Therapeutic Potential of Nrf2 Activators in Streptozotocin-Induced Diabetic Nephropathy

CA, STZ, and tert-butylhydroquinone (tBHQ), Periodic Acid-Schiff (PAS) kit, and Trichrome stain (Masson) kit were purchased from Sigma (St. Louis, MO). Sulforaphane (SF, l-isomer) was obtained from LKT laboratories (St. Paul, MN). Blood glucose levels were measured using AlphaTRAK system (Abbott Laboratory, North Chicago, IL). Insulin was detected using mouse insulin ELISA kit (Millipore, St. Charles, MO). Urine was collected for analysis in metabolic cages (Hazleton Systems Inc., Aberdeen, MD). Urinary albumin or urinary creatinine was measured by ELISA (Bethyl Laboratories, Houston, TX; Exocell, Philadelphia, PA). Antibodies for immunoblot and immunohistochemical (IHC) analysis included anti-7,8-dihydro-8-oxo-2’-deoxyguanosine (8-oxo-dG) (Trevigen, Gaithersburg, MD); anti-fibronectin (FN) (Calbiochem, San Diego, CA); anti-collagen IV (Abcam, San Francisco, CA); anti-Ki67 (Vector Laboratories, Burlingame, CA); and anti-Nrf2, NAD(P)H dehydrogenase, quinone 1 (NQO1), γ-glutamylcysteine synthetase (γ-GCS), p21, TGF-β1, and β-actin (Santa Cruz Biotechnology, Santa Cruz, CA).

Nrf2^+/+ and Nrf2^−/− C57BL/6 mice (described previously in 25) were obtained through breeding of Nrf2^+/−. All animals received water and food ad libitum. Eight-week-old mice received either sodium citrate (control) or STZ (50 mg/kg, pH 4.5, dissolved in sodium citrate) through intraperitoneal (i.p.) injection for 5 consecutive days. Two weeks following STZ injection, fasting glucose levels (4-h fast) were measured, and mice with a fasting glucose level above 250 mg/dL were considered diabetic and used for this study. Mice were randomly allocated into four groups (n = 8 per group) to receive treatment: 1) control = corn oil, 2) CA1 = 25 mg/kg CA, 3) CA2 = 50 mg/kg CA, 4) SF = 12.5 mg/kg SF. Each of the compounds was administered (i.p., diluted in corn oil) three times per week for 16 weeks. Doses of SF and CA were guided by published literature (14–16) and tested in pilot studies to ensure Nrf2 activation within 48 h after i.p. injection. At 18 weeks, mice were killed and the kidney, blood, and urine were isolated for analysis.

Tissue sections from paraffin-embedded kidney were stained with hematoxylin-eosin (HE), periodic acid Schiff (PAS), and Masson’s trichrome. Glomerulosclerotic index was assessed on PAS-stained tissue sections using a five-grade method described previously (11). The thickness of the glomerular basement membrane (GBM) was measured by electron microscopy (Arizona Health Sciences Center, Imaging Core, University of Arizona, Tucson, AZ). Images were captured by CCD camera system (Advanced Microscopy Techniques, Woburn, MA), and thickness was measured using the ruler function of the software. Average thickness for each mouse was calculated from 50 measurements obtained from different sites of GBM.

IHC analysis was performed as described previously (26). Oxidative DNA damage was measured by the amount of 8-oxo-dG, using two methods: 1) deparaffinized kidney sections were treated with proteinase K (10 μg/mL), RNase A (100 μg/mL), and 2N HCl then stained by IHC with an anti-8-oxo-dG antibody and 2) urinary 8-oxo-dG was measured by sampling 50 μl urine against standards generated from 0.7067 mmol/L stock using liquid chromatography-mass spectometry/mass spectormetry (Arizona Cancer Center Proteomics Core) according to published protocols (27,28).

HRMCs were purchased and maintained according to distributor guidelines from ScienCell Research Laboratories (Carlsbad, CA). For experiments, HRMCs were starved in serum- and supplement-free Dulbecco’s modified Eagle’s medium (DMEM) containing normal glucose (NG, 5.5 mmol/L) for 24 h. Cells were then either maintained in serum-free NG DMEM (plus 19.5 mmol/L mannitol to control for osmolality) or switched to serum-free high glucose (HG, 25 mmol/L) DMEM supplied with supplement, along with treatment (SF [1.25 μmol/L, empirically determined]), or CA (5 μmol/L [29]), or tBHQ (6.25 μmol/L, empirically determined). Cells were cultured in HG with or without treatments for 48 or 96 h. Reactive oxygen species (ROS) in cultured HRMCs were detected using dichlorodihydrofluorescein diacetate (Invitrogen, Carlsbad, CA), and the fluorescence intensity was measured by flow cytometry. Small interfering RNAs (siRNAs) specific for Nrf2, Keap1, or p21, as well as HiPerfect transfection reagent were purchased from Qiagen (Valencia, CA), and transfection of siRNAs was performed according to manufacturer’s instructions.

Tissue lysates were prepared as previously described (11). Protein concentration was determined using Quant-IT Protein Assay (Invitrogen). Cultured cells were lysed in sample buffer (50 mmol/L Tris–HCl [pH 6.8], 2% SDS, 10% glycerol, 100 mmol/L dithiothreitol, 0.1% bromophenol blue). Total lysates were loaded and electrophoresed through SDS–polyacrylamide gels and subjected to immunoblot analysis. The rate of cell proliferation was measured two ways: 1) 8,000 HRMCs per well were seeded in NG or HG media with or without Nrf2 activator and cell growth was monitored for 96 h using the xCELLigence system (Roche) (30) and 2) detection of Ki67 using indirect immunofluorescence as described previously (31). The In Situ Cell Death Detection Kit (Roche) was used according to manufacturer’s instructions to detect apoptosis. Cell size was measured by forward light scatter using flow cytometry and the average area per cell in green fluorescent protein (GFP)-transfected living cells as described below. Cells were transfected with GFP and maintained in NG or HG media with or without an Nrf2 activator. Fluorescent images were taken using a Zeiss Observer.Z1 microscope. Cell size was calculated using the Slidebook 4.2.0.11 software (Intelligent Imaging Innovations, Inc., Denver, CO).

Results are expressed as mean ± SD. Statistical tests were performed using SPSS 10.0. Unpaired Student t tests were used to compare the means of two groups. One-way ANOVA was applied to compare the means of three or more groups. P < 0.05 was considered to be significant.

The effectiveness of dietary Nrf2 activators in alleviating metabolic disorder was assessed in the STZ-induced diabetic model in both Nrf2^+/+ and Nrf2^−/− mice, and common metabolic disorder indices associated with diabetes were measured. In response to STZ, mice showed significantly increased blood glucose levels (Fig. 1A), reduced weight gain (Fig. 1B), increased urine production (Fig. 1C), and increased water uptake (Fig. 1D). Importantly, treatment of diabetic animals with SF or CA significantly alleviated all indices of metabolic dysfunction in Nrf2^+/+ animals only, demonstrating the Nrf2-dependent response to SF and CA. In addition, STZ injection decreased insulin levels significantly, which was unaltered by treatment with SF or CA regardless of genotype (Fig. 1E), indicating that the protective effects of SF and CA were not insulin dependent. Since treatment with Nrf2 activators began 2-weeks post-STZ injection, after the pancreatic β-cells had been destroyed and diabetes was established, we did not expect SF or CA treatments to influence circulating insulin levels in this study. Taken together, these results indicate that SF or CA treatment significantly attenuated metabolic disorder induced by STZ, and that this protective effect was independent of circulating insulin levels.

To investigate the therapeutic effects of Nrf2 activation on improvement of kidney performance in the STZ diabetic model, functional and pathological changes of the kidney were measured. The ratio of kidney weight-to-body weight was higher in all STZ-injected groups compared with untreated, which was significantly attenuated by treatment with SF or CA in Nrf2^+/+ mice (Fig. 2A). Next, as indices of albuminuria, both urinary albumin excretion (UAE) and the urine albumin-to-creatinine ratio (UACR) were determined at 0, 10, and 18 weeks. Injection with STZ increased UAE and UACR in all treated groups; however, this increase was blunted by SF or CA treatment in Nrf2^+/+ mice but not Nrf2^−/− mice (Fig. 2B and C). These results demonstrate that SF or CA is able to improve albuminuria in Nrf2^+/+ diabetic mice.

In agreement with the urinalysis, SF or CA treatment alleviated the pathological alterations of the glomerulus as revealed by histological examination. Glomerular lesions, including K-W (Kimmelstiel Wilson) nodules, were observed in HE-stained tissue sections from STZ-injected mice (Fig. 2D, HE panel, arrows). Treatment with SF or CA2 effectively restored the normal morphology of glomeruli in Nrf2^+/+ but not Nrf2^−/− mice (Fig. 2D, HE panel). Glycogen deposition in the glomeruli was measured by PAS staining and scored to indicate the severity of glomerulosclerosis. PAS staining showed STZ-induced glomerulosclerosis, which was significantly improved in SF- or CA2-treated Nrf2^+/+ but not Nrf2^−/− mice (Fig. 2D, PAS panel, and Fig. 2E). Since ECM deposition is another hallmark of glomerulosclerosis, collagen was visualized using Masson’s trichrome staining. STZ treatment resulted in collagen accumulation inside glomeruli or in the periglomerular area, which was reduced in SF- or CA-treated Nrf2^+/+ but not Nrf2^−/− groups (Fig. 2D, trichrome panel). As an independent criterion for glomerulosclerosis, thickening of the GBM was measured by electron microscopy (Fig. 2F). STZ injection increased thickness of the GBM, which was reverted to control by SF or CA treatment only in Nrf2^+/+ mice (Fig. 2F). These results clearly demonstrate the ability of SF and CA treatment to attenuate STZ-induced pathological alterations both in renal function and structure during the progression of diabetic nephropathy.

Next, activation of Nrf2, its downstream targets, and oxidative damage in the kidney were assessed to demonstrate that the beneficial effect of SF or CA against renal damage is attributed to activation of the Nrf2 pathway and thus reduction of oxidative damage. Expression of Nrf2 and Nrf2 downstream targets NQO1 and γ-GCS was slightly increased after STZ injection, indicating induction of the Nrf2 pathway by renal oxidative stress. As expected, treatment with SF and CA markedly increased the protein levels of Nrf2, NQO1, and γ-GCS (Fig. 3A and B). Importantly, SF or CA treatment significantly reduced oxidative damage in the Nrf2^+/+ kidney, as measured by the formation of 8-oxo-dG either locally in glomerular tissues or systemically from the urine (Fig. 3C and D). Treatment with SF or CA did not increase NQO1 or γ-GCS or reduce oxidative damage in Nrf2^−/− mice. Collectively, these results indicate that the kidney of STZ-induced diabetic mice is under oxidative stress, and that the renal protection mediated by SF or CA in the diabetic model is, at least in part, derived from specific activation of the Nrf2-mediated antioxidant response and reduction of oxidative damage.

Given that treatment with SF or CA diminished renal disorders resulting from STZ-induced diabetes, the molecular mechanisms underlying Nrf2-dependent protection of the kidney by SF or CA were explored. Previously, our laboratory reported a negative association between Nrf2 and TGF-β1. In accordance with that report, the basal expression of TGF-β1 was higher in Nrf2^−/− mice compared with Nrf2^+/+ mice (Fig. 4A and B). Diabetes induced by STZ significantly upregulated protein levels of TGF-β1 and its downstream effectors, FN, collagen IV, and p21, which were significantly abrogated by treatment with SF or CA in the Nrf2^+/+ animals only (Fig. 4A and B). Treatment with SF and CA in the Nrf2^−/− mice did not alter TGF-β1, ECM, or p21, indicating that the negative regulation of TGF-β1 and subsequent downstream effectors (ECM and p21) observed with SF and CA treatment is Nrf2 dependent.

Mesangial cells play a crucial role in dictating the function of glomeruli. To understand the molecular mechanism(s) by which activation of Nrf2 is able to preserve renal function during the progression of diabetic nephropathy, cultured primary HRMCs were used. HRMCs growing in NG (5.5 mmol/L) media were shifted to either NG (plus 19.5 mmol/L mannitol) or HG (25 mmol/L) media to mimic hyperglycemic conditions in the presence or absence of an Nrf2 activator. Similar to the results reported above in the animal model of STZ-induced diabetes, culturing HRMCs under hyperglycemic conditions also activated the Nrf2 pathway (Fig. 5A and B), presumably through oxidative stress.

To support the in vivo findings reported above, HRMCs grown in HG media were treated with SF, CA, and tBHQ, a well-characterized Nrf2 activator. As expected, higher Nrf2 expression and predominate nuclear localization of Nrf2 were observed in response to tBHQ, SF, or CA treatment, indicating activation of the Nrf2 pathway (Fig. 5A). Consistently, protein levels of Nrf2 were enhanced after 48-h incubation in HG media and further induced by tBHQ, SF, or CA treatment (Fig. 5B). Similarly, protein levels of Nrf2 downstream targets including NQO1, GCS, or Mrp2, were increased in response to HG and further enhanced by treatment with an Nrf2 activator (Fig. 5B). To test if HRMCs growing in HG media were under oxidative stress, the level of ROS was measured. Indeed, HRMCs growing in HG media had ROS levels approximately 3.5-fold higher compared with HRMCs growing in NG media (Fig. 5C). As predicted, activation of Nrf2 by tBHQ, SF, or CA reduced ROS levels significantly (Fig. 5C). These results demonstrate ROS generation as a likely source of mesangial cell damage in diabetes and further demonstrate the beneficial effects of Nrf2 pathway activation in combating oxidative stress.

As a chronic disease, diabetic nephropathy is characterized by sequential pathological changes including cell growth inhibition and glomerular hypertrophy. First, we measured cell growth rate of HRMCs in differential glucose media using the xCELLigence system. As expected, the growth curve for HRMCs growing in HG media fell below those either growing in NG or in HG and supplemented with an Nrf2 activator (Fig. 6A). No cell death was observed under any condition and staining with Ki67 showed that hyperglycemia inhibited cell proliferation, which was counteracted by activation of Nrf2 (Fig. 6B). Next, the size of HRMCs under NG and HG conditions were measured by two independent methods. First, forward light scattering analysis by flow cytometry revealed a dramatic shift in size distribution (Fig. 6C), indicating an increase in the size of HRMCs growing in HG compared with NG media. Conversely, the light scattering curves for HRMCs growing in HG media in the presence of an Nrf2 activator were completely shifted back to the left where they overlapped with the curve of NG cells (Fig. 6C). Second, HRMCs were transfected with GFP to mark the entire area of cells. Incubation of HRMCs in HG media resulted in an increase in cell size as measured by the average area per GFP⁺ cell. The hypertrophy observed in HG media was suppressed by treatment with tBHQ, SF, or CA (Fig. 6D). Collectively, these results demonstrate that hyperglycemia induced cell growth inhibition, and hypertrophy can be attenuated by activation of the Nrf2 pathway.

To gain molecular understanding of how Nrf2 activation relieves HRMC hypertrophy and growth inhibition under hyperglycemic conditions, the expression of TGF-β1 and its downstream effectors were analyzed. Consistent with our in vivo data, hyperglycemia upregulated TGF-β1, FN, collagen IV, and p21 in HRMCs, which was suppressed by treatment with tBHQ, SF, or CA (Fig. 7A). Next, the functional contribution of the cell cycle regulator, p21, in mesangial cell growth inhibition and hypertrophy was investigated. Knockdown of p21 expression by transient transfection with p21-siRNA reduced the protein level of p21 and recovered the growth rate of HRMCs in HG media to that of NG conditions (Fig. 7B), while control siRNA had no effects on HRMC growth (Fig. 7B). Furthermore, hypertrophy of HRMCs growing in HG was no longer observed when expression of p21 was reduced by siRNA (Fig. 7C), indicating p21 controls mesangial growth inhibition and hypertrophy.

To further confirm the negative regulation of Nrf2 on TGF-β1 and its downstream response, endogenous Nrf2 levels were modulated either by Nrf2-siRNA or by siRNA to Keap1, a negative regulator of Nrf2. As expected, Nrf2 levels were reduced by Nrf2-siRNA and increased by Keap1-siRNA (Fig. 7D). The resulting effects of Nrf2 modulation display a negative Nrf2 regulation of TGF-β1 and its downstream effectors including FN, collagen IV, and p21. Specifically, downregulation of Nrf2 resulted in enhanced expression of these proteins, whereas upregulation of Nrf2 reduced expression of the TGF-β1 pathway under both NG and HG media (Fig. 7D). Taken together, these results suggest that Nrf2 activation is able to confer renal protection in diabetic conditions through blockage of the TGF-β1 downstream response, specifically by reducing ECM overproduction and p21-mediated growth inhibition and hypertrophy.

In the current study, the therapeutic potential of dietary Nrf2 activators in diabetic nephropathy is clearly demonstrated both in vivo and in vitro. Treatment with Nrf2 activators, SF, or CA, significantly reduced common metabolic disorder indices associated with diabetes, including hyperglycemia, polydipsia, polyuria, and weight loss. In the kidney, SF or CA treatment retarded pathological characteristics of diabetic nephropathy including oxidative damage, albuminuria, renal hypertrophy, ECM deposition, and thickening of basement membranes. All these effects were only observed in Nrf2^+/+ mice, indicating that the beneficial effects of SF or CA is derived from specific activation of the Nrf2 pathway. Furthermore, treatment with CA, SF, and tBHQ is able to protect mesangial cells from high glucose–induced alterations such as ROS and ECM production, cell growth inhibition, and hypertrophy in vitro, which is likely through Nrf2’s negative regulation of TGF-β1 and p21.

In addition to demonstrating the therapeutic potential of SF or CA in suppressing diabetic nephropathy pathogenesis, we also sought the mechanism by which Nrf2 activation by these compounds is able to provide beneficial effects. One contributing factor is possibly the mild reduction of blood glucose in SF- or CA-treated Nrf2^+/+ mice. Importantly, Nrf2 activation had no effects on insulin levels, indicating that the observed reduction of blood glucose is insulin independent. Recently it was reported that gluconeogenesis and glycolysis are altered in Nrf2^−/− mice (32) suggesting that the reduction of blood glucose in our SF- or CA-treated Nrf2^+/+ mice may be partially due to Nrf2-mediated glucose metabolism and glycogen formation in their livers. Alternatively, administration of SF or CA may affect insulin sensitivity in peripheral tissues, thereby altering systemic insulin resistance. Literature supports the notion of at least CA influencing glucose uptake in fibroblasts (33); and Nrf2 was shown to be central in the acquisition of insulin resistance induced by oxidative stress in cardiomyocytes (34). Future studies should include closer inspection of the effects of SF and CA on the major peripheral tissues contributing to insulin resistance in diabetes (i.e., skeletal muscle and adipocytes).

Another contributing factor to Nrf2 protection in diabetic nephropathy might be the Nrf2-dependent antioxidant response. The role of oxidative stress in the pathogenesis of diabetic nephropathy has gained increasing attention in the field (3,35). Previously, we detected oxidative damage in the glomeruli of both human diabetic nephropathy patients and STZ-injected mice (11). In the current study, SF or CA treatment significantly reduced oxidative damage in the glomerular tissues and urine of STZ-treated Nrf2^+/+ mice. Collectively, these results indicate the importance of the Nrf2-dependent antioxidant response in neutralizing ROS and alleviating oxidative damage in the kidney.

Nrf2-mediated protection in diabetic nephropathy may also rely on the negative regulation of Nrf2 on TGF-β1, a major profibrotic mediator of diabetic nephropathy. We have previously shown a negative association between Nrf2 and TGF-β1 promoter activity (11), however the precise mechanism remains unknown. Here, Nrf2 activation by SF or CA decreased protein levels of TGF-β1 in the kidney of diabetic Nrf2^+/+ mice and in HRMCs growing in HG media in vitro. Subsequently, we found that the beneficial effects of Nrf2 activation through TGF-β1 downregulation are due to suppression of both ECM production and p21 expression. ECM overproduction is a hallmark of diabetic nephropathy, and the function of TGF-β1 signaling ECM production is well established (rev. in (36,37). Alterations in the rate of proliferation and enlargement of renal mesangial cells are other typical features in early diabetic nephropathy, and upregulation of p21 by high glucose or other factors is associated with mesangial hypertrophy in vitro and in vivo (38–41). A critical role of p21 in mesangial hypertrophy was demonstrated by the fact that diabetic p21^−/− mice do not develop glomerular hypertrophy (39,42). Interestingly, these p21^−/− mice exhibited no alterations in their histology or function of the kidney, providing evidence that chronic decrease in p21 through Nrf2 activation may not be detrimental in long-term therapies.

Recently, several attempts have been made at using Nrf2 activators as treatment for diabetes. Specific to the kidney, treatment with bardoxolone of type 2 diabetes patients having chronic kidney disease elicited significant improvements in the estimated glomerular filtration rate with mild to moderate side effects (20,43). Additionally, activation of Nrf2 with SF in cultured microvascular endothelial cells protected them from the biochemical dysfunction induced by hyperglycemia (21). Alternate therapies that addressed triosephosphate accumulation resulting from hyperglycemia also showed that high doses of thiamine reduced albuminuria, ROS, and dyslipidemia without reducing circulating glucose levels after STZ-induced diabetes (44,45). Indeed, diabetic nephropathy is a multifaceted disease with several contributing factors that are both systemic and local to the kidney, making combination therapy or the discovery of an equally versatile treatment essential.

Collectively, our findings indicate that the therapeutic benefit of Nrf2 activation in diabetic nephropathy is multifactorial. In addition to its antioxidant function, Nrf2 also negatively regulates TGF-β1, ECM, and p21 expression. The reported results provide convincing experimental evidence that dietary compounds targeting Nrf2 activation can be used therapeutically to improve metabolic disorder and relieve kidney damage induced by diabetes. This study lays the foundation for clinical evaluation and ultimately the development of new Nrf2 activators in therapeutic use to prevent the onset and progression of diabetic nephropathy.

This study was supported by grants ES015010 (National Institutes of Health) and RSG-07-154 (American Cancer Society) (to D.D.Z.) and National Natural Science Foundation of China (number 30700384) (to H.Z.).

No potential conflicts of interest relevant to this article were reported.

H.Z. and W.W. researched data. H.Z., S.A.W., and D.D.Z. wrote the manuscript. G.T.W., P.K.W., and D.F. contributed to discussion and provided critical reagents.

The authors would like to thank Tony Day for his contributions to electron microscopy services and Wade Chew and the Arizona Cancer Center Analytical Core Shared Service for assistance with urinary analysis.