NS-398 and Piroxicam Suppress UVB-induced Activator Protein 1 Activity by Mechanisms Independent of Cyclooxygenase-2*

Cyclooxygenases (COX) are rate-limiting enzymes that catalyze the conversion of arachidonic acid to prostaglandins, which are involved in many physiological and pathophysiological responses. COX-2, one of two isoforms of COX, was recently found to play an important role in carcinogenesis in many cell and tissue types. COX-2 inhibitors, which belong to the family of nonsteroidal anti-inflammatory drugs, are believed to be effective in many biological activities such as tumor chemo-prevention because of their inhibition of COX-2. However, in the present study we found that both piroxicam, a general COX inhibitor, and NS-398, a COX-2 selective inhibitor, effectively suppressed the activation of transcription factor activator protein 1 (AP-1) induced by ultraviolet B (UVB) or 12- O -tetradecanoyl-phorbol-13-acetate in mouse epidermal JB6 cells. These COX-2 inhibitors could also inhibit 12- O -tetradecanoyl-phorbol-13-acetate-induced cell transformation. UVB significantly increased AP-1 activity in Cox -2 (cid:1) / (cid:1) fibroblasts transfected with an AP-1 luciferase reporter gene, and this increase was blocked by NS-389 or piroxicam. In JB6, Cox -2 (cid:1) / (cid:1) , or wild-type Cox -2 (cid:2) / (cid:2) cells, both NS-398 and piroxicam piroxicam or (20 independently. an additional 12 treated with UVB) or 24 h (for treated cells lysis buffer (0.1 M buffer m m M luciferase activity relative Anchorage-independent Transformation Assay— The effect of NS-398, piroxicam, or SP600125 on TPA-induced cell transformation was investigated in JB6 C1 41 cells. Cells 10 /ml) were exposed to TPA with or without the chemicals at the indicated in 1 ml of 0.33% basal medium Eagle (BME) agar containing 10% FBS over 3.5 ml of 0.5% BME agar medium containing 10% FBS. The cultures were maintained in a 37 °C, 5% CO weeks, and cell

Cyclooxygenases (COX) 1 are rate-limiting enzymes that catalyze the conversion of arachidonic acid to prostaglandins, which are involved in many normal and pathophysiological responses (1)(2)(3). Of the two known COX enzymes, COX-1 is expressed in nearly all cells, whereas COX-2 is primarily considered an inducible immediate-early gene product (4). COX-2 is suggested to play an important role in tumorigenesis because of epidemiological evidence showing that various types of cancers and transformed cells tend to overexpress COX-2 constitutively (5)(6)(7)(8)(9)(10). Moreover, the tumor promoter 12-O-tetradecanoylphorbol-13-acetate (TPA), epidermal growth factor, and ultraviolet (UV) irradiation can induce COX-2 protein expres-sion in many tissues including the epidermis of skin and also in epidermal keratinocytes (11).
COX inhibitors are known to function through the inhibition of the COX enzymes. However, other potential mechanisms are still being considered (12). COX inhibitors were reported to protect neurons against hypoxia/reperfusion through mechanisms independent of COX (13). We and others have shown that nonsteroidal anti-inflammatory drugs can block mitogeninduced transcription activator protein 1 (AP-1) activity and transactivation (14). Considering the important role of AP-1 in tumoroigensis, the inhibition of AP-1 is likely to be one of the major mechanisms involved in nonsteroidal anti-inflammatory drugs chemopreventive effects.
Luciferase Assay for AP-1 Transactivation-Confluent monolayers of cells were trypsinized, and 8000 viable cells suspended in 100 l of 5% FBS MEM were added to each well of a 96-well plate. Plates were incubated at 37°C in a humidified atmosphere of 5% CO 2 , and 12-24 h later, cells were starved by culturing in 0.1% FBS MEM for 24 h before being treated or not treated with different concentrations of NS-398 or piroxicam for 30 min. The cells were then exposed to UVB (4 kJ/m 2 ) or TPA (20 ng/ml) independently. After an additional 12 h (for AP-1 treated with UVB) or 24 h (for AP-1 treated with TPA) of culturing, the cells were extracted with lysis buffer (0.1 M potassium phosphate buffer (pH 7.8), 1% Triton X-100, 1 mM dithiothreitol, 2 mM EDTA), and luciferase activity was measured using a luminometer (Monolight 2010). The results are expressed as relative AP-1 (17).
Anchorage-independent Transformation Assay-The effect of NS-398, piroxicam, or SP600125 on TPA-induced cell transformation was investigated in JB6 C1 41 cells. Cells (8 ϫ 10 3 /ml) were exposed to TPA with or without the chemicals at the concentration indicated in 1 ml of 0.33% basal medium Eagle (BME) agar containing 10% FBS over 3.5 ml of 0.5% BME agar medium containing 10% FBS. The cultures were maintained in a 37°C, 5% CO 2 incubator for 4 weeks, and the cell colonies were scored by the methods described by Colburn et al. (20). The effect of NS-398, piroxicam, or SP600125 on transformation of JB6 C1 41 cells is presented as a percentage inhibition compared with control.
Assay of AP-1 Activity in Vivo-AP-1-luciferase reporter transgenic mice were established as previously described (18). All the mice were characterized by testing both the basal level and UVB-induced level of luciferase activity. The AP-1-luciferase reporter gene bearing male and female mice (6 -9 weeks old) were randomly divided into six groups (22 mice/group) for the treatment indicated in the relevant figure legend. Five topical doses of NS-398 or piroxicam dissolved in 300 l of acetone were applied to the dorsal skin of the mice over 8 days. The last of the five topical doses of NS-398 or piroxicam was given 3 h before 10 kJ/m 2 of UVB irradiation. 48 h after UVB exposure the level of AP-1 luciferase activity was measured from a skin biopsy. Negative control mice were treated with acetone alone, and the luciferase activity of the biopsied epidermis was measured as described previously (19).
Western Immunoblotting-Cells growing in 100-mm cell culture plates were lysed with 0.6 ml radioimmune precipitation assay buffer (1ϫ phosphate-buffered saline, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, and freshly added 100 g/ml phenylmethanesulfonyl fluoride, 10 g/ml aprotinin, 1 mM sodium orthovanadate), and the samples for electrophoresis were prepared according to the manufacturer's instructions (Santa Cruz Biotechnology, Santa Cruz, CA). The After another 48 h, AP-1 luciferase activity was measured as described previously. The results are presented as relative AP-1 activity. Each bar indicates the mean and S.D. of six assay wells from three independent experiments. *, a significant (p Ͻ 0.01) inhibition was observed compared with TPA treatment and no inhibitor present. B, 10 4 JB6 Cl 41 cells in 0.33% agar were exposed for 4 weeks simultaneously to TPA with or without NS-398 or piroxicam at the concentration indicated and scored for colonies at the end of the experiment. Results are expressed as the mean and S.D. of triplicate experiments. A significant (*, p Ͻ 0.01) inhibition of AP-1 activity was observed in cells treated with TPA plus inhibitors compared with cells treated only with TPA.

FIG. 2. NS-398 or piroxicam suppresses UVB-induced AP-1 activity in JB6 cells in vitro and in vivo.
A, Cl 41 Pϩ1Ϫ1 cells were prepared and serum-starved as in Fig. 1. The cells were then treated with NS-398 or piroxicam for 30 min at the concentrations indicated before exposure to UVB (4 kJ/m 2 ). After another 12 h, the AP-1 luciferase activity was measured as described previously. The results are presented as relative AP-1 activity. Each bar indicates the mean and S.D. of six assay wells from three independent experiments. A significant (*, p Ͻ 0.01) inhibition of AP-1 activity was observed in cells treated with UVB plus inhibitors compared with cells treated only with UVB. B, AP-1-luciferase reporter gene-bearing mice were grouped and treated with NS-398 or piroxicam as described under "Materials and Methods." Each bar indicates the mean and S.D. of the AP-1 activity for each group of 22 mice. A significant (*, p Ͻ 0.01) inhibition of AP-1 activity was observed in mice treated with UVB plus inhibitors compared with mice treated only with UVB.
protein concentration of each sample was determined, and equal amounts of protein were loaded for SDS gel electrophoresis. Immunoblotting for the proteins of COX-2, total JNKs, and phosphorylated JNKs was carried out using antibodies against COX-2 (Upstate Biotechnology, Lake Placid, NY), total JNKs, and the phosphorylated sites of JNKs, respectively (Cell Signaling Technology, Beverly, CA).
Prostaglandin E 2 (PGE 2 ) Assay-Cells were treated with 30 M arachidonic acid for 15 min, and PGE 2 release in the culture medium was determined using the PGE 2 enzyme immunoassay (EIA) system (Amersham Biosciences) according to the protocol provided.
Kinase Assay-In vitro phosphorylation of JNK1 and JNK2 by active MKK4 was determined according to the manufacturer's protocol (Up- The cells were then exposed to UVB (4 kJ/m 2 ) followed by culturing for another 30 min. Western immunoblotting for detection of phosphorylation of JNKs was carried out using phospho-specific mitogen-activated protein kinase antibodies against phosphorylated sites of JNKs. Levels of JNKs nonphosphorylated protein indicates equal protein loading. B, Cl 41 cells were serum-starved for 24 h and treated or not treated for 30 min with NS-398 or piroxicam at the concentrations indicated. The cells were then exposed to UVB at a dose of 4 kJ/m 2 . After exposure, the cells were cultured for another 8 -10 h. The cells were then harvested, and gelshift assays were performed as described under "Materials and Methods." Each bar indicates the mean and S.D. of the densitometry analysis of the gels from three independent experiments. A significant (*, p Ͻ 0.01) inhibition of AP-1 DNA binding activity was observed. Incubation with an excess of unlabeled probe indicates that the AP-1 DNA binding was specific. state Biotechnology) except with the modification that NS-398 or piroxicam at the concentration indicated was added to the samples before incubation for 30 min at 30°C. For JNK kinase activity assay, Cl 41 cells growing in 100-mm cell culture plates were exposed to UVB (4 kJ/m 2 ). After a 30-min incubation at 37°C, 5% CO 2 , the cells were lysed, and JNKs were selectively pulled down by N-terminal c-Jun fusion protein bound to glutathione-Sepharose beads (stress-activated protein kinase (SAPK)/JNK assay kit from Cell Signaling). In vitro analysis of phosphorylation of c-Jun by pull-down JNK kinase assay was determined according to the manufacturer's protocol except with the modification that NS-398 or piroxicam at the concentration indicated was added to the samples before incubation for 30 min at 30°C.
Statistical Analysis-Significant differences in AP-1 activity were determined with the Student's t test. The results are expressed as the means Ϯ S.D.

NS-398 or Piroxicam Inhibits TPA-induced AP-1 Activity and Cell
Transformation-Increased AP-1 activity leads to malignant transformation, and thus, the suppression of AP-1 activity is suggested to be involved in the mechanisms of action of many potential chemopreventive agents (25)(26)(27). We therefore investigated whether NS-398 or piroxicam could block malignant cell transformation. We used TPA, one of the most potent experimental stimuli used to activate AP-1 activity and tumor promotion in skin, to test the effects of the COX-2 inhibitors on AP-1 activity and cell transformation. NS-398 or piroxicam significantly suppressed both TPA-induced AP-1 activity (Fig.  1A) and TPA-induced cell transformation on soft agar (Fig. 1B) in a concentration-dependent manner. As we have previously reported, increased AP-1 activity is required for tumor promoter-induced transformation (24,28). Therefore, the inhibition of AP-1 activity may be functionally linked to the anti-cancer effect of these chemicals.

NS-398 or Piroxicam Suppresses UVB-induced AP-1 Transactivation Both in Vitro and in Vivo-Solar
UVB is a strong etiological factor in human skin cancer (29). UVB-induced AP-1 and other signal transduction pathways play a role in skin carcinogenesis (23). To determine the effect of NS-398 or piroxicam on UVB-induced AP-1 transactivation, AP-1luciferase reporter gene-bearing JB6 C1 41 cells were incubated with the COX-2 inhibitors and then exposed to UVB. Both NS-398 and piroxicam markedly inhibited UVB-induced AP-1 luciferase activity, and the inhibitory effects appeared to be dose-dependent (Fig. 2A). The doses of NS-398 and

FIG. 6. NS-398 or piroxicam suppresses UVB-induced JNKs phosphorylation in Cox-2-null MEFs.
Cox-2 Ϫ/Ϫ and Cox-2 ϩ/ϩ MEFs were serum-starved for 12 h and then exposed to UVB (4 kJ/m 2 ) and harvested at the different time points indicated (A), or the cells were exposed to different doses of UVB as indicated and then harvested 30 min after treatment (B). The cells were also treated for 30 min with NS-398 or piroxicam at the concentrations indicated and then exposed to UVB (4 kJ/m 2 ) followed by culturing for another 30 min (C). Western immunoblotting for phosphorylation of JNKs was carried out using phospho-specific mitogen-activated protein kinase antibodies against phosphorylated sites of JNKs. JNKs non-phosphorylated protein level indicates that an equal amount of protein was loaded.

FIG. 7. Mutants (S516M or S516Q) of COX-2 do not attenuate the inhibitory effects of NS-398 or piroxicam on AP-1 activity.
Cox-2 Ϫ/Ϫ and Cox-2 ϩ/ϩ MEFs were transiently transfected with the AP-1 luciferase reporter plasmid. Cox-2 Ϫ/Ϫ MEFs bearing the AP-1 luciferase reporter gene were also transfected with COX-2, COX-2-S516M, or COX-2-S516Q mutant plasmids as described under "Materials and Methods." A, the transfected cells were treated and assessed for PGE 2 production as described under "Materials and Methods." The asterisks indicate a significantly (p Ͻ 0.01) lower production of PGE 2 in COX-2 null cells compared with cells having COX-2 gene expression. B, COX-2 protein expression in each cell type was assessed as described under "Materials and Methods." ␤-Actin was applied as an inner control of the equal protein volume loaded. C, the transfected cells were also treated with NS-398 or piroxicam 30 min before UVB exposure. The results are presented as relative AP-1 activity. Each bar indicates the mean and S.D. of three assay wells from three independent experiments. The asterisks indicate that a significant (p Ͻ 0.01) inhibition of AP-1 activity occurred in cells treated with UVB plus inhibitors compared with cells treated with UVB only. piroxicam used in the experiment were below the observed cytotoxic range (30,31).
NS-398 or Piroxicam Blocks UVB-activated JNKs and AP-1 DNA Binding Activity-Mitogen-activated protein kinases comprise the most common pathways known to mediate AP-1 function (32). Many reports indicated that JNK, a member of the mitogen-activated protein kinase family, is critical in mediating AP-1 transactivation and malignant transformation (25,(33)(34)(35)(36). Inhibition of JNK leads to suppression of AP-1 activity and cell transformation (23). In our results, both NS-398 and piroxicam inhibited UVB-induced phosphorylation of JNKs (Fig. 3A), indicating that the blocking of JNK activation is involved in the inhibitory effects of NS-398 or piroxicam on these signal transduction pathways.
AP-1 functions as a transcription factor by binding to the transactivation promoter region (TPA response elements) of specific genes. To determine whether NS-398 or piroxicam functions through the attenuation of AP-1 binding to its target DNA, we performed electrophoretic mobility shift assay as an indicator of AP-1 DNA binding activity. UVB (4 kJ/m 2 ) induced a significant increase in AP-1 DNA binding activity (Fig. 3B, lane 3) compared with the unexposed control (Fig. 3B, lane 2). The elevated AP-1 DNA binding was suppressed by NS-398 or piroxicam in a dose-dependent manner (Fig. 3B, lanes 4 -7). The DNA binding was specific for AP-1 because a 10-fold excess of unlabeled AP-1 probe suc-cessfully competed with the labeled probe (Fig. 3B, lane 1).
The Inhibitory Effects of NS-398 or Piroxicam on AP-1 Are Independent of COX-2-Although we have observed that both NS-398 and piroxicam could efficiently block AP-1 activation, whether the effects of the COX-2 inhibitors on AP-1 occurred through their inhibition of COX-2 is important to determine because of the key role that AP-1 plays in tumorigenesis. Cells transfected with the AP-1 luciferase reporter gene showed that an increase in AP-1 activity could also be induced by UVB exposure in Cox-2 Ϫ/Ϫ MEFs (Fig. 4). NS-398 and piroxicam suppressed the UVB-induced increase in AP-1 activity in these MEFs (Fig. 4). We observed that similar to Cox-2 ϩ/ϩ cells, either UVB irradiation or TPA exposure could still induce a significant increase in AP-1 DNA binding in Cox-2 Ϫ/Ϫ MEFs (Fig. 5, A and B), and NS-398 and piroxicam suppressed the increases in AP-1 DNA binding in these cells (Fig. 5, A and B). These results indicated that COX-2 is not required in the signaling pathways mediating UVB-induced AP-1 activation because the COX-2 inhibitors NS-398 and piroxicam suppressed AP-1 activity by mechanisms independent of their inhibition of COX-2. To further investigate whether the absence of COX-2 affects JNK activation, we exposed Cox-2 Ϫ/Ϫ MEFs to different doses of UVB irradiation and harvested the cells at different time points after UVB treatment. The results showed that UVB activated JNKs in Cox-2 Ϫ/Ϫ as well as in Cox-2 ϩ/ϩ MEFs in a time (Fig. 6A)-and dose (Fig. 6B)-dependent manner, and both NS-398 and piroxicam effectively blocked UVB-induced JNK activation in both cell types (Fig. 6C). These results indicated that COX-2 is not involved in the inhibitory mechanism of NS-398 or piroxicam affecting UVB-induced AP-1 activity. Some reports suggest that JNKs and the downstream AP-1 were required in mediating COX-2 expression (37,38), but whether COX-2 is involved in the regulation of JNKs or AP-1 activation is still unclear.
COX inhibitors block the catalysis of arachidonic acid to PGE 2 (39 -41). Several studies suggest that PGE 2 signals through AP-1 and, thus, may play a definitive role in tumor development (42)(43)(44). To elucidate whether NS-389 or piroxicam blocks AP-1 activation by attenuating PGE 2 production, 33% agar for 4 weeks were exposed simultaneously to TPA with or without SP600125 at the concentration indicated and scored for colonies at the end of the experiment. Results are expressed as the mean and S.D. of triplicate experiments. The asterisks indicate that a significant (p Ͻ 0.01) inhibition of cell growth occurred in cells exposed to TPA plus SP600125 compared with cells treated with TPA only. DMSO, Me 2 SO. B, the same cell type and soft agar procedure as for A, except that the cells were exposed to TPA with or without SP600125 plus NS-398 or piroxicam at the concentrations indicated. The double asterisks indicate that a significant (p Ͻ 0.01) inhibition of cell growth occurred in cells treated with TPA plus SP600125 and COX-2 inhibitors compared with cells treated with TPA plus SP600125 indicated by *. we transfected two acetylation active site mutants of COX-2, S516 M and S516Q, together with the AP-1 luciferase reporter gene into Cox-2 Ϫ/Ϫ cells. The two mutant cell lines did not produce PGE 2 (Fig. 7A), although they could encode the COX-2 protein (Fig. 7B). NS-398 and piroxicam blocked UVB-induced AP-1 activation in both COX-2 mutant-transfected Cox-2 Ϫ/Ϫ cells (Fig. 7C). The results indicated that PGE 2 production was not the major effect involved in the inhibition of AP-1 by NS-398 or piroxicam.
The Blocking of JNK Kinase Activation Inhibits TPAinduced Cell Transformation-Although COX-2 has been well recognized as an important target for therapy/ prevention of cancer and COX-2 selective inhibitors are already applied as an approach for prevention or treatment of cancer, many reports suggest that some tumors can still occur or grow in COX-2-negative mice or cell lines. COX-2 inhibitors exhibit identical tumor growth inhibitory effects on either Cox-2 ϩ/ϩ , Cox-2 ϩ/Ϫ , or Cox-2 Ϫ/Ϫ mice or cell lines (45)(46)(47)(48), indicating that alternative pathways exist. So far in this paper, we have demonstrated that NS-398 or piroxicam inhibits AP-1 transactivation independently of COX-2, and blocking of JNK kinase phosphorylation likely plays a role in the function of the chemicals. However, we still don't know if JNKs are really important in mediating mitogen-induced cell transformation. To answer the question we utilized a highly selective JNK inhibitor, SP600125, to observe its effect on TPA-induced JB6 cell transformation on soft agar. SP600125 inhibits JNKs with more than a 20-fold selectivity versus other kinases such as extracellular signal-regulated kinases, p38 kinase, or protein kinase C, etc. (49), and it is not cytotoxic for JB6 cells, as indicated by the 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2Htetrazolium, inner salt assay at the concentration range used in our experiment (CellTiter 96 ® AQ ueous One Solution Cell Proliferation assay kit, Promega) (data not shown). Our results showed that SP600125 significantly suppressed TPAinduced JB6 cell transformation (Fig. 8A), indicating that the activation of JNKs is required in the signaling pathway leading to cell transformation. Similar observations have also been reported previously (23). However, further investigation revealed a synergistic inhibition when both SP600125 and NS-398 or piroxicam were included with TPA on soft agar (Fig. 8B). The results indicated that although JNKs might be a major target of COX-2 inhibitors, other molecules may play a role in the mechanisms of COX-2 inhibition of cell transformation.
NS-398 and Piroxicam Block JNK Phosphorylation in Vitro-To further confirm that the COX-2 inhibitors act on JNK function, we studied the effect of NS-398 or piroxicam on active MKK4-induced JNK1/2 phosphorylation in vitro. Our results indicated that the COX-2 inhibitors effectively suppressed JNK1/2 phosphorylation in vitro (Fig. 9A). In another investigation as to whether the COX-2 inhibitors target mediators other than JNKs existing in the pathway, we used Nterminal c-Jun fusion protein-bound glutathione-Sepharose beads to pull down the JNK proteins from UVB-irradiated cells and incubated them with ATP at the presence of NS-398 or piroxicam in vitro. Piroxicam at the high concentration of 10 M in vitro had a weak inhibitory effect on active JNK-induced phosphorylation of c-Jun, the downstream oncogene target of JNKs and a component of AP-1, whereas NS-398 had little effect (Fig. 8B). This result combined with the observation that NS-398 or piroxicam blocks JNK phosphorylation, indicates that the chemicals may work on JNK signaling from within the JNKs themselves.
In summary, our data indicated that the COX-2 "selective" inhibitors NS-398 or piroxicam effectively blocked UVB-induced AP-1 activity both in vivo and in vitro. They could also suppress TPA-induced AP-1 activity and cell transformation. The blocking of JNK activation by the chemicals appears to be involved in their inhibition of AP-1 activity. We further demonstrated that the inhibition of AP-1 by these compounds was COX-2-independent. Considering the critical role that AP-1 plays in tumorigenesis, these results may supply some novel insights regarding the mechanism of COX-2 inhibitory effects and provide a biological basis for the development of new chemopreventive agents for cancer.