The PPARγ agonist pioglitazone prevents TGF-β induced renal fibrosis by repressing EGR-1 and STAT3

Background It has been proposed that peroxisome proliferator-activated receptor-γ (PPARγ) agonists might reduce renal fibrosis, however, several studies had contradictory results. Moreover, the possible interaction of TGF-β1, PPARγ, and transcription factors in renal fibrosis have not been investigated. We hypothesized that oral pioglitazone treatment would inhibit TGF-β–driven renal fibrosis and its progression, by modulating profibrotic transcription factors in TGF-β1 transgenic mice. Methods Male C57Bl/6 J mice (control, CTL, n = 14) and TGF-β overexpressing transgenic mice (TGFβ, n = 14, having elevated plasma TGF-β1 level) were divided in two sets at 10 weeks of age. Mice in the first set were fed with regular rodent chow (CTL and TGFβ, n = 7/group). Mice in the second set were fed with chow containing pioglitazone (at a dose of 20 mg/kg/day, CTL + Pio and TGFβ+Pio, n = 7/group). After 5 weeks of treatment, blood pressure was assessed and urine samples were collected, and the kidneys were analyzed for histology, mRNA and protein expression. Results TGF-β1 induced glomerulosclerosis and tubulointerstitial damage were significantly reduced by pioglitazone. Pioglitazone inhibited renal mRNA expression of all the profibrotic effectors: type-III collagen, TGF-β1, CTGF and TIMP-1, and alike transcription factors cFos/cJun and protein expression of EGR-1, and STAT3 protein phosphorylation. Conclusions Oral administration of PPARγ agonist pioglitazone significantly reduces TGF-β1-driven renal fibrosis, via the attenuation of EGR-1, STAT3 and AP-1. This implies that PPARγ agonists might be effective in the treatment of chronic kidney disease patients. Electronic supplementary material The online version of this article (10.1186/s12882-019-1431-x) contains supplementary material, which is available to authorized users.


Background
Renal fibrosis is the hallmark of several chronic kidney diseases (diabetes mellitus, arterial hypertension) that progress to end-stage renal disease (ESRD), and thus produce a serious health burden worldwide [1]. The dynamic balance of extracellular matrix protein (ECM) synthesis and degradation is mainly regulated by matrix metalloproteinases (MMPs) and their inhibitors, the tissue inhibitors of metalloproteinases (TIMPs). In fibrosis that occurs in multiple organs, this MMP/TIMP balance is disrupted. Transforming growth factor-β 1 (TGF-β 1 ) is a multifunctional cytokine involved in various pathologic conditions, including carcinogenesis, and tissue fibrosis [2,3]. TGF-β 1 signaling can induce TIMP-1 transcription at promoter level via activator protein-1 complex (AP-1) [4]. TGF-β 1 transgenic mice with elevated plasma levels of TGF-β 1 develop severe renal fibrosis, underscoring the central role of TGF-β 1 in the pathogenesis of fibrotic kidney diseases [5].
Peroxisome proliferator-activated receptor-γ (PPARγ) is an ubiquitary, anti-fibrotic and vasoprotective nuclear hormone receptor and transcription factor. PPARγ agonists (e.g. pioglitazone) have been widely used in the treatment of diabetes mellitus [17] and the prevention of macrovascular events and the development of diabetes in patients with insulin resistance [18].
Moreover, PPARγ agonists have been also shown to reduce renal damage in animal models of ischemia reperfusion injury, autosomal dominant polycystic kidney disease (ADPKD) or nondiabetic chronic kidney diseases [19]. Interestingly, despite a beneficial effect of the (otherwise liver-toxic) PPARγ agonist troglitazone was reported in the unilateral ureter obstruction (UUO) model of renal fibrosis [20], others could not confirm these results using pioglitazone [21]. In addition to these conflicting results, the possible interaction of PPARγ, TGF-β 1 and EGR-1 in renal fibrosis has not been investigated.
Here, we aimed to study whether the chronic administration of PPARγ agonist pioglitazone could influence renal EGR-1, STAT3 and AP-1 expression, and ameliorate renal fibrosis in TGF-β 1 overexpressing mice. We report that oral administration of pioglitazone effectively reduced glomerulosclerosis, tubular injury and interstitial fibrosis by inhibiting EGR-1 and TIMP-1 expression. Our results might help to clarify the beneficial role of PPARγ in preventing or reversing TGFβ 1 -driven renal fibrosis.

Animals and experimental design
Male C57Bl6/J (B6) and B6-Alb/TGF-β 1 (cys 223,225 ser) [5] (TGF-β1 transgenic mice with elevated circulating TGFβ 1 level) were housed at the Semmelweis University NET GMO facility under standard specific pathogen free (SPF) conditions, with a 12/12 h light/dark cycle. In TGF-β1 transgenic mice, the transgene construct consists of the full-length porcine TGF-β1 cDNA (with mutated cysteine 223 and 225 to serine which leads to production of active TGF-β1) driven by the albumin promoter and enhancer, and the transgene was incorporated into the Y chromosome therefore only male mice are transgenic. The expression of the transgene starts after birth in the hepatocytes, and leads to constitutive secretion of active TGF-β1 into the bloodstream. The mouse strains were bred and maintained at the NET GMO SPF facility. The animals had free access to standard rodent chow (Altromin 1314 TPF) and filtered drinking water. 10 week old male C57Bl/6 J control (CTL) and TGF-β 1 transgenic mice (TGFβ) were randomly divided in two sets. The number of animals was calculated by the Ethical Committee to provide the minimum sample size needed for relevant results. The first set of mice received regular chow. The second set of mice were treated orally with pioglitazone (20 mg/kg/day) for 5 weeks (CTL + Pio and TGFβ+Pio). Pioglitazone was incorporated in the chow, and average food consumption was assessed at the beginning of the study. Sample size was n = 7/group. At 15 weeks of age, urine spot samples were obtained by sterile punction of the bladder under isoflurane (2%) anesthesia, blood pressure was measured and mice were perfused by intracardiac cannula under narcosis for 20 min with 4°C physiological saline. Then, mice were euthanized with 5% isoflurane and kidneys were harvested for histology and mRNA expression analyses. All assays were performed in duplicates for each sample to ensure reliability of single values.

Animal genotyping
The presence of TGF-β 1 transgene was verified from genomic DNA (1-2 mm tail samples) by polymerase chain reaction (PCR). DNA was extracted by Tris-NaOH method. Briefly, samples were incubated in 200 μl of 0.1 N NaOH in a microcentrifuge tube at 96°C for 10 min, mixed and cooled on ice. After briefly spun, 50 μl Tris (pH 8) was added to the lysates, mixed and centrifuged at 20000 g for 6 min at 4°C. Supernatants were pipetted into a new tube, and 2.5 μl of each supernatant was used for the PCR.

Measurement of blood pressure
Systemic blood pressure was measured invasively under isoflurane analgesia, followed by organ harvest. A 1.4F microtip catheter (Millar Instruments, USA) inserted into the right carotid artery/ aortic arch where the systemic blood pressure was measured. Blood pressure curves were recorded and analyzed using a PowerLab recording unit and LabChart software (AD Instruments, Colorado Springs, USA).

Determination of plasma TGF-β 1 levels
During harvest, blood samples from the aorta were harvested in siliconized microcentrifuge tubes containing EDTA as anticoagulant (100 μl/mouse). Blood samples were centrifuged at 1000 g for 20 min and the isolated plasma was centrifuged again in new siliconized tubes at 10000 g for 10 min. Supernatants (10 ul) were used to determine plasma TGF-β 1 level using a commercial ELISA kit (R&D Quantikine TGF-β 1 ELISA, Minneapolis, USA).

Determination of proteinuria
The amount of excreted protein (BCA assay, Thermo Fisher, Waltham, MA, USA) and creatinine (Creatinine Kit, Diagnosticum ZRt, Budapest, Hungary) were measured from spot urine samples according to the manufacturer's protocols. Proteinuria was expressed as urinary protein to creatinine ratio (UPCR), in order to normalize for GFR.

Histology and immunohistochemistry
Formaline fixed paraffin embedded kidney sections were evaluated after Masson's trichrome staining. The degree of glomerulosclerosis and tubulointerstitial damage was determined blinded on a semiquantitative scale as previously described [23].
Briefly, glomerulosclerosis index (GSI) of each animal was determined with a light microscope at 400x magnification from the arithmetic mean of 100 evaluated glomeruli. The tubulointerstitial damage index (TDI) scores were evaluated at 100x magnification as follows: score 0: no change; score 1-5 depending on the criteria have been met in the given field of view: tubular dilatation, tubular atrophy, hyalin in tubular lumen, interstitial infiltration of mononuclear cells, interstitial fibrosis.
Immunohistochemical staining of paraffin embedded sections was performed using avidin-biotin method as previously described [23] using citrate buffer pH 6.0 for heat induced antigen retrieval. The primary antibodies were rabbit polyclonal anti-fibronectin at 1:1000 (Sigma-Aldrich, Budapest), and rabbit polyclonal anti-EGR-1 at 1:500 (Cell Signaling, USA).
Immunostaining reactivitiy scores were evaluated in a blinded manner at 400x magnification using a semiquantitative scoring method: score 0: no staining; score 1: light staining; score 2: moderate staining; score 3: strong staining; score 4: very strong staining. Renal EGR-1 expression was assessed by counting the amount of EGR-1 positive nuclei at 400x magnification (high power field, HPF) and expressed as positive cells / HPF.

Statistics
Experimental data are presented as mean ± SD and statistical analysis was performed using SPSS 10 for Windows (SPSS Inc). The data were analyzed using Kruskal-Wallis test followed by Dunn's multiple comparison test. The level of significance was set to p < 0.05. The data and statistical analysis comply with the recommendations on experimental design and analysis in pharmacology [24].

Results
Murine plasma TGF-β 1 levels, body weight and blood pressure Both wild type control mice and TGF-β 1 transgenic animals were evaluated for the absence or presence of the transgene. Control mice did not show the amplification of transgene, while transgenic mice showed the porcine TGF-β 1 PCR product (Additional file 1: Figure S1). Accordingly, the plasma TGF-β 1 levels of transgenic mice was significantly elevated as compared to control mice (Additional file 1: Figure S1).
Both control and TGF-β 1 transgenic mice had similar body weight regardless of the treatment (Fig. 1b), and the mean arterial blood pressure was also similar in all groups (Fig. 1a).
Pioglitazone ameliorates TGF-β 1 induced proteinuria, glomerular and tubular injury Urinary protein/creatinine ratio (UPCR) was similar in both the pioglitazone treated and non-treated controls. In contrast, UPCR was elevated by 40% in untreated TGF-β 1 transgenic mice, as compared to controls, but was significantly reduced to near normal levels in pioglitazone treated transgenic mice (Fig. 1c.).
Kidneys of control animals depicted normal histology without signs of glomerular or tubular damage, independent of the therapy (Fig. 1d and e). In contrast, and consistent with the UPCR results, untreated TGF-β 1 transgenic kidneys depicted extensive glomerulosclerosis and significant tubulointerstitial damage, accompanied by marked renal fibronectin immunoreactivity (Fig. 1f). Pioglitazone treatment reduced the TGF-β 1 induced glomerulosclerosis and tubulointerstitial damage by 30% and by 50%, respectively ( Fig. 1d and e), and normalized fibronectin expression (Fig. 1f).
In line with our histological findings, mRNA expression levels of Lipocalin-2 (Lcn2, a sensitive biomarker of tubular damage [25]) were similar in the pioglitazone treated and non-treated controls. In contrast, Lcn2 expression was significantly elevated in untreated TGF-β 1 Fig. 1 The values of mean arterial pressure, body weights, urinary protein/creatinine ratio and renal histology. Both mean arterial blood pressure (MAP, a) and body weight of control and transgenic mice (b) was similar in all groups regardless of treatment. Urinary protein/creatinine ratio (UPCR) was elevated in untreated TGF-β 1 transgenic mice (c), but was significantly ameliorated by pioglitazone treatment. Renal histology depicted significant glomerulosclerosis and tubulointerstitial damage in untreated TGF-β 1 mice (d and e) that were reduced by pioglitazone (Masson's trichrome staining, 400X magnification, bar represents 50 μm). Fibronectin immunostaining was markedly stronger in both glomeruli and tubulointerstitium of untreated TGF-β 1 mice (f), but reduced to control levels in pioglitazone treated mice (n = 7/group, *:p < 0.05, **:p < 0.01, ***:p < 0.001, Kruskal-Wallis test) transgenic mice but reduced to near normal levels in pioglitazone treated transgenic mice (Fig. 2a).
Among the profibrotic growth factors, renal expression of both TGF-β 1 (Tgfb1) and CTGF (Ctgf ) mRNA were significantly reduced by pioglitazone treatment (Fig. 2c and d). Similarly, we found strong EGR-1 immunoreactivity in TGF-β 1 transgenic kidneys that was attenuated by pioglitazone treatment (Fig. 2e and f ).
We observed marked STAT3 phosphorylation in the kidneys of untreated TGF-β 1 transgenic mice, as compared to control groups, that was normalized by pioglitazone administration (Fig. 2g).
The observed Timp1 overexpression in TGF-β 1 transgenic kidneys was accompanied by 4-fold and 2-fold increase in mRNA expression of the AP-1 components cFos and cJun, respectively ( Fig. 2e and f ), all of which were completely blocked (to control levels) by oral pioglitazone treatment (Fig. 2e and f ).

Discussion
In the present study, we examined the effect of PPARγ activation on the transcriptional regulation of renal fibrosis in TGF-β 1 transgenic mice. To the best of our knowledge, this is the first report demonstrating that the PPARγ agonist pioglitazone reduced kidney fibrosis in association with dampened renal EGR-1 and TIMP-1 expression.
In our study, neither the several fold increased plasma TGF-β 1 concentration, nor pioglitazone treatment had any influence on the mean arterial blood pressure of mice. This is in contrast to some reports on PPARγ agonists having antihypertensive effects, caused by increased endothelial NO levels and inhibition of ANG II [27]. However, several other studies did not find changes in systemic blood pressure with PPARγ agonist treatment (rosiglitazone) in wild type mice on regular and high fat diet [28]. Our study does support the beneficial effect of PPARγ agonists on reducing proteinuria, as previously shown in hypertensive and diabetes animal models [20], as well as T2DM patients [27]. Accordingly, pioglitazone significantly ameliorated both glomerular and tubulointerstitial fibrotic changes in the TGF-β 1 transgenic mice, associated with decreased expression of lipocalin-2, a sensitive biomarker of tubular damage [25]). Along with these findings, mRNA expression of fibronectin, type I and type III collagens, and TGF-β 1 were reduced, all known hallmarks of renal fibrosis [29][30][31]. Increased renal TGF-β 1 level promotes endothelial-mesenchymal transition (EMT) and progressive renal interstitial fibrosis [32]. CTGF, as one of the downstream TGF-β 1 mediators, participates in fibroblast proliferation and ECM production [19]. In our study, oral pioglitazone administration inhibited the TGF-β 1 -driven CTGF mRNA expression. The anti-fibrotic and antiproliferative effect of pioglitazone in our studies is supported by animal models of 5/6 nephrectomy [33], passive Heymann nephritis [34], acute mesangial proliferative glomerulonephritis, UUO [20] and ischemia reperfusion injury [35]. We have recently demonstrated that pioglitazone attenuates TGF-β 1 induced pulmonary arterial hypertension and remodeling in the TGF-β 1 overexpressing mouse, via decreasing TGF-β 1 , CTGF and α-smooth muscle actin (Acta2) expression in pulmonary arterial SMC [16]. Furthermore, PPARγ agonists might directly inhibit TGF-β 1 expression and reduce interstitial myofibroblast accumulation [20]. In our current study, the reduced α-smooth muscle actin and renal TGF-β 1 expressions support the proposed inhibitory effect of PPARγ agonists on myofibroblast activation and ECM production in fibrotic kidneys [36].
TGF-β 1 not only induces the synthesis of ECM components but also reduces matrix degradation via tissue inhibitors of metalloproteinases (TIMPs) [37]. Untreated TGF-β 1 transgenic mice in our study had disrupted renal MMP/TIMP balance that, intriguingly, was normalized by pioglitazone treatment. The latter could be, on one hand, a direct consequence of the reduced renal TGF-β 1 expression. On the other hand, pioglitazone might have indirect effect on the MMP/TIMP balance via transcriptional modulation. As TIMP-1 transcription is partly regulated by the activator protein-1 (AP-1) complex, we investigated the expression of the AP-1 components cFos and cJun [4], that were significantly increased in untreated TGF-β 1 transgenic kidneys. Consistent with our proposal, both cFos and cJun expression was normalized by pioglitazone treatment. Our findings not only support the regulatory importance of AP-1 transcription complex, but also show that PPARγ counteracts AP-1 activation.
Other transcription factors, such as the immediate early gene EGR-1, might be induced by a variety of fibrogenic stimuli, directly stimulating collagen production, matrix accumulation and myofibroblast differentiation [6,7]. We found boosted EGR-1 expression in untreated TGF-β 1 transgenic kidneys that was attenuated by pioglitazone treatment. Although the relationship of EGR-1 and PPARγ in the kidney has not been investigated yet, the PPARγ agonist rosiglitazone decreased skin fibrosis and EGR-1 levels in a mouse model of scleroderma [38]. This latter finding supports our results and implicates that EGR-1 is a potential anti-fibrotic target of activated PPARγ.
In addition, activation of the JAK/STAT pathways may be responsible for the increased myofibroblast transdifferentiation in the injured kidney [39]. STAT3 can induce TGF-β 1 expression [15], and STAT3 inhibition was reported to suppress tubulointerstitial fibrosis in UUO model [40]. Furthermore, we have recently demonstrated in human pulmonary arterial smooth muscle cells that TGF-β 1 induces STAT3 phosphorylation, and this STAT3 activation is inhibited by pioglitazone [16]. Our present study shows an additional potential interplay of STAT3 and PPARγ in renal fibrosis, as STAT3 was activated in kidneys of untreated TGF-β 1 transgenic mice, but STAT3 activation was absent in kidneys of TGF-β 1 transgenic mice chronically treated with pioglitazone.
Based on our study, we postulate that PPARγ agonist pioglitazone exerts its antifibrotic effect in the kidney by repressing STAT3 activation as well as decreasing the expression of EGR-1 and AP-1 components (cFos, cJun) (Fig. 4).

Conclusions
To the best of our knowledge, this is the first study to show that oral administration of PPARγ agonist pioglitazone in fibrosis prone TGF-β 1 transgenic mice antagonized the profibrotic effects of TGF-β 1 by repressing EGR-1, STAT3 and AP-1. In addition, pioglitazone normalized the MMP/TIMP imbalance in the kidneys without changing the systemic blood pressure. Our observations and previous reports suggest that PPARγ agonists might be effective in the future treatment of chronic kidney disease patients, especially in the context of heightened TGF-β 1 signaling.
As a limitation, we did not test the renal effect of direct STAT3 or EGR-1 inhibition in our model, eg. by using in vivo gene silencing approach. However, such expansion of our study would have increased the amount of animals needed, and would introduce a significant bias due to the intrarenal delivery problems of gene silencing oligonucleotides.

Additional file
Additional file 1: Figure S1. (a) Representative picture of genotyping the TGF-β transgenic mice (samples 3,4 and 6 are transgenic, showing the 370 bp PCR product of the transgene; samples 1,2 and 5 are wild type controls). (b) Levels of circulating TGF-β1 in wild type B6 control mice (CTL) and transgenic mice (TGFβ) at the beginning of the study clearly shows that transgenic mice had 3-fold elevated plasma TGF-β1 levels (n = 14/group, p < 0.001, Mann-Whitney test). (PDF 209 kb)