinant human protein (62% purity and 276 units/mg activity) and 7,3Ј,4Ј-THIF (20, 40, or 60 M) were reacted at 30°C for 10 min. For each reaction, 5 l of 2ϫ kinase buffer, 1 g/ml of p65 as substrate, and 5 l of 0.16 Ci/l [ 32 P]ATP solution were added. The mixture was reacted at 30°C for 15 min. Then, 20-l aliquots were transferred onto p81 filter paper and washed three times with 1% phosphoric acid for 5 min per wash and once with acetone for 2 min. Radioactive incorporation was determined using a scintillation counter (LS6500, Beckman Coulter). Each experiment was performed separately three times.
Ex Vivo MKK4 and Cot Immunoprecipitation and Kinase Assays-JB6 Pϩ cells were cultured to 80% confluence and then serum-starved in 0.1% FBS/MEM for 24 h at 37°C. Cells were either treated with 7,3Ј,4Ј-THIF or left untreated for 1 h, exposed to 4 kJ/m 2 UVB, harvested after 30 min, disrupted with lysis buffer (20 mM Tris-HCl (pH 7.4), 1 mM EDTA, 150 mM NaCl, 1 mM EGTA, 1% Triton X-100, 1 mM ␤-glycerophosphate, 1 mg/ml leupeptin, 1 mM Na 3 VO 4 , 1 mM PMSF), and finally centrifuged at 14,000 rpm for 10 min in a microcentrifuge. The lysates, each containing 500 g of protein, were used for immunoprecipitation with an antibody against MKK4 or Cot and then incubated at 4°C for 4 h. Protein A/G Plus-agarose beads were then added, and the mixture was continuously rotated overnight at 4°C. The beads were then washed three times with kinase buffer (20 mM MOPS (pH 7.2), 25 mM ␤-glycerol phosphate, 5 mM EGTA, 1 mM sodium orthovanadate, 1 mM DTT), and then a kinase reaction was performed in the same manner as for in vitro kinase assays.
Mouse Skin Tumorigenesis Analysis-SKH-1 hairless mice (5 weeks of age; mean body weight, 25 g) were purchased from the Institute of Laboratory Animal Resources at Seoul National University. Animals were acclimated for 1 week before the study and had free access to food and water. The animals were housed in climate-controlled quarters (24°C at 50% humidity) with a 12-h light/12-h dark cycle. Skin carcinogenesis was induced in mice receiving UVB irradiation over the course of 27 weeks. The UVB radiation source (Bio-Link cross-linker, Vilber Lourmat) emitted wavelengths with peak emission at 312 nm. SKH-1 mice were divided into four groups of 12 animals each. In the control group, the dorsal skin was topically treated with 200 l of acetone only. In the UVB group, the dorsal skin was topically treated with 200 l of acetone 1 h before UVB exposure. The mice in the third and fourth groups received a topical application of 7,3Ј,4Ј-THIF (10 or 40 nmol, respectively) in 200 l of acetone 1 h before UVB irradiation. The UVB dose was 0.18 J/cm 2 given three times/week for 27 weeks. The incidence of skin tumors was recorded weekly, with a tumor defined as an outgrowth of Ͼ1 mm in diameter that persisted for 2 weeks or more. Tumor incidence, multiplicity, and volume were recorded every week until the end of the experiment at the 27th week.
In Vivo Western Blot Analysis-For in vivo Western blotting, mice received a topical application of 7,3Ј,4Ј-THIF (10 or 40 nmol) in 200 l acetone on their backs 1 h before UVB irradiation. To isolate protein from mouse skin, the dorsal skin of each mouse was excised and placed on ice. Any fat was removed, and the skin was placed in liquid nitrogen and immediately pulverized with a mortar and pestle. The pulverized skin was blended on ice with a homogenizer (IKA T10 basic, IKA Laboratory Equipment), and skin lysates were centrifuged at 12,000 rpm for 20 min. After the protein content was determined using the Bio-Rad protein assay kit, 100 g protein from mouse skin extract was subjected to 10% SDS-PAGE and transferred to a PVDF membrane (Amersham Biosciences). Membranes were processed, and proteins were analyzed as described above for the in vitro Western blot assay.
Short-term in Vivo Model-Imprinting control region mice (5 weeks of age; mean body weight, 25 g) were purchased from the Institute of Laboratory Animal Resources at Seoul National University. ICR mice were maintained under the same condition as SKH-1 hairless mice described above. The dorsal areas of ICR mice were shaved before sample treatment. Each group was topically treated with 7,3Ј,4Ј-THIF (40 or 160 nmol) or daidzein (160 nmol) in 200 l of acetone 1 h before UVB (0.5 J/cm 2 ) irradiation. The protein extract for Western blot analysis was prepared as described above.
In Vivo Kinase and Pull-down Assays-For the in vivo Cot or MKK4 immunoprecipitation and kinase assay, mice were treated with 7,3Ј,4Ј-THIF (10 or 40 nmol) in 200 l of acetone, and dorsal skin was prepared as for in vivo Western blotting. Proteins were extracted as above and centrifuged at 12,000 rpm for 20 min. 700 g of protein from mouse skin extract was mixed with protein-A/G beads (20 l) for 1 h at 4°C. The mixture was processed, and radioactive incorporation was determined as for the ex vivo assay described above. Data are presented as the mean of data points from five mice in each group.
For the in vivo pull-down assay, mice received a topical application of 200 l of acetone alone or 7,3Ј,4Ј-THIF (10 or 40 nmol, respectively) in 200 l of acetone on their backs 1 h before UVB irradiation. Dorsal skin was prepared as described above for the in vivo Western blotting, and proteins were extracted as described above for the Cot or MKK4 immunoprecipitation and kinase assays. Then 500 g of protein from mouse skin extract were incubated with 7,3Ј,4Ј-THIF-Sepharose 4B (or Sepharose 4B alone as a control) beads (100 l, 50% slurry) in reaction buffer as described for the ex vivo pull-down assay. Beads were incubated and washed, and proteins bound to the beads were analyzed by Western blotting as described above.
Molecular Modeling-The homology model structures of Cot and MKK4 were generated by Geno3D using the coordinates of Mst1 and MKK7 (PDB codes 3COM and 2DYL), respectively. Insight II (Accelrys, Inc., San Diego, CA) was used for the docking study and structure analysis.
Statistical Analysis-When necessary, data were expressed as mean Ϯ S.D., and analysis of variance was used for multiple

7,3,4-Trihydroxyisoflavone Suppresses Skin Cancer
statistical comparisons. A probability value of p Ͻ 0.05 was used as the criterion for statistical significance.

7,3Ј,4Ј-THIF Inhibits UVB-induced COX-2 Expression by Suppressing COX-2 Promoter Activity and NF-B Transcriptional Activity in JB6 Pϩ Cells, Whereas Daidzein Has No Effect-
As aberrant COX-2 expression is intimately involved in multiple malignancies, including skin cancer (17), we first examined the possible inhibitory effects of 7,3Ј,4Ј-THIF on UVB-induced COX-2 expression in JB6 Pϩ cells. Treatment with 7,3Ј,4Ј-THIF dose-dependently inhibited UVB-induced COX-2 expression in these cells (Fig. 1B, left panel), whereas daidzein, even up to 60 M, had no effect (Fig. 1B, right panel). Consistent with the Western blotting results, 7,3Ј,4Ј-THIF strongly suppressed UVB-induced COX-2 promoter activity in JB6 Pϩ cells stably transfected with a COX-2 luciferase reporter plasmid in a dose-dependent manner (Fig. 1C, left panel), but daidzein had The cells were treated with 7,3Ј,4Ј-THIF or daidzein at the indicated concentrations for 1 h before exposure to UVB (4 kJ/m 2 ) and harvested 4 h later. The levels of COX-2 expression were then determined by Western blot analysis, as described under "Materials and Methods," using specific antibodies against the corresponding COX-2 and ␤-actin proteins. Data are representative of three independent experiments. The protein levels were quantified using an image analysis program to evaluate the density of each band on the immunoblot, and the fold value was calculated. C and D, effects of 7,3Ј,4Ј-THIF or daidzein on UVB-induced COX-2 (C) or NF-B (D) activity in JB6 Pϩ cells. In the luciferase assay, JB6 P ϩ stably transfected with the COX-2 or NF-B luciferase reporter plasmid were cultured as described under "Materials and Methods." After reaching 80% confluence, the cells were cultured in 0.1% FBS/MEM and then treated with 7,3Ј,4Ј-THIF or daidzein at the indicated concentrations or left untreated for 1 h before exposure to UVB (4 kJ/m 2 ) and harvested 24 h later. COX-2 or NF-B activity is expressed as the percent inhibition relative to cells treated with only UVB. Data are represented as the mean Ϯ S.D. of the luciferase activity calculated from three separate experiments. For C and D, the asterisk indicates significant differences between groups treated with UVB alone and the 7,3Ј,4Ј-THIF-or daidzein-treated groups. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001). APRIL

7,3,4-Trihydroxyisoflavone Suppresses Skin Cancer
no effect (Fig. 1C, right panel). NF-B is an important transcription factor regulating COX-2 expression in the skin (15). Thus, to investigate whether 7,3Ј,4Ј-THIF down-regulates UVB-induced COX-2 expression through the inhibition of the NF-B transcription factor, we measured NF-B transcription activity by using cells stably transfected with the NF-B luciferase reporter plasmid. 7,3Ј,4Ј-THIF significantly attenuated UVBinduced NF-B transcription activity (Fig. 1D, left panel). Compared with daidzein, treatment with 7,3Ј,4Ј-THIF at low concentrations inhibited NF-B transcription activity more effectively (Fig. 1D, right panel). 7,3Ј,4Ј-THIF Suppresses UVB-induced Phosphorylation of JNKs and p38 MAPK in JB6 Pϩ Cells-JNKs and p38 are generally referred to as stress-activated MAP kinases. We examined whether 7,3Ј,4Ј-THIF blocks the activation of the JNKs and p38 pathways stimulated by UVB in JB6 Pϩ cells. Western blotting data showed that 7,3Ј,4Ј-THIF inhibited UVB-induced phosphorylation of JNKs and p38 dose-dependently but had no effect on the phosphorylation of ERKs ( Fig. 2A, left panels). Also, 7,3Ј,4Ј-THIF did not affect IB kinase beta activity (data not shown). Recently, several studies indicated that stimulation of Cot leads to the activation of JNKs, p38, and NF-B (7). MKK3 and MKK6 activate p38, whereas MKK4 can activate both JNKs and p38 (18). Thus, we further investigated the effects of 7,3Ј,4Ј-THIF on the upstream regulatory proteins of the JNKs and p38 pathways. Our results revealed that 7,3Ј,4Ј-THIF inhibited both Cot and MKK4 activity in vitro (Fig. 2B, upper panels), but it did not affect the phosphorylation level of MKK4 or Cot in JB6 Pϩ cells ( Fig. 2A, right panels). Consistent with the results from the in vitro kinase assay, an ex vivo kinase assay also revealed that 7,3Ј,4Ј-THIF inhibited UVB-induced MKK4 and Cot activity in JB6 Pϩ cells (Fig. 2C). In contrast, 7,3Ј,4Ј-THIF had no effect on either the phosphorylation level of MKK3/6 ( Fig. 2A, right panels) or the in vitro kinase activity of MKK3 and MKK6 (Fig. 2B, lower panels). We next examined whether 7,3Ј,4Ј-THIF had an effect on the activity of various MAP kinases. The in vitro kinase assay showed that 7,3Ј,4Ј-THIF had no effect on ERKs, p38␣, or JNK1 kinase activity as h before being exposed to UVB (4 kJ/m 2 ) and harvested 30 min later. The cells were disrupted, and the levels of phosphorylated and total proteins were determined by Western blot analysis, as described under "Materials and Methods," using specific antibodies against the respective phosphorylated and total proteins. Data are representative of three independent experiments that gave similar results. B, 7,3Ј,4Ј-THIF inhibited both Cot and MKK4 in vitro. In contrast, MKK3 and MKK6 activity was not affected by 7,3Ј,4Ј-THIF. The in vitro kinase assay was performed as described under "Materials and Methods," and kinase activity is expressed as percent inhibition relative to the activity of the untreated kinase control. C, 7,3Ј,4Ј-THIF inhibited both Cot and MKK4 activity ex vivo. In the ex vivo Cot or MKK4 kinase assay, cells were pretreated with 7,3Ј,4Ј-THIF at the indicated concentrations (0, 20, 40, or 60 M) for 1 h and then exposed to UVB (4 kJ/m 2 ) and harvested after 30 min. Cells were used for immunoprecipitation, and the kinase assay was performed. Kinase activity is expressed as percent inhibition relative to cells treated with UVB only. The average 32 P count was determined from three separate experiments, and the data are presented as mean Ϯ S.D. In the ex vivo kinase assays, the asterisks indicate a significant decrease in kinase activity between cells treated with 7,3Ј,4Ј-THIF and cells treated with UVB only. **, p Ͻ 0.01; ***, p Ͻ 0.001). D, 7,3Ј,4Ј-THIF did not affect ERKs or MSK1 activity in vitro. The in vitro kinase assay was performed as described under "Materials and Methods," and kinase activity is expressed as percent inhibition relative to the activity of the untreated kinase control. For the in vitro kinase assays (B and D), the asterisk indicates a significant decrease in kinase activity between the groups treated with active Cot (or MKK4) and 7,3Ј,4Ј,-THIF and the group treated with active Cot (or MKK4) alone. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001.  APRIL 22, 2011 • VOLUME 286 • NUMBER 16 well as no effect on activity of MSK1, a downstream protein of p38 and ERKs (Fig. 2D). These results indicate that the inhibition of UVB-induced COX-2 expression by 7,3Ј,4Ј-THIF was mainly caused by the suppression of both Cot and MKK4 activity.

7,3Ј,4Ј-THIF Inhibits UVB-induced COX-2 Expression and Cot and MKK4 Activity by Directly Binding with Cot or MKK4
in SKH-1 Hairless Mouse Skin-To further confirm the inhibitory effect of 7,3Ј,4Ј-THIF on UVB-induced COX-2 expression in an in vivo model, we examined the level of COX-2 expression in SKH-1 hairless mouse skin. Consistent with the results from the JB6 Pϩ cells, the Western blotting data showed that the levels of COX-2 expression in the 7,3Ј,4Ј-THIF-treated groups The mice in the third and fourth groups received a topical application of 7,3Ј,4Ј-THIF (10 or 40 nmol, respectively, per mouse in 200 l acetone) on the dorsal surface 1 h before UVB (0.18 J/cm 2 ) irradiation 3 days/week for 27 weeks. The incidence of skin tumors was recorded weekly, and tumors were defined as an outgrowth of Ͼ1 mm in diameter persisting for 2 weeks or longer. Tumor incidence and multiplicity were recorded every week until the end of the experiment at 27 weeks. A, appearance of skin tumors. B, 7,3Ј,4Ј-THIF retarded the incidence of skin tumors compared with the UVB only-treated group. C, 7,3Ј,4Ј-THIF strongly inhibits UVB-induced tumor multiplicity in SKH-1 hairless mice. D, 7,3Ј,4Ј-THIF reduces UVB-induced tumor volume in SKH-1 hairless mice. At the end of the study, the dimensions of all tumors on each mouse were recorded. Tumor volumes were calculated using the hemiellipsoid model formula: tumor volume ϭ 1/2(4/ 3)(l/2)(w/2)h, wherein l is length, w is width, and h is height. The asterisks indicate a significant decrease in tumor incidence or volume between the group treated with UVB alone and the group treated with 7,3Ј,4Ј-THIF. **, p Ͻ 0.05; ***, p Ͻ 0.001.

DISCUSSION
Isoflavones are most abundant in soybeans but also exist in significant amounts in various other beans, legumes, sprouts, and clover. In plants, isoflavones are present as glycoside compounds, and on ingestion, the aglycons are easily detached from glucoside following reductive metabolism by intestinal bacteria (19). Several cellular and mouse studies have reported that soy extract (17) or isoflavones (20) have a photoprotective effect in skin. Recently, numerous studies have shown that isoflavones are subject to oxidative biotransformation in the rat liver (21) and in humans (22) through hepatic metabolism. One of the major products of the hepatic metabolism of daidzein is 7,3Ј,4Ј-THIF (19). However, the effects of this hepatic metabolite on UVB-induced skin cancer have not been reported. We found that 7,3Ј,4Ј-THIF had a stronger inhibitory effect on UVB-induced COX-2 expression in JB6 Pϩ cells than did 6,7,4Ј-THIF, another metabolite of daidzein (data not shown). In contrast, daidzein had no effect on UVB-induced COX-2 expression in JB6 Pϩ cells. In the present study, we demonstrated a strong chemopreventive effect of 7,3Ј,4Ј-THIF on Five mouse skin samples were randomly selected from 12 mouse skin samples of each group and analyzed for COX-2 expression by immunoblotting. B, COX-2 immunoblot results were normalized to ␤-actin followed by statistical analysis of relative image density. ##, significant difference (p Ͻ 0.01) between the control group and the untreated group; *, significant difference at p Ͻ 0.05 between 7,3Ј,4Ј-THIFtreated groups and untreated groups. C, 7,3Ј,4Ј-THIF directly binds with either Cot or MKK4 in SKH-1 hairless mouse skin. Cot or MKK4 -7,3Ј,4Ј-THIF binding was confirmed by immunoblotting using an antibody against Cot (upper panel) or MKK4 (lower panel). 1st lane (input control), whole skin lysates from SKH-1 hairless mice; 2nd lane (control), a lysate of mouse skin precipitated with Sepharose 4B beads; 3rd lane, whole skin lysates from SKH-1 hairless mice precipitated by 7,3Ј,4Ј-THIF-Sepharose 4B beads. D, 7,3Ј,4Ј-THIF inhibits UVB-induced Cot or MKK4 kinase activity in hairless mouse skin extracts. In the Cot (or MKK4) kinase assay, dorsal skin protein lysates were prepared from the epidermis, and immunoprecipitation and kinase assays were performed as described under "Materials and Methods." Columns, mean of the 32 P count from three separate experiments; bars, S.D., #, significant difference (p Ͻ 0.05); ##, significant difference (p Ͻ 0.001) between the control group and the group exposed to UVB; *, significant difference at p Ͻ 0.05 between groups treated with 7,3Ј,4Ј-THIF and untreated groups.
UVB-induced skin cancer and suggested a molecular mechanism and targets.
Therefore, the development of a natural inhibitor suppressing the aberrant expression of COX-2 is a promising strategy in the chemoprevention of skin cancer. Our results showed that 7,3Ј,4Ј-THIF exerts potent anti-tumor and anti-inflammatory effects. NF-B is an important transcription factor in tumorpromoting processes such as inflammation and proliferation, and it is an important factor in the development of skin cancer.

7,3,4-Trihydroxyisoflavone Suppresses Skin Cancer
Two estimated NF-B binding sites are present in the promoter region of COX-2, and COX-2 expression is positively regulated by NF-B. Our results showed that 7,3Ј,4Ј-THIF inhibits UVBinduced COX-2 expression by attenuating COX-2 promoter activity, which was attributed to the inhibition of NF-B transcription activation.
Research data indicate that MAP kinase phosphorylation triggers signals from the cell surface, activating various transcription factors, thereby contributing to the regulation of target gene expression. JNKs and p38 are generally recognized as stress-activated MAP kinases. Several kinds of MKKs can phosphorylate JNKs, and p38-MKK3 and MKK6 activate p38, whereas MKK4 can activate both JNKs and p38 MAPK (18). Increased levels of MKK4 are positively related to the increased proliferation and invasion of several cancer cell lines (26). Our results showed that 7,3Ј,4Ј-THIF suppressed UVB-induced phosphorylation of p38 and JNKs and was also effective at inhibiting MKK4 activity in vitro and ex vivo, resulting in the inhibition of COX-2 expression.
The serine/threonine protein kinase Cot has been studied mainly as a mediator of the inflammation or macrophage signaling pathways (27,28). Recent studies showed that Cot is up-regulated in a variety of cancers and that Cot positively regulates COX-2 expression and neoplastic cell transformation through transcriptional transactivation. Our previous study showed that Cot represents a novel serine/threonine kinase directly interacting with histone H3, resulting in the increased cell transformation (29). Cot regulates NF-B transcriptional activity by directly binding to p65 and triggering its phosphorylation (30). Our preliminary data showed that the Cot inhibitor down-regulated UVB-induced COX-2 expression by suppressing COX-2 promoter activity and NF-B transcription activity (data not shown). We then examined whether 7,3Ј,4Ј-THIF affected Cot activity, and the results showed that 7,3Ј,4Ј-THIF clearly suppressed Cot1 activity in vitro and ex vivo by directly binding with Cot1 competitively with ATP. Taken together, these results indicate that the inhibition of COX-2 expression by 7,3Ј,4Ј-THIF was attributable to the suppression of Cot and MKK4 activity.
The SKH-1 hairless mouse is an excellent model to study the appearance of skin tumors induced by chronic UV radiation (31). The topical application of 7,3Ј,4Ј-THIF strongly suppressed the incidence, multiplicity, and volume of UVB-induced skin cancer development in the SKH-1 hairless mouse model. Induction of COX-2 expression after skin exposure to UVB was substantial, and 7,3Ј,4Ј-THIF strongly inhibited the UVB-induced COX-2 expression in hairless mouse skin. Furthermore, 7,3Ј,4Ј-THIF directly bound with Cot or MKK4, resulting in suppression of Cot or MKK4 activity in both JB6 Pϩ cells and hairless mouse skin. Collectively, these results suggest that 7,3Ј,4Ј-THIF suppresses UVB-induced skin cancer by directly targeting Cot and MKK4.
To investigate the molecular basis of Cot and MKK4 inhibition by 7,3Ј,4Ј-THIF, we carried out a docking study using the homology model structures of the kinase domain of Cot, derived from the crystal structure of Mst1, which has 50% homology in amino acid sequence, and MKK4, derived from the crystal structure of MKK7 that has 66% homology. Homol-ogy modeling was used because neither the structure of Cot nor MKK4 is available as yet. The kinase domains of Cot and MKK4 consist of an N-lobe and a C-lobe. The N-and C-lobes are linked through a loop, which is called a "hinge region." The backbone of this loop interacts with the adenine moiety of ATP through hydrogen bonding. Considering the experimental result showing that 7,3Ј,4Ј-THIF is an ATP-competitive inhibitor of Cot and MKK4, we docked the compound to the ATP binding site of each of the two kinases. 7,3Ј,4Ј-THIF easily docked to the Cot or MKK4 ATP binding site, which is located between the Nand C-lobes of the kinase domain. In the model structure of Cot, complexed with 7,3Ј,4Ј-THIF, the hydroxyl group at the 3Ј and 4Ј positions of 7,3Ј,4Ј-THIF can form hydrogen bonds with the backbone of Glu-208 and Gly-210 in the hinge region of Cot (Fig. 6A). The hydroxyl group at the 7 position could form a hydrogen bond with the backbone carbonyl group of Gln-173. In addition, the inhibitor would be sandwiched in the ATP binding site by the side chains of the hydrophobic residues, including Ala-165, Met-207, and Val-152 from the N-lobe and Met-262, Val-260, and Val-269 from the C-lobe. In the model structure of MKK4, complexed with 7,3Ј,4Ј-THIF, the hydroxyl groups at the 3Ј and 4Ј positions on 7,3Ј,4Ј-THIF could form hydrogen bonds with the backbone of Glu-179 and Met-181 in the hinge region of MKK4 (Fig. 6B). The carbonyl group at the 4 position could make a hydrogen bond with the side chain of Lys-187. 7,3Ј,4Ј-THIF could form van der Waals interactions at the ATP binding site with hydrophobic residues, including Ala-120, Met-178, Ile-108, Val-116, Cys-156, and Leu-236. Because of the lack of a hydroxyl group at the 3Ј position of daidzein, its interaction with the hinge regions of Cot and MKK4 would be weaker than that of 7,3Ј,4Ј-THIF, and thus daidzein could not effectively inhibit either of these kinases. Further studies by x-ray crystallography to determine the inhibitor complex structures could elucidate the exact binding mode of 7,3Ј,4Ј-THIF to Cot and MKK4.
In summary, 7,3Ј,4Ј-THIF, but not daidzein, inhibits UVBinduced COX-2 expression in JB6 Pϩ cells. The inhibition is mediated mainly by the blockage of the JNKs and p38 pathways activation and subsequent suppression of NF-B activity. 7,3Ј,4Ј-THIF binds with Cot and MKK4 and strongly inhibits their kinase activity. The chemopreventive effect of 7,3Ј,4Ј-THIF was confirmed in the hairless mouse model because the in vivo data showed that 7,3Ј,4Ј-THIF strongly suppressed UVBinduced tumor incidence, multiplicity, and tumor volume. Consistent with the tumor data, 7,3Ј,4Ј-THIF clearly inhibited UVB-induced COX-2 expression in hairless mouse skin, and 7,3Ј,4Ј-THIF directly bound with Cot or MKK4, resulting in the suppression of Cot or MKK4 activity in hairless mouse skin. Collectively, these results suggest that Cot and MKK4 are the main molecular targets of 7,3Ј,4Ј-THIF in the suppression of UVB-mediated skin cancer. These results provide insight into the biological actions of 7,3Ј,4Ј-THIF and the molecular basis for the development of new chemoprotective agents.