The conversion of arachidonic acid (AA) to prostaglandin H2 (PGH2 ), the committed step in prostanoid biosynthesis, is catalyzed by the COX isoenzymes, COX-1 and COX-2, each of which is encoded by a unique gene, located on different chromosomes (Smith et al., 1996). COX-1 is expressed constitutively in most tissues and is thought to perform primarily “housekeeping” functions, such as gastric cytoprotection, regulation of renal blood flow, and platelet aggregation. In contrast, COX-2 message and protein are normally undetectable in most tissues, but its expression can be induced rapidly as part of the inflammatory reaction in response to a wide variety of extracellular stimuli (Herschman, 1996; Smith et al., 1996). In addition to its association with inflammation, COX-2 expression clearly plays a role in regulating cellular proliferation, differentiation and tumorigenesis (Bornfeldt et al., 1997; DuBois et al., 1996). There are indications that intestinal epithelial cells overexpressing the COX-2 gene develop altered adhesion properties and become resistant to undergoing apoptosis (Tsujii and DuBois, 1995). Direct genetic evidence that COX-2 plays a key role in tumorigenesis was provided with a murine model system, which showed that knocking out the COX-2 gene caused a marked reduction in the number and size of intestinal polyps in ApcΔ716 knockout mice, a murine model of familial adenomatous polyposis (Oshima et al., 1996). In humans, increased COX-2 expression is found in carcinomas of various organs including colon, breast, prostate, lung, esophagus, pancreas and mucous membranes of head and neck (Subbaramaiah et al., 1996; Zimmermann et al., 1999). Treatment with highly selective COX-2 inhibitors results in a reduction in the size and number of colonic lesions in several animal models (Kawamori et al., 1998) and suppression of carcinogenesis of human colon, prostate and esophageal cancers (Zimmermann et al., 1999). The mechanism by which COX-2 inhibitors suppress carcinogenesis is attributed to its modulation of prostanoid production which affects cell proliferation, tumor growth and immune responsiveness. PGE2, a major product of COX-2, induces bcl-2 expression and inhibits apoptosis and, conversely, that COX-2 inhibitors induce apoptosis (Sheng et al., 1998a; Zimmermann et al., 1999). While the involvement of prostanoids in carcinogenesis is clear, another study showed that selective inhibition of COX-2 activity suppresses the production of PGE2 but does not alter the progress of the morphological transformation of rat fibroblasts (Sheng et al., 1998b), suggesting the existence of distinct signal pathways, PGE2-dependent and PGE2-independent, in the regulation of carcinogenesis.
Since we and others reported that COX-2 is associated with differentiation and transformation of keratinocytes (Kanekura et al., 1998; Leong et al., 1996), we examined the expression of COX-2 and its role in regulating cell growth using two different skin epidermal cancer cell lines and examined whether prostaglandins and COX-2 expression have selective effects on cell growth using either a COX-2 inhibitor which inhibits its catalytic activity or a COX-2 antisense oligonucleotide that inhibits its expression.
MATERIAL AND METHODS
Cell culture reagents were purchased from Life Technologies, Inc. (Grand Island, NY). Anti-PGE2 antibody was purchased from Upstate Biotechnology (Lake Placid, NY). [3H]PGE2 was obtained from DuPont-NEN (Boston, MA). Polyclonal antibodies to human COX-1, COX-2 and cytosolic phospholipase A2 (cPLA2) were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). NS398 was purchased from Cayman Chemicals (Ann Arbor, MI). The COX-2 antisense oligonucleotide (GGAAACATCGACAGT) was synthesized by SAWADY (Tokyo, Japan). The sequence of the antisense oligonucleotide corresponded to nucleotides 2263 to 2277 of the sense sequence (Hla and Neilson, 1992), which was determined according to the previous report (Tsujii et al., 1998).
HaCaT cells were kindly provided by Dr. L. Matrisian (Vanderbilt University, Nashville, TN) with the permission of Dr. N. E. Fusenig (German Cancer Research Center, Heidelberg, Germany) and the HSC-5 cell line was a gift from Dr. Y. Hozumi (Yamagata University, Yamagata, Japan). HaCaT, HSC-5 and EcCa were grown as described previously (Boukamp et al., 1988) with some modifications. Cells were cultured at 37°C in a humidified atmosphere containing 5% CO2 in Dulbecco's modified Eagle's medium (DMEM) containing high glucose and 100,000 units/l Penicillin G, 100 mg/l Streptomycin, 1 mg/l Fungizone, 0.1 mM nonessential amino acids, 292 mg/l glutamine, 50 mg/l ascorbic acid and 10% FCS.
Prostaglandin E2 measurement
Cells were seeded at 1×105 cells/ml in 24-well tissue culture plates (0.5 ml/well) and incubated for 72 hr until confluent. The media was then replaced with fresh DMEM containing 1% FCS. The culture media were collected after 24 hr and analyzed for PGE2 production. PGE2 produced by transfected cells was measured with a little modification because of the suppressed cell growth and the results were normalized by cell numbers of the dishes. PGE2 was measured by the radioimmunoassay described previously (Kanekura et al., 1998). The method is based on the competition of PGE2 in the test sample with [3H]-labeled PGE2 for binding to anti-PGE2 antibody. A 10 to 100 μl aliquot of culture medium was added to RIA assay buffer (0.1 mM phosphate buffer, pH 7.4, containing 0.9% sodium chloride, 0.1% sodium azide, and 0.1% gelatin), mixed with the appropriate amounts of labeled PGE2 and reconstituted antiserum, and incubated overnight at 4°C. The assay tubes were then placed on ice and 1 ml of cold charcoal-dextran suspension was added. After 15 min, the tubes were centrifuged at 2,200g for 10 min at 4°C. The supernatants were decanted into scintillation vials and radioactivity was determined by scintillation spectrometry. Percent binding was compared to a standard curve and the amounts of PGE2 in the samples were calculated.
Cells grown in 10-cm plastic dishes were pelleted in PBS containing 1 μg/ml leupeptin, 0.03% Aprotinin, 2 mM PMSF and 2 mM EDTA, pH 7.2. Proteins were obtained by suspending the pellet in extraction buffer (62.5 mM Tris, 2% SDS, 0.5 mM DTT, 1 mM fresh PMSF, pH 6.8) and denaturing the proteins by incubating the tubes at 95°C for 5 min. Identical amounts of protein were electrophoresed in a 7.5% SDS/polyacrylamide gel and then transferred onto a nitrocellulose membrane. The membrane was blocked with 1% BSA and 4% non-fat dry milk in PBS containing 0.1% Tween 20 and then incubated with either anti-COX-1, anti-COX-2 or anti-cPLA2 antibodies for 1 hr at room temperature. After incubation with anti-goat (COX) or anti-rabbit (cPLA2) alkaline phosphatase linked immunoglobulin for 30 min at room temperature, the immunoreactive bands were detected using an ECL kit (Amersham, Arlington Heights, IL).
Transfection was carried out using the Lipofectin reagent (Gibco BRL, Gaithersburg, MD) according to the manufacturer's instructions. HSC-5 and EcCa were seeded at 2× 105 cells/plate into 6-cm dishes and grown in DMEM containing 10% FCS until the cells were 50–70% confluent. Two micrograms of COX-2 antisense oligonucleotide and 20 μl Lipofectin were combined and incubated for 15 min at room temperature. Cells were incubated with the DNA lipid complexes for 12 to 18 hr and fed with DMEM containing 10% fetal calf serum.
Cell growth assays
Cell growth was monitored by a colorimetric [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] (MTT) assay with minor modifications. Cells were seeded at 5 × 103 into 96 well plates. At the indicated times (culture days), 10 μl of MTT solution (2.5 mg/ml in PBS) was added to each well and the plates were incubated at 37°C for 4 hr. Acid-SDS (100 μl of 0.01 N HCl in 10% SDS) was added to all the wells and mixed thoroughly to dissolve the dark blue crystals. The plates were kept at room temperature over night in the dark to ensure that all crystals were dissolved. They were read with a Microplate Reader, using a test wavelength of 570 nm and a reference wavelength of 630 nm.
Increased levels of COX-2 protein expression in skin epidermal cancer cells
The levels of COX-2 expression were studied in 2 different skin epidermal cancer cells, HSC-5 and EcCa. As shown in Figure 1, the skin epidermal cancer cells expressed elevated levels of COX-2 protein compared to the nontumorigenic human keratinocyte cell line, HaCaT, as detected by Western immunoblotting. COX-2 expression was higher in EcCa than in HSC-5 (Fig. 1).
Increased PGE2 production proportional to COX-2 expression levels in HSC-5 and EcCa
Because PGE2 is a major COX-2 product, the levels of PGE2 produced by these cells were measured. As shown in Figure 2, HSC-5 and EcCa synthesized 17-fold and 90-fold more PGE2 than HaCaT, respectively. The amounts of PGE2 produced correlated well with the levels of expression of COX-2 shown in Figure 1.
Decrease in expression of COX-2 protein by transfection with COX-2 antisense oligonucleotide
In order to elucidate the functional role of COX-2 in regulating cell growth, we examined the effect of suppressing its expression on cell growth. We employed a COX-2 antisense oligonucleotide which specifically inhibits COX-2 expression but has no biological function. As shown in Figure 3, Western blotting of extracts confirmed that COX-2 expression was suppressed in HSC-5 and EcCa cells treated with the antisense oligonucleotide.
Suppression of growth by inhibition of COX-2 expression
When COX-2 expression was inhibited by the antisense oligonucleotide, HSC-5 and EcCa cell growth was suppressed with statistical significance (Fig. 4).
Suppression of cell growth by inhibition of COX-2 catalytic activity
The effect of a highly selective COX-2 inhibitor, NS398, on PGE2 production and cell growth was examined. COX-2 expression was not reduced by 20 μM NS398 in both HSC-5 and EcCa (Fig. 5). In contrast, as shown in Figure 6, cell growth was suppressed by NS398, which at 20 μM decreased PGE2 production more significantly than the antisense oligonucleotide (Fig. 7).
Elevated COX-2 expression has previously been observed in various carcinomas including skin cancers (Buckman et al., 1998; Muller-Decker et al., 1999; Subbaramaiah et al., 1996; Zimmermann et al., 1999). While COX-2 expression in normal skin was usually very low and restricted to regions of differentiated epidermis (Buckman et al., 1998; Muller-Decker et al., 1999), studies on mouse and human skin carcinogenesis revealed that overexpression of COX-2 contributes to the development of skin cancers and COX-2 expression is constitutively enhanced in skin cancers (Buckman et al., 1998; Muller Decker et al., 1998, 1999). In the present report, we showed that the skin epidermal cancer cells, HSC-5 and EcCa expressed elevated levels of COX-2 protein. PGE2, a major COX-2 product, was synthesized much more by HSC-5 and EcCa than HaCaT. PGE2 is believed to play a role in cell proliferation, differentiation and development of tumors. For example, PGE2 treatment of human colon cancer cells induces bcl-2 expression and inhibits apoptosis (Sheng et al., 1998a). In human skin keratinocytes, PGE2 modulates calcium-induced differentiation (Evans et al., 1993). Our findings support these studies showing that COX-2 and PGs are involved in carcinogenesis.
Conversion of arachidonic acid to PGs is catalyzed by 2 isoenzymes, COX-1 and COX-2. COX-1 is expressed constitutively in most tissues to generate PGs for physiological functions, whereas COX-2 expression is induced by various extracellular stimuli, including growth factors and tumor promoters (Herschman, 1996). In addition to COX, a primary enzyme in PG production is cytosolic phospholipase A2 (cPLA2), which releases arachidonic acid from membrane phospholipids into the cytoplasm. Previous studies revealed that cPLA2 also mediates cell proliferation through mitogen-activated protein kinase and/or protein kinase C pathways (Bornfeldt et al., 1997). Therefore, we also examined COX-1 and cPLA2 expression in HSC-5 and EcCa, but no increase was found compared with their expression in HaCaT cells (data not shown).
When HSC-5 and EcCa were treated with the COX-2 antisense oligonucleotide, COX-2 expression and cell growth were suppressed. This clearly indicates that COX-2 plays an important role in tumor cell growth. One possible mechanism by which COX-2 is thought to function in carcinogenesis is via increased PG biosynthesis resulting from enhanced COX-2 expression, since PGE2 induces cell proliferation (Sheng et al., 1998b), promotes angiogenesis (Tsujii et al., 1998) and inhibits immune surveillance in human cancers.
While it is clear from a number of previous reports that PGs are involved in carcinogenesis, a study by Sheng et al. (1998b) indicates the existence of a pathway of cell transformation which is independent of PGE2. Ras proteins are key molecules in cell proliferation and differentiation, and COX-2 is thought to be an important target gene of Ras (Sheng et al., 1998b; Subbaramaiah et al., 1996). Sheng et al. (1998b) showed that COX-2 was induced by Ras and produced PGE2 in Ras-mediated transformation of rat fibroblasts. In that study, a selective COX-2 inhibitor suppressed PGE2 production but neither delayed the progress nor reduced the extent of morphological transformation, indicating that increased PGE2 production was not required for Ras-mediated morphological transformation. This finding suggests that 2 distinct signal pathways, PGE2-dependent and PGE2-independent, may regulate carcinogenesis. In order to examine whether PGE2 and COX-2 expression have a specific effect on cell growth, the effect of a highly selective COX-2 inhibitor, NS398 on PGE2 production and cell growth was compared with the effect of the antisense oligonucleotide. NS398 inhibits COX-2 catalytic activity but does not suppress its expression (Chinery et al., 1998). COX-2 expression was not reduced by 20 μM NS398 in both HSC-5 and EcCa whereas cell growth was suppressed by NS398, and the extent of growth suppression was more pronounced with the antisense oligonucleotide than with NS398 (see Figs. 4 and 6), despite the fact that 20 μM NS398 inhibited PGE2 production more than the antisense oligonucleotide. These results show that inhibiting COX-2 expression reduces cell growth more effectively than inhibiting its catalytic activity suggesting that there is a PGE2-independent pathway modulating cell growth, which is downstream of COX-2. This proposal may be supported by a previous study that demonstrated that combined treatment of colon cancer cells with an antioxidant and a selective COX-2 inhibitor, each of which reduces tumor growth independently, had an additive or synergistic effect on tumor regression (Chinery et al., 1998). COX is a bifunctional enzyme possessing cyclooxygenase activity, which converts arachidonic acid to PGG, and peroxidase activity, which generates PGH2 , a direct precursor of PGE2, from PGG. In addition to its role in prostaglandin synthesis, the peroxidase activity of COX contributes to superoxide production and subsequent alterations in intracellular redox status, which are associated with perturbations of cell growth and transformation (Chinery et al., 1998). Inhibitors such as NS398 inhibit the cyclooxygenase activity but not the peroxidase activity of COX (Subbaramaiah et al., 1996), whereas the suppression of COX-2 expression might result in the inhibition of both activities. These facts can explain why the antisense oligonucleotide reduced cell growth more effectively than NS398.
Our results show that COX-2 expression is enhanced in skin cancer cells and that cell growth is suppressed by inhibition of COX-2 expression.