Inflammatory bowel disease (IBD), such as ulcerative colitis (UC) and Crohn’s disease, is associated with chronic relapsing inflammation of the intestinal tract of unknown aetiology.^1–3 Evidence suggests that IBD is triggered by a disturbance of the barrier to luminal antigens and pathogens in colonic epithelium and an aberrant immune response to enteric flora, leading to intestinal inflammation. The role of peroxisome proliferator activated receptor γ (PPARγ) in the aetiology and treatment of IBD has been of great interest as ligands for this receptor have been in use for the treatment of type II diabetes and these ligands can attenuate inflammation under certain experimental conditions.

PPARγ is a member of the nuclear receptor superfamily of transcription factors, most of which are ligand dependent transcriptional activators. PPARγ is abundantly expressed in adipose tissue and colonic epithelium. Expression has also been observed in muscle, macrophage, and in T and B cells of humans and rodents.^4,5 Ligands for PPARγ include natural compounds with relatively low affinity such as polyunsaturated fatty acids, oxidised low density lipoprotein, certain eicosanoids, and 15deoxy-Δ12,14-PGJ₂, and drugs including the thiazolidinedione derivatives troglitazone, rosiglitazone, and pioglitazone used for the treatment of type II diabetes.^6–8 PPARγ is the key transcription factor controlling adipogenesis associated with lipid storage, and ligand activation of PPARγ is used to improve insulin sensitivity in type II diabetics.^9–12 Genetic studies in mouse models have revealed a role for this receptor in the control of blood pressure¹³ and renal fluid retention^14,15 in the kidney. PPARγ was found to have tumour suppressor effects in the colon and other tissues^16–21 although controversy still exists regarding its role in the colon in mouse models of carcinogenesis.²² In humans, an allelic variant of PPARγ that exhibits increased sensitivity to ligand activation and is associated with a reduced risk of type II diabetes²³ appears to protect against colon adenomas.^24,25 The transcriptional coactivator Hic-5 was found to promote PPARγ associated differentiation of colon epithelia thus suggesting a key role for PPARγ in gut homeostasis.²⁶

Recently, thiazolidinediones were suggested to be of clinical benefit in the treatment of IBD. Indeed, 5-aminosalicylic acid, a drug used in the treatment of IBD patients, may exert its effects through PPARγ.²⁷ Studies performed in vivo have shown that PPARγ ligands suppress the inflammatory response by attenuating the production of chemokines and cytokines secreted from macrophage,²⁸ T and B lymphocytes,²⁹ and epithelial cells.³⁰ PPARγ ligands also decrease the severity of colitis activity induced in mouse models.^31–39 Furthermore, the observation that PPARγ^+/− heterozygous mice exhibit an increased susceptibility to experimentally induced colitis suggests that PPARγ may have an important role in homeostasis within the gastrointestinal tract that affects the severity of colitis.³³ Other studies revealed that colonic epithelial cells from UC patients displayed dramatically impaired expression of PPARγ, in particular reduced expression and dependency on intestinal flora,³⁴ indicating that PPARγ expressed in gut epithelium may have a protective effect against colon inflammation in humans. However, the colon contains several cell types that express PPARγ and the thus the role of epithelial cell PPARγ in IBD and other diseases of the colonic remains unknown.

To assess the role of PPARγ expressed in intestinal epithelium in the protection against experimentally induced IBD, conditional null mice with intestinal epithelial cell specific disruption of PPARγ, designated PPARγ^ΔIEpC, were created, using a floxed PPARγ allele⁴⁰ and Cre recombinase under the control of the villin gene promoter.⁴¹ PPARγ^ΔIEpC mice displayed a significantly enhanced susceptibility to dextran sodium sulphate (DSS) induced colitis compared with their wild-type littermates, PPARγ^F/F mice. mRNA levels of interleukin (IL)-6, IL-1β, and tumour necrosis factor α (TNF-α) in the colon in PPARγ^ΔIEpC mice were also higher than in PPARγ^F/F mice. These studies reveal that PPARγ in the colonic epithelium has a major protective effect against DSS induced colitis.

METHODS

Intestine specific PPARγ^ΔIEpC

PPARγ^F/F mice, produced as described previously,⁴⁰ were crossed with mice harbouring the Cre DNA recombinase under the control of the villin promoter (villin-Cre mice).⁴¹ Mice homozygous for the PPARγ floxed allele and hemizygous for the villin-Cre transgene (designated PPARγ^ΔIEpC) and littermate control mice (designated PPARγ^F/F) were generated. Subsequently, PPARγ^ΔIEpC and PPARγ^F/F were interbred for at least six generations in order to produce littermates of the same mixed genetic background. Mice were reared on a 12 hour light/dark cycle and fed water and a pellet chow diet (NIH-07) ad libitum. All animal studies were carried out in accordance with Institute of Laboratory Animal Resources guidelines and approved by the National Cancer Institute Animal Care and Use Committee.

Quantitative real time polymerase chain reaction (PCR)

Total RNA was prepared from colon epithelium using Trizol reagent (Life Technology Inc., Rockville, Maryland, USA) following the manufacturer’s instructions. cDNA was generated using 2.5 μg total RNA with the MultiScribe Reverse Transcriptase kit (Applied Biosystems, Foster City, California, USA). Primers were designed for real time PCR using Primer Express software (Applied Biosystems). Sequence and Genbank accession numbers for the forward and reverse primers used to quantify mRNAs were: FABP (NM_017399) forward 5′- CCA TGA ACT TCT CCG GCA AGT-3′ and reverse 5′-TCC TTC CCT TTC TGG ATG AGG T-3′; ADRP (NM_007408) forward 5′-CAC AAA TTG CGG TTG CCA AT-3′ and reverse 5′-ACT GGC AAC AAT CTC GGA CGT-3′; KLF4 (NM_010637) forward 5′-CCA GAC CAG ATG CAG TCA CAA-3′ and reverse 5′-ACG ACC TTC TTC CCC TCT TTG-3′; keratin 20 (NM_023256) forward 5′-ATG AAG TCC TGG CCC AGA AGA-3′ and reverse 5′-TTT CAT TTC AGC TCC TCC GTG TTC ACT-3′; PPARβ/δ (NM_011145) forward 5′- TTG AGC CCA AGT TCG AGT TTG CTG-3′ and reverse 5′- ATT CTA GAG CCC GCA GAA TGG TGT-3′; and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (BC_095932) forward 5′ -CAT GGC CTT CCG TGT TCC TA-3′ and reverse 5′-GCG GCA CGT CAG ATC CA-3′. Real time PCR reactions were carried out using SYBR green PCR master mix (Finnzymes, Espoo, Finland) in the PTC-200 DNA Engine Cycler and detected using the CFD-3200 Opticon Detector (MJ Research, Waltham, Massachusetts, USA). The following conditions were used for PCR: 95°C for 15 seconds, 94°C for 10 seconds, 60°C for 30 seconds, and 72°C for 30 seconds, and repeated for 45 cycles. PCR included a no template control reaction to control for contamination and/or genomic amplification. All reactions had >90% efficiency. Relative expression levels of mRNA were normalised to GAPDH and analysed for statistical significance using the Student’s t test (Prism 4.0).

Induction of colitis with DSS and rosiglitazone treatment

Female mice, 8–10 weeks old, PPARγ^ΔIEpC mice or PPARγ^F/F mice, were administered 2.5% (wt/vol) DSS (molecular weight 35 000–44 000; ICN Biomedicals, Aurora, Ohio, USA) in the drinking water for seven days. For rosiglitazone (SmithKline Beecham Pharmaceuticals, West Sussex, UK) treatment, a powdered diet (AIN-93G; Dyets Inc., Bethlehem, Pennsylvania, USA) blended with the drug was administered for two days prior to DSS treatment to the end of DSS treatment to yield a dose of drug of approximately 20 mg/kg/day.

Assessment of colitis

After DSS treatment was started, daily changes in body weight and clinical signs of colitis, such as rectal bleeding, diarrhoea, and piloerection, were examined. Disease activity index consisted of scoring for rectal bleeding (0–4) and diarrhoea (0–3) as previously reported.⁴² Hemoccult SENSA (Beckman Coulter, Inc., Fullerton, California, USA) was used for examination of rectal bleeding. Macroscopic colonic damage was evaluated as previously described.⁴³

Southern blot analysis

Southern blot analysis to assess the floxed and recombined PPARγ alleles was performed as previously described.⁴⁰ Genomic DNA isolated from epithelial cells from the colon, caecum, and small intestine, and total lung and kidney, were digested with BamHI, subjected to electrophoresis on a 0.5% agarose gel, and transferred to a GeneScreen Plus Hybridisation Transfer membrane (NEN Life Sciences, Boston, Massachusetts, USA). A 0.6 kb DNA fragment (SalI-EcoRI) derived from intron 2 of the PPARγ gene (3′-probe) was used as a probe after labelling with ³²P by random priming in the presence of α³²P-dATP. Blots were exposed to a phosphorimager screen cassette followed by visualisation using a Molecular Dynamics Storm 860 Phosphorimager system (Sunnyvale, California, USA).

RNA analysis

RNA was extracted from total colon after DSS treatment using TRIzol reagent (Life Technology Inc.). Northern blot analysis was performed as previously described using standard protocols.⁴⁰ cDNA probes for PPARγ (exon 2) and ribosomal protein 36B4 were amplified from a mouse cDNA library by PCR using gene specific primers and cloned into a pGEM-T Easy Vector (Promega Corp., Madison, Wisconsin, USA). The sequences were confirmed using a CEQ 2000XL DNA Analysis System (Beckman Coulter, Fullerton, California, USA). For ribonuclease protection assays (RPA), the Multiprobe RNase protection assay was performed according to the manufacturer’s (Pharmingen, Philadelphia, Pennsylvania, USA) directions with modifications as previously described.⁴⁴ RNA was fractionated by electrophoresis through a formaldehyde-agarose gel and transferred to GeneScreen Plus membranes (DuPont, Wilmington, Delaware, USA) and the blots were hybridised at 42°C in Ultrahyb (Ambion, Austin, Texas, USA) with random primed labelled cDNA probes. Northern blots and RPA gels were visualised using a Molecular Dynamic Storm 860 Phosphorimager system and the signals quantified by ImageQuant software package (Molecular Dynamics, Sunnyvale, California, USA).

Western blot analysis

Epithelial cells were isolated from colon and cell extracts prepared by sonication. Monkey kidney CV-1 cells transfected with a mouse PPARγ1 expression vector⁴⁵ were used as a positive control. Nuclear extracts were prepared as described previously⁴⁵ and protein concentrations were determined using the BCA Kit (Pierce Biotechnology Inc, Rockford, Illinois, USA). Nuclear extract proteins (10 μg) from each sample were separated by sodium dodecyl sulphate-polyacrylamide gel electrophoresis through 10% polyacrylamide gels and transferring onto nitrocellulose membranes (Schleicher and Schuell, Keene, New Hampshire, USA). Staining was carried out using mouse anti-PPARγ monoclonal antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, California, USA) and peroxidase conjugated goat antimouse IgG horseradish peroxidase secondary antibody (Jackson ImmmunoResearch Laboratories, West Grove, Pennsylvania, USA). Staining for mouse GAPDH was carried out using a goat antimouse GAPDH (Chemicon, Temecula, California, USA). Blots were developed with an enhanced chemiluminescence detection kit (Pierce) and visualised using autoradiographic film.

Histological analysis

For histological analysis, tissue samples were fixed in 10% neutral buffered formalin overnight at 4°C and paraffin embedded. Thereafter, 4 μm sections were cut, deparaffinised with xylene, and stained with haematoxylin-eosin. Alcian blue staining was carried out using the MicroProbe System Alcian blue Stain Kit (Fisher Scientific Company, Virginia, USA). Nuclei were counterstained by Nuclear Fast Red (Kernechtrot).

In situ hybridisation

In situ hybridisation was carried out using digoxigenin (DIG) labelled cRNA probes and DIG RNA Labelling Mix (Roche Applied Science, Indianapolis, Indiana, USA) according to the manufacturer’s instruction with a slight modification. In brief, 5 μM frozen sections were incubated in 2 μg/ml proteinase K for 30 minutes at 37°C and then 4% paraformaldehyde for 15 minutes to stop the reaction, followed by incubation in 0.2 M HCl for 10 minutes. Sections were washed twice in PBS containing 0.1% active diethyl pyrocarbonate (Sigma-Aldrich Co., St Louis, Missouri, USA) for 15 minutes and equilibrated in 5×SSC for 15 minutes. Sections were then subjected to prehybridisation with hybridisation buffer (5×SSC, 50% formamide, 40 μg/ml salmon sperm DNA) for two hours at 50°C, followed by hybridisation with a PPARγ (exon 2) probe overnight at 50°C. A PPARγ specific probe was generated from mouse PPARγ exon 2 cDNA inserted into pGEM (Promega Corp). The antisense and sense probes were synthesised by T7 and SP6 polymerases, respectively, after linearisation of the vector with HindIII and EcoRI, respectively. Sections were serially washed in 2×SSC, 1×SSC, 0.5×SSC, and 0.1×SSC for one hour each at room temperature. They were then equilibrated in buffer I (100 mM Tris, 150 mM NaCl, pH 7.5) for five minutes and treated with 1:1000 dilution of anti-DIG antibody (Roche Applied Science) in buffer I containing 0.5% blocking reagent for two hours at room temperature. Sections were washed twice in buffer I for five minutes and incubated with biotinylated anti-sheep IgG (Vector Laboratories, Burlingame, California, USA) for 30 minutes at room temperature. Sections were then incubated with alkaline phosphatase conjugated avidin and biotin complex for 30 minutes at room temperature and subjected to incubation with buffer II (100 mM Tris, 150 mM NaCl, 50 mM MgCl₂, pH 9.5) for five minutes. Buffer II containing 450 μg/ml of 4-nitroblue tetrazolium chloride and 175 μg/ml of 5-bromo-4-chloro-3-indolyl-phosphate was applied to each section and the reaction was stopped in TE buffer.

Statistical analysis

All values in the figures and text are expressed as mean (SEM). Statistical comparisons were examined with ANOVA followed by Fisher’s protected least significant difference test. Results were considered significantly different at p<0.05.

RESULTS

Generation of intestinal epithelial cell specific PPARγ null mice

To study the functional role of PPARγ in intestinal epithelial cells in IBD, PPARγ-floxed mice (PPARγ^F/F) mice were crossed with villin-Cre transgenic mice to generate mice lacking expression of the receptor in the intestinal epithelia. PPARγ^ΔIEpC mice were born at the expected Mendelian frequencies and exhibited no overt abnormalities compared with PPARγ^F/F littermates. To estimate the extent of intestinal epithelium specific deletion of the PPARγ floxed exon 2, recombination was examined by PCR and Southern blot analyses, as described previously.⁴⁰ By PCR analysis, the null allele from the recombined PPARγ floxed allele, amplified as a 400 bp product, was detected in epithelial cells of the colon, caecum, and small intestine isolated from PPARγ^ΔIEpC mice and was not detected in DNA isolated from the lung, kidney, and brown adipose tissue of these mice, other known sites of significant PPARγ expression (fig 1A).

The intact floxed allele, which was amplified as a 285 bp product, was not detected in epithelial cells of the colon, caecum, or small intestine isolated from PPARγ^ΔIEpC mice. By Southern blot analysis, an 8 kb fragment that represents the PPARγ gene lacking exon 2 was only detected in DNA isolated from epithelial cells of the colon, caecum, and small intestine isolated from PPARγ^ΔIEpC mice; this fragment was not found in DNA isolated from the lung, kidney, and brown adipose tissue (fig 1B). The intact PPARγ exon 2 floxed allele, represented by a 10 kb hybridising band, was not detected in DNA isolated from epithelial cells of the colon, caecum, and small intestine of PPARγ^ΔIEpC mice but was found in DNA from the lung, kidney, and brown adipose tissue. These data suggest that epithelial cells of the colon, caecum, and small intestine of PPARγ^ΔIEpC mice had almost complete recombination of the PPARγ floxed allele, while no recombination was found in the lung, kidney, or brown adipose.

This degree of specificity is in agreement with an earlier report showing, using a villin-beta gal as well as a villin-Cre transgenic mouse crossed with the R26R indicator strain, that expression of the villin gene promoter was restricted to intestinal epithelial cells.⁴¹ As shown by northern blot analysis using a probe for PPARγ exon 2, the colon and caecum of PPARγ^ΔIEpC mice had essentially no expression of PPARγ mRNA whereas BAT of PPARγ^ΔIEpC mice showed a similar degree of expression of PPARγ mRNA to brown adipose tissue of PPARγ^F/F mice (fig 1C). The reason that no hybridising bands were detected in the colon and caecum of PPARγ^ΔIEpC mice is likely due to the fact that cells other than epithelial cells in the colon do not express significant amounts of PPARγ mRNA.

Western blot analysis revealed that PPARγ expression was extinguished in colon epithelial cells from PPARγ^ΔIEpC mice; the receptor protein was readily detected in PPARγ^F/F mice and mobility corresponded with PPARγ1 (fig 1D). GAPDH protein was used as a loading control. The identity of the non-specific band (fig 1D, NS) that reacts with the anti-monoclonal antibody is not known.

To further demonstrate that deletion of PPARγ in the colon resulted in functional changes in PPARγ target gene expression, quantitative real time PCR was performed using RNA isolated from colon epithelium. Indeed, deletion of PPARγ mediated by villin-Cre resulted in a significant decrease in the constitutive expression of mRNA encoding fatty acid binding protein (FABP), a PPARγ target gene (fig 1E). Constitutive expression of mRNA encoding adipocyte related protein (ADRP) had a trend towards lower expression without statistical significance (p = 0.06) in PPARγ^ΔIEpC compared with PPARγ^F/F mice (fig 1D). Expression of mRNA encoding two epithelial cell differentiation markers, keratin 20 and kruppel-like factor 4 (KLF4), was unchanged by deletion of PPARγ in the colon epithelia (fig 1D); expression of PPARβ/δ mRNA was also not different between genotypes.

In situ hybridisation analysis revealed that PPARγ mRNA was not expressed in colon epithelial cells of PPARγ^ΔIEpC mice (fig 2A) compared with clear expression found in the colon of PPARγ^F/F mice (fig 2B). Histological observations with haematoxylin-eosin staining revealed no gross differences in whole colon epithelia, including goblet cells, between PPARγ^ΔIEpC mice and PPARγ^F/F mice in the absence of DSS treatment. Occasionally, colon epithelia from PPARγ^ΔIEpC mice exhibited abnormal narrowing with apparent accumulation of mucin in cells towards the lumen (fig 2Ev 2D). In fact, Alcian blue staining uncovered regions of the colon epithelia in PPARγ^ΔIEpC mice that secreted more mucin (fig 2E, arrows) compared with PPARγ^F/F mice (fig 2F). Furthermore, a few small inflammatory lesions were detected in the colon of PPARγ^ΔIEpC mice without DSS treatment (fig 2G, arrow), whereas there were no inflammatory lesions in PPARγ^F/F mice. These inflammatory lesions were distinct from lesions having more mucin staining in the colon of PPARγ^ΔIEpC mice.

Susceptibility of PPARγ^ΔIEpC to DSS induced IBD

PPARγ^ΔIEpC mice showed increased susceptibility to DSS induced colitis in comparison with PPARγ^F/F mice. Administration of 2.5% DSS in drinking water for seven days induced colitis, as revealed by clinical symptoms (loss of body weight, diarrhoea, and bloody faeces) in PPARγ^ΔIEpC and PPARγ^F/F mice (fig 3). PPARγ^ΔIEpC mice showed significant body weight loss in comparison with mice PPARγ^F/F mice at 4, 5, 6, and 7 days from the beginning of DSS treatment (19.1 (1.1)% and 11.2 (1.0)% of initial body weight at day 7 for PPARγ^ΔIEpC and PPARγ^F/F mice, respectively) (fig 3A, B). Colon length of PPARγ^ΔIEpC mice on treatment with DSS for seven days was considerably shortened compared with PPARγ^F/F mice, which indicates that PPARγ^ΔIEpC mice exhibited more severe colitis than PPARγ^F/F mice (fig 3C, 4A). Compared with PPARγ^F/F mice, PPARγ^ΔIEpC mice also showed worsening of clinical symptoms, represented by the relatively poor diarrhoea score and bleeding score (fig 3D, E) and increased severity of the macroscopic observations and histological signs of colon injury (fig 3F, 4B–E). Histological analysis of PPARγ^ΔIEpC mice showed absence of epithelium and intensive submucosal infiltration of inflammatory cells in comparison with PPARγ^F/F mice (fig 4B–E). These findings indicate that, in the absence of an exogenous ligand, PPARγ in colonic epithelium has a protective effect against DSS induced colitis.

Attenuation of DSS induced IBD by PPARγ ligands

Administration of the PPARγ ligand rosiglitazone (20 mg/kg/day), beginning two days before commencing DSS treatment, resulted in a decrease in the severity of symptoms in both PPARγ^F/F and PPARγ^ΔIEpC mice. In PPARγ^F/F mice, rosiglitazone reduced body weight loss (10.8 (1.3)% and 7.8 (0.4)% of initial body weight at day 7, without and with rosiglitazone, respectively) (fig 5A). In PPARγ^ΔIEpC mice, rosiglitazone treatment also reversed the body weight loss (20.5 (1.4)% and 15.2 (1.1)% of initial body weight at day 7, without and with rosiglitazone, respectively) and decreased the severity of other symptoms associated with DSS induced colitis. Colon length, diarrhoea, and bleeding scores as well as histological signs of colon injury were improved by ligand treatment (fig 5B–D). However, after administration of rosiglitazone, PPARγ^ΔIEpC mice still showed more severe disease symptoms than PPARγ^F/F mice treated with rosiglitazone. These data indicate that ligand activated PPARγ attenuates colitis regardless of the presence or absence of PPARγ in the colonic epithelium.

Analysis of cytokine expression in DSS induced IBD and effect of PPARγ ligands

As another indication of the extent of IBD, cytokines were analysed. IL-1β, IL-6, and TNF-α, secreted from colonic epithelium as well as macrophage and T cells, are increased during the course of IBD.^1,46,47 A ribonuclease protection assay for tissue associated cytokines was performed on RNA isolated from whole colon after seven days of DSS treatment. Marked induction of IL-6 (fig 6A, B, F, G), IL-1β (fig 6C, H) and TNF-α (p = 0.065) (fig 6D, I) in the colon of PPARγ^ΔIEpC mice was observed while no differences in expression of IL-15 (fig 6A, E) or macrophage migration inhibitory factor (MIF) (fig 6D, J) mRNAs were noted between PPARγ^ΔIEpC and PPARγ^F/F mice. Thus lack of PPARγ expression in the colonic epithelium resulted in enhanced induction of IL-6, IL-1β, and TNF-α mRNA expression by DSS treatment. Administration of rosiglitazone reduced IL-6, IL-1β, and TNF-α mRNA levels in both PPARγ^ΔIEpC and PPARγ^F/F mice. However, PPARγ^ΔIEpC mice treated with rosiglitazone had higher IL-6, IL-1β, and TNF-α mRNA levels than did similarly treated PPARγ^F/F mice after DSS treatment.

DISCUSSION

Activation of PPARγ has been reported to ameliorate IBD disease activity in rodent DSS, trinitrobenzene sulphonic acid, and ischaemic colitis models.^32,33,35,36 The role of PPARγ in attenuating inflammation has been investigated^34,48 and a clinical trial of rosiglitazone in UC patients has been carried out.⁴⁹ During the progress of inflammation, PPARγ is expressed in epithelial cells as well as macrophages and T and B cells infiltrating colon tissue. It is generally thought that alterations of epithelial cells and activation of macrophages cause the extensive tissue injury associated with experimental colitis, followed by a dysregulated immune response of CD4⁺ T cells. Although PPARγ is expressed in colonic epithelium, the role of this receptor in colonic epithelial cells has remained unclear. Evidence for a role of PPARγ in colitis has emerged from the use of PPARγ ligands in colon cancer cell lines.³⁰ PPARγ ligands reduce nuclear factor κB signalling, leading to a decrease in the production of cytokines in colon cancer cell lines that are derived from the colon epithelium.³⁰ In addition, in the clinical setting, expression of PPARγ in colonic epithelium is impaired in UC patients while expression of PPARγ in inflammatory cells remains normal.⁵⁰

However, the precise contribution of PPARγ in epithelia versus immune cells had not been elucidated prior to this study.^32,35 PPARγ in macrophage may also have a role in the inflammation associated with IBD. Indeed, PPARγ ligands have been shown to reduce expression of cytokines in activated macrophage,²⁸ an effect that appears to be independent of PPARγ expression at high ligand concentrations.⁵¹ Of interest, PPARγ^+/− heterozygous mice exhibit an increased susceptibility to experimental colitis, as revealed by loss of body weight data, indicating that endogenous expression of PPARγ, in the absence of endogenous ligands, may have an effective role in the control of colitis.³³ However, in another study using mice with disruption of the PPARγ gene in epithelial tissues using a mouse mammary tumour virus promoter, there were no apparent differences in the extent of colitis symptoms between the floxed and knockout mice.³⁶ It should be noted that these mice do not have disruption of the PPARγ gene specifically in gut epithelial cells and thus could have functional PPARγ expression in T and B cells.^52,53

Thus the issue of site of action of PPARγ in ameliorating the effect of ligands in IBD remains a point of contention.⁵⁴ In addition, the role of this receptor in protecting against induced colitis is still unclear; is it due to an anti-inflammatory effect of PPARγ in macrophage or due to its expression in the colon epithelia?

To directly examine the role of PPARγ expressed in gut epithelial in the inflammatory colitis model, intestinal epithelium specific PPARγ null mice were generated using the Cre/loxP strategy and the PPARγ floxed allele described in an earlier report.⁴⁰ PPARγ^ΔIEpC mice exhibited almost complete deletion of PPARγ expression in colonic epithelium. The functional significance of this deletion was demonstrated by showing that expression of mRNA encoding FABP, a known PPAR target gene, was significantly lower in PPARγ^ΔIEpC mice compared with PPARγ^F/F mice. While others have shown that heterozygous PPARγ^+/− mice exhibit reduced constitutive expression of mRNAs encoding keratin 20 and KLF4,²⁶ expression of keratin 20 and KLF4 were unchanged in PPARγ^ΔIEpC mice compared with PPARγ^F/F mice. The reason for this difference is uncertain but may be due to the influence of PPARγ in other cell types.

There were some lesions with more mucin in total colon tissue of PPARγ^ΔIEpC mice. The role of PPARγ in mucin accumulation is not known and requires further investigation. In addition, it is not clear from these studies whether mucin influences the response to DSS induced colitis. There is a possibility that hypersecretion of mucin induced the increased activity of proteinase for mucin, which interferes with defence in the colon mucosa. PPARγ^ΔIEpC mice without DSS treatment also had a few small inflammatory lesions in the colon, suggesting that PPARγ^ΔIEpC mice had a weakened or abnormal immune system in colon mucosa possibly against certain types of commensal bacteria. Further analysis is needed.

After receiving DSS treatment for seven days, PPARγ^ΔIEpC mice exhibited more severe colitis, as revealed by loss of body weight, shortened colon length, diarrhoea, and bleeding scores, and macroscopic and histological analysis in comparison with PPARγ^F/F mice with or without rosiglitazone administration. This finding suggests that endogenous PPARγ in colonic epithelial cells might be involved in maintaining the homeostasis of the colon and preventing initiation of colon injury and subsequent development of inflammation. Thus the possibility exists that UC patients with impaired expression of PPARγ have more severe inflammation in the colon. It is especially noteworthy that attenuation of disease activity was noted in both PPARγ^F/F and PPARγ^ΔIEpC mice on rosiglitazone treatment. This suggests that the response to rosiglitazone may be independent of PPARγ, as suggested by earlier studies.⁵¹ This may not be surprising in view of a recent study revealing that troglitazone is able to effectively inhibit the activity of myeloperoxidase through a PPARγ independent mechanism.⁵⁵ This could account for the PPARγ independent effects of rosiglitazone on PPARγ^ΔIEpC mice observed in the current study. While recombination is virtually complete and PPARγ mRNA protein not detectable in epithelial cells, the possibility cannot be excluded that rosiglitazone activated PPARγ expressed in non-colon epithelial cells, such as macrophage, and that attenuation of colitis may be due to PPARγ in infiltrating macrophages that inhibit the production of cytokines²⁸ resulting in a decrease in the severity of colitis.

The finding that PPARγ^ΔIEpC mice exhibited higher levels of IL-6, IL-1β, and TNF-α than PPARγ^F/F mice either without or with rosiglitazone treatment suggests the possibility that PPARγ in epithelial cells normally suppresses the pathway producing these cytokines. This hypothesis is supported by previous data showing that PPARγ ligands inhibit the activation of nuclear factor κB,⁴⁸ and that TLR4-activated PPARγ in epithelium attenuates nuclear factor κB signalling in an IκB dependent fashion.⁵⁰ However, the composition of inflammatory cells between PPARγ^ΔIEpC and PPARγ^F/F mice remains to be determined. Epithelial PPARγ may also modulate mucosal barrier function and alter the control of apoptosis of epithelial cells resulting in the accumulation and activation of immune cells and upregulation of proinflammatory cytokines.

In summary, disruption of PPARγ in colonic epithelial cells resulted in increased susceptibility to DSS induced colitis associated with an elevation in IL-6, IL-1β, and TNF-α levels. These data suggest that endogenous PPARγ in colonic epithelium has a protective effect against IBD. Further studies are warranted to determine the mechanism of this protective effect.