Troglitazone Induces Extracellular Matrix and Cytoskeleton Remodeling in Mouse Collecting Duct Cells

Peroxisome proliferator-activated receptor (PPARγ) has been shown to have a protective role in the nephron through its ability to inhibit a transforming growth factor- (TGF-β) mediated fibrotic response. In contrast, PPARγ was also shown to induce a mesenchymal transformation in epithelial intestinal cells. A fibrotic response in the collecting duct has only recently been established; however, the entire collecting duct has not been fully examined. Inner medullary collecting duct cells (IMCD-K2) and mouse cortical collecting duct cells (M1), representing the cortical and medullary collecting duct, were exposed to 5–10 μM troglitazone for 24 hours. Troglitazone resulted in an elongated morphology, 60% decreases in E-cadherin and β-catenin, a 35% decrease in α-catenin, and a 1.5-fold increase in fibronectin. These effects were not reversed with PPARγ antagonists or affected with PPARγ overexpression. Our results indicate that troglitazone induced a mesenchymal-like transformation in M1 and IMCD-K2 epithelial cells independently of PPARγ.

In the kidney, PPARγ ligands have been shown to have a protective effect in various segments of the nephron including the glomerulus (podocytes and mesangial cells), proximal tubule, and distal tubule. Although PPARγ expression is among the highest in the collecting duct (CD), its role in the CD remains unclear. PPARγ activation by TZDs has been shown to result in oedema through enhanced epithelial sodium cotransporter (ENaC) activity in the CD [6]. These effects were blocked by amiloride, a CD-specific diuretic. The deletion of PPARγ from the collecting duct, using PPARγ flox/flox mice, was also successful in reversing the fluid retention. The full impact of PPARγ ligands in the collecting duct requires further characterization as the potential renal protective and antidiabetic effects are promising. Whether PPARγ is an antifibrotic transcription factor is unclear as there have been studies indicating an induction of fibrotic responses by PPARγ ligands. An example of this was found in the intestine where activation of PPARγ was shown to induce a mesenchymal transition in epithelial intestinal cell line [7].
Epithelial cells form organized cell-cell adhesion junctions preventing movement of individual cells and producing a tight epithelial surface. Mesenchymal cells, however, tend to be highly motile as they lack this level of organization. The transformation of EMT is described as the process in which epithelial cells that function as ion and fluid transporters lose their epithelial polarity, cellular adhesion molecules, and reorganize their actin cytoskeleton from a cortical bundle formation that supported adhesion molecules into stress fibers containing de novo expressed α-smooth muscle actin (α-SMA) that supports migration.
EMT has been described in the collecting duct where TGF-β induced activation of Smad3 and ERK1/2 leading to a mesenchymal morphology, increased vimentin and α-SMA, and a decrease in E-cadherin and β-catenin [8]. Expression of E-cadherin has been shown to vary inversely with fibronectin in several diseased states including a unilateral ureteral obstruction (UTO) model, a model which invokes scarring in the collecting duct [9]. This study, however, does not account for a full representation of the collecting duct as it focused on the terminal section.
E-cadherin is an integral, Ca 2+ -dependent transmembrane adhesion protein that is generally localized at the adheren junctions of epithelial cells and plays an important role in the development and maintenance of renal epithelial polarity. The intracellular domain of E-cadherin is bound to β-catenin that is bound to α-catenin, which in turn links the entire complex to the actin cytoskeleton. Unbound βcatenin can also function as a component of the Wnt nuclear signaling pathway acting as a coactivator of T-cell factor (TCF)/lymphocyte enhancer factor (LEF) transcription factors [9,10].
PPARγ has been shown to interact with members of the cadherin-catenin cytoskeletal complex through the functional peroxisome proliferator response element (PPRE) described in E-cadherin [11], as well as its catenin-binding domain (CDB) which interacts directly with the TEC/LEF domain in β-catenin [10]. Studies have shown that PPARγ activation induces proteasomal degradation of β-catenin as well as a reduction in E-cadherin in several cell types, including hepatocytes [12]. In normal cells, PPARγ can function to suppress Wnt signaling by targeting phosphorylated βcatenin to the proteasome through a process involving the above-mentioned catenin-binding domain [1].
The overall purpose of this study is to clarify the mechanisms by which TRO/PPARγ alters collecting duct structure and function. We report that TGF-β was unable to initiate an epithelial to mesenchymal transformation in the IMCD-K2 and M1 collecting duct cell lines. In contrast, Troglitazone caused morphological changes, decreased E-cadherin, αcatenin and β-catenin, and increased fibronectin. These effects were not reversed with PPARγ antagonists or altered by PPARγ overexpression.

Chemicals and Reagents.
Troglitazone (Sigma) is a synthetic PPARγ agonist and was used at 5 μM and 10 μM. Troglitazone is a well-studied synthetic PPARγ agonist. It has high specificity for PPARγ as it does not activate the PPARδ or PPARα isoforms at a concentration of 25 mM and higher. Troglitazone is a member of a class of antidiabetic drugs that are currently used to improve insulin sensitivity in type 2 diabetics. GW9662 (Cayman Chemicals) is a PPARγ antagonist that covalently modifies Cys 285 in PPARγ's LBD. PPARγ antagonist (T0070907) (Cayman Chemical) covalently modifies Cys 313 thereby modulating PPARγ cofactor recruitment. Both antagonists were used alongside Troglitazone at 10 μM. PGJ 2 (Cayman Chemical) is an endogenous PPARγ agonist. All reagents were dissolved in DMSO to their suggested stock concentrations. Antibodies for α-SMA, α-catenin, β-actin, β-catenin, fibronectin, and Ecadherin were obtained from Sigma-Aldrich (ON, Canada). Anti-PPARγ was obtained from Santa Cruz Biotech (CA, USA). AlexaFluor Phalloidin and DAPI were gifts from Dr. Chris Kennedy (University of Ottawa).

Western Blot.
Following stimulations, protein isolations were performed using RIPA lysis buffer consisting of: RIPA buffer, 10 mM Na fluoride, 1 mM Na pyrophosphate, 100 μM Na orthovanadate, 1 : 100 protease inhibitor cocktail (Sigma), and 500 μM PMSF. The solution was centrifuged at 10,000 ×g for 10 minutes, and protein was quantified by spectrophotometry using Bradford reagent and denatured by boiling the samples at 95 • C for 5-10 minutes. Equal amounts of lysates (50-100 μg) were electrophoresed and transferred onto a nitrocellulose membrane (Amersham). The nitrocellulose membrane was blocked in 10% milk in TBS/T for 1 to 3 hours, and a primary antibody was applied directly to the milk and left overnight. Following three 20minute washes, the primary antibody solution was removed, and the appropriate secondary antibody was applied in 10% milk-TBS/T for 90 minutes, after which the membranes were washed with TBS/T. The HRP-conjugated secondary antibodies were developed with Pierce SuperSignal chemiluminescent reagent for 5 minutes and exposed to Kodak X-Omat Blue XB-1 film as well as digitally exposed using the Alpha Innotech Fluorchem HD2 imaging system.

2.4.
Immunofluorescence. Cells were plated on coverslips and allowed to adhere overnight. The culture media was replaced for 24 hours with serum-free DMEM : F-12 and stimulated with TRO and GW9662 for an additional 24 hrs. The coverslips were washed with PBS to remove traces of media and fixed for 30 minutes at room temperature in 4% paraformaldehyde or at −20 • C in a (1 : 1) methanol-acetone solution. Excess fixative was washed with PBS/Mg 2+ /Ca 2+ for 2 × 5 minutes. Triton-X at 0.1% in PBS was used for permeabilization for 30 minutes at room temperature followed by 3 × 5-minute washes in PBS/Mg 2+ /Ca 2+ . The coverslips were blocked in 5% BSA in 0.1% Triton-X in PBS/Mg 2+ /Ca 2+ for 1 hour prior to adding the antibody of choice. All antibodies were diluted in 5% BSA in 0.1% Triton-X in PBS/Mg 2+ /Ca 2+ . Cells were incubated with Phalloidin-FITC for 1 hour at RT in the dark. Next, 3 × 5minute PBS/Mg 2+ /Ca 2+ washes were performed prior to the addition of the secondary, α-mouse FITC (Sigma) 1 : 300 for 90 minutes or DAPI (1 : 1000) for 15 minutes. The coverslips were then mounted in Fluoromount G-mounting media (Southern Biotech). Images were captured using the Zeiss AxioCam of an Axioskop2 MOT Fluorescent microscope.

PPARγ
Overexpression. pcDNA and pcDNA-PPARγ vectors were obtained from Addgene as bacterial stabs. The sizes of each insert were verified with XhoI/HindIII digests. M1 cells were grown to 25% confluence prior to transfection. 4 μg of plasmid DNA was diluted to a volume of 150 μL in DMEM and 15 μL of Polyfect Transfection Reagent (Qiagen) was added to the solution. The suspension was then incubated at room temperature for 10 minutes, mixed with 1 mL of culture media containing 0.1% ampicillin, and added to the culture plate. This was repeated 48 hours later for the final 24 hours of the 72 hours transfection period.
2.6. Statistics. GraphPad Prism v4.03 was used to plot and analyse the data collected. Values are expressed as means ± standard error of the mean (S.E.M.). An unpaired t-test was used to assess statistical significance between selected experimental groups. A one-sample t-test was performed against a hypothetical value of 1.0 for all groups as the values represented fold controls where the control group all had values of 1.0. Statistical analysis for multiple comparisons was done using one-way ANOVA corrected with Bonferroni's posttest. A P value ≤ 0.05 was regarded as statistically significant.

TRO Induces Morphological Changes in M1 Cells.
The presence of PPARγ in both the M1 and IMCD-K2 cell lines was verified through Western blot and RT-PCR ( Figure 1) with PPARγ being successfully overexpressed in M1 cells as seen in Figure 1(a). The general effects of TZDs in the kidney have been protective; however, the treatment of M1 cells with TRO resulted in a decrease in cell-to-cell contact, disorganization of the F-actin cytoskeleton filament network, and a less rounded, elongated spindle-like shape at 5 μM and 10 μM, visualized through FITC-conjugated Phalloidin ( Figure 2). These morphological changes were not attenuated by the PPARγ antagonist GW9662 at 10 μM with the same order of disorganization and cell elongation being observed after cotreatment in Figures 2(d) and 2(f). in several epithelial adhesion markers. Figure 3 shows the analysis of α-smooth muscle actin (α-SMA) in M1 cells and vascular smooth muscle cells (VSMC, positive control  Figure 5 with a 60% reduction in β-catenin levels after 24-hour TRO treatment, and a 30% reduction in α-catenin ( Figure 6). Conversely, an upregulation of fibronectin was obtained in response to 24-hour TRO (Figures 7(b) and 7(c)), near 2-fold in M1 cells, and 1.5-fold in IMCD-K2 cells. In comparison, Figure 8 illustrates the response of M1 cells to PGJ 2 . As shown, 24-hour treatment with 0.5 μM PGJ 2 increased E-cadherin and fibronectin levels to 1.5-2-fold of control. A numerical increase in β-catenin was also observed.

TRO-Mediated Effects in the M1 Collecting Duct Cell Line
Are PPARγ Independent. TZDs are known to induce several PPARγ-independent effects. In our study PPARγ antagonists GW9662 and T0070907 had no effect on E-cadherin (Figure 4(d)), β-catenin ( Figure 5(c)), or fibronectin (data not shown). However, as shown in Figure 6, T0070907 resulted in a further 2-fold reduction in α-catenin levels, but this was not statistically significant. The overexpression of PPARγ in M1 cells was performed to see whether an enhancement of the abovementioned effects could be observed. However, PPARγ overexpression did not significantly change the 50% decrease in E-cadherin in nontransfected M1 cells compared to cells transfected with the pcDNA plasmid control and the pcDNA-PPARγ vector insert (Figure 4(e)). As seen in Figure 7(d) the 1.5-fold increase in fibronectin in nontransfected and plasmid control transfected cells was not increased with PPARγ overexpression in pcDNA-PPARγ transfected cells, instead yielding a modest 1.3-fold increase in fibronectin, lower than in nontransfected cells. Taken together these results portray a system in which PPARγ does not mediate TRO responses in collecting duct cells. It should also be noted that the transformation of the epithelial CD cell lines via a cytotoxic response is unlikely. An inhibition of growth along with p38 activation was described in a previous study; however, no oxidative stress (ROS generation) or cell death (cleaved caspase-3 expression) were observed [13]. The reduction in cell number seen with TRO compared to control can be primarily attributed to the aforementioned decrease in proliferation. TRO is also known to drop pH levels [14]; however, we did not observe any significant levels of acidosis in our cell culture media in a 48 hr period.

PPARγ in the Collecting Duct.
PPARγ has previously been detected in vivo in the medullary and cortical collecting duct [13,15] and was present in both IMCD-K2 and  M1 cells in our study. To our knowledge the only other studies on PPARγ and the collecting duct indicate that it may be involved in sodium and fluid balance, as seen with the PPARγ knockout model reversing the oedemic effects resulting from TZD treatment [6,16]. The use of PPARγ agonists, specifically TZDs, on the other hand, has raised interest in several fields of research due to their potential antifibrotic and antitumorigenic effects. Troglitazone has been shown to suppress TGF-β-induced SMAD2/3 phosphorylation, thereby inhibiting EMT in human retinal pigment epithelial cells independent of PPARγ [17]. It has also been suggested that the antiproliferative effects of TZDs in cancer cells are also independent of PPARγ activation as TZD analogs, unable to activate PPARγ, retain the ability to induce apoptosis at an equal potency in several cancer cell lines. Nonetheless, there is an emerging view that some antitumorigenic effects are partially PPARγ dependent, whereas most are independent of the receptor [1,18].
In our studies PPARγ was successfully overexpressed in M1 cells. The increased PPARγ levels did not appear to alter cell morphology alone; however, stimulation with TRO in over-expressed cells resulted in a similar phenotype as in nontransfected M1 cells. Generally, PPARγ overexpression alone is sufficient to induce a cellular response without the use of an agonist as seen in pancreatic islets where overexpression of PPARγ protein suppressed insulin secretion induced by stimulatory concentrations of glucose [19,20]. This was also seen in rat livers where PPARγ was over-expressed using an additional adenoviral vector (AdPPARγ), resulting in significantly lower levels of fibrosis compared to controls [21]. Troglitazone has been shown to act through both PPAR-dependent and -independent mechanisms in epithelial cells. For instance, in porcine renal epithelial tubules, TZDs can modulate the pathways influencing glucose and glutamine metabolism, as well as cellular acidosis favoring reduced plasma glucose in vivo [14,22]. This alteration in cellular acidosis and glutamine metabolism was verified in pH-sensitive lung carcinoma cells where TZD-dependent PPARγ activation was verified; however, the primary effects of the study were still independent of this activation, instead involving PKC/ERK. The task remains to decipher which effects obtained with TZDs are PPARγ mediated and how they are interconnected with the PPARγindependent pathways to fully understand how these drugs operate. been previously characterized in both cell lines [23,24]. The inability to induce EMT prevented a full assessment of whether PPARγ activation could have reversed these effects, which would have confirmed a protective role for the transcription factor in collecting duct cells.

Troglitazone and Cell
Morphology. TRO caused a loss in cell-to-cell contact and an elongated spindle-like phenotype in a percentage of the treated cell population. The changes in cell shape appear to be PPARγ independent as GW9662 did not restore normal morphology or any cell-to-cell contact.
Additionally the agonist may have resulted in a further disruption of cell integrity with TRO cotreatment. TRO may be acting through both PPARγ-dependent and -independent pathways in inducing the aforementioned changes in morphology; however, with no partial reversal upon antagonism, it is more likely that a PPARγ-independent pathway is able to fulfill the necessary steps to obtain the phenotype observed. PKC/Akt and MAP kinase activation may be involved as TZDs have been shown to enhance PKC/Akt phosphorylation in both diabetic and nondiabetic rats and MAP kinase activation is a well-characterized PPARγ-independent response to TZDs [10]. This elongated phenotype was absent in populations that were able to reach confluence. This generally occurred if cells were stimulated with TRO at a confluence greater than 75%. Even with an inhibition in growth, highly populated plates were able to proliferate in close proximity. TRO does not appear to be able to alter morphology and initiate cell-to-cell contact disruption in confluent layers of IMCD-K2 and M1 cells. Similarly, intact tubular epithelial layers do not undergo EMT upon TGF-β stimulation either; however, loss of epithelial integrity (subconfluence, wounding, and contact disassembly by Ca 2+ removal) restores the ability of the monolayer to undergo a mesenchymal transformation [25,26]. This suggests that there may be a similar phenomenon observed in our study, with Troglitazone requiring a disruption in the cell monolayer in order to induce the morphological changes characterized. α-catenin, and β-catenin. TRO also caused an increase in fibronectin protein but did not trigger de novo α-SMA expression, the hallmark of myofibroblast generation. PGJ 2 on the other hand increased E-cadherin, β-catenin, and fibronectin levels. This suggests that PPARγ activation in vivo, as PGJ 2 is a supposed endogenous ligand of PPARγ, may counteract the decreases in epithelial markers observed in vitro with a synthetic ligand. These inconsistencies among PPARγ ligands are mirrored in MCF-7 breast cancer cells where conjugated linoleic acid, which acts as a PPARγ agonist, also caused an upregulation and redistribution of β-catenin and E-cadherin [27]. E-cadherin has been shown to have a functional PPRE [11] suggesting that PPARγ stimulation may result in an increase in E-cadherin levels.

Troglitazone and the ECM/
Although the PGJ 2 and TRO results contradict one another, TRO stimulation may still involve a PPAR response element since a PPARγ co-repressor complex could have bound to E-cadherin's PPRE, preventing further transcription and thereby translation of the E-cadherin protein. A core function of p120-catenin is to regulate cadherin stability and turnover by controlling cadherin entry into the degradative endocytic pathways [1,28]. The decrease in protein levels could have been due to the prevention of E-cadherin turnover and enhanced degradation. A closer look at p120catenin would clarify whether TRO's effects on E-cadherin were secondary, preventing the regeneration of E-cadherin protein, as opposed to causing the direct degradation of the adheren marker. Nonetheless, the decrease in E-cadherin in our study occurs independently of PPARγ. The PPARγ antagonists T0070907 and GW9662 did not reverse the decrease observed. Overexpression of PPARγ did not alter the drop in E-cadherin levels upon TRO stimulation either. The opposing effects of the two agonists, TRO and PGJ 2 , create another argument that the effects of at least one of the ligands are PPARγ independent. Another explanation is that one ligand activates PPARγ at a higher capacity than the other as this paradox has been shown to induce opposing effects in breast cancer tissue, depending on the magnitude of PPARγ activation in terms of cellular proliferation, differentiation, and apoptosis of cancer cells and tumors [1]. For example, lower concentrations increased cellular proliferation, whereas higher concentrations blocked cell growth.
It should be noted that the 50% reduction in β-catenin may have not only led to the destabilization of adheren junctions and loss of cell-to-cell contact but also to decreases in β-catenin activity as an LEF/TCF transcription factor, possibly accounting for some of the other effects observed upon TRO stimulation. Whether any one effect can be directly linked to catenin signaling would require a full restoration of β-catenin levels. TRO also caused a decrease in α-catenin in M1 cells. This decrease was not reversed by T0070907 suggesting the degradation of α-catenin to be independent of PPARγ activation as well. Generally αcatenin is thought of as a static link between the cadherincatenin chain and actin cytoskeleton; however, new roles for α-catenin are being described as α-catenin has been shown to exist as a monomer or a dimer with monomeric α-catenin binding more strongly to E-cadherin-β-catenin and the dimer preferentially binding to actin filaments [29]. Determining the proportion of monomeric to dimerized αcatenin affected by TRO stimulation may shed light on the local regulation of actin assembly and organization at sites of cadherin-mediated cell-cell adhesion.
Fibronectin was upregulated by TRO and PGJ 2 . In a previous study we showed that lower concentrations of TRO had no effect on fibronectin in collecting duct cells [13].
This provides a further argument for the concentration dependence of TRO actions though the effect of fibronectin seems to be mainly PPARγ independent (no effect of PPARγ antagonism or overexpression). Generally TZDs have been shown to reverse fibronectin expression in diseased states, as in human lung carcinoma cells [3], in human cortical fibroblasts [30], in TGF-β-treated glomerular mesangial cells [31], and in diabetic mice [32]. Nonetheless, the complex nature and inconsistencies of PPARγ activation and function in terms of its ligands are seen throughout the literature. An inverse relation between the adhesion molecule E-cadherin and the matrix protein fibronectin has been established in many diseased states, a trend observed in this study suggesting that the collecting duct may be altered upon stimulation with Troglitazone.

Conclusion
PPARγ is present in the collecting duct; however, the use of TRO, a potent synthetic ligand, resulted in structural changes independent of its target receptor. These effects were not replicated with the use of PGJ 2 , a proposed endogenous PPARγ ligand. PPARγ overexpression did not accentuate any of the effects obtained from stimulation with TRO. Taken together the data suggest that TRO can alter ECM and cytoskeletal marker expression in the CD leading to a partial epithelial to mesenchymal-like transformation, independent of PPARγ. The beneficial effects of PPARγ ligands in other diseased systems highlight the relevance of this study and underline the importance of fully understanding the effects of these drugs to improve their use and specificity.