Dimethylfumarate Attenuates Renal Fibrosis via NF-E2-Related Factor 2-Mediated Inhibition of Transforming Growth Factor-β/Smad Signaling

Firstly, the effect of DMF on TGF-β-stimulated profibrotic genes and on ECM protein expression was examined in a normal rat renal fibroblast cell line (NRK-49F). As shown in Figure 1A–D, DMF inhibited TGF-β-stimulated PAI-1, α-SMA and fibronectin mRNA and protein expression in a dose-dependent manner. The expression of type I collagen mRNA and protein had not increased at 6 h after treatment with TGF-β (data not shown); however, its expression was induced at 24 h after TGF-β treatment, after which time DMF was seen to inhibit TGF-β-stimulated type I collagen expression in a dose-dependent manner (Figure 1B). The inhibitory effect of DMF on TGF-β-stimulated PAI-1 protein expression was confirmed further in RMCs, a rat mesangial cell line (Figure 1E). These data demonstrated that DMF inhibits TGF-β-stimulated profibrotic genes and ECM protein expression in rat kidney cell lines.

To determine whether DMF represses profibrotic genes and ECM protein expression through the inhibition of the TGF-β-activated Smad signaling pathway, we examined the effect of DMF on a luciferase reporter construct carrying the PAI-1 promoter (PAI-1-Luc), which contains three binding sites for Smad3. Its effect on (CAGA)₉MLP-Luc promoter activity, a reporter construct that contains nine copies of the Smad3 binding site, was also investigated. As shown in Figure 2A and B, DMF inhibited TGF-β-stimulated PAI-1 promoter activity in AD-293 cells, a derivative of the commonly used HEK-293 human embryonic kidney cell line. DMF also inhibited the activity of (CAGA)₉MLP-Luc promoter stimulated by either TGF-β or a constitutively active TGF-β type I receptor (ALK5).

Because phosphorylation of Smad3 is a critical step in ensuring the proper TGF-β signaling [33], we evaluated the effect of DMF on the phosphorylation of Smad3 (p-Smad3). As shown in Figure 2C–D, DMF treatment inhibited TGF-β-stimulated Smad3 phosphorylation in NRK-49F (Figure 2C) and RMC cells (Figure 2D). These data indicate that DMF negatively affects TGF-β-mediated transcription via the inhibition of Smad3 phosphorylation.

Since DMF is a well-known activator of the transcription factor Nrf2, which has been implicated in the pathogenesis of renal fibrosis, we investigated whether Nrf2 mediates the suppression of DMF on TGF-β-stimulated profibrotic genes and ECM protein expression. As expected, treatment with DMF resulted in a rapid increase in the protein expression of Nrf2 (Figure 3A). This increased expression of Nrf2 was maintained up to 24 h after co-treatment of DMF with TGF-β in NRK-49F cells (Figure S1), although its expression levels gradually decreased from 1 h after treatment of DMF. DMF also induced the nuclear accumulation of Nrf2 in a dose dependent manner (Figure S2). Because the p62-mediated stabilization of Nrf2 has recently been suggested as an antioxidant-independent mechanism for Nrf2 activation [34], we investigated the effect of DMF on p62 expression and the involvement of p62 in DMF-induced Nrf2 expression. Unlike rapid induction of Nrf2 expression by DMF, p62 expression was augmented at 6 h and further enhanced at 12 h after DMF treatment (Figure S3A), suggesting that DMF increases Nrf2 expression by a p62-independent mechanism. Consistently, a small interfering RNA (siRNA) against p62 (p62-siRNA) had little effect on DMF-increased Nrf2 expression, while it significantly diminished both basal and DMF-induced p62 expression in NRK-49F cells (Figure S3B). Moreover, down-regulation of p62 expression in AD-293 cells did not reverse the inhibitory effects of DMF on (CAGA)₉MLP-Luc activity and profibrotic gene expression stimulated by TGF-β (Figure S3C–E).

Next, we examined whether Nrf2 suppresses TGF-β-stimulated ECM expression. The results showed that adenovirus-mediated overexpression of Nrf2 (Ad-Nrf2) decreased PAI-1, α-SMA and fibronectin mRNA and protein expression in NRK-49F cells (Figure 3B and C). Ad-Nrf2 also decreased type I collagen mRNA expression at 24 h after TGF-β treatment (Figure 3D). Additionally, we confirmed that DMF causes a rapid increase in expression of the Nrf2 protein in RMC cells (Figure 3E) and that Ad-Nrf2 inhibits TGF-β-stimulated ECM expression in RMCs (Figure 3F). Moreover, transient transfection with Nrf2 decreased PAI-1 promoter activities stimulated by either TGF-β treatment or ALK5 cotransfection in a dose-dependent manner (Figure 3G). In contrast, promoter activity of NQO1, a well-known target gene of Nrf2 was markedly activated by transfection with Nrf2. These data demonstrate that Nrf2 activated by DMF suppresses the TGF-β-induced expression of profibrotic genes.

To determine whether Nrf2 inhibits TGF-β/Smad signaling, the effect of Ad-Nrf2 on TGF-β-stimulated Smad3 phosphorylation was investigated. The results revealed that Ad-Nrf2 inhibited TGF-β-stimulated Smad3 phosphorylation but had no effect on total Smad3 and Smad4 protein expression in NRK-49F cells (Figure 4A) and RMC cells (Figure 4B). To confirm further that the suppression of TGF-β/Smad3 activity and ECM protein expression by DMF is mediated by Nrf2, endogenous Nrf2 expression was down-regulated by transfecting AD-293 cells with a small interfering RNA (siRNA) against Nrf2 (Nrf2-siRNA). The Nrf2-siRNA successfully inhibited the expression of Nrf2 (Figure 4C) and significantly blocked the DMF-induced suppression of TGF-β-stimulated (CAGA)₉MLP-Luc promoter activity (Figure 4D). Moreover, the Nrf2-siRNA reversed the inhibitory effects of DMF on TGF-β-stimulated type 1 collagen expression (Figure 4E). These data suggest that Nrf2 mediates the inhibitory effect of DMF on TGF-β/Smad signaling pathway and TGF-β-stimulated ECM protein expression.

A growing body of evidence indicates that reactive oxygen species (ROS) mediate TGF-β-induced renal fibrosis, while Nrf2 target genes, such as NQO1 and HO-1, prevent ROS-induced renal fibrosis [29]–[31]. As expected, DMF increased NQO1 and HO-1 mRNA expression in NRK-49F cells (Figure 5A). Thus, we aimed to investigate the involvement of these antioxidant enzymes in the inhibition of TGF-β/Smad signaling and TGF-β-stimulated ECM protein expression by DMF. To determine whether the induction of NQO1 and HO-1 is required for the suppressive effect of DMF on the TGF-β/Smad signaling pathway, DMF-induced NQO-1 and HO-1 expression was knocked down by siRNAs against NQO-1 (NQO1-siRNA) or HO-1 (HO-1-siRNA) (Figure 5B). However, neither NQO1 nor HO-1 siRNAs abolished the inhibitory effects of DMF on the TGF-β-stimulated (CAGA)₉MLP-Luc promoter activity and ECM mRNA expression (Figure 5C and 5D). In a good agreement with knock-down experiments, ES936 or SnPP, chemical inhibitors of NQO1 or HO-1, respectively did not reverse the suppressive effects of DMF on TGF-β/Smad signaling and ECM expression (Figure S4). Furthermore, down-regulation of glutathione S-transferase (GST), one of other antioxidant response element (ARE)-dependent Nrf2 target genes using siRNA did not block the inhibition of TGF-β-stimulated expression of profibrotic genes by DMF (Figure S5). Collectively, these data indicated that the effects of DMF on TGF-β-stimulated Smad signaling and expression of profibrotic genes may not be mediated by the induction of ARE-driven Nrf2 target genes.

Finally, an experimental UUO-induced renal fibrosis model was employed to determine whether DMF attenuates renal fibrosis in vivo. As expected, the architecture of the tubules was dramatically altered in UUO kidneys, resulting in the development of interstitial fibrosis.

Expression of Nrf2 and NQO1 was markedly increased in the renal tubules and interstitial area in the DMF-treated UUO kidneys, compared with those of sham-operated or UUO control mice (Figure 6A). Sirius Red and trichrome staining showed, in comparison with UUO control kidneys, that UUO kidneys of DMF-treated mice exhibited a marked decrease in renal fibrosis seven days after UUO (Figure 6B). Immunohistochemical staining for type 1 collagen and α-SMA revealed that renal fibrosis in UUO kidneys of DMF-treated mice was less extensive than that in UUO control kidneys (Figure 6C). Quantitative real-time PCR results showed that UUO markedly increased mRNA levels of α-SMA, fibronectin and type 1 collagen. In contrast, the UUO kidneys of DMF-treated mice showed a significant decrease in expression of all three ECM proteins analyzed (Figure 6D). Moreover, DMF markedly reduced the UUO-induced nuclear staining of phosphorylated Smad3 in the renal tubular epithelial cells (Figure 6E). Collectively, these results indicated that DMF prevents the progression of UUO-induced renal fibrosis in mice.

Dimethylfumarate (DMF) was purchased from Sigma-Aldrich (St. Louis, MO). As previously described [33], recombinant human TGF-β1 and tin protoporphyrin IX (SnPP) were purchased from R&D Systems (Minneapolis, MN) and Frontier Scientific, Inc. (Logan, UT), respectively. ES936 was kindly donated by Mazence, Inc. (Suwon, Korea).

NRK-49F, a normal rat kidney fibroblast cell line, and RMC, a rat kidney mesangial cell line immortalized with the plasmid pSV3-Neo were purchased from the American Type Culture Collection (Manassas, VA). AD-293, a derivative of the commonly used HEK293 human embryonic kidney cell line, was purchased from Stratagene (La Jolla, CA). NRK-49F cells were cultured in Dulbecco's Modified Eagle's Medium (DMEM; Invitrogen, Carlsbad, CA) supplemented with 5% fetal bovine serum (FBS; Hyclone, Logan, UT) and antibiotics (Gibco, Grand Island, NY). RMC cells were cultured in DMEM (Invitrogen) supplemented with 15% FBS (Hyclone) and 0.4 mg/ml G418 (Gibco). AD-293 cells were cultured in DMEM (Invitrogen) supplemented with 10% FBS and antibiotics (Gibco).

As described previously [33], the reporter plasmids, (CAGA)₉MLP-Luc and PAI-1-Luc (−850 to +20) were kindly donated by Drs. Jean-Michel Gauthier (Laboratoire GlaxoWellcome, France) and Bernd R. Binder (University of Vienna, Austria), respectively. The mammalian expression plasmids pcDNA3HA-ALK5TD, which encoded the constitutively active activin-like receptor kinase 5 (ALK5), and pcDNA3-Nrf2 were kind gifts from Drs. Carl-Henrik Heldin (Ludwig Institute for Cancer Research, Sweden) and Mi-Kyoung Kwak (Yeungnam University, Korea), respectively. For the promoter assay, AD-293 cells were seeded on 24-well plates, incubated for 24 h, and transfected with 300 ng/well of reporter plasmids and mammalian expression plasmids encoding Nrf2 (100 or 300 ng/well) or ALK5 (50 ng/well) using the TransIT-LT1 transfection reagent (Mirus Bio Incorporation, Madison, WI). AD-293 cells were cotransfected with (CAGA)₉MLP-Luc and 50 nmol/l human Nrf2-siRNA, 10 nmol/l human NQO1-siRNA, 10 nmol/l human HO-1-siRNA and control siRNA duplexes (Table 1, Bioneer Corporation, Daejeon, Korea) using both TransIT-LT1 and TransIT-TKO transfection reagents (Mirus Bio Incorporation). Cytomegalovirus (CMV)-β-galactosidase plasmids were cotransfected as an internal control. Luciferase activity was normalized to that of β-galactosidase.

A full length mouse Nrf2 cDNA was inserted into the KpnI and XhoI sites of the pAdTrack-CMV shuttle vector. The recombinant adenoviral plasmid was generated as described previously [44], and recombinant adenoviruses were amplified in HEK-293 cells and subsequently purified.

For siRNA transfection, 10 nmol/l rat Nrf2-siRNA, 10 nmol/l rat NQO1-siRNA, 10 nmol/l rat HO-1-siRNA and control siRNA duplexes were purchased from the Bioneer Corporation (Table 1). Cells were seeded onto 60 mm plates and simultaneously transfected with Lipofectamine™ RNAiMax reagent (Invitrogen). After incubation for 24 h, cells were starved for 12 h, and then pretreated with DMF for 1 h, cells were stimulated with TGF-β. Total RNA was extracted using Trizol reagent (Invitrogen) according to the manufacturer's instructions, and semi-quantitative RT-PCR analysis was performed as described previously [44]. An aliquot (1 µg) of total RNA was reverse transcribed using the First Strand cDNA synthesis kit (Fermentas, EU) according to the manufacturer's protocol. The first strand cDNAs were amplified by PCR using gene-specific primers [33] to determine mRNA expression levels. Quantitative real-time PCR (qRT-PCR) was carried out using Power SYBR Green PCR Master Mix (Applied Biosystems, Warrington, UK) with the StepOnePlus Real-Time PCR System (Applied Biosystems). The expression levels of β-actin (semi-quantitative RT-PCR) and GAPDH (qRT-PCR) were used as internal controls.

Western blot analysis was performed as described previously [44] using specific primary antibodies [anti-p-Smad3 (Ser423/425); anti-Smad3 and anti-Smad4 (Cell Signaling, Beverly, MA); anti-type I collagen; anti-α-SMA and anti-β-actin (Abcam, Cambridge, MA); anti-Nrf2 (Santa Cruz Biotechnologies, Santa Cruz, CA); and anti-PAI-1 and anti-fibronectin (BD Biosciences, San Jose, CA). For Western blot analysis of type I collagen, cell lysates were subjected to native polyacrylamide gel electrophoresis (native PAGE: non-heating and non-denaturing conditions). Quantitative analysis of p-Smad3 to total Smad3 ratio was conducted with NIH Image J software.

All experimental procedures were carried out in accordance with the appropriate institutional guidelines for animal research. The protocol was approved by the Committee on the Ethics of Animal Experiments of the Keimyung University School of Medicine (Permit Number: KM-2010–29).

Eight-week-old male C57BL/6 mice (Hyochang Science, Daegu, Korea) were divided into three groups (sham-operated control, UUO mice and UUO mice treated with 25 mg/kg of DMF, n = 5/group) for each experimental regime. For UUO surgery, the animals were anesthetized with pentobarbital (50 mg/kg) and the left ureter was doubly ligated. Sham animals underwent the same surgical procedures except for ligation. Following UUO, mice in the DMF-treated group were dosed with DMF (25 mg/kg/day) by gavage daily for seven days. All mice were sacrificed seven days after UUO surgery.

Kidney tissues were fixed with 4% paraformaldehyde (PFA, Sigma-Aldrich) and paraffin embedded. Kidney sections (4 mm) were deparaffinized in xylene, rehydrated through descending grades of ethanol and subjected to immunohistochemical analysis using antibodies against Nrf2, NQO1 (Santa Cruz), type I collagen, α-SMA and p-Smad3 [33], [45]. To evaluate renal fibrosis, sections were stained with Sirius Red (Sigma-Aldrich) and trichrome (Sigma-Aldrich) according to the manufacturer's instructions.

Data are expressed as means ±SEM. Statistical analyses were performed using an unpaired Student's t-test and a value of P<0.05 was considered statistically significant.