15-PGJ2, but not thiazolidinediones, inhibits cell growth, induces apoptosis, and causes downregulation of Stat3 in human oral SCCa cells

Activation of peroxisome proliferator-activated receptor gamma (PPARγ) has been linked to induction of differentiation, cell growth inhibition and apoptosis in several types of human cancer. However, the possible effects of PPARγ agonists on human oral squamous cell carcinoma have not yet been reported. In this study, treatment with 15-deoxy-Δ12,14-PGJ2 (15-PGJ2), a natural PPARγ ligand, induced a significant reduction of oral squamous cell carcinoma cell growth, which was mainly attributed to upregulation of apoptosis. Interestingly, rosiglitazone and ciglitazone, two members of the thiazolidinedione family of PPARγ activators, did not exert a growth inhibitory effect. Given the critical role that the oncogene signal transducer and activator of transcription 3 (Stat3) plays in head and neck carcinogenesis, its potential regulation by PPARγ ligands was also examined. Treatment of oral squamous cell carcinoma cells with 15-PGJ2 induced an initial reduction and eventual elimination of both phosphorylated and unphosphorylated Stat3 protein levels. In contrast, other PPARγ did not induce similar effects. Our results provide the first evidence of significant antineoplastic effects of 15-PGJ2 on human oral squamous cell carcinoma cells, which may be related to downmodulation of Stat3 and are at least partly mediated through PPARγ-independent events. British Journal of Cancer (2002) 87, 1396–1403. doi:10.1038/sj.bjc.6600618 www.bjcancer.com © 2002 Cancer Research UK

Peroxisome proliferator-activated receptors (PPARs) are members of the nuclear hormone receptor family, which function as ligand-dependent, sequence-specific activators of transcription. The PPAR family consists of three distinct molecules, termed a, d (b, FFAR or NUC-1), and g, encoded by separate genes and characterised by specific tissue and developmental distribution patterns (Mangelsdorf et al, 1995;Lemberger et al, 1996;Gelman et al, 1999;Kersten et al, 2000). A variety of natural and pharmacological ligands, including prostaglandins, the fibrate class of hypolipidaemic drugs, the anti-diabetic drugs thiazolidinediones, and certain non-steroidal anti-inflammatory drugs, are able to bind to and activate PPARs (Forman et al, 1995Yu et al, 1995;Lehmann et al, 1997). Upon activation, PPARs heterodimerise with the retinoic X receptor (RXR) and bind to peroxisome proliferator response elements (PPREs), located in the promoter region of target genes, driving their transcription (Mangelsdorf et al, 1995).
PPARs were initially described as molecular targets for compounds that induce peroxisome proliferation (Issemann and Green, 1990). However, shortly after their discovery, it became apparent that the physiologic role of PPARs extends far beyond peroxisome proliferation, involving such diverse processes as lipid homeostasis, insulin sensitisation, inflammation, and cell proliferation (Lemberger et al, 1996;Gelman et al, 1999;Kersten et al, 2000). Today, PPARs are recognised as key regulators of lipid homeostasis, playing fundamental roles in adipogenesis and fat catabolism (Lemberger et al, 1996;Kersten et al, 2000). Potential roles of PPARs in the treatment of diabetes mellitus (Vamecq and Latruffe, 1999), in inflammation control (Gelman et al, 1999), and in the regulation of atherosclerosis and thrombosis (Vamecq and Latruffe, 1999;Duez et al, 2001) have also been described.
The role of PPARg in the acquisition of an adipocyte phenotype, through the control of the expression of genes that promote cell cycle withdrawal, drive differentiation and induce apoptosis (Shao and Lazar, 1997;Debril et al, 2001), prompted many investigators to study the potential function of PPARg in neoplasia. Multiple lines of evidence suggest that PPARs, especially PPARg, play an important role in modulating cell proliferation and tumour growth (Gelman et al, 1999;Kersten et al, 2000;Debril et al, 2001;Park et al, 2001;Rosen and Spiegelman, 2001). Ligand-induced PPARg activation has been shown to promote differentiation and to induce cell growth inhibition and apoptosis in several types of human cancer, including colon cancer Kitamura et al, 1999;Yang and Frucht, 2001), breast cancer (Elstner et al, 1998;Mueller et al, 1998), lung cancer (Chang and Szabo, 2000), prostate cancer (Kubota et al, 1998;Butler et al, 2000), gastric cancer (Sato et al, 2000), liposarcoma (Tontonoz et al, 1997;Demetri et al, 1999), and leukaemia (Sugimura et al, 1999). Histo-logical and biochemical evidence indicate that PPARg ligands induce tumour cell differentiation in patients with advanced liposarcoma (Demetri et al, 1999). Consistent with its possible function as a tumour suppressor gene, PPARg has been reported to be functionally mutated in sporadic cases of colon cancer ; nonetheless, the overall incidence of PPARg mutations in human malignancies seems to be very rare (Ikezoe et al, 2001).
We recently showed that the non-steroidal anti-inflammatory drug (NSAID) sulindac induces cell growth inhibition and apoptosis in human oral squamous cell carcinoma (SCCa) cells, accompanied by upregulation of the mRNA and protein expression of PPARg. Treatment with antisense PPARg oligonucleotides abolished the cell growth inhibitory effect of the NSAID sulindac. These results suggest that upregulation of PPARg expression and activation may be, at least partially, responsible for sulindac's antineoplastic effect (Nikitakis et al, 2002a). However, it is unknown whether ligand-induced PPARg activation may alter the cell growth of oral SCCa. Here, we assessed the effects that the natural PPARg ligand 15-deoxy-D 12,14 -PGJ 2 (15-PGJ 2 ) and the synthetic PPARg ligands rosiglitazone and cigliatazone have on cell growth, apoptosis and cell proliferation in oral SCCa cells. Moreover, we explored the possibility that PPARg activation may affect the expression and activation of Stat3, an oncogene that plays a critical role in head and neck carcinogenesis (Grandis et al, 1998(Grandis et al, , 2000Bromberg et al, 1999;Bowman et al, 2000) and is downregulated by sulindac in oral SCCa cells (Nikitakis et al, 2002b).

Quantitative reverse transcriptase polymerase chain reaction
To estimate the mRNA levels of PPARg, real-time quantitative reverse transcriptase polymerase chain reaction (RT -PCR) was performed using a PE Applied Biosystems protocols. Total RNA was isolated using the TRIZOL Reagent (BRL/Life Technologies) and the concentration of RNA was determined using spectrophotometry. The forward and reverse PPARg primers were selected using Primer Express software (PE Applied Biosysystems, Foster City, CA, USA) as follows: 5'-TATCGACCAGCTGAATCCAGAG-3' (forward) and 5'-TCGCCTTTGCTTTGGTCA-3' (reverse). For the PCR reaction, a SYBR 1 Green PCR kit (PE Applied Biosystems) was used and the analyses were performed in triplicate. For each well, 5 ml of 25 nM RNA extract were added to a solution consisting of: 25 ml Master Mix solution (SYBR Green PCR Buffer, AmpErase 1 UNG, AmpliTaq Gold 1 DNA Polymerase, dATP, dCTP, dGTP, dUTP and 25 mM MgCl 2 ), 1 ml of RNAse inhibitor, 0.25 ml of reverse transcriptase (MultiScribe), 1 ml of each primer and 16.75 ml of water. The amplification process included 30 min at 488C, 10 min at 958C, followed by 40 cycles of 15 s at 958C and 1 min at 608C. Thermal cycling and fluorescence detection were performed using an ABI 5700 Prism (PE Applied Biosystems). Relative quantitation of the signal of PPARg mRNA followed. The signal of the target mRNA was normalised by comparison with the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA signal. The normalised amount of PPARg mRNA present in each cell line was calculated by arbitrarily designating SCC25 cells as a calibrator using a comparative Ct method following PE Applied Biosystems protocols.

Immunocytochemistry
Cells were plated on chamber slides (LabTech Nalge/Nunc) at a density of 5610 4 cells/chamber and were allowed to grow until almost confluent. The cells were then rinsed with Hanks' balanced salt solution (HBSS) and fixed with 95% ethanol for 20 min. PPARg protein expression was ascertained by immunocytochemical analysis (Santa Cruz Biotechnology, Santa Cruz, CA, USA, sc-1984) 1 : 50. The presence of antibody staining was determined after incubation of the sections with a secondary antibody (Biogenex, San Ramon, CA, USA, HK327-UG) 1 : 20, followed by the application of StrepABComplex/HRP (Dako, Carpinteria, CA, USA, K0377), and diaminobenzidine (DAB). The slides were counterstained with Harris's haematoxylin. The intensity of the immunostaining was classified as: 0 (negative), 1 (weakly positive), 2 (moderately positive), and 3 (strongly positive).

Immunohistochemistry
Ten cases of oral squamous cell carcinoma were randomly selected from the files of the Department of Diagnostic Sciences and Pathology. No patients were identified for these studies. The histological slides were reviewed to confirm diagnosis.
Five-micron sections of paraffin-embedded tissue were mounted on glass slides, deparaffinised and rehydrated. An antigen retrieval procedure was performed by placing the sections in Citra-solution (HK086-9K; Biogenex, San Ramon, CA, USA) inside a plastic pressure cooker, which was positioned in a microwave oven (Kenmore; Sears, Chicago, IL, USA). The specimens were treated by two cycles, 15 min each, at a high level and at level 4, respectively. Endogenous peroxidase activity was blocked with 3% hydrogen peroxide and non-specific protein was blocked with a universal blocking reagent (HK085-5K; Biogenex). Sections were then treated with PPARg antibody, followed by incubation with secondary antibody, application of StrepABComplex/HRP and staining with diaminobenzidine (DAB) and Harris haematoxylin, using the aforementioned reagents and dilutions.

Cell growth inhibition
Cells were plated on 24-well plates at a density of 5610 4 cells well 71 . After 24 h, the growth medium was supplemented with DMSO at a concentration of 0.1% or with either one of the following: 15-deoxy-D 12,14 -Prostaglandin J 2 (15-PGJ 2 ) (Cayman Chemical, Ann Arbor, MI, USA) at concentrations of 10 or 20 mM, rosiglitazone (Cayman Chemical) at concentrations of 25 or 50 mM, and ciglitazone (Biomol, Plymouth Meeting, PA, USA) at concentrations of 25 or 50 mM. All three compounds were dissolved in 100% DMSO, so that the final concentration of DMSO did not exceed 0.1%. Following incubation for 24, 48 or 72 h, either treated or untreated cells were removed enzymatically and counted using a Coulter Counter (Coulter Model ZI, Coulter Corporation, Miami, FL, USA). The per cent of growth was determined setting as 100% the growth of cells treated only with the vehicle (0.1% DMSO). All analyses were performed in triplicate.

Cell cycle analysis
Cells were treated either with the vehicle alone (0.1% DMSO) or with 10 or 20 mM of 15-PGJ 2 dissolved in 100% DMSO for 72 h. Cells were dissociated using trypsin-EDTA in PBS, pelleted and resuspended in 1 ml of PBS. While vortexing, 5 ml of 70% ethanol in distilled water were added dropwise. The cells were incubated at 48C for 30 min and then were centrifuged at 2500 g for 5 min. The pellet was resuspended in 200 ml of two-fold propidium iodide (PI) dye and 200 ml of RNAse A (2 mg ml 71 ) and the tubes were incubated for 45 min in the dark at 48C. Cells were filtered through nylon mesh prior to analysis. The DNA content of cells stained with PI and was measured with a FACScan instrument using Cell Quest software (Becton/Dickinson). To determine the proportion of cells in G0-G1, S and G2-M, cell cycle analysis of DNA histograms was performed using ModFitLTV2.0 (PMac). All analyses were performed in duplicate.

Apoptosis analysis
Apoptosis was assessed by Annexin V-FITC staining (BD Biosciences). In essence, cells were treated either with the vehicle alone (0.1% DMSO) or with 10 or 20 mM of 15-PGJ 2 dissolved in 100% DMSO for 72 h. Then, cells were washed twice with cold PBS and resuspended in 16 binding buffer at a concentration of 1610 6 cell ml 71 . One hundred microlitres of the solution were transferred to a 5 ml culture tube and 5 ml of Annexin V-FITC and 5 ml of PI were added. The mixture was gently vortexed and incubated for 15 min in the dark. Four hundred microlitres of 16 binding buffer were added to each tube and FACScan analysis of the samples was performed within 1 h using a FACS scan equipped with Cell Quest software (Becton Dickenson). The following controls were used to set up compensation and quadrants: (1) unstained cells, (2) cells stained with Annexin V-FITC alone, and (3) cells stained with PI alone. The percentage of cells that have been induced to undergo apoptosis was determined by subtracting the percentage of apoptotic cells in the untreated population from the percentage of apoptotic cells in the treated population. All analyses were performed in duplicate.

Protein lysate preparation and Western blotting
Cells were plated on 6-well plates at a density of 5610 4 cells well 71 and were allowed to grow to 80% confluency. Then, 15-PGJ 2 at 20 mM concentration, rosiglitazone at 50 mM concentration, or ciglitazone at 50 mM concentration, was added to the medium. All three compounds were dissolved in 100% DMSO, so that the final concentration of DMSO did not exceed 0.1%. Following incubation for various time periods, the cells were washed twice with cold PBS, lysed in RIPA buffer (50 mM Tris (pH 7.4), 150 mM NaCl, 1% Triton X-100, 1% deoxycholic acid, sodium salt, 0.1% sodium dodecyl sulphate (SDS), 100 mg ml 71 phenylmethylsulphonyl fluoride, 1 mg ml 71 aprotinin, 1 mM dithiothreitol, and 1 mM sodium orthovanadate) for 10 min, and scraped. The extracts were centrifuged at 40 000 g for 15 min at 48C. Protein concentrations were measured and equalised using Bio-Rad protein assay (Bio-Rad Laboratories, Richmond, CA, USA) according to the manufacturer's instructions.

Statistical analysis
Data were submitted to an analysis of variance with repeated measures (time and dosage), using Newman -Keuls post hoc comparisons where appropriate (Statistica for Windows, StatSoft, Inc., Tulsa, OK, USA). An alpha value of P40.05 was used in all models and post hoc comparisons.

Expression of PPARg in oral SCCa cell lines and tissues
Quantitative RT -PCR analysis showed mRNA expression for PPARg in all four oral SCCa cell lines. The lowest levels of PPARg mRNA were expressed in SCC25 and the highest in SCC9 (Table  1). Protein expression of PPARg was detected by means of immunocytochemistry; PPARg immunostaining, primarily in a cytoplasmic location, was evident in all four oral SCCa cell lines ( Figure 1A,B). Immunohistochemistry for PPARg in tumour specimens of patients with oral SCCa revealed PPARg production by the tumour cells, which was limited to the well-differentiated areas of the tumours ( Figure 1C).

Effect of PPARg agonists on cell growth inhibition
Oral SCC25 cells treated with 10 or 20 mM of 15-PGJ 2 exhibited reduction in cell growth ( Figure 2); similar results were obtained from the other oral SCCa cell lines. Combining the results from all four cell lines, a significant main effect for time (F 2,6 =20.6, P40.01) and interaction of time vs dosage (F 4,12 =17.4, P40.0001) was observed, reflecting that increases in dosage and time of treatment were associated with decreases in cell growth.
In contrast, treatment with rosiglitazone or ciglitazone did not induce significant cell growth inhibition in any of the tested cell lines (P40.05), even when the highest dosage of treatment (i.e. 50 mM) for the longest duration (i.e. 72 h) was employed ( Figure  2C).
Effect of 15-deoxy-D 12,14 -PGJ 2 on apoptosis and cell proliferation We tested next whether the cell growth inhibition induced by 15-PGJ 2 was due to alterations in apoptosis or cell proliferation rates.
Treatment of oral SCCa cells with 10 or 20 mM of 15-PGJ 2 did not significantly affect the percentage of cells in the S phase of the cell cycle, which remained stable or was only slightly increased; Results were obtained by real-time quantitative RT-PCR; b All cell lines were compared to SCC25, whose level of PPARg mRNA expression was set as 1. however, an increase in the percentage of cells in the G2 phase of the cell cycle was observed, accompanied by a corresponding reduction of cells in the G1 phase (Table 2). These results indicate a relative accumulation of cells in the G2 phase, which may interfere with cell cycle progression. On the other hand, significant 5 -8-fold increases in the levels of apoptosis resulted following treatment with 15-deoxy-D 12,14 -PGJ 2 for 72 h at concentrations ranging from 10 to 20 mM (F 2,4 =9.5, P40.05) (Figure 3).

Effects of PPAR agonists on Stat3 phosphorylation and expression
Because of the critical role of Stat3 in head and neck carcinogenesis (Grandis et al, 1998(Grandis et al, , 2000 and our recent demonstration of sulindac-mediated downmodulation of Stat3 in oral SCCa (Nikitakis et al, 2002b), we explored the hypothesis that ligand-mediated PPARg activation causes changes in Stat3 expression and activation. Forty-five minutes of treatment with 20 mM of 15-PGJ 2 resulted in a significant reduction of phosphorylated Stat3 in SCC9 cells. Longer treatment, up to 9 h, did not induce further reduction in phosphorylated Stat3 levels. In contrast, phosphorylated Stat3 levels were eliminated after 24, 48 or 72 h of treatment with 20 mM of 15deoxy-PGJ 2 ( Figure 4A). Stat3 protein expression levels in SCC9 cells also exhibited a small decrease after 45 min of treatment with    Figure 4A). In accordance with the aforementioned results, phosphorylated Stat3 levels were eliminated and Stat3 protein expression levels were either eliminated or severely reduced by 72 h of 15-deoxy-PGJ 2 treatment with 20 mM in all tested cell lines ( Figure 4B). These effects could not be attributed to a nonspecific reduction of protein expression, as the protein expression levels of actin were not significantly affected by treatment ( Figure 4A,B). On the contrary, treatments with rosiglitazone or ciglitazone at 50 mM concentrations failed to affect the phosphorylation or expression of Stat3 ( Figure 4B).
Notably, no changes in PPARg protein expression were induced by either 15-deoxy-PGJ 2 or thiazolidinediones (Data not shown).

DISCUSSION
The capacity of PPARg to promote differentiation, cell cycle withdrawal and apoptosis has encouraged extensive investigation of its potential anticancer activity in multiple types of human cancer (see Introduction). We recently showed that NSAID sulindac upregulates PPARg expression and activity and relies on PPARg availability for its antineoplastic activities on oral SCCa (Nikitakis et al, 2002a). These observations prompted us to investigate the possibility that oral SCCa cells are also responsive to treatment with direct PPARg ligands.
15-PGJ 2 , a natural PPARg ligand, exerted a statistically significant growth inhibitory effect on oral SCCa cells. Significant increases in the levels of apoptosis were observed, suggesting that 15-PGJ 2 cell growth inhibitory effect is primarily mediated through induction of apoptosis. Nonetheless, accumulation of cells in the G2 phase of the cell cycle was also noted, supporting the concept that interference of 15-PGJ 2 with cell cycle progression may also contribute to its activity. In contrast, the well-characterised PPARg ligands rosiglitazone and ciglitazone did not inhibit the cell growth of oral SCCa cells, implying that PPARg activation is not sufficient for inducing antineoplastic effects on oral SCCa cells. These results raise the possibility that the antineoplastic properties of 15-PGJ 2 are, at least partially, mediated through PPARg-independent mechanisms.
PPARg-independent effects of 15-PGJ 2 have been previously reported in other systems, including chondrocytes, myofibroblasts, mesangial cells, inflammatory cells and cells of the nervous system (Petrova et al, 1999;Castrillo et al, 2000;Rossi et al, 2000;Straus et al, 2000;Boyault et al, 2001;Li et al, 2001;Janabi, 2002;Ward et al, 2002). Several mechanisms have been implicated as responsible for the effects of 15-PGJ 2 on these cells and may also partly explain its antineoplastic properties. Negative regulation of the NF-kB pathway through inhibition of IkB kinase (IKK) and abrogation of the DNA binding ability of NF-kB, has emerged as a major pathway of PPARg-independent 15-PGJ 2 activity (Petrova et al, 1999;Castrillo et al, 2000;Rossi et al, 2000;Straus et al, 2000;Boyault et al, 2001;Janabi, 2002;). 15-PGJ 2 -mediated NF-kB inhibition has been linked to downregulation of inducible nitric oxide synthase (iNOS) and abrogation of cyclooxygenase-2 (COX-2) transactivation (Petrova et al, 1999;Castrillo et al, 2000;Rossi et al, 2000;Straus et al, 2000;Boyault et al, 2001;Janabi, 2002). Similarly, Ward et al (2002) recently showed that 15-PGJ 2 exploits PPARg-independent inhibition of NF-kB activation to induce caspase-dependent apoptosis in granulocytes. Other mediators that have been implicated in PPARg-independent properties of 15-PGJ 2 and can potentially have a similar function in cancer cells include AP-1 (Boyault et al, 2001), MAP kinase (Harris et al, 2002;Lennon et al, 2002), and reactive oxygen species (Li et al, 2001;Lennon et al, 2002). The latter has been shown to act as intermediates for the induction of apoptosis caused by 15-PGJ 2 in human myofibroblasts (Li et al, 2001). Identification of the molecular pathways that mediate the PPARg-independent antineoplastic effects of 15-PGJ 2 should be thoroughly addressed in future studies. In this respect, the recent observations of Clay et al (2001) that early de novo gene expression is necessary for 15-PGJ 2 -induced apoptosis in breast cancer cells may be of particular relevance.
Although our data strongly support the existence of PPARgindependent effects of 15-PGJ 2 on oral SCCa cells, recruitment of PPARg-mediated pathways cannot be ruled out. However, the precise molecular mechanisms that are responsible for the antineo-  , or 1.5, 3, 4.5, 9, 24, 48, 72 h, as indicated. Cells were lysed, blotted with antibody to phosphorylated Stat3 (phospho Stat3), and then sequentially stripped and reprobed with antibodies to Stat3 and to actin. (B) 15-deoxy-PGJ 2 , but not rosiglitazone or ciglitazone, inhibits phosphorylation and production of Stat3 in oral SCCa cells. Oral SCCa cells, derived from cell lines SCC4, 15, or 25, were treated for 48 h with normal medium (NM), 50 mM ciglitazone (CI), 50 mM rosiglitazone (RO), or 20 mM 15-deoxy-PGJ 2 (PG), as indicated. Cells were lysed, blotted with antibody to phosphorylated Stat3 (phospho Stat3), and then sequentially stripped and reprobed with antibodies to Stat3 and to actin.

Molecular and Cellular Pathology
plastic properties of PPARg are not well understood. An association has been suggested between PPARg and COX-2, which has also been implicated in various human cancers, including head and neck SCCa (Chan et al, 1999;Dannenberg et al, 2001). PPARg activators may inhibit COX-2 expression, possibly through negative interference with NF-kB and/or AP-1 activation (Inoue et al, 2000;Subbaramaiah et al, 2001;Yang and Frucht, 2001). There is also evidence that supports the function of PPARg ligands as potent inhibitors of angiogenesis in vivo and in vitro, providing an additional mechanism that may partially account for the anticancer properties of PPARg (Bishop-Bailey and Hla, 1999;Xin et al, 1999). Finally, cross-talk between PPARg and other signalling molecules, such as NF-kB, AP-1 and STAT (Ricote et al, 1998;Zhou and Waxman, 1999a,b), may contribute significantly to the effects of PPARg on tumour growth. The possible contribution, if any, of PPARg activation to growth inhibition induced by 15-PGJ 2 treatment warrants further exploration. Constitutive activation of Stat3 plays an important role in the tumorigenesis of various types of human cancer (Catlett-Falcone et al, 1999;Fernandes et al, 1999), and abrogation of Stat3 signalling has been correlated with stimulation of cell proliferation, prevention of apoptosis and tumour formation (Bromberg et al, 1999;Bowman et al, 2000). Based on accumulating evidence that Stat3 is upregulated in head and neck SCCa (Grandis et al, 1998(Grandis et al, , 2000, we investigated whether treatment with PPARg ligands exerts an effect on Stat3 protein expression and Stat3 tyrosine phosphorylation. Treatment of oral SCCa cells with 15-PGJ 2 at concentrations that cause significant reduction of cell growth resulted in downregulation of both Stat3 expression and phosphorylation. Given that targeting of Stat3 in head and neck SCCa has been linked to significant growth inhibition and induction of apoptosis both in vitro and in vivo (Grandis et al, 1998(Grandis et al, , 2000, our results suggest that the ability of 15-PGJ 2 to downregulate Stat3 may be aetiologically related to its growth inhibitory and apoptotic effects. Interestingly, the mode of Stat3 downregulation (i.e. reduction of Stat3 phosphorylated levels after 45 min and elimination of both Stat3 phosphorylated and unphosphorylated levels after 72 h) was very similar to that induced by sulindac sulphide in the same cell lines (Nikitakis et al, 2002b). Although sulindac's growth inhibitory effect was dependent on PPARg availability, its ability to induce Stat3 downmodulation was independent of its ability to act as a PPARg ligand. Similarly, the inability of PPARg activation trough rosiglitizone and ciglitazone stimulation to affect the phosphorylation and expression levels of Stat3 entails that activation of PPARg-independent mechanisms is necessary for 15-PGJ 2 -mediated Stat3 downmodulation.
Both PPARg-dependent and PPARg-independent mechanisms may come into play during Stat3 downregulation by 15-PGJ 2 . Direct protein -protein interactions or indirect mechanisms, such as competition for common co-activators or modulation of inhibitors of transcriptional activity, have been suggested as possible mediators of STAT-PPAR cross-talk and could account for the Stat3 inhibitory effect of PPARg activation (Zhou and Waxman, 1999a,b). Cytokine stimulation results in phosphorylation of STATs through the mediation of the Janus kinase (JAK) family of protein tyrosine kinases (Darnell et al, 1994). Accordingly, the recently proposed association between PPARg and cytokine expression may represent another possible connection between Stat3 and PPARg. Indeed, PPARg agonists have been shown to suppress monocyte elaboration of inflammatory cytokines (Jiang et al, 1998) and to inhibit IL-1b-induced expression of IL-8 in colon cancer cell lines (Su et al, 1999). 15-PGJ 2 -mediated inhibition of cytokine production and function may also ensue through PPARg-independent pathways; for example, 15-PGJ 2 has been shown to modulate IL-1b effects in human chondrocytes (Boyault et al, 2001) and to inhibit TNF-a and IL-6 production in human macrophages without PPARg mediation. Inhibition of cytokine expression by 15-PGJ 2 would conceivably result in decreased activation of STAT molecules that mediate the transduction of the cytokine signal from the cell surface to the nucleus. Head and neck cancer cells have been shown to express a variety of pro-inflammatory and pro-angiogenic cytokines (Chen et al, 1999;Ondrey et al, 1999) and to respond to IL-6 stimulation with upregulation of Stat3 phosphorylation and promotion of cell growth; reversal of this process could participate in the observed effects of 15-PGJ 2 on oral SCCa cells. In that aberrant tumour growth factor-a/ epidermal growth factor receptor signalling has been demonstrated to play a major role in Stat3 constitutive activation of head and neck SCCa cells, possible interference of 15-PGJ 2 with this pathway, at the level of the ligand or the receptor, constitutes another distinct possibility.
In summary, we demonstrated that 15-PGJ 2 , a specific natural PPARg ligand, inhibits growth of oral SCCa cells, which may be related to its capacity to downregulate the oncogene Stat3. The ineffectiveness of rosiglitazone and ciglitazone to cause similar effects strongly suggests that 15-PGJ 2 effects are at least partly mediated through PPARg-independent mechanisms. Delineation of these mechanisms, as well as determination of the potential contribution of PPARg activation, not only will provide an explanation for 15-PGJ 2 anticancer qualities but it will also enhance our understanding of critical signalling pathways for oral SCCa carcinogenesis.