GW501516, a PPARδ Agonist, Ameliorates Tubulointerstitial Inflammation in Proteinuric Kidney Disease via Inhibition of TAK1-NFκB Pathway in Mice

Peroxisome proliferator-activated receptors (PPARs) are a nuclear receptor family of ligand-inducible transcription factors, which have three different isoforms: PPARα, δ and γ. It has been demonstrated that PPARα and γ agonists have renoprotective effects in proteinuric kidney diseases; however, the role of PPARδ agonists in kidney diseases remains unclear. Thus, we examined the renoprotective effect of GW501516, a PPARδ agonist, in a protein-overload mouse nephropathy model and identified its molecular mechanism. Mice fed with a control diet or GW501516-containing diet were intraperitoneally injected with free fatty acid (FFA)-bound albumin or PBS(−). In the control group, protein overload caused tubular damages, macrophage infiltration and increased mRNA expression of MCP-1 and TNFα. These effects were prevented by GW501516 treatment. In proteinuric kidney diseases, excess exposure of proximal tubular cells to albumin, FFA bound to albumin or cytokines such as TNFα is detrimental. In vitro studies using cultured proximal tubular cells showed that GW501516 attenuated both TNFα- and FFA (palmitate)-induced, but not albumin-induced, MCP-1 expression via direct inhibition of the TGF-β activated kinase 1 (TAK1)-NFκB pathway, a common downstream signaling pathway to TNFα receptor and toll-like receptor-4. In conclusion, we demonstrate that GW501516 has an anti-inflammatory effect in renal tubular cells and may serve as a therapeutic candidate to attenuate tubulointerstitial lesions in proteinuric kidney diseases.


Introduction
Proteinuria is not only a clinical indicator of chronic kidney disease (CKD), but also a detrimental factor in progressive CKD [1,2,3]. In proteinuric kidney diseases, excessive albumin and several macromolecules bound to albumin, including free fatty acid (FFA), are filtered from glomeruli and reabsorbed into proximal tubular cells. This causes an inflammatory vicious cycle via secretion of pro-inflammatory cytokines, such as monocyte chemoattractant protein-1 (MCP-1), and subsequent infiltration of inflammatory cells, which secrete tumor necrotic factor-a (TNFa) [1,2,3,4,5]. Continuous inflammation triggers tubular cell damage, fibrosis and eventually progression to the final stage of kidney disease [5,6]. Thus, suppression of inflammation caused by all these factors should serve as an important means to protect proximal tubular cells and maintain renal function in proteinuric kidney diseases.
Peroxisome proliferator-activated receptors (PPARs) are a nuclear receptor family of ligand-inducible transcription factors, composed of three different isoforms: PPARa, PPARd and PPARc [7,8]. Acting as transcription factors, PPARs regulate various metabolic processes involving lipid metabolism, glucose homeostasis, cell differentiation and inflammation [7,8,9,10,11]. Recent reports demonstrated that PPARa and -c agonists show renoprotective effects in a PPAR-dependent and independent manner [12,13,14,15,16,17]; however, the physiological role and therapeutic potential of PPARd in the kidney remain unclear. Several reports have shown that PPARd agonists are potentially anti-inflammatory agents in some cell types and animal models [18,19,20,21]. In summary, PPARd agonists may show antiinflammatory effects in renal proximal tubular cells and could be promising new therapeutic candidates for proteinuric kidney diseases.
To certify this hypothesis, we investigated whether GW501516, a highly selective PPARd agonist [22], could attenuate tubulointerstitial lesions in a protein-overload mouse model (an established mouse model for studying interstitial lesions in proteinuric kidney diseases [4,23]), and whether it could inhibit albumin-, saturated FFA (palmitate)-and TNFa-induced inflammation in cultured mouse proximal tubular cells. Our results showed that GW501516 exerted an anti-inflammatory effect in proximal tubular cells in both in vivo and in vitro studies, and it may be a promising new candidate for slowing down the progression of proteinuric kidney diseases.

Results
GW501516 attenuates interstitial inflammation and proximal tubular cell damage in a protein-overload mouse nephropathy model To determine whether PPARd agonists exert anti-inflammatory effects in vivo, we examined the effect of GW501516 on a proteinoverload mouse model. First, we confirmed that bovine serum albumin (BSA) injection in mice increased urinary protein excretion, and that treatment with GW501516 did not affect basal and BSA-induced urinary protein excretion [(PBS(2) group: 3.2660.66, PBS(2) + GW501516 group: 3.2160.70, BSA group: 6.3661.26, BSA + GW501516 group: 6.6061.93 mg/day; data are shown as means 6 SEM]. In mice fed with the control diet, intraperitoneal protein overload induced significant and severe tubular injury, characterized by diffused tubular cell vacuolation, tubular cell flattening, tubular lumen dilation and cast formation ( Figure 1A and B). In the mice fed with the GW501516-containing diet, these tubulointerstitial damages were significantly attenuated ( Figure 1A and B). The PBS(2) injection resulted in no apparent changes in proximal tubular cells of either control or GW501516treated groups ( Figure 1A and B).
Protein overload induced the prominent infiltration of F4/80positive macrophages into the renal interstitial area, which was markedly attenuated by GW501516 treatment (Figure 2A). Infiltration of F4/80-positive macrophages into glomeruli was not observed in the kidney of any group of mice ( Figure 2A). The changes in F4/80 expression were confirmed by mRNA levels, determined by real-time PCR ( Figure 2B). Furthermore, protein overload significantly increased mRNA expression levels of inflammatory cytokines such as MCP-1, TNFa and IL-6. These increases were significantly attenuated by GW501516 treatment ( Figure 2C, D and E). Protein overload and GW501516 treatment did not affect PPARd expression levels in mice (data not shown).
These results suggest that GW501516 shows an anti-inflammatory effect and protection of proximal tubular cells in the proteinuric kidney disease mouse model.

GW501516 exerts anti-inflammatory effects in mouse cultured proximal tubular (mProx) cells
Because proximal tubular cells are the direct victims of overflowed protein [1,2,3], we examined the anti-inflammatory effect of GW501516 on cultured mouse proximal tubular cells, treated with pro-inflammatory molecules associated with proteinuria-induced inflammation, such as FFA-free albumin [24,25], albumin-bound saturated FFA (palmitate) [26] and TNFa. GW501516 failed to inhibit FFA-free albumin-induced MCP-1 expression ( Figure 3A). However, GW501516 inhibited palmitate-and TNFa-induced increases in MCP-1 mRNA expression in a dose-dependent manner ( Figure 3B and C). These stimuli and treatment did not affect PPARd expression levels (data not shown). These results suggest that the GW501516-mediated suppression of MCP-1 expression is specific to palmitate and TNFa in cultured mProx cells. PPARd is not required for GW501516-mediated antiinflammatory effects in cultured proximal tubular (mProx) cells The molecular mechanism of PPAR agonist-mediated antiinflammatory effects is cell-dependent [11,27]. Also, some PPAR agonists have been shown to have an anti-inflammatory effect via direct inhibition of an intracellular signaling pathway, but not via ligand-activation of PPARs and subsequent upregulation of their target genes [11,27]. Thus, we investigated whether PPARd itself is required for the anti-inflammatory effect of GW501516 in cultured mProx cells, using siRNA-mediated PPARd knockdown and retrovirus-mediated PPARd overexpression. We confirmed the decreased PPARd protein expression in siRNA-mediated PPARd-knockdown mProx cells ( Figure 4A). Furthermore, the mRNA level of PPARd target genes was decreased in these cells ( Figure 4B). GW501516 significantly attenuated both palmitateand TNFa-induced MCP-1 expression levels in mProx cells transfected with control siRNA ( Figure 4C and D). This response was not affected by siRNA-mediated PPARd knockdown ( Figure 4C and D). Moreover, we generated retrovirus-mediated PPARd-overexpressing mProx cells and confirmed the functional overexpression of PPARd ( Figure 5A and B). The GW501516 inhibitory effect on both palmitate-and TNFa-induced MCP-1 expression was not enhanced in retrovirus-mediated PPARdoverexpressing mProx cells ( Figure 5C and D). Thus, neither knockdown nor overexpression of PPARd changed the antiinflammatory effect of GW501516. These results suggest that PPARd itself is not essential for the anti-inflammatory effect of GW501516.

GW501516 exerts its anti-inflammatory effect by inhibition of the TAK1-NFkB pathway
We continued to elucidate the identity of the signaling molecules involved in the anti-inflammatory effect of GW501516, specifically focusing on signaling molecules associated with the NFkB pathway, which is a main inflammation regulatory pathway. In cultured mProx cells incubated with palmitate, phosphorylation of IkB and subsequent nuclear translocation of    the p65 component of NFkB were observed after 1 hour of stimulation ( Figure 6A). Palmitate-induced phosphorylation of IkB and nuclear translocation of p65 were attenuated by pretreatment with GW501516 ( Figure 6B). Furthermore, we found that GW501516 inhibited palmitate-induced DNA binding of p65 to the MCP-1 promoter, as determined by chromatin immunoprecipitation (ChIP) analysis ( Figure 6C). Next, we checked the profiles of upstream and regulatory molecules associated with receptor-mediated NFkB activation, such as mitogen activated protein kinases (MAPKs; ERK, p38 and JNK) [28,29] and TGF-b activated kinase 1 (TAK1) [30,31]. The phosphorylation levels of the MAPKs and TAK1 were increased after 1 hour of palmitate stimulation ( Figure 6D). GW501516 inhibited the palmitateinduced phosphorylation of TAK1 ( Figure 6E and F), but not of the MAPKs ( Figure 6E).
Furthermore, we observed similar results with cultured mProx cells incubated with TNFa. TNFa increased IkB phosphorylation and subsequent nuclear translocation of p65 after 5 minutes of stimulation ( Figure 7A and 7B). These changes and DNA binding of p65 to the MCP-1 promoter were significantly attenuated by pretreatment with GW501516 ( Figure 7C and D, respectively). Finally, TAK1 phosphorylation was increased after 5 minutes of TNFa stimulation ( Figure 7E and F), which was suppressed by pretreatment with GW501516 ( Figure 7E and F).
GW501516 inhibited palmitate-and TNFa-induced phosphorylation of TAK1 and IkB, and subsequent nuclear translocation of p65 even in mProx cells transfected with siRNA against PPARd ( Figure 8A and B), suggesting that GW501516 inhibited the TAK1-IkB pathway directly.
To investigate whether PPARd agonist-mediated inhibition of TAK1 activation was specific to GW501516, we examined the anti-inflammatory effect and mechanism of another PPARd agonist, GW0742, in mProx cells. Similarly to GW501516, GW0742 inhibited both palmitate-and TNFa-induced MCP-1 expression ( Figure 9A and B) and suppressed TAK1 phosphorylation ( Figure 9C, D, E and F). These results suggest that PPARd agonist-mediated inhibition of TAK1 activation is not specific to GW501516.

Discussion
The present study shows that GW501516 attenuated tubulointerstitial lesions in a protein-overload nephropathy mouse model and that its renoprotective effect was achieved by its antiinflammatory action in the kidney. These results highlight the possibility that GW501516 may be a novel therapeutic candidate for the improvement of renal prognosis in patients with proteinuric kidney diseases.
The protein-overload model is a well-established model for studying the renal tubulointerstitial lesions caused by proteinuria [4,23]. In this model, upregulation of MCP-1 was demonstrated and the anti-MCP-1 gene therapy attenuated interstitial lesions [36], which verified that attenuation of inflammation through inhibition of MCP-1 should serve as a therapeutic target in proteinuric kidney diseases. In our study, we confirmed that protein overload induced MCP-1 expression and subsequent renal inflammation, and these inflammation-associated renal damages were attenuated by treatment with GW501516. To our knowledge, this is the first report showing that a pharmacological agent can improve tubulointerstitial lesions in this model.
Our in vitro study has identified the detailed molecular mechanism underlying the anti-inflammatory effect of GW501516 in the kidney. In proteinuric kidney diseases, albumin, FFAs bound to albumin and inflammatory cytokines coordinately mediate interstitial inflammation. In this study, we confirmed that all of these factors induced overexpression of MCP-1 in cultured proximal tubular cells via NFkB activation. Interestingly, GW501516 attenuated MCP-1 expression induced by palmitate and TNFa, but not albumin. These results suggest that the antiinflammatory effect of GW501516 is not universal in NFkBassociated inflammatory processes; hence we focused on the upstream signaling to NFkB. NFkB activation is regulated by a variety of molecular pathways such as MAPKs and TAK1   [28,29,30]. Our group previously demonstrated that albumininduced MCP-1 expression was MAPK (ERK)-dependent [25]. Here we show that GW501516 could not attenuate albumininduced MCP-1 expression, suggesting that inhibition of ERK activity is not the target of the anti-inflammatory effect of GW501516. In contrast, GW501516 inhibited palmitate-and TNFa-induced MCP-1 expression via the inhibition of TAK1 activation, without affecting MAPK activity. Furthermore, albumin failed to induce phosphorylation of TAK1 (data not shown). Thus, for the first time, we concluded that inhibition of TAK1, but not MAPKs, is involved in the mechanism underlying the anti-inflammatory effect of GW501516 ( Figure 10). To our knowledge, few reports have shown the interaction between PPAR agonists and TAK1. Our current findings may enrich the knowledge of anti-inflammatory mechanisms of PPAR agonists.
To date, the mechanism underlying the anti-inflammatory effect of PPARd agonists has been focused on BCL6 protein. The release of BCL6 from PPARd is known to contribute to any of the anti-inflammatory actions of PPARd agonists in macrophages, both in vitro [21,37] and in vivo [21]. In other tissues it is not so clear; the inhibitory actions of PPARd on adhesion molecules expression in endothelial cells have been reported to be both BCL6-dependent [38] and BCL6-independent [39], suggesting that the BCL6 theory is cell-specific and ligand-specific. We showed that suppression of PPARd gene expression had no effect on MCP-1 expression in mProx cells, indicating that BCL6 is not involved in the anti-inflammatory mechanism of GW501516 in proximal tubular cells.
Also in our study, we demonstrated that PPARd itself is not essential for the anti-inflammatory effect of GW501516. The knockdown or overexpression of PPARd could neither impede nor enhance the anti-inflammatory action of GW501516. GW501516 has been shown to exert an anti-inflammatory effect in many cell types [18,19,20,21], but whether or not this effect is mediated through the PPARd receptor has not been clarified. Although we demonstrated that two types of PPARd agonists, GW501516 and GW0742, did not require PPARd itself for their anti-inflammatory effects in cultured proximal tubular cells, it remains to be confirmed whether all PPARd agonists function similarly, to conclude whether this mechanism is specific for GW501516 and GW0742 or global for PPARd agonists. Furthermore, to elucidate whether GW501516 can improve renal inflammation in a PPARdindependent manner in vivo, further examination of the effect of GW501516 on the progression of tubulointerstitial lesions induced by protein overload in PPARd knockout mice is required.  In conclusion, we demonstrated that GW501516, a PPARd agonist, showed a renoprotective effect by exerting an antiinflammatory effect in proximal tubular cells, both in vivo and in vitro. Thus, it may be a promising candidate for slowing down the progression of chronic kidney diseases.

Animals.
Animal care and treatment were performed in accordance with the guidelines of the Research Center for Animal Life Science of Shiga University of Medical Science (approval ID: 2010-2-13). Seven-week-old male C57BL/6 mice (CLEA Japan, Tokyo, Japan) were housed at a constant room temperature, 12-h dark/12-h light cycle, and freely fed a standard diet and tap water. Mice were randomly allocated to different groups and received therapeutic diet and treatment. The GW501516-containing rodent diet was made by evenly adding GW501516 to the control diet to a final concentration of 0.04% w/w. In the control diet, 10% of the total calories were from fat (5.5% from soybean oil and 4.5% from lard). All foods were purchased from Research Diets (New Brunswick, NJ, USA).

Protein-overload mouse nephropathy model
Animal experiments were performed as described previously, with minor modifications [4,23]. FFA-bound BSA was administered by intraperitoneal injection to induce renal tubulointerstitial damages. Mice were divided into the following four groups: PBS(2), PBS(2) + GW501516, protein-overload and proteinoverload + GW501516 (n = 3, 4, 6 and 6, respectively). Mice were fed either a control diet or a GW501516-containing diet. Each group was given its diet for 10 days before starting treatments. Once a day for the following 4 days, the mice of each group were given an intraperitoneal injection of either 10 mg/g body weight FFA-bound BSA diluted in sterile PBS (2) or an equal volume of sterile PBS(2) as a control. On the fifth day, mice were anesthetized with intraperitoneal injection of pentobarbital sodium, then blood was drawn from the right atrium with a heparinized syringe and serum was separated for biochemical assays. Kidneys were removed for mRNA assays, histological examination and immunohistochemical staining, as previously reported [40,41]. Urinary protein concentrations were measured with the Bradford method.

Histological analyses
Fixed kidneys embedded in paraffin were sectioned to 3 mm thickness [40,41]. Semiquantitative histological analysis of the tubular damage was performed using HE-stained sections from each mouse. Tubular damage scores, ranging from 0 to 4, were assessed according to both the severity and range of proximal tubular damages; an average was taken from the final score of 20 randomly selected, non-overlapping cortical areas under 6200 magnification [42]. Immunohistochemical staining was performed with the F4/80-specific monoclonal anti-rat antibody (Serotec, Oxford, UK), as previously described [40].

Cell culture
Immortalized murine proximal tubular cells (mProx) were established as described previously [25]. First, we confirmed that the dosage of GW501516 used in the following experiments was not cytotoxic to the cells by using the AlamarBlue TM Assay (Alamar Bioscience, Sacramento, CA, USA; data not shown). Cells were cultured in DMEM containing 10% FBS, 100 U/ml penicillin and 100 mg/ml streptomycin at 37uC and 5% CO 2 thermostat. Subconfluent cells were starved by incubation in 0.2% FCS DMEM for 9 h, then preincubated with GW501516 [dissolved in 10% dimethyl sulfoxide (DMSO)], at a final concentration of 2.5 and 5 mM, or 0.05% DMSO as control for 3 hours, followed by stimulation with 150 mM palmitate bound to 8.0% BSA for 12 h. We also performed the same experiment under stimulation with 10 nM TNFa and 30% BSA.

Quantitative real-time PCR
Total RNAs were isolated from the kidney samples and cultured cells, and cDNAs were synthesized as previously reported [41]. The iQSYBR Green Supermix (Bio-Rad Laboratories, Hercules, CA, USA) was used for real-time PCR (ABI Prism TM 7500 Sequence Detection System; Perkin-Elmer Applied Biosystems, Foster City, CA, USA). The mRNA expression levels were quantified using the standard curve method [41]. Analytical data were normalized to the mRNA expression level of GAPDH as an internal control. Primer sequences are described in Table 1.

Immunoblotting
Total cell lysates for immunoblotting were collected as previously described [43,44]. For detecting nuclear translocation of p65, nuclear extracts were prepared as previously described [43,44]. Equal amounts of protein samples were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to polyvinylidene fluoride membranes (Immobilon, Bedford, MA, USA). The membranes were incubated with the appropriate antibodies, washed and incubated with horseradish peroxidase-coupled secondary antibodies (Amersham, Buckinghamshire, UK). After washing, the blots were visualized by using an enhanced chemiluminescence detection system (Perkin Elmer Life Science, Boston, MA, USA).

Small interference RNA transfections against PPARd
Control siRNA and siRNA against PPARd were purchased from Ambion (Austin, TX, USA). Transient transfections of siRNA were performed using Lipofectamin 2000 (Invitrogen, Carlsbad, CA, USA) [43,44]. To determine the effect of siRNA against PPARd on palmitate-and TNFa-induced MCP-1 expression, all experiments were performed 24 h post-siRNA transfection.

ChIP analysis
The ChIP experiment was performed using the ChIP-IT kit from Active Motif (Carlsbad, CA, USA) according to the manufacturer's instructions [44]. The primer sequences are listed in Table 1.
Briefly, HEK293T cells were transfected with pBABE and pBABE-PPARd using Lipofectamin 2000 reagent. Forty-eight hours post-transfection, the media containing retroviruses were collected, centrifuged, and transferred into the cultured proximal tubular cells, which were pre-treated with polybrene (1 mg/mL) for 30 min. Retrovirus-infected cells were selected by treatment with puromycin (2.5 mg/mL) for several days. We confirmed overexpression of PPARd by immunoblotting and real-time PCR.

Statistical analysis
Results are expressed as mean 6 SEM. ANOVA with subsequent Scheffe test was used to determine the significance of differences in multiple comparisons. A p value ,0.05 was considered statistically significant.