Diallyl Trisulfide Inhibits Phorbol Ester–Induced Tumor Promotion, Activation of AP-1, and Expression of COX-2 in Mouse Skin by Blocking JNK and Akt Signaling

A perilous loop exists between inflammation and cancer (1, 2). One of the well-defined molecular links between inflammation and cancer is the proinflammatory enzyme cyclooxygenase-2 (COX-2), which is a rate-limiting enzyme in the biosynthesis of prostaglandins (3). The levels of COX-2 are transiently elevated in cells or tissues stimulated with different growth factors, proinflammatory cytokines, endotoxin, and tumor promoters (4). COX-2 is aberrantly overexpressed in various premalignant and malignant tissues (4). Mice genetically engineered to overexpress cox-2 are highly susceptible to spontaneous skin tumor formation (5), whereas cox-2 knockout animals are less prone to develop chemically induced skin tumors (6). Selective COX-2 inhibitors have been shown to possess antitumor-promoting activities in different organ-specific cancers including that of skin (7). As evidenced by numerous epidemiologic and laboratory studies, targeted inhibition of COX-2 represents one of the most pragmatic approaches to prevent tumor promotion (3, 4).

Topical application of a prototype tumor promoter 12-O-tetradecanoylphorbol-13-acetate (TPA) induces the expression of COX-2 in mouse skin (8). Mechanistically, the expression of COX-2 in TPA-stimulated mouse skin involves excessive activation of intracellular signal transduction pathways comprising proline-directed serine/threonine kinases and their downstream transcription factors (4, 8–12). The 5′-flanking region of murine cox-2 contains binding sites for various transcription factors, including NF-κB and activator protein-1 (AP-1; ref. 13). TPA has been shown to activate NF-κB (8, 10) and AP-1 (9, 14) in mouse skin through activation of mitogen-activated protein kinases (MAPK). Of the MAPK family members, extracellular signal-regulated kinase (ERK; ref. 8) and p38 MAPK (10) predominantly regulate the activation of NF-κB, whereas c-Jun NH₂-terminal kinase (JNK; ref. 14) and p38 MAPK (9) regulate AP-1 DNA binding in TPA-treated mouse skin. TPA also phosphorylates another upstream kinase Akt, thereby leading to enhanced DNA binding of NF-κB and elevated expression of COX-2 in mouse skin (15). In the present study, we report that activated Akt also transmits signals to the downstream transcription factor AP-1 and regulates the subsequent expression of COX-2 in TPA-stimulated mouse skin.

One of the potential sources of chemopreventive phytochemicals is garlic, which contains organosulfur compounds, such as diallyl sulfide (DAS), diallyl disulfide (DADS), and diallyl trisulfide (DATS; refs. 16, 17). Among the garlic-derived allyl compounds, DATS is the most potent in suppressing constitutive expression of COX-2 in immortalized human embryonic kidney (HEK-293T) cells (18). Whereas DAS and DADS have been reported to inhibit chemically induced mouse skin tumor promotion (19), the anti-inflammatory and antitumor-promoting effects of DATS in mouse skin have not been investigated yet. Here, we report that topical application of DATS inhibits TPA-induced tumor promotion and COX-2 expression in mouse skin, which seemed to be mediated through blockade of AP-1 activation via downregulation of Akt and JNK signaling pathways.

DAS, DADS, and DATS (LKT Laboratories), TPA (Alexis Biochemicals), LY294002 and SP600125 (Tocris), and 7,12-dimethylbenz(a)anthracene (DMBA) and β-actin antibody (Sigma Chemical Co.) were purchased. Primary antibodies for IκBα, phospho-IκBα, ERK, p38 MAPK, JNK, c-Jun, p65, and purified glutathione S-transferase (GST)–c-Jun fusion protein were procured from Santa Cruz Biotechnology. COX-2 antibody was from Cayman Chemical Co. Akt and phospho-Akt antibodies were from Cell Signaling Technology. c-Fos antibody was obtained from Lab Vision. Lamin B antibody and horseradish peroxidase–conjugated anti-rabbit and anti-mouse secondary antibodies were purchased from Zymed Laboratories. Oligonucleotide probes containing NF-κB or AP-1 consensus sequence were acquired from Promega. Enhanced chemiluminescence detection kit and [γ-³²P]ATP were purchased from Amersham Pharmacia Biotech.

Female ICR mice (6–7 wk) were supplied from Sankyo Laboservice Corp., Inc. and housed in climate-controlled quarters with a 12-h light/12-h dark cycle. DATS (5 or 25 μmol in 0.2 mL vehicle) and SP600125 (4 μmol in 0.2 mL vehicle) were applied topically to the shaved dorsal skin 30 min before TPA (10 nmol), whereas LY294002 (10 μmol in 0.2 mL vehicle) was cotreated with TPA. Control animals were treated with 10% DMSO in acetone. All experiments were performed using three animals per group.

Dorsal skins of mice treated with vehicle or TPA in the presence or absence of DATS, SP600125, or LY294002 were excised and fat was removed on ice. Collected epidermis was used to prepare whole tissue lysate, cytosolic, and nuclear protein extracts as described (11). Protein samples were separated by SDS-PAGE and immunoblot analysis was performed according to the procedure described earlier (11). Immunoblot membranes were incubated for 4 h at room temperature with 1:1,000 dilution of primary antibodies for COX-2, Akt, and JNK and for 12 h at 4°C with 1:1,000 dilution of primary antibodies for lamin B, p65, phospho-JNK, phospho-Akt, c-Jun, and c-Fos. Immunoblots were then probed with horseradish peroxidase–conjugated rabbit or mouse secondary antibodies for 1 h and visualized according to the procedure described previously (11).

Sections (4 μm) of 10% formalin-fixed, paraffin-embedded skin tissues from mice (n = 3 per group) treated with vehicle or TPA in the presence or absence of DATS (25 μmol), SP600125 (4 μmol), or LY294002 (10 μmol) were subjected to immunohistochemical analysis for detecting the epidermal expression of COX-2, phospho-JNK, c-Jun, and c-Fos following the procedure described earlier (11). The quantification of immunohistochemical data was achieved by counting the positively stained cells as percent of total epidermal cells from 10 microscopic fields of immunostained tissues.

Epidermal nuclear extracts were prepared from mice treated with vehicle or TPA in the presence or absence of DATS, SP600125, or LY294002 for 2 h. Oligonucleotides for NF-κB (5′-GAGGGGATTCCCTTA-3′) and AP-1 (5′-CGCTTGATGAGTCAGCCGGAAC-3′) were labeled with [γ-³²P]ATP and incubated with nuclear proteins from different treatment groups. The DNA binding of NF-κB and AP-1 was assessed by electrophoretic mobility shift assay (EMSA) as described (11).

The catalytic activities of JNK and Akt in mouse skin treated with TPA for 2 h in the presence or absence of DATS, SP600125, or LY294002 were examined by in vitro kinase assay using GST–c-Jun as the substrate protein for both JNK and Akt following the protocol described previously (11). Briefly, epidermal extract (200 μg) was immunoprecipitated with Akt (for Akt kinase assay) or JNK (for JNK kinase assay) antibody. The immunoprecipitate was suspended in 50 μL of reaction mix containing 47 μL of 1× kinase buffer, 1 μg of GST–c-Jun fusion protein, and 10 μCi [γ-³²P]ATP and incubated at 30°C for 45 min. The kinase reaction was stopped by adding 2.5× SDS loading dye, boiled at 99°C for 5 min, and centrifuged at 5000 rpm for 2 min. The supernatant was separated by gel electrophoresis followed by gel staining and destaining as described (11). The destained gel was dried and exposed to X-ray film to detect the phosphorylated GST–c-Jun in the autoradiogram.

Female ICR mice were randomly divided into five groups, each consisting of 18 animals. Mice from groups II, III, and IV were treated on their shaven backs with single topical application of DMBA (0.2 μmol) dissolved in 0.2 mL of acetone/DMSO (85:15, v/v) and animals in groups I and V were treated only with vehicle. One week after initiation with DMBA, animals in groups II, III, and IV were treated topically with TPA (10 nmol) twice a week for 20 wk. DATS was applied topically at doses of 5 and 25 μmol 30 min before each TPA treatment to animals in groups III and IV, respectively, until the termination of experiment at 20th week. Animals in groups I and V were treated with vehicle alone and DATS (25 μmol), respectively, twice a week for 20 wk. Tumors of at least 1 mm diameter were counted every week until 20 wk.

Values were expressed as the mean ± SEM of at least three independent experiments. The intensity of different immunoblots was measured by using Gel-Pro image density analyzer, and statistical analysis was performed by using Microsoft Excel.

As an initial approach, the relative potency of DAS, DADS, and DATS (25 μmol each) in suppressing COX-2 expression in TPA-treated mouse skin was evaluated. Pretreatment with DADS or DATS, but not DAS, significantly attenuated TPA-induced expression of COX-2. DATS was found to be more potent than DADS (Fig. 1A), and its effect was dose dependent (Fig. 1B). Immunohistochemical analysis (Fig. 1C) verified the pronounced inhibitory effect of DATS on TPA-induced COX-2 expression. Thus, DATS significantly reduced the epidermal COX-2 positivity compared with TPA treatment alone (Fig. 1D). DATS alone had no effect on COX-2 induction in mouse skin.

One of the major transcription factors that regulate the expression of COX-2 is AP-1, which on activation bind to the 5′-flanking region of cox-2 gene promoter (13, 20). Pretreatment with DATS negated the AP-1 DNA binding (Fig. 2A) in TPA-stimulated mouse skin. DATS also diminished TPA-induced expression of c-Jun and c-Fos in mouse skin as determined by immunoblot (Fig. 2B) as well as immunohistochemical analyses (Fig. 2C).

The activation of AP-1 in TPA-treated mouse skin often involves amplification of signaling mediated through upstream p38 MAPK (9) and JNK (14). Although pretreatment with DATS significantly attenuated the phosphorylation of JNK (Fig. 3A), it barely affected the phosphorylation of p38 MAPK (data not shown) in TPA-treated mouse skin. The inhibitory effect of DATS on TPA-induced phosphorylation of JNK was further confirmed by immunohistochemical analysis. As shown in Fig. 3B, pretreatment with DATS significantly diminished TPA-induced phosphorylation of JNK in comparison with TPA treatment alone. Assessment of JNK activity by an in vitro JNK kinase assay using GST–c-Jun fusion protein as a substrate showed that TPA enhanced the catalytic activity of JNK in mouse skin, which was significantly attenuated by pretreatment with DATS (Fig. 3C).

To examine the role of JNK in TPA-induced activation of AP-1 and expression of COX-2 in mouse skin, we used the pharmacologic inhibitor (SP600125) of JNK. Pretreatment with a topical dose (4 μmol) of SP600125 abrogated TPA-induced phosphorylation of JNK as assessed by immunoblot (Fig. 4A, left) and immunohistochemical analyses (Fig. 4A, right). Likewise, SP600125 significantly attenuated TPA-induced JNK activity in mouse skin (Fig. 4B). We attempted to explore a mechanistic link between JNK and the transcription factor AP-1 with regard to upregulation of COX-2 expression in TPA-treated mouse skin. As shown in Fig. 4C, topical application of SP600125 negated TPA-induced DNA binding of AP-1. Moreover, the immunoblot (Fig. 4D) and immunohistochemical (Fig. 4E) analyses revealed that the inactivation of JNK with SP600125 significantly inhibited TPA-induced COX-2 expression in mouse skin.

TPA-induced expression of COX-2 in mouse skin is regulated by upstream kinase Akt (15). Pretreatment with DATS inhibited TPA-induced phosphorylation of Akt in a dose-dependent manner (Fig. 5A, left; Supplementary Fig. S1A). The in vitro Akt kinase assay revealed that DATS pretreatment also attenuated TPA-induced Akt activity by reducing the phosphorylation of the target substrate GST–c-Jun (Fig. 5A, right; Supplementary Fig. S1B). Likewise, topical application of LY294002, a pharmacologic inhibitor of Akt, abrogated TPA-induced phosphorylation (Fig. 5B, left) and the activity (Fig. 5B, right; Supplementary Fig. S1C) of Akt in mouse skin. Furthermore, topically applied LY294002 significantly attenuated TPA-induced DNA binding of AP-1 in mouse skin (Fig. 5C). In addition, the expression of c-Jun and c-Fos in TPA-treated mouse skin was markedly attenuated by LY294002 as assessed by immunoblot (Fig. 5D, top) and immunohistochemical (Fig. 5D, bottom) analyses. Moreover, there was pronounced suppression of TPA-induced COX-2 expression in mouse skin following cotreatment with LY294002 as determined by immunoblot (Fig. 5E, left; Supplementary Fig. S1D) and immunohistochemical (Fig. 5E, right; Supplementary Fig. S1E) analyses.

The inhibitory effects of DATS on TPA-induced COX-2 expression prompted us to investigate the effect of this organosulfur compound on mouse skin tumor promotion. The onset of papillomagenesis in DMBA-initiated mouse skin occurred 6 weeks after TPA treatment, which resulted in a 100% incidence (Fig. 6A) with an average of 21.89 ± 1.16 papillomas per mouse at the 20th week (Fig. 6B). Although DATS (5 μmol) did not affect the tumor incidence, it reduced the multiplicity of skin tumors by 25.63%. Pretreatment with the higher dose (25 μmol) of DATS reduced both the incidence (Fig. 6A) and the multiplicity (Fig. 6B) of papillomas by 22.22% and 71.08%, respectively.

The discovery of intracellular signaling pathways involved in aberrant COX-2 expression and their modulation by dietary phytochemicals provided one of the rational approaches for molecular target-based cancer chemoprevention (7, 21). Garlic has been used as a food additive since the prehistoric time. The oil-soluble fraction of garlic contains organosulfur compounds, which possess chemopreventive properties (16, 17). Among the organosulfur compounds, DATS bearing three sulfur atoms and two allyl moieties exhibits the most potent chemopreventive and anticarcinogenic activities (18, 22–25). In the present study, we examined the anti-inflammatory and antitumor-promoting effects of DATS in TPA-stimulated mouse skin with special focus on its underlying molecular mechanisms. We initially assessed the effect of DATS, at different dose ranges, on TPA-induced COX-2 expression in mouse skin. Based on our preliminary data, we selected 5 or 25 μmol DATS as suitable doses for assessing its anti-inflammatory and antitumor-promoting activities.

Because aberrant expression of COX-2 is causally linked to tumor promotion (12, 26), the inhibitory effect of DATS on TPA-induced COX-2 expression in mouse skin delineates its anti-inflammatory and antitumor-promoting properties. Our finding is in good agreement with the study by Elango and colleagues (18) who showed that DATS suppressed the constitutive expression of cox-2 mRNA in HEK-293T cells. The proximal promoter region of COX-2 gene contains a canonical TATA box and various consensus sequences for interacting with a variety of transcription factors, including NF-κB and AP-1 (13, 20). Although DATS inhibited the activation of AP-1, it barely blocked the DNA binding and nuclear localization of NF-κB in TPA-treated mouse skin.⁵ However, DATS has been reported to inhibit NF-κB activation in cultured cells in vitro (24). Although it is generally considered that the nuclear translocation and DNA binding of NF-κB are critical events for the induction of NF-κB–dependent gene expression (27), several other studies have shown that the downregulation of NF-κB DNA binding is not necessarily associated with its reduced transcriptional activity (28, 29). Moreover, the efficient transcriptional activation of NF-κB depends on the phosphorylation of its active subunit p65/RelA (30). Thus, it is worthwhile to further investigate the effects of DATS on the phosphorylation and transcriptional activation of NF-κB in TPA-stimulated mouse skin.

In addition to the inhibition of AP-1 DNA binding, the downregulation of TPA-induced c-Jun and c-Fos expression suggests AP-1 as one of the molecular targets of DATS in attenuating inflammation. Because certain organosulfur compounds have been found to directly interact with free sulfhydryl groups (31), DATS may modify critical cysteine residues present in the DNA binding domains of c-Jun and c-Fos, thereby inactivating AP-1.

It has been reported that the activation of AP-1 and expression of COX-2 in TPA-treated mouse skin are regulated by various upstream kinases, including p38 MAPK (9) and JNK (14). In accordance with our previous report (14), the present study further confirmed the role of JNK in regulating AP-1 activation and COX-2 expression in mouse skin treated with TPA. Although DATS failed to modulate TPA-induced p38 MAPK activation (data not shown), the compound significantly attenuated the phosphorylation and the catalytic activity of JNK in TPA-stimulated mouse skin. In contrast, DATS induces apoptosis in various cancer cells through upregulation of JNK (32, 33).

Recent reports have shown that Akt, another upstream kinase, regulates COX-2 expression in various cells in culture and in animal studies (34, 35). The significant decrease in TPA-induced COX-2 expression in female ICR mouse skin treated with LY294002 agrees with our previous study showing that pharmacologic inhibition of Akt abrogated TPA-induced COX-2 expression in male HR-1 hairless mouse skin (15). Our findings that the pharmacologic inhibition of Akt abrogated the DNA binding of AP-1, the expression of c-Jun and c-Fos, and the level of COX-2 suggest that Akt functions as an upstream signaling molecule to activate AP-1 and induce COX-2 in TPA-stimulated mouse skin. The inhibition of Akt-mediated phosphorylation of c-Jun in TPA-treated mouse skin by DATS, therefore, supports the notion that the compound inhibits TPA-induced AP-1 activation and COX-2 expression in mouse skin by blocking the Akt signaling. DATS also induced apoptosis in human prostate cancer PC-3 and DU-145 cells through inactivation of Akt (36). Therefore, Akt may act as a potential upstream target for antiproliferative as well as anti-inflammatory and antitumor-promoting activities of DATS.

In the present study, SP600125 and LY294002 were applied to mouse skin following a pretreatment and cotreatment protocol, respectively. According to our previous studies, pretreatment with SP600125 (14), but not with LY294002 (37), significantly attenuated TPA-induced COX-2 expression in mouse skin. We therefore examined if cotreatment with LY294002 may have any effect on TPA-induced AP-1 activation and COX-2 expression. We found that cotreatment with LY294002 not only downregulated Akt phosphorylation but also attenuated the activity of Akt kinase, with subsequent inhibition of AP-1 activation and COX-2 expression in TPA-stimulated mouse skin.

Several studies have shown the involvement of Akt (38, 39) and JNK (14, 40) in TPA-induced AP-1 activation. Our findings that JNK and Akt regulate TPA-induced AP-1 activation in mouse skin in vivo and that DATS inhibits both JNK- and Akt-mediated AP-1 activation suggest a possible cross talk between the JNK and Akt signaling pathways in TPA-treated mouse skin. Whereas JNK has been reported to act downstream of Akt (41), our study shows no such inhibitory effect of LY294002 or SP600125 on the phosphorylation of JNK or Akt, respectively (data not shown), suggesting that TPA-induced JNK or Akt signaling converges independently on the downstream transcription factor AP-1 in mouse skin. In line with our observation, Beales and Ogunwobi (42) showed that pharmacologic inhibition of either JNK or Akt failed to inhibit the phosphorylation of Akt or JNK, respectively, in glycine-extended gastrin-stimulated human colon cancer (HT-29) cells.

The inhibition of COX-2 expression or activity is critical for not only alleviating inflammation but also preventing tumor promotion (4, 21). Thus, the inhibitory effects of DATS on TPA-induced COX-2 expression may account for its antitumor-promoting potential in mouse skin in vivo. The significant decrease in the incidence and the multiplicity of papillomas in DMBA-initiated and TPA-promoted mouse skin suggest DATS as a potential cancer chemopreventive agent. In conclusion, DATS suppressed TPA-induced expression of COX-2 by inactivating AP-1 via blockade of upstream JNK and Akt signaling pathways, which provides a mechanistic basis of anti-inflammatory and antitumor-promoting activities of DATS in mouse skin in vivo.

No potential conflicts of interest were disclosed.

Grant Support: 21C Frontier Functional Human Genome Project grant FG07-21-21 and National Research Foundation, Ministry of Education, Science and Technology, Republic of Korea ERC grant R11-2007-107-01002-0.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.