Abstract

Inhibition of cyclooxygenase-2 (COX2) by non-steroidal anti-inflammatory drugs (NSAID) is known to suppress skin carcinogenesis. It was further suggested that inhibition of COX2-derived prostaglandins by NSAIDs could reduce levels of putative endogenous ligands of peroxisome proliferator-activated receptor-β (PPARβ), and these ligands could potentiate tumorigenesis. However, it is currently unclear whether ligand activation of PPARβ either inhibits or potentiates carcinogenesis. The present studies were designed to examine the mechanism of NSAID-mediated chemoprevention in skin, and, in particular, to determine the role of PPARβ in this process. A two-stage skin carcinogenicity bioassay was performed using wild-type and PPARβ-null mice that were fed either a control diet or one containing 0.32 g sulindac/kg diet. Significant inhibition of chemically induced skin carcinogenesis was observed in both wild-type and PPARβ-null mice, and this was associated with a marked decrease in the concentration of skin prostaglandins including PGE 2 and PGI 2 . Results from these studies demonstrate that inhibition of COX2 by dietary sulindac in mouse skin can effectively inhibit chemically induced skin carcinogenesis, and suggest that the mechanism underlying this chemopreventive effect is independent of PPARβ. Additionally, results from these studies do not support the hypothesis that ligand activation of PPARβ by COX-derived metabolites potentiates chemically induced skin carcinogenesis.

Introduction

The mechanisms mediating skin carcinogenesis are not completely known but include events required to initiate DNA damage, promote cell proliferation of DNA-damaged cells, inhibit apoptosis and facilitate angiogenesis. Delineating the specific mechanisms that mediate skin carcinogenesis could identify more effective means to inhibit and/or prevent this disease. There is a strong relationship between the induction of cyclooxygenase (COX) and skin carcinogenesis ( 1 ). Two isoforms of COX exist, COX1 and COX2, with the latter being induced by a variety of stimuli, including UV light and tumor promoters. COX catalyzes the formation of prostaglandins from arachidonic acid released from plasma membranes by phospholipases. The major prostaglandins produced in skin are PGE 2 , PGF and small amounts of PGI 2 /prostacyclin—as measured by its stable degradation product, 6-keto PGF ( 2 ). Prostaglandins produced from this pathway can then bind to specific receptors and influence signaling pathways that modulate cell proliferation and apoptosis. Specific receptors have been identified and characterized that mediate the biological responses to prostaglandins including the EP (e.g. EP1, EP2, EP3 and EP4), FP and IP receptors that mediate many of the biological effects induced by PGE 2 , PGF and PGI 2 , respectively ( 3 ).

There is good evidence supporting a causal relationship between COX2 expression and skin carcinogenesis. For example, COX2 is over-expressed in a number of pre-neoplastic and epithelial tumors ( 4 ), transgenic mice over-expressing COX2 exhibit increased sensitivity to chemically induced skin carcinogenesis ( 5 ) and chemically induced skin carcinogenesis is significantly reduced in COX2-null mice ( 6 ). Further, treatment with non-steroidal anti-inflammatory drugs (NSAIDs) that target both COX1 and COX2 can effectively inhibit skin tumorigenesis ( 711 ). Collectively, there is good reason to conclude that inhibition of COX activity can inhibit skin carcinogenesis; however, increased production of prostaglandins has also been linked to attenuation of tumorigenesis. For example, increased expression of prostacyclin synthase inhibits lung carcinogenesis ( 12 , 13 ); prostacyclin can inhibit melanoma cell proliferation in vitro ( 14 ) and is known to be anti-metastatic through a number of possible mechanisms ( 15 ). Further, prostaglandins from the A, D and J series have also been shown to be anti-tumorigenic in both in vitro and in vivo models ( 1624 ). Thus, while inhibition of prostaglandin synthesis by inhibiting COX activity has proven to be of benefit for a number of different cancers, it remains possible that this also reduces the levels of prostaglandins that could also inhibit tumorigenic mechanisms and could explain why the efficacy of NSAIDs therapy for cancer chemoprevention is not 100%.

In addition to the EP, FP and IP receptors that are known to mediate many of the biological effects of skin-derived prostaglandins, there are some reports suggesting that peroxisome proliferator-activated receptors (PPARs) could also participate in prostaglandin-mediated signaling. PPARs are members of the nuclear receptor superfamily and include three isoforms, PPARα, PPARβ (also referred to as PPARδ) and PPARγ ( 25 ). In response to ligand activation, PPARs heterodimerize with another nuclear receptor, RXRα; recruit transcriptional co-factors; and modulate transcription of target genes ( 25 ). Specific ligands have been identified for each of the three PPARs, but there is some promiscuity associated with ligand activation of each receptor (e.g. in some cases, ‘specific’ ligands can activate more than one PPAR isoform with varying affinities). Early studies identified potential endogenous ligands including fatty acid derivatives and eicosanoids, which were capable of activating PPAR-dependent reporter gene activity. For example, 15 deoxy-Δ 12,14 -PGJ 2 can activate PPARγ, 8( S )-HETE can activate PPARα, and carbaprostacyclin (cPGI) and PGA 1 can activate PPARβ ( 2628 ). The functional significance of eicosanoid-mediated PPAR-dependent interactions is still relatively unclear. On the basis of the observations that PPARs can be activated by prostaglandins, and that in some cases expression of PPARβ and/or PPARγ is reportedly increased in epithelial cancers and correlates with increased presence of COX2 ( 29 , 30 ), it has been hypothesized that prostaglandins could potentiate carcinogenesis via interactions with PPARs and that inhibition of tumorigenesis by NSAIDs could be due, at least in part, to reduced PPAR signaling activity.

The hypothesis that ligand activation of PPARβ potentiates carcinogenesis is supported by earlier studies showing that administration of a PPARβ ligand caused increased intestinal tumorigenesis in genetically predisposed mice ( 31 ). In contrast, PPARβ-null mice exhibit enhanced skin and colon carcinogenesis as compared with controls ( 3234 ), indirectly suggesting that ligand activation of PPARβ would actually attenuate carcinogenesis. However, these paradoxical observations could also be explained by the fact that PPARβ could exhibit both ligand-independent and ligand-dependent mechanisms of regulation. For example, PPARβ can physically interact with other proteins such as NF-κB in the absence of ligand ( 35 , 36 ) and possibly other transcription factors ( Figure 1 ). PPARβ is also found associated with the co-repressor SMRT in the absence of ligand and can repress expression of genes in the absence of ligand ( 37 ). Thus, it is possible that the observed exacerbation of carcinogenesis observed in PPARβ-null mice is due to the absence of this ligand-independent regulation ( Figure 1 ). Ligand-dependent activation of PPARβ is also known to regulate processes including keratinocyte differentiation, and this occurs through classic receptor-mediated transcriptional regulation of target genes necessary for differentiation ( 36 , 38 , 39 ). Therefore, the possibility exists that the enhanced carcinogenesis observed by administration of PPARβ ligands ( 31 ) could be mediated by ligand-dependent regulation of yet unidentified target genes. This idea is consistent with the hypothesis that COX-derived metabolites could function as PPARβ ligands and potentiate tumorigenesis ( Figure 1 ). The hypothesis that PPARβ could have ligand-independent effects that prevent chemically induced carcinogenesis and ligand-dependent effects that potentiate chemically induced carcinogenesis is supported by another study showing that PPARβ has both pro-inflammatory and anti-inflammatory effects in macrophages, depending on whether the receptor is bound to ligand or not ( 40 ). In the present study, the hypothesis that COX-derived metabolites function as PPARβ ligands and potentiate tumorigenesis was examined by performing a two-stage carcinogenesis bioassay using wild-type and PPARβ-null mice. The probable outcomes and general interpretation from this analysis are summarized in Table I .

Fig. 1.

Hypothetical regulation of PPARβ-dependent skin carcinogenesis. PPARβ could interact with other proteins ( A ), or modulate transcription ( B ), in the absence of ligand and lead to ligand-independent attenuation of cell growth and/or carcinogenesis as shown by previous studies using PPARβ-null mice. In contrast, there is also evidence that ligand activation of PPARβ can transcriptionally modulate gene expression leading to cell growth and carcinogenesis, and that COX-derived prostaglandins may function as PPARβ ligands to modulate this latter putative mechanism ( C ).

Fig. 1.

Hypothetical regulation of PPARβ-dependent skin carcinogenesis. PPARβ could interact with other proteins ( A ), or modulate transcription ( B ), in the absence of ligand and lead to ligand-independent attenuation of cell growth and/or carcinogenesis as shown by previous studies using PPARβ-null mice. In contrast, there is also evidence that ligand activation of PPARβ can transcriptionally modulate gene expression leading to cell growth and carcinogenesis, and that COX-derived prostaglandins may function as PPARβ ligands to modulate this latter putative mechanism ( C ).

Table I.

Expected outcomes and likely interpretations

Observation
 
Interpretation(s)
 
Sulindac inhibits chemically induced skin cancer in wild-type mice, but not in PPARβ-null mice COX-derived ligands potentiate tumorigenesis by activating PPARβ. Indirectly suggests that the known exacerbation of chemically induced skin carcinogenesis in PPARβ-null mice is due to ligand-independent mechanism(s). 
Sulindac inhibits chemically induced skin cancer in both wild-type mice and PPARβ-null mice COX-derived ligands do not potentiate tumorigenesis by activating PPARβ. Indirectly suggests that chemically induced skin carcinogenesis could be inhibited by PPARβ ligands by ligand-dependent mechanism(s). 
Observation
 
Interpretation(s)
 
Sulindac inhibits chemically induced skin cancer in wild-type mice, but not in PPARβ-null mice COX-derived ligands potentiate tumorigenesis by activating PPARβ. Indirectly suggests that the known exacerbation of chemically induced skin carcinogenesis in PPARβ-null mice is due to ligand-independent mechanism(s). 
Sulindac inhibits chemically induced skin cancer in both wild-type mice and PPARβ-null mice COX-derived ligands do not potentiate tumorigenesis by activating PPARβ. Indirectly suggests that chemically induced skin carcinogenesis could be inhibited by PPARβ ligands by ligand-dependent mechanism(s). 

Materials and methods

Two-stage chemical carcinogenesis bioassay

The generation of wild-type and PPARβ-null mice on a C57BL/6 genetic background has been described previously ( 41 ). Male and female mice, aged 8–10 weeks, in the resting phase of the hair cycle were shaved of back hair and 24 h later initiated with 50 µg of 7,12 dimethylbenz[ a ]anthracene (DMBA) dissolved in 200 µL of acetone. One week post-DMBA application, mice were treated topically with 5 µg of 12- O -tetradecanoylphorbol-13-acetate (TPA) dissolved in 200 µL of acetone, three times per week for 22 weeks. Wild-type and PPARβ-null mice were fed either a control diet or one containing sulindac at a concentration of 0.32 g sulindac/kg diet for one week prior to DMBA treatment, and then for the duration of the experiment. This concentration of sulindac was chosen as it has been shown previously to significantly inhibit colon tumor formation in mice, and does not result in overt signs of toxicity or animal distress ( 4244 ). A total of 21 wild-type mice (14 male and 7 female) and 19 (13 male and 6 female) PPARβ-null mice were fed the sulindac diet, and a total of 12 (5 male and 7 female) wild-type mice and 9 (5 male and 4 female) PPARβ-null mice were fed the control diet. Mice were monitored daily for discomfort and stress. The onset of papilloma formation, and the number and size of papillomas were recorded weekly for each mouse. Statistical comparisons of all endpoints were performed between male and female mice for all treatment groups, and after demonstrating that no statistically significant differences existed between sexes within a given experimental group and endpoint, the data obtained from mice of both sexes were pooled for statistical comparisons.

Measurement of skin prostaglandins

PGE 2 , PGF and PGI 2 levels in skin were determined using enzyme-linked immunoassays (Assay Designs, Ann Arbor, MI, USA) for PGE 2 , PGF and 6-keto PGF (the stable breakdown product of PGI 2 ). Briefly, tissue samples from mice were homogenized in buffer (20 mM MOPS, pH 7.2; 5 mM EGTA; 2 mM EDTA and protease inhibitors), and cytosolic fractions were obtained after ultracentrifugation. Tissue samples were collected within 8 h of the last TPA treatment. Cytosol fractions were further used to isolate prostaglandins and subsequently processed for measuring prostaglandins using the manufacturer's recommended procedures, or used for protein quantification. The concentration of respective prostaglandin was normalized to protein concentration and these values were analyzed for significance using ANOVA and Tukey post-testing (GraphPad Prism 4.0c).

Prostaglandin-mediated alterations in keratinocyte growth and apoptosis

Primary keratinocytes were obtained and cultured as described previously ( 45 ). For analysis of cell growth, primary keratinocytes from either wild-type or PPARβ-null mice were cultured for 2 days in control medium and then in the presence or absence of either PGE 2 , (2.0, 10.0 or 20.0 nM), PGF (20.0, 100.0 or 200.0 nM) or cPGI (1.0, 5.0 or 10.0 µM) for 3 days. Cell proliferation was determined over a 5-day culture period by counting cells using a Coulter counter as described previously ( 45 ). Each concentration was tested in triplicate and analyzed for significance by ANOVA and Tukey post-testing (GraphPad Prism 4.0c). To determine the effect of prostaglandins on apoptosis, primary keratinocytes from wild-type or PPARβ-null mice were cultured to ∼75% confluency and then treated with either PGE 2 (10 nM), cPGI (5 µM), or sulindac (50 µM) for 12 h. For a positive control, keratinocytes were irradiated with 100 mJ/cm 2 of UVC. Relative apoptosis was determined after the 12-h treatment period by labeling cells with FITC-conjugated annexin V and propidium iodide (PI) followed by analysis on a flow cytometer (Coulter XL-MCL). Cells that were annexin V-positive and PI-negative were considered apoptotic. The percentage of cells undergoing apoptosis was normalized to control wild-type average, and these values were used for analysis by ANOVA and Tukey post-testing (GraphPad Prism 4.0c).

Western blot analysis of COX2

Microsomal protein fractions from skin samples were obtained from the previously described ultracentrifugation step, and resuspended in buffer (50 mM Tris–HCl, pH = 7.5; 1 mM EDTA; 20% glycerol and 0.1 M DTT). After quantifying protein concentration, 50 µg of microsomal protein was electrophoresed using SDS–PAGE, transferred to PVDF membranes and probed for COX2 or lactic dehydrogenase (LDH), as a loading control, as described previously ( 45 ). After incubation in biotinylated secondary antibody, membranes were incubated in 125 I-streptavidin, washed and exposed to phosphorimager screens. Hybridization signals were normalized to LDH and analyzed for significance using ANOVA and Tukey post-testing (GraphPad Prism 4.0c).

Results

Sulindac inhibits skin carcinogenesis in both wild-type and PPARβ-null mice

The onset of papilloma formation occurred sooner, and the percentage of mice with papillomas was greater in PPARβ-null mice as compared with wild-type mice ( Figure 2A ). By 10 weeks post-initiation, 100% of the PPARβ-null mice had papillomas as compared with 17% of wild-type mice ( Figure 2A ). After 22 weeks, 83% of control wild-type mice had papillomas. The average number of papillomas and average size of papillomas were significantly greater in PPARβ-null mice than wild-type mice ( Figure 2B,C ). These results are consistent with previous work ( 33 ). The onset of papilloma formation was delayed by one week as a result of sulindac feeding in PPARβ-null mice, as grossly visible papillomas were found after 7 or 8 weeks in control and sulindac-fed PPARβ-null mice, respectively ( Figure 2A ). The percentage of mice with papillomas was significantly lower in both wild-type and PPARβ-null mice fed the sulindac diet as compared with controls ( Figure 2A ). In particular, the percentage of mice with papillomas was significantly lower in sulindac-fed PPARβ-null mice as compared with control PPARβ-null mice at week 7 and 10, and significantly lower in sulindac-fed wild-type mice as compared with control wild-type at week 14 through week 17, and at week 20 ( Figure 2A ). Interestingly, sulindac feeding also resulted in a significant decrease in the average number of papillomas per mouse ( Figure 2B ) and the average size of papillomas ( Figure 2C ) in both PPARβ-null mice and wild-type mice. The average number of papillomas per mouse was significantly lower in sulindac-fed PPARβ-null mice as compared with control PPARβ-null mice from week 9 through week 22, and significantly lower in sulindac-fed wild-type mice as compared with control wild-type at week 21 and 22 ( Figure 2B ). The average size of papillomas per mouse was significantly lower in sulindac-fed PPARβ-null mice as compared with control PPARβ-null mice from week 10 through week 22 and significantly lower in sulindac-fed wild-type mice as compared with control wild-type at week 16 ( Figure 2C ). Since the average size of papilloma per mouse does not account for the variation of size, a comparison of size distribution was also performed to determine if sulindac treatment caused a shift in the size of papillomas observed ( Table II ). Indeed, sulindac treatment caused a significant shift in the distribution of papilloma size in both genotypes. For example, there were fewer papillomas with an average size ≥ 1.1 mm and more papillomas with an average size ≤ 1 mm in wild-type mice fed sulindac as compared with control wild-type mice ( Table II ). Similarly, there were fewer papillomas with an average size ≥ 2.1 mm and more papillomas with an average size ≤ 2 mm in PPARβ-null mice fed sulindac as compared with control PPARβ-null mice ( Table II ).

Fig. 2.

Effect of dietary sulindac on chemically induced skin carcinogenesis. ( A ) Wild-type (+/+) and PPARβ-null (−/−) mice were fed either a control diet or one containing 0.32 g sulindac/kg diet beginning 1 week prior to initiation with DMBA and for the remainder of the experiment. Mice were treated topically with TPA three times per week, one week after initiation with DMBA for the duration of the experiment. The percentage of mice with papillomas was quantified weekly. The percentage of mice with papillomas was significantly lower in control (+/+) mice as compared with control (−/−) from week 7 until week 20 with P ≤ 0.05 as determined by χ-square analysis. * Significantly lower in sulindac-fed (−/−) mice as compared with control (−/−) mice with P ≤ 0.05 as determined by χ-square analysis. #Significantly lower in sulindac-fed (+/+) mice as compared with control (+/+) mice with P ≤ 0.05 as determined by χ-square analysis. ( B ) The average number of papillomas per mouse was quantified weekly. Data are presented as the mean ± SEM. The average number of papillomas per mouse was significantly lower in control (+/+) mice as compared with control (−/−) from week 10 until week 22 with P ≤ 0.05 as determined by ANOVA. * Significantly lower in sulindac-fed (−/−) mice as compared with control (−/−) mice with P ≤ 0.05 as determined by ANOVA. #Significantly lower in sulindac-fed (+/+) mice as compared with control (+/+) mice with P ≤ 0.05 as determined by ANOVA. ( C ) The average size of papillomas was quantified weekly. Data are presented as the mean ± SEM. The average size of papillomas per mouse was significantly lower in control (+/+) mice as compared with control (−/−) mice from week 14 until week 22 with P ≤ 0.05 as determined by ANOVA. * Significantly lower in sulindac-fed (−/−) mice as compared to control (−/−) mice with P ≤ 0.05 as determined by ANOVA. #Significantly lower in sulindac-fed (+/+) mice as compared with control (+/+) mice with P ≤ 0.05 as determined by ANOVA.

Fig. 2.

Effect of dietary sulindac on chemically induced skin carcinogenesis. ( A ) Wild-type (+/+) and PPARβ-null (−/−) mice were fed either a control diet or one containing 0.32 g sulindac/kg diet beginning 1 week prior to initiation with DMBA and for the remainder of the experiment. Mice were treated topically with TPA three times per week, one week after initiation with DMBA for the duration of the experiment. The percentage of mice with papillomas was quantified weekly. The percentage of mice with papillomas was significantly lower in control (+/+) mice as compared with control (−/−) from week 7 until week 20 with P ≤ 0.05 as determined by χ-square analysis. * Significantly lower in sulindac-fed (−/−) mice as compared with control (−/−) mice with P ≤ 0.05 as determined by χ-square analysis. #Significantly lower in sulindac-fed (+/+) mice as compared with control (+/+) mice with P ≤ 0.05 as determined by χ-square analysis. ( B ) The average number of papillomas per mouse was quantified weekly. Data are presented as the mean ± SEM. The average number of papillomas per mouse was significantly lower in control (+/+) mice as compared with control (−/−) from week 10 until week 22 with P ≤ 0.05 as determined by ANOVA. * Significantly lower in sulindac-fed (−/−) mice as compared with control (−/−) mice with P ≤ 0.05 as determined by ANOVA. #Significantly lower in sulindac-fed (+/+) mice as compared with control (+/+) mice with P ≤ 0.05 as determined by ANOVA. ( C ) The average size of papillomas was quantified weekly. Data are presented as the mean ± SEM. The average size of papillomas per mouse was significantly lower in control (+/+) mice as compared with control (−/−) mice from week 14 until week 22 with P ≤ 0.05 as determined by ANOVA. * Significantly lower in sulindac-fed (−/−) mice as compared to control (−/−) mice with P ≤ 0.05 as determined by ANOVA. #Significantly lower in sulindac-fed (+/+) mice as compared with control (+/+) mice with P ≤ 0.05 as determined by ANOVA.

Table II.

Size distribution of papillomas in wild-type (+/+) and PPARβ-null (−/−) mice fed sulindac on week 22

Genotype
 
Group
 
≤1 mm
 
1.1–2.0 mm
 
2.1–5.0 mm
 
5.0–10.0 mm
 
Total
 
(+/+) Control 1.3 ± 0.2 2.7 ± 0.8 0.7 ± 0.2 0.1 ± 0.1 4.8 ± 0.8 
   (38 ± 7 a )   (45 ± 7 a )   (16 ± 5 a )   (1 ± 1 a )   
(+/+) Sulindac 1.7 ± 0.3 0.5 ± 0.1 0.1 ± 0.1 0.1 ± 0.1 2.3 ± 0.3 
   (66 ± 8 b )   (24 ± 7 b )   (2 ± 2 b )   (8 ± 6 a )   
(−/−) Control 0.7 ± 0.3 3.4 ± 1.0 7.7 ± 0.9 0.9 ± 0.5 12.7 ± 1.4 
   (5 ± 2 c )   (24 ± 6 b )   (64 ± 7 c )   (7 ± 4 a )   
(−/−) Sulindac 1.4 ± 0.2 3.9 ± 0.5 1.3 ± 0.4 0.0 ± 0.0 6.6 ± 0.9 
   (26 ± 5 a )   (60 ± 6 c )   (15 ± 5 a )   (0 ± 0 a )   
Genotype
 
Group
 
≤1 mm
 
1.1–2.0 mm
 
2.1–5.0 mm
 
5.0–10.0 mm
 
Total
 
(+/+) Control 1.3 ± 0.2 2.7 ± 0.8 0.7 ± 0.2 0.1 ± 0.1 4.8 ± 0.8 
   (38 ± 7 a )   (45 ± 7 a )   (16 ± 5 a )   (1 ± 1 a )   
(+/+) Sulindac 1.7 ± 0.3 0.5 ± 0.1 0.1 ± 0.1 0.1 ± 0.1 2.3 ± 0.3 
   (66 ± 8 b )   (24 ± 7 b )   (2 ± 2 b )   (8 ± 6 a )   
(−/−) Control 0.7 ± 0.3 3.4 ± 1.0 7.7 ± 0.9 0.9 ± 0.5 12.7 ± 1.4 
   (5 ± 2 c )   (24 ± 6 b )   (64 ± 7 c )   (7 ± 4 a )   
(−/−) Sulindac 1.4 ± 0.2 3.9 ± 0.5 1.3 ± 0.4 0.0 ± 0.0 6.6 ± 0.9 
   (26 ± 5 a )   (60 ± 6 c )   (15 ± 5 a )   (0 ± 0 a )   

The number and (percentage) of papillomas within each size category are presented as the mean ± SEM. The percentages within a given column with different superscripts are significantly different from other percentages at P ≤ 0.05 as determined by ANOVA.

Sulindac effectively inhibits prostaglandin production in both genotypes

Prostaglandin levels were measured in skin to confirm that the sulindac treatment inhibited synthesis of these COX-dependent bioactive molecules. Indeed, sulindac feeding during the two-stage bioassay caused a significant decrease in the levels of PGE 2 and 6-keto-PGF (a marker of PGI 2 ) in both genotypes ( Figure 3 ). While the average concentration of PGF was lower in sulindac-fed mice of both genotypes, this difference was not significantly different ( Figure 3B ). Interestingly, the level of PGE 2 in skin was significantly higher in PPARβ-null mice as compared with wild-type mice ( Figure 3A ). The level of COX2 was significantly higher in PPARβ-null mouse skin as compared with wild-type ( Figure 3D ), consistent with the higher level of PGE 2 ( Figure 3A ). The observed higher expression level of COX2 in TPA-treated PPARβ-null mouse skin is in agreement with previous findings ( 45 ).

Fig. 3.

Effect of dietary sulindac on the prostaglandin concentration and COX2 expression in skin. The major prostaglandins present in skin ( A ) PGE 2 , ( B ) PGF and ( C ) PGI 2 (as determined by measuring the stable breakdown product 6-keto-PGF ) were quantified from control and sulindac-fed wild-type (+/+) and PPARβ-null (−/−) mice using an EIA. Data are presented as the mean ± SEM. Values with different letters are significantly different at P ≤ 0.05 as analyzed by ANOVA. ( D ) Quantified Western blot analysis of COX2 expression in skin from control and sulindac-fed (+/+) and (−/−) mice. Normalized, quantified hybridization values are presented as the mean ± SEM from three independent samples. Values with different letters are significantly different at P ≤ 0.05 as analyzed by ANOVA.

Fig. 3.

Effect of dietary sulindac on the prostaglandin concentration and COX2 expression in skin. The major prostaglandins present in skin ( A ) PGE 2 , ( B ) PGF and ( C ) PGI 2 (as determined by measuring the stable breakdown product 6-keto-PGF ) were quantified from control and sulindac-fed wild-type (+/+) and PPARβ-null (−/−) mice using an EIA. Data are presented as the mean ± SEM. Values with different letters are significantly different at P ≤ 0.05 as analyzed by ANOVA. ( D ) Quantified Western blot analysis of COX2 expression in skin from control and sulindac-fed (+/+) and (−/−) mice. Normalized, quantified hybridization values are presented as the mean ± SEM from three independent samples. Values with different letters are significantly different at P ≤ 0.05 as analyzed by ANOVA.

Prostaglandins modulate cell growth similarly in both wild-type and PPARβ-null keratinocytes

Results from the bioassay suggest that COX-derived prostaglandins function similarly in the absence of PPARβ expression. To directly examine this hypothesis, keratinocytes from wild-type and PPARβ-null mice were cultured in the presence of prostaglandins, and relative cell proliferation and apoptosis were measured. Treatment of keratinocytes with PGE 2 resulted in a marginal increase (≤23%) in cell proliferation in both wild-type and PPARβ-null cells after three, four or five days of culture ( Figure 4A,B ). In contrast, wild-type primary keratinocytes cultured in the presence of 5 µM cPGI exhibited a marginal decrease in cell proliferation on day three of culture (24 h after exposure to cPGI), and this effect did not occur in PPARβ-null keratinocytes where increased cell growth was observed at this time point ( Figure 4C,D ). Cell growth was not different between wild-type or PPARβ-null keratinocytes cultured in cPGI after 4 or 5 days of culture. Culturing keratinocytes in PGF had no significant effect on cell growth in either wild-type or PPARβ-null cells (data not shown). To determine if prostaglandins or sulindac could influence apoptosis in primary keratinocytes, cells were cultured in either PGE 2 , cPGI or sulindac, and apoptosis was measured by flow cytometry of annexin V-positive/PI-negative cells. While irradiation of both wild-type and PPARβ-null keratinocytes resulted in a significant increase in the number of apoptotic cells, PGE 2 , cPGI and sulindac had no effect ( Figure 5 ).

Fig. 4.

Effect of skin prostaglandins on cell proliferation of keratinocytes. Primary keratinocytes were cultured for two days in control medium and then switched to a medium containing either ( A, B ) PGE 2 or ( C, D ) carbaprostacyclin (stable PGI 2 analog) and the cell growth was measured. Data are presented as the mean number of cells per well ± SEM ( A and C ) and as the percentage of respective control ( B and D ). Data were analyzed for significance using ANOVA for each respective time point. * Significantly different ( P ≤ 0.05) than wild-type control at each respective time point (e.g. Day 3, 4 or 5) as analyzed by ANOVA.

Fig. 4.

Effect of skin prostaglandins on cell proliferation of keratinocytes. Primary keratinocytes were cultured for two days in control medium and then switched to a medium containing either ( A, B ) PGE 2 or ( C, D ) carbaprostacyclin (stable PGI 2 analog) and the cell growth was measured. Data are presented as the mean number of cells per well ± SEM ( A and C ) and as the percentage of respective control ( B and D ). Data were analyzed for significance using ANOVA for each respective time point. * Significantly different ( P ≤ 0.05) than wild-type control at each respective time point (e.g. Day 3, 4 or 5) as analyzed by ANOVA.

Fig. 5.

Effect of prostaglandins or sulindac on apoptosis in keratinocytes. Primary keratinocytes were cultured in the presence of either 10 nM PGE 2 , 5 µM carbaprostacyclin (cPGI) or 50 µM sulindac. Keratinocytes irradiated with ultraviolet (UV) light served as the positive control. Data are presented as the mean ± SEM. Values with different letters are significantly different at P ≤ 0.05 as analyzed by ANOVA.

Fig. 5.

Effect of prostaglandins or sulindac on apoptosis in keratinocytes. Primary keratinocytes were cultured in the presence of either 10 nM PGE 2 , 5 µM carbaprostacyclin (cPGI) or 50 µM sulindac. Keratinocytes irradiated with ultraviolet (UV) light served as the positive control. Data are presented as the mean ± SEM. Values with different letters are significantly different at P ≤ 0.05 as analyzed by ANOVA.

Discussion

Feeding sulindac to mice at a concentration that has previously been shown to inhibit colon carcinogenesis ( 4244 ) is also effective at inhibiting chemically induced skin carcinogenesis as shown from the present studies. This is consistent with other studies showing that inhibition of COX by NSAIDs can inhibit UV-induced ( 79 , 11 ) and chemically induced skin cancer ( 10 ). Since inhibition of chemically induced skin carcinogenesis was observed in both wild-type and PPARβ-null mice, this demonstrates that COX-derived prostaglandins are unlikely to mediate potentiation of tumorigenesis through suppression of production of endogenous ligands that could modulate PPARβ-dependent signaling ( Table II , Figure 1 ). Interestingly, while the efficacy of the inhibitory effect of sulindac in wild-type mice was relatively less as compared with PPARβ-null mice, significant differences in the incidence, tumor multiplicity and notably the distribution of papilloma size were detected in both wild-type and PPARβ-null mice fed sulindac. This difference in the efficacy of COX inhibition on tumorigenesis is consistent with the previous finding that inhibition of COX2 by NS-398 significantly inhibited cell proliferation of primary keratinocytes stimulated with TPA more effectively in cells from PPARβ-null mice as compared with wild-type ( 45 ). It is possible that inhibition of COX by NSAIDs is more effective in PPARβ-null mice because expression of COX2 is significantly greater in PPARβ-null mouse skin and keratinocytes treated with TPA ( 45 ), which may be due to reduced PPARβ-dependent ubiquitination of kinases that control COX2 gene expression ( 33 , 45 ). Collectively, since inhibition of chemically induced skin carcinogenesis by sulindac occurs in the absence of PPARβ expression, these results strongly suggest that COX-derived ligands are unlikely to serve as signaling molecules that activate PPARβ and potentiate carcinogenesis. Indeed, previous work and the current study strongly suggest that ligand activation of PPARβ would actually attenuate chemically induced carcinogenesis since PPARβ-null mice exhibit enhanced sensitivity to chemical carcinogens ( 32 , 33 ).

The finding that inhibition of COX metabolism by sulindac inhibits chemically induced skin carcinogenesis is of interest because there is evidence that COX-derived metabolites could both potentiate and attenuate carcinogenesis. For example, it is thought that one of the primary mechanisms underlying NSAID-mediated inhibition of skin cancer is the decrease in PGE 2 -dependent signaling, which is known to regulate cell migration, cell proliferation and angiogenesis ( 46 ). Specific receptors have been identified that are known to mediate PGE 2 -dependent tumor growth and angiogenesis ( 4751 ). Thus, results from the present study suggest that PGE 2 -dependent pathways that modulate cell growth are inhibited by sulindac administration, and that this effect is not influenced by PPARβ expression. This is consistent with the observation that cell proliferation is increased modestly in both wild-type and PPARβ-null keratinocytes cultured in the presence of PGE 2 . In contrast, others have suggested that PGE 2 can transactivate PPARβ through PI3Kinase/Akt signaling that promotes cell survival and tumor growth during intestinal tumorigenesis ( 52 ). Results from the present study are inconsistent with this hypothesis, which suggests that this hypothetical PGE 2 /PPARβ-dependent pathway described in intestinal cells does not function similarly in keratinocytes. This idea is supported by the lack of PGE 2 modulation of apoptosis in keratinocytes observed in the present study.

It is also possible that NSAIDs could inhibit tumorigenesis by preventing or limiting COX2-mediated DNA damage ( 53 ). Since COX2 expression is exacerbated in TPA-treated PPARβ-null mouse skin ( 45 ), it is possible that the significant decrease in papilloma multiplicity in PPARβ-null mice fed sulindac could be due to decreased COX-mediated DNA damage resulting from bioactivation of endogenous compounds ( 53 ) or from bioactivation of DMBA by COX2 ( 5456 ). Although not within the scope of the present study, this hypothesis should be examined in the future to determine whether inhibition of COX-dependent bioactivation of either endogenous or exogenous chemicals can functionally modulate chemically induced skin carcinogenesis.

While there is good evidence suggesting that inhibition of COX metabolism can inhibit tumorigenesis, it is curious to note that COX-derived metabolites have also been linked to anti-tumorigenic functions. In particular, it is noteworthy that there are studies suggesting that prostacyclin inhibits carcinogenesis. For example, over-expression of prostacyclin synthase inhibits chemically induced lung cancer ( 12 , 13 ), prostacyclin can inhibit growth of cells including human keratinocytes ( 14 , 15 , 5759 ), and it is known that prostacyclin is anti-metastatic ( 15 ). Results from the present study suggest that prostacyclin can inhibit cell growth of keratinocytes and that this effect requires PPARβ since this was not observed in PPARβ-null cells during early periods of culture where cPGI actually increased proliferation. These combined observations are of interest because stable prostacyclin derivatives (e.g. carbaprostacyclin) can bind to and activate PPARβ ( 27 ). This suggests that activation of PPARβ by COX-derived prostacyclin could function to attenuate tumor growth, which is inconsistent with the known anti-tumorigenic effect of COX inhibition. However, this could explain why inhibition of COX with selective inhibitors is not 100% effective. Consistent with this hypothesis, inhibition of lung adenocarcinoma cell growth is more effective in the presence of both a COX inhibitor (indomethacin) and a PPARβ ligand ( 60 ). The hypothesis that the combined inhibition of COX coupled with ligand activation of PPARβ could be more effective in the inhibition of carcinogenesis should thus be examined.

In summary, results from these studies clearly demonstrate that sulindac feeding can effectively inhibit chemically induced skin carcinogenesis, consistent with previous findings. Since sulindac inhibits the production of COX-derived metabolites (which are thought to potentiate tumorigenesis by activating PPARβ) and inhibition of chemically induced carcinogenesis occurred in mice fed sulindac that do not express PPARβ, these findings suggest that COX-derived metabolites are unlikely to facilitate potentiation of cell growth during tumorigenesis by activating PPARβ. No evidence for differential activation of a cell proliferative response in keratinocytes in response to prostaglandins was observed between wild-type and PPARβ-null keratinocytes, other than a moderate inhibition of cell growth by carbaprostacyclin. These findings support the hypothesis that a combinational approach of inhibiting COX coupled with ligand activation of PPARβ may effectively inhibit carcinogenesis.

The authors gratefully acknowledge Susan Magargee, Elaine Kunze, Luowei Li and Amanda Burns for providing technical help and Xiaoxuan Fan and Robert Schlegel for providing FITC-labeled annexin V. This work was supported by The National Institutes of Health, CA89607, CA97999 (J.M.P.).

Conflict of Interest Statement : None declared.

References

1.
Trifan,O.C. and Hla,T. (
2003
) Cyclooxygenase-2 modulates cellular growth and promotes tumorigenesis.
J. Cell Mol. Med.
  ,
7
,
207
–222.
2.
Pentland,A.P., Mahoney,M., Jacobs,S.C. and Holtzman,M.J. (
1990
) Enhanced prostaglandin synthesis after ultraviolet injury is mediated by endogenous histamine stimulation. A mechanism for irradiation erythema.
J. Clin. Invest.
  ,
86
,
566
–574.
3.
Tsuboi,K., Sugimoto,Y. and Ichikawa,A. (
2002
) Prostanoid receptor subtypes.
Prostaglandins Other Lipid Mediat.
  ,
68–69
,
535
–556.
4.
Gupta,R.A. and Dubois,R.N. (
2001
) Colorectal cancer prevention and treatment by inhibition of cyclooxygenase-2.
Nat. Rev. Cancer,
 
1
,
11
–21.
5.
Muller-Decker,K., Neufang,G., Berger,I., Neumann,M., Marks,F. and Furstenberger,G. (
2002
) Transgenic cyclooxygenase-2 overexpression sensitizes mouse skin for carcinogenesis.
Proc. Natl Acad. Sci. USA
  ,
99
,
12483
–12488.
6.
Tiano,H.F., Loftin,C.D., Akunda,J. et al . (
2002
) Deficiency of either cyclooxygenase (COX)-1 or COX-2 alters epidermal differentiation and reduces mouse skin tumorigenesis.
Cancer Res.
  ,
62
,
3395
–3401.
7.
Wilgus,T.A., Breza,T.S.Jr, Tober,K.L. and Oberyszyn,T.M. (
2004
) Treatment with 5-fluorouracil and celecoxib displays synergistic regression of ultraviolet light B-induced skin tumors.
J. Invest. Dermatol.
  ,
122
,
1488
–1494.
8.
Fischer,S.M., Conti,C.J., Viner,J., Aldaz,C.M. and Lubet,R.A. (
2003
) Celecoxib and difluoromethylornithine in combination have strong therapeutic activity against UV-induced skin tumors in mice.
Carcinogenesis,
 
24
,
945
–952.
9.
Pentland,A.P., Schoggins,J.W., Scott,G.A., Khan,K.N. and Han,R. (
1999
) Reduction of UV-induced skin tumors in hairless mice by selective COX-2 inhibition.
Carcinogenesis
  ,
20
,
1939
–1944.
10.
Muller-Decker,K., Kopp-Schneider,A., Marks,F., Seibert,K. and Furstenberger,G. (
1998
) Localization of prostaglandin H synthase isoenzymes in murine epidermal tumors: suppression of skin tumor promotion by inhibition of prostaglandin H synthase-2.
Mol. Carcinog.
  ,
23
,
36
–44.
11.
Fischer,S.M., Lo,H.H., Gordon,G.B., Seibert,K., Kelloff,G., Lubet,R.A. and Conti,C.J. (
1999
) Chemopreventive activity of celecoxib, a specific cyclooxygenase-2 inhibitor, and indomethacin against ultraviolet light-induced skin carcinogenesis.
Mol. Carcinog.
  ,
25
,
231
–240.
12.
Keith,R.L., Miller,Y.E., Hoshikawa,Y., Moore,M.D., Gesell,T.L., Gao,B., Malkinson,A.M., Golpon,H.A., Nemenoff,R.A. and Geraci,M.W. (
2002
) Manipulation of pulmonary prostacyclin synthase expression prevents murine lung cancer.
Cancer Res.
  ,
62
,
734
–740.
13.
Keith,R.L., Miller,Y.E., Hudish,T.M., Girod,C.E., Sotto-Santiago,S., Franklin,W.A., Nemenoff,R.A., March,T.H., Nana-Sinkam,S.P. and Geraci,M.W. (
2004
) Pulmonary prostacyclin synthase overexpression chemoprevents tobacco smoke lung carcinogenesis in mice.
Cancer Res.
  ,
64
,
5897
–5904.
14.
Honn,K.V. and Meyer,J. (
1981
) Thromboxanes and prostacyclin: positive and negative modulators of tumor growth.
Biochem. Biophys. Res. Commun.
  ,
102
,
1122
–1129.
15.
Schneider,M.R., Tang,D.G., Schirner,M. and Honn,K.V. (
1994
) Prostacyclin and its analogues: antimetastatic effects and mechanisms of action.
Cancer Metastasis Rev.
  ,
13
,
349
–364.
16.
Bregman,M.D., Funk,C. and Fukushima,M. (
1986
) Inhibition of human melanoma growth by prostaglandin A, D, and J analogues.
Cancer Res.
  ,
46
,
2740
–2744.
17.
Fukushima,M., Kato,T., Ueda,R., Ota,K., Narumiya,S. and Hayaishi,O. (
1982
) Prostaglandin D2, a potential antineoplastic agent.
Biochem. Biophys. Res. Commun.
  ,
105
,
956
–964.
18.
Fukushima,S., Kishimoto,S., Takeuchi,Y. et al . (
1998
) [Pharmaceutical and pharmacological development of antitumor prostaglandins.] (in Japanese).
Nippon Rinsho
  ,
56
,
663
–669.
19.
Kikuchi,Y., Kita,T., Hirata,J., Kuki,E., Nagata,I. and Fukushima,M. (
1992
) Modulation of human lymphocyte response to phytohemagglutinin by antineoplastic prostaglandins.
Int. J. Immunopharmacol.
  ,
14
,
105
–110.
20.
Kikuchi,Y., Kita,T., Miyauchi,M., Hirata,J., Sasa,H., Nagata,I. and Fukushima,M. (
1992
) Adjuvant effects of antineoplastic prostaglandins to cisplatin in nude mice bearing human ovarian cancer cells.
J. Cancer Res. Clin. Oncol.
  ,
118
,
453
–457.
21.
Narumiya,S. and Fukushima,M. (
1986
) Site and mechanism of growth inhibition by prostaglandins. I. Active transport and intracellular accumulation of cyclopentenone prostaglandins, a reaction leading to growth inhibition.
J. Pharmacol. Exp. Ther.
  ,
239
,
500
–505.
22.
Narumiya,S. and Fukushima,M. (
1987
) Active transport and cellular accumulation of cyclopentenone prostaglandins: a mechanism of prostaglandin-induced growth inhibition.
Adv Prostaglandin Thromboxane Leukot Res.
  ,
17B
,
972
–975.
23.
Narumiya,S., Ohno,K., Fujiwara,M. and Fukushima,M. (
1986
) Site and mechanism of growth inhibition by prostaglandins. II. Temperature-dependent transfer of a cyclopentenone prostaglandin to nuclei.
J. Pharmacol. Exp. Ther.
  ,
239
,
506
–511.
24.
Sasaki,H. and Fukushima,M. (
1994
) Prostaglandins in the treatment of cancer.
Anticancer Drugs,
 
5
,
131
–138.
25.
Michalik,L., Desvergne,B. and Wahli,W. (
2004
) Peroxisome-proliferator-activated receptors and cancers: complex stories.
Nat. Rev. Cancer
  ,
4
,
61
–70.
26.
Yu,K., Bayona,W., Kallen,C.B., Harding,H.P., Ravera,C.P., McMahon,G., Brown,M. and Lazar,M.A. (
1995
) Differential activation of peroxisome proliferator-activated receptors by eicosanoids.
J. Biol. Chem.
  ,
270
,
23975
–23983.
27.
Forman,B.M., Chen,J. and Evans,R.M. (
1997
) Hypolipidemic drugs, polyunsaturated fatty acids, and eicosanoids are ligands for peroxisome proliferator-activated receptors alpha and delta.
Proc. Natl Acad. Sci. USA
  ,
94
,
4312
–4317.
28.
Kliewer,S.A., Sundseth,S.S., Jones,S.A. et al. (
1997
) Fatty acids and eicosanoids regulate gene expression through direct interactions with peroxisome proliferator-activated receptors alpha and gamma.
Proc. Natl Acad. Sci. USA
  ,
94
,
4318
–4323.
29.
DuBois,R.N., Gupta,R., Brockman,J., Reddy,B.S., Krakow,S.L. and Lazar,M.A. (
1998
) The nuclear eicosanoid receptor, PPARgamma, is aberrantly expressed in colonic cancers.
Carcinogenesis,
 
19
,
49
–53.
30.
Gupta,R.A., Tan,J., Krause,W.F., Geraci,M.W., Willson,T.M., Dey,S.K. and DuBois,R.N. (
2000
) Prostacyclin-mediated activation of peroxisome proliferator-activated receptor delta in colorectal cancer.
Proc. Natl Acad. Sci. USA
  ,
97
,
13275
–13280.
31.
Gupta,R.A., Wang,D., Katkuri,S., Wang,H., Dey,S.K. and DuBois,R.N. (
2004
) Activation of nuclear hormone receptor peroxisome proliferator-activated receptor-delta accelerates intestinal adenoma growth.
Nat. Med.
  ,
10
,
245
–247.
32.
Harman,F.S., Nicol,C.J., Marin,H.E., Ward,J.M., Gonzalez,F.J. and Peters,J.M. (
2004
) Peroxisome proliferator-activated receptor-delta attenuates colon carcinogenesis.
Nat. Med.
  ,
10
,
481
–483.
33.
Kim,D.J., Akiyama,T.E., Harman,F.S., Burns,A.M., Shan,W., Ward,J.M., Kennett,M.J., Gonzalez,F.J. and Peters,J.M. (
2004
) Peroxisome proliferator-activated receptor beta (delta)-dependent regulation of ubiquitin C expression contributes to attenuation of skin carcinogenesis.
J. Biol. Chem.
  ,
279
,
23719
–23727.
34.
Reed,K.R., Sansom,O.J., Hayes,A.J., Gescher,A.J., Winton,D.J., Peters,J.M. and Clarke,A.R. (
2004
) PPARdelta status and Apc-mediated tumourigenesis in the mouse intestine.
Oncogene
  ,
23
,
8992
–8996.
35.
Jove,M., Laguna,J.C. and Vazquez-Carrera,M. (
2005
) Agonist-induced activation releases peroxisome proliferator-activated receptor beta/delta from its inhibition by palmitate-induced nuclear factor-kappaB in skeletal muscle cells.
Biochim. Biophys. Acta
  ,
1734
,
52
–61.
36.
Westergaard,M., Henningsen,J., Johansen,C. et al . (
2003
) Expression and localization of peroxisome proliferator-activated receptors and nuclear factor kappaB in normal and lesional psoriatic skin.
J. Invest. Dermatol.
  ,
121
,
1104
–1117.
37.
Shi,Y., Hon,M. and Evans,R.M. (
2002
) The peroxisome proliferator-activated receptor delta, an integrator of transcriptional repression and nuclear receptor signaling.
Proc. Natl Acad. Sci. USA
  ,
99
,
2613
–2618.
38.
Kim,D.J., Bility,M.T., Billin,A.N., Willson,T.M., Gonzalez,F.J. and Peters,J.M. (
2006
) PPARβ/δ selectively induces differentiation and inhibits cell proliferation.
Cell Death Differ.
  ,
13
,
53
–60.
39.
Schmuth,M., Haqq,C.M., Cairns,W.J. et al . (
2004
) Peroxisome proliferator-activated receptor (PPAR)-beta/delta stimulates differentiation and lipid accumulation in keratinocytes.
J. Invest. Dermatol.
  ,
122
,
971
–983.
40.
Lee,C.H., Chawla,A., Urbiztondo,N., Liao,D., Boisvert,W.A., Evans,R.M. and Curtiss,L.K. (
2003
) Transcriptional repression of atherogenic inflammation: modulation by PPARdelta.
Science
  ,
302
,
453
–457.
41.
Peters,J.M., Lee,S.S.T., Li,W., Ward,J.M., Gavrilova,O., Everett,C., Reitman,M.L., Hudson,L.D. and Gonzalez,F.J. (
2000
) Growth, adipose, brain and skin alterations resulting from targeted disruption of the mouse peroxisome proliferator-activated receptor β(δ).
Mol. Cell Biol.
  ,
20
,
5119
–5128.
42.
Beazer-Barclay,Y., Levy,D.B., Moser,A.R., Dove,W.F., Hamilton,S.R., Vogelstein,B. and Kinzler,K.W. (
1996
) Sulindac suppresses tumorigenesis in the Min mouse.
Carcinogenesis
  ,
17
,
1757
–1760.
43.
Chiu,C.H., McEntee,M.F. and Whelan,J. (
1997
) Sulindac causes rapid regression of preexisting tumors in Min/+ mice independent of prostaglandin biosynthesis.
Cancer Res.
  ,
57
,
4267
–4273.
44.
McEntee,M.F., Chiu,C.H. and Whelan,J. (
1999
) Relationship of beta-catenin and Bcl-2 expression to sulindac-induced regression of intestinal tumors in Min mice.
Carcinogenesis
  ,
20
,
635
–640.
45.
Kim,D.J., Murray,I.A., Burns,A.M., Gonzalez,F.J., Perdew,G.H. and Peters,J.M. (
2005
) Peroxisome proliferator-activated receptor-β/δ (PPARβ/δ) inhibits epidermal cell proliferation by down-regulation of kinase activity.
J. Biol. Chem.
  ,
280
,
9519
–9527.
46.
Subbaramaiah,K., Zakim,D., Weksler,B.B. and Dannenberg,A.J. (
1997
) Inhibition of cyclooxygenase: a novel approach to cancer prevention.
Proc. Soc. Exp. Biol. Med.
  ,
216
,
201
–210.
47.
Konger,R.L., Scott,G.A., Landt,Y., Ladenson,J.H. and Pentland,A.P. (
2002
) Loss of the EP2 prostaglandin E2 receptor in immortalized human keratinocytes results in increased invasiveness and decreased paxillin expression.
Am. J. Pathol.
  ,
161
,
2065
–2078.
48.
Mutoh,M., Watanabe,K., Kitamura,T. et al . (
2002
) Involvement of prostaglandin E receptor subtype EP(4) in colon carcinogenesis.
Cancer Res.
  ,
62
,
28
–32.
49.
Seno,H., Oshima,M., Ishikawa,T.O., Oshima,H., Takaku,K., Chiba,T., Narumiya,S. and Taketo,M.M. (
2002
) Cyclooxygenase 2- and prostaglandin E(2) receptor EP(2)-dependent angiogenesis in Apc(Delta716) mouse intestinal polyps.
Cancer Res.
  ,
62
,
506
–511.
50.
Sonoshita,M., Takaku,K., Sasaki,N., Sugimoto,Y., Ushikubi,F., Narumiya,S., Oshima,M. and Taketo,M.M. (
2001
) Acceleration of intestinal polyposis through prostaglandin receptor EP2 in Apc(Delta 716) knockout mice.
Nat. Med.
  ,
7
,
1048
–1051.
51.
Watanabe,K., Kawamori,T., Nakatsugi,S. et al . (
1999
) Role of the prostaglandin E receptor subtype EP1 in colon carcinogenesis.
Cancer Res.
  ,
59
,
5093
–5096.
52.
Wang,D., Wang,H., Shi,Q., Katkuri,S., Walhi,W., Desvergne,B., Das,S.K., Dey,S.K. and DuBois,R.N. (
2004
) Prostaglandin E(2) promotes colorectal adenoma growth via transactivation of the nuclear peroxisome proliferator-activated receptor delta.
Cancer Cell
  ,
6
,
285
–295.
53.
Lee,S.H., Williams,M.V., Dubois,R.N. and Blair,I.A. (
2005
) Cyclooxygenase-2-mediated DNA damage.
J. Biol. Chem.
  ,
280
,
28337
–28346.
54.
Marnett,L.J. (
1981
) Polycyclic aromatic hydrocarbon oxidation during prostaglandin biosynthesis.
Life Sci.
  ,
29
,
531
–546.
55.
Sivarajah,K., Anderson,M.W. and Eling,T.E. (
1978
) Metabolism of benzo(a)pyrene to reactive intermediate(s) via prostaglandin biosynthesis.
Life Sci.
  ,
23
,
2571
–2578.
56.
Smith,B.J., Curtis,J.F. and Eling,T.E. (
1991
) Bioactivation of xenobiotics by prostaglandin H synthase.
Chem. Biol. Interact.
  ,
79
,
245
–264.
57.
Kaneko,F., Zhang,J.Z., Maruyama,K., Nihei,Y., Ono,I., Iwatsuki,K. and Yamamoto,T. (
1995
) Prostaglandin I1 analogues, SM-10902 and SM-10906, affect human keratinocytes and fibroblasts in vitro in a manner similar to PGE1: therapeutic potential for wound healing.
Arch. Dermatol. Res.
  ,
287
,
539
–545.
58.
Li,R.C., Cindrova-Davies,T., Skepper,J.N. and Sellers,L.A. (
2004
) Prostacyclin induces apoptosis of vascular smooth muscle cells by a cAMP-mediated inhibition of extracellular signal-regulated kinase activity and can counteract the mitogenic activity of endothelin-1 or basic fibroblast growth factor.
Circ. Res.
  ,
94
,
759
–767.
59.
Phillips,P.G., Long,L., Wilkins,M.R. and Morrell,N.W. (
2005
) cAMP phosphodiesterase inhibitors potentiate effects of prostacyclin analogs in hypoxic pulmonary vascular remodeling.
Am. J. Physiol. Lung Cell Mol. Physiol.
  ,
288
,
L103
–L115.
60.
Fukumoto,K., Yano,Y., Virgona,N., Hagiwara,H., Sato,H., Senba,H., Suzuki,K., Asano,R., Yamada,K. and Yano,T. (
2005
) Peroxisome proliferator-activated receptor delta as a molecular target to regulate lung cancer cell growth.
FEBS Lett.
  ,
579
,
3829
–3836.

Author notes

1Department of Veterinary and Biomedical Sciences and The Center for Molecular Toxicology and Carcinogenesis, The Pennsylvania State University, University Park, PA, 16802, USA, 2Graduate Program in Molecular Toxicology, The Huck Institutes of the Life Sciences, The Pennsylvania State University, University Park, PA, 16802, USA and 3Laboratory of Metabolism, National Cancer Institute, Bethesda, MD, 20892, USA