Materials and Methods

Cell Culture.

The human sporadic CRC cell lines SW480 and HCT116 (European Collection of Animal Cell Cultures, Porton Down, UK) were cultured in RPMI 1640 medium containing Glutamax-I, supplemented with 10% (v/v) fetal bovine serum, 1000 U/ml penicillin, and 500 U/ml streptomycin (all Invitrogen, Paisley, UK), on tissue culture plastic, as described previously (Smith et al., 2000; Hawcroft et al., 2002).

Drugs, Antibodies, and DNA Plasmids.

Indomethacin (Sigma-Aldrich, Poole, Dorset, UK) was prepared as a 100 mM stock solution in dimethyl sulfoxide (DMSO; Sigma Chemical). Troglitazone (a kind gift from Parke-Davis Pharmaceutical Research, Ann Arbor, MI) was prepared as a 10 mM stock solution in DMSO. In experiments testing the effects of these drugs, control cell cultures always contained an equivalent v/v dilution of DMSO to that in cultures that contained the highest concentration of drug.

Mouse monoclonal anti-human PPARγ antibody (E8) and mouse monoclonal anti-human cyclin D1 antibody (A-12) were obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Mouse monoclonal anti-human β-actin antibody (AC-15) and anti-FLAG (M2) antibody were obtained from Sigma Chemical. Horseradish peroxidase-conjugated rabbit anti-mouse secondary antibody was obtained from DAKO Ltd. (Ely, UK).

A pCMX vector containing a peroxisome proliferator response element (x3)-luciferase reporter gene (PPRE3-tk-luciferase) was a kind gift from R. Evans (The Salk Institute, San Diego, CA) (Forman et al., 1995). Plasmid pcDNA3 containing FLAG epitope-tagged human PPARγ1 with a L468A/E471A double mutation in the AF-2 domain (thus producing dominant-negative PPARγ activity) was a kind gift from K. Chatterjee (Cambridge, UK) (Gurnell et al., 2000). pRL-TK Renilla luciferase-reporter gene plasmid was obtained from Promega (Madison, WI).

Reverse Transcription-Polymerase Chain Reaction Analysis of PPARγ mRNA Expression.

Total RNA was prepared from SW480 and HCT116 cells using RNeasy columns (QIAGEN, Crawley, UK) as per manufacturer's instructions and then reverse-transcribed. Polymerase chain reaction (PCR) amplification for human PPARγ was performed as described previously (Brockman et al., 1998), giving an amplicon size of 234 base pairs. PCR products were subjected to electrophoresis on 2% agarose in the presence of 0.5 mg ml⁻¹ ethidium bromide (Sigma Chemical).

Western Blot Analysis.

Cell monolayers were lysed using 50 mM Tris-HCl, pH 7.2, 0.137 M NaCl containing 1% (v/v) Brij 96 (all Sigma Chemical) as described previously (Smith et al., 2000; Hawcroft et al., 2002) or passive lysis buffer (Promega). The total protein concentration of lysate supernatants was determined using the DC protein assay (Bio-Rad, Hemel Hempstead, UK). NUPAGE NOVEX 10% bis-Tris 1-mm gels in 1× NU-PAGE 3-(N-morpholino) propanesulfonic acid running buffer (all Invitrogen) were used to resolve 20 μg of total protein samples and a MagicMark Western molecular weight standard (Invitrogen). Proteins were transferred to Hybond P polyvinylidene difluoride membranes (Amersham Biosciences UK Ltd., Little Chalfont, Buckinghamshire, UK) using a XCell II wet blot module with 1× NU-PAGE transfer buffer (Invitrogen). Membranes were blocked with a 5% (w/v) solution of nonfat skimmed milk powder in PBS containing 0.02% (v/v) Tween 20 (PBS/T) for 1 h at 20°C, before incubation with primary antibodies [anti-PPARγ, 1:200; anti-β-actin, 1:1000; or anti-cyclin D1, 1:1000 in PBS/T plus 5% (w/v) nonfat skimmed milk powder] for 1 h at 20°C. Subsequently, three washes with PBS containing 0.05% (v/v) Tween 20 were followed by incubation with secondary antibody (1:5000 in PBS/T plus 5% nonfat skimmed milk powder) for 1 h at 20°C. After three further washes with PBS containing 0.05% (v/v) Tween 20, enhanced chemiluminescence was detected as per manufacturer's instructions (Perbio Science, Tattenhall, UK).

Transient DNA Transfection and Dual Luciferase Reporter Assays.

GeneJuice transfection reagent (Novagen, Madison, WI) was added to serum-free RPMI 1640 medium containing Glutamax-I and incubated for 5 min before addition of the appropriate DNA. The GeneJuice-DNA mix was incubated for 30 min at 20°C before addition to 35-mm-well cell cultures (30–50% confluence) in RPMI 1640 medium containing Glutamax-I, supplemented with 2.5% (v/v) fetal bovine serum, 250 U/ml penicillin, and 125 U/ml streptomycin. Medium was removed and fresh medium [RPMI 1640 medium containing Glutamax-I, supplemented with 10% (v/v) fetal bovine serum, 1000 U/ml penicillin, and 500 U/ml streptomycin] containing drug or carrier control was added 24 h later. After a further 24 h, dual luciferase reporter assays (Promega) were performed as described previously (Hawcroft et al., 2002). Experiments were performed in triplicate, and all data are expressed as the mean + S.E.M. Firefly luciferase activity relative to Renilla luciferase activity. Cell lysates were used for subsequent Western blot analysis.

Cell Proliferation Assay.

Cells were plated at 5 × 10⁴ cells/well in 24-well plates. At 72 h, cells were transfected with PPARγΔAF-2 (4 μg) or mock transfected as described above. After 24 h, indomethacin (600 μM) or carrier control was added, and the cells were incubated for a further 24 h. Adherent cells were then harvested using 0.25% (w/v) trypsin and 1 mM ethylenediaminetetra-acetic solution (Invitrogen). Cell number and viability were measured using a hemocytometer and exclusion of 0.4% trypan blue (Sigma Chemical) in PBS. All conditions were assayed in triplicate.

Statistical Analysis.

Student's independent samplet test was used for pairwise comparisons. One-way analysis of variance (ANOVA), with post hoc Bonferroni tests, was used for multiple comparisons. Statistical significance was assumed if thep value was less than 0.05. All analyses were performed using SPSS (version 11) computer software (SPSS Science, Chicago, IL).

Results

HCT116 and SW480 Human CRC Cells Express Functional PPARγ.

First, we determined whether HCT116 and SW480 human CRC cells expressed functional PPARγ. These two cell lines were chosen for these PPARγ studies because we had previously demonstrated that they were sensitive to the growth inhibitory and proapoptotic effects of indomethacin (100–600 μM; Smith et al., 2000). HCT116 and SW480 cells contain COX-1 but do not express COX-2 constitutively (Smith et al., 2000), which mirrors the phenotype of intestinal epithelial cells at the earliest stages of colorectal carcinogenesis of relevance to CRC chemoprevention (Chapple et al., 2002). HCT116 and SW480 cell lines both expressed PPARγ mRNA and protein (Fig.1, a and b). The HCT15 human CRC cell line also expresses PPARγ protein, but it exists in a functionally inactive state in these cells (Brockman et al., 1998). Therefore, we confirmed the functional status of PPARγ in HCT116 and SW480 cells by testing that activation of PPARγ by the TZD troglitazone led to increased PPRE-luciferase reporter gene activity in HCT116 and SW480 cells. Troglitazone (10–50 μM) significantly increasedPPRE-luciferase reporter gene activity in both cell lines, in a concentration-dependent manner (Fig. 1c).PPRE-luciferase gene trans-activation induced by troglitazone was significantly greater in HCT116 cells than SW480 cells (Fig. 1c).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

SW480 and HCT116 human CRC cells express functional PPARγ. a, reverse transcription-PCR for PPARγ mRNA using SW480 and HCT116 cell total RNA. b, Western blot analysis of PPARγ (48-kDa) and β-actin (42-kDa) protein expression in SW480 and HCT116 cell lysates. c, SW480 and HCT116 cells contain functional PPARγ. APPRE3-tk-luciferasereporter (1 μg) was transiently cotransfected with aRenilla-tk-luciferasereporter (0.5 μg) into SW480 or HCT116 cells, which were then treated with the PPARγ ligand troglitazone (0–50 μM) for 24 h. Cells were lysed and dual luciferase assays were performed. Data from triplicate experiments are expressed as the mean + S.E.M. of the ratio of PPRE reporter Firefly luciferase activity to controlRenilla luciferase activity. ∗, p< 0.05; ∗∗, p < 0.01 compared with control values (ANOVA).

Indomethacin Activates PPARγ in HCT116 and SW480 Human CRC Cells.

Indomethacin induces G₁ arrest and apoptosis of human CRC cells, including HCT116 and SW480 cells in a concentration-dependent manner, at concentrations greater than 100 μM (Hixson et al., 1994; Smith et al., 2000; Turchanowa et al., 2001). These effects are demonstrable from 24 h onwards (Smith et al., 2000). In our previous studies, maximal effects of indomethacin on proliferation and apoptosis of human CRC cells were observed at 600 μM (Smith et al., 2000). At this concentration of indomethacin, we have also previously observed specific down-regulation of β-catenin protein levels (but not of its homolog γ-catenin) in SW480 and HCT116 cells (Hawcroft et al., 2002). Therefore, we initially tested the ability of this concentration of indomethacin to activate PPARγ in these human CRC cell lines. Indomethacin treatment (600 μM) significantly increased PPRE-luciferase activity in both cell lines (Fig. 2a). In a similar manner to troglitazone (Fig. 1c), the fold increase in PPARγ activation by indomethacin was greater in HCT116 cells than SW480 cells (Fig. 2a). Therefore, we used HCT116 cells as model human CRC cells in all subsequent experiments. Indomethacin induced PPARγ activation in a concentration-dependent manner (Fig. 2b) with maximal activation occurring at a concentration of 300 μM. Importantly, 10 μM indomethacin (a concentration below that required for significant, direct PPARγ activation (Lehmann et al., 1997; Jaradat et al., 2001;Yamazaki et al., 2002), but at which profound COX inhibition in human CRC cells occurs (Kokoska et al., 1999) did not significantly alter basal PPARγ activity in HCT116 cells, with only a slight increase inPPRE-luciferase activity being apparent (Fig. 2b). This suggests that down-regulation of PPARγ activity, via inhibition of COX-1-derived PG PPARγ ligand synthesis, does not occur at lower concentrations of indomethacin in HCT116 cells.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Indomethacin activates PPARγ in SW480 and HCT116 human CRC cells. a,PPRE3-tk-luciferasereporter (1 μg) was transiently cotransfected with aRenilla-tk-luciferasereporter (0.5 μg) into SW480 or HCT116 cells. Cells were then treated with carrier control (■) or 600 μM indomethacin (▪) for 24 h. b, PPRE3-tk-luciferasereporter (1 μg) was transiently cotransfected with aRenilla-tk-luciferasereporter (0.5 μg) into HCT116 cells, which were then treated with 30 μM troglitazone (trog) or indomethacin (10–600 μM) for 24 h. In both a and b, cells were lysed and dual luciferase assays were performed 24 h after drug treatment. Data from triplicate experiments are expressed as the mean + S.E.M. of the ratio of PPRE reporter Firefly luciferase activity to controlRenilla luciferase activity. ∗, p< 0.05; ∗∗, p < 0.01 compared with control values (a, Student's t test; b, ANOVA).

Antiproliferative Activity of Indomethacin Is Not Dependent on PPARγ Activation.

We then used mutant dominant-negative human PPARγ (PPARγΔAF-2) to antagonize endogenous PPARγ activity and test the functional relationship between direct PPARγ activation and the antiproliferative effects of indomethacin. First, we confirmed that PPARγΔAF-2 had dominant-negative activity in HCT116 cells. PPARγΔAF-2 expression significantly decreased basal and TZD-inducedtrans-activation of the PPRE-luciferase reporter gene in HCT116 cells by approximately 80% (Fig.3). We tested the antiproliferative activity of 600 μM indomethacin on HCT116 cells, because this concentration of indomethacin induces maximal growth arrest and apoptosis of HCT116 cells at 24 h (the most relevant time point for the transiently transfected cells in the current experiments; Smith et al., 2000). Indomethacin treatment induced a 65% decrease in control HCT116 cell number at 24 h (Fig.4). The presence of the transfection reagent alone did not alter HCT116 cell proliferation (Fig. 4). Expression of PPARγΔAF-2 in HCT116 cells was confirmed by Western blot analysis of the FLAG epitope (Fig. 4). Dominant-negative PPARγ expression was associated with an increase in HCT116 cell proliferation (Fig. 4), which approached statistical significance (p= 0.11). However, the antiproliferative effect of indomethacin was not altered in the presence of dominant-negative PPARγ (Fig. 4). Indomethacin treatment has been associated with down-regulation of the cell cycle gene cyclin D1 in human CRC cells (Hawcroft et al., 2002), and PPARγ activation has been demonstrated to represscyclin D1 expression in HeLa cells (Wang et al., 2001). Therefore, we also investigated whether decreased cyclin D1expression associated with indomethacin treatment was dependent on endogenous PPARγ activation. In keeping with the lack of effect of dominant-negative PPARγ on the antiproliferative effects of indomethacin, PPARγΔAF-2 did not abrogate the decrease in cyclin D1 protein levels associated with indomethacin treatment (Fig. 4).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Dominant-negative PPARγ inhibits endogenous PPARγ activity in HCT116 human CRC cells. APPRE3-tk-luciferasereporter (1 μg) and aRenilla-tk-luciferasereporter (0.5 μg) were transiently cotransfected in the absence (−) or presence (+) of PPARγΔAF-2 (4 μg) into HCT116 cells. Cells were treated with carrier control (■) or 30 μM troglitazone (▪) for 24 h. Cells were then lysed and dual luciferase assays performed. Data from triplicate experiments are expressed as the mean + S.E.M. of the ratio of PPRE reporterFirefly luciferase activity to control Renilla luciferase activity. ∗, p < 0.01 compared with respective cells treated with carrier control or troglitazone, in the absence of PPARγΔAF-2(Student's t test).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Indomethacin decreases HCT116 human CRC cell proliferation and cyclin D1 protein levels despite inhibition of endogenous PPARγ activation by dominant-negative PPARγ. HCT116 cells were transiently transfected with (4 μg) dominant-negative FLAG epitope-tagged PPARγ (PPARγΔAF-2) or mock-transfected and then treated with carrier control (■) or 600 μM indomethacin (▪) for 24 h. The viable cell number was counted in triplicate wells, and the data are expressed as the mean + S.E.M. cell number. Western blot analysis was performed for cyclin D1 (34 kDa), FLAG-PPARγΔAF-2 (54 kDa), and β-actin (42 kDa). ∗,p = 0.11 compared with the number of carrier control-treated cells in the presence of transfection reagent (Student's t test).

Discussion

We have demonstrated that indomethacin can activate PPARγ in human CRC cells but that this pharmacological mode of action does not contribute to the antiproliferative activity of indomethacin on human CRC cells in vitro.

Previous data showing direct binding and transcriptional activation of PPARγ by indomethacin has been obtained from experiments on cells (CV-1 and COS-1 cells transfected with human PPARγ, as well as human rheumatoid synoviocytes) with little relevance to CRC (Lehmann et al., 1997; Jaradat et al., 2001; Adamson et al., 2002; Yamazaki et al., 2002). This study has provided definitive evidence that indomethacin also has similar activity in human CRC cells in vitro, at similar (>100 μM) concentrations to those previously reported to activate PPARγ (Lehmann et al., 1997; Jaradat et al., 2001; Adamson et al., 2002; Yamazaki et al., 2002). Importantly, negative regulation of PPARγ activity at lower concentrations of indomethacin (10 μM), compatible with significant COX inhibition in human CRC cells (Kokoska et al., 1999) and other cultured cell types (Kirtikara et al., 2001), but minimal direct PPARγ activation (Lehmann et al., 1997; Jaradat et al., 2001; Yamazaki et al., 2002), did not occur in HCT116 cells. The possibility that lower concentrations of indomethacin and other NSAIDs may decrease cyclopentenone PG (e.g., 15-d-PGJ₂) synthesis, and hence abrogate PPARγ activity, requires further investigation in other human CRC cell lines with different COX-1 and COX-2 expression profiles.

Both troglitazone and indomethacin produced more potent PPARγ activation in HCT116 cells than SW480 cells, despite similar levels of PPARγ mRNA and protein in these two human CRC cell lines. This suggests a human CRC cell line-specific difference in PPARγ function, which has previously been described in HCT15 human CRC cells (Brockman et al., 1998). Differential PPARγ activity may be related to the phosphorylation status of PPARγ (Adams et al., 1997) or heterozygous “loss-of-function” mutation of PPARγ (Sarraf et al., 1999).

There was a biphasic response of the PPRE-luciferasereporter gene to indomethacin with maximal PPARγ activation at a concentration of 300 μM. A similar biphasic PPARγ activation pattern, with diminished PPRE-reporter gene activity at NSAID concentrations above 300 μM, has previously been described in transfected CV-1 cells treated with indomethacin or ibuprofen (Jaradat et al., 2001). At present, the explanation for this phenomenon is unclear.

It remains controversial what NSAID concentrations are achievable in colorectal mucosa and whether in vitro studies using high NSAID concentrations are relevant to NSAID chemoprevention in vivo (Marx, 2001). Plasma indomethacin levels between 1 and 10 μM are obtained after acute dosing (200 mg) in humans (Hucker et al., 1966). However, indomethacin may undergo enterohepatic circulation, which could lead to intestine luminal drug levels significantly higher than plasma values (Hucker et al., 1966; Kwan et al., 1976). For example, the sulfone metabolite of the related NSAID sulindac can attain colorectal mucosal levels of 50 to 100 μM in humans (R. Pamukcu, personal communication). Therefore, indomethacin concentrations capable of significant, direct PPARγ activation could be generated in human colonic mucosa.

We used a dominant-negative approach to test the relevance of PPARγ activation to the antineoplastic effects of indomethacin on human CRC cells in vitro. Mutant dominant-negative PPARγΔAF-2 has potent inhibitory activity in human CRC cells and is a potent tool for studying PPARγ-mediated gene trans-activation in a host of relevant human cell types. Although dominant-negative PPARγ did not affect the antiproliferative activity of indomethacin, the basal proliferation rate of HCT116 cells was increased in cells expressing dominant-negative PPARγΔAF-2. This implies that PPARγ signaling may down-regulate the proliferation rate of HCT116 cells and is in keeping with data that demonstrate that TZD-induced PPARγ activation leads to growth arrest and apoptosis of human CRC cells (Brockman et al., 1998; Shimada et al., 2002). These data are relevant to the continuing controversy about the role of PPARγ during colorectal carcinogenesis and its potential as a chemoprevention/chemotherapy target (Gupta and DuBois, 2002). Girnun and colleagues have recently described a complementary approach to our dominant-negative PPARγ strategy in which they tested the effect of Pparγhaplo-insufficiency (homozygous deletion of Pparγ is embryonic lethal) on colorectal carcinogenesis in mice (Girnun et al., 2002). Ppar^+/− mice had an increased incidence of azoxymethane-induced colonic tumors compared with Ppar^+/+ littermates. Therefore, our data on human CRC cell proliferation are consistent with the tumor suppressor activity of PPARγ described in this in vivo model.

Several other effects of indomethacin on human CRC cells have been described previously (Tegeder et al., 2001). These include not only COX inhibition but also alterations in WNT signaling (Dihlmann et al., 2001; Hawcroft et al., 2002), BAX-mediated apoptosis (Zhang et al., 2000), increased p38 mitogen-activated protein kinase signaling (Kim et al., 2001), induction of NSAID-associated gene-1 expression (Baek et al., 2002; Kim et al., 2002), decreased ornithine decarboxylase activity (Turchanowa et al., 2001), and induction of nerve growth factor-induced gene B (Kang et al., 2000). In only some instances, the functional relevance of these findings to the antineoplastic activity of indomethacin in vitro and in vivo has been proven (Zhang et al., 2000; Kim et al., 2002). Our data strongly suggest that PPARγ activation is not necessary for the antiproliferative action of indomethacin on human CRC cells in vitro. It will now be important to compare the chemopreventive activity of indomethacin in azoxymethane-treatedPpar^+/− andPpar^+/+ mice (Girnun et al., 2002) to confirm our findings using an in vivo model.

In summary, we have provided evidence that PPARγ activation does not underlie the antiproliferative activity of indomethacin on human CRC cells in vitro. We have also confirmed previous data, based on PPARγ activation experiments, that PPARγ has tumor suppressor activity in human CRC cells, using a novel dominant-negative PPARγ strategy.