The increasing incidence of secondary complications associated with diabetes highlights the need to improve existing treatment regimens and preventive measures. Current data indicate that approx. 43% of new cases of ESRDs (end-stage renal diseases) are due to diabetes, suggesting the need for novel therapies to address diabetes-induced kidney damage [¹]. Diabetic nephropathy is characterized by decreased glomerular filtration rate, proteinuria, mesangial expansion, tubulointerstitial fibrosis and glomerulosclerosis [²,³].

Both Type 1 and Type 2 diabetes mellitus have complex pathophysiologies, including insulin resistance syndrome and hyperglycaemia, which are associated with abnormalities in fat oxidation and increased formation of ROS (reactive oxygen species) and RNS (reactive nitrogen species) [⁴,⁵]. Kidney mitochondria from STZ (streptozotocin)-induced diabetic rats show increased superoxide production and post-translational modification of mitochondrial complex III [⁶,⁷]. The production of ROS from the mitochondria has received a great deal of attention, and it is now becoming clear it may be regulated under physiological conditions and play an important role in redox cell signalling. For example, it has been shown that mitochondrial ROS produced at complex III can activate proliferative signalling pathways in cancer cells [⁸,⁹]. ROS-mediated mechanisms have also been reported to lead to activation of matrix metalloproteinases 2 and 9 via a decrease in Mn-SOD (manganese superoxide dismutase) activity [¹⁰]. Mitochondrial ROS production occurs at several sites in the respiratory chain, including complexes I and III. These major sites of ROS production both involve the low-molecular-mass redox shuttle CoQ (coenzyme Q). Decreased levels of CoQ are associated with increased ROS production and it is reasonable to suggest that such conditions will result in dysregulation of mitochondrial-dependent redox signalling [¹¹]. Interestingly, decreased levels of CoQ in the kidneys have been reported in the rat model of STZ-induced diabetes [¹²]. As has been shown previously, the CoQ antioxidant system can be supplemented with a mitochondria-targeted analogue MitoQ (mitoubiquinone) [¹³]. This molecule is orally bioavailable, accumulates in many tissues in the body, including the kidney, and can protect organs from insults such as ischaemia/reperfusion, endothelial cell dysfunction, cardiac hypertrophy and sepsis [¹⁴–¹⁹]. We therefore hypothesized that treatment with MitoQ could protect mitochondrial redox signalling from hyperglycaemia-induced alterations and this would ameliorate diabetic nephropathy.

To test the potential role of mitochondria in diabetic nephropathy we used the genetic model of Type 1 diabetes, the Akita mouse (Ins2^+/−AkitaJ mice). Akita mice develop hyperglycaemia because the A chain of mature insulin is mutated, resulting in misfolding of the insulin protein [²⁰]. Symptoms in the heterozygous mutant mice include hyperglycaemia, hypoinsulinaemia, polydipsia and polyuria, beginning at around 3–4 weeks of age. Akita mice undergo progressive organ dysfunction in the kidney and retina after the onset of hyperglycaemia, exhibiting increased proteinuria, vascular permeability, increased apoptosis and alterations in the retina [²¹–²³]. Mitochondria in this model show morphological changes, up-regulation of some tricarboxylic acid cycle enzymes including aconitase, but no major changes in the respiratory chain complexes or respiratory function [²⁴,²⁵]. In the present study we found increased proteinuria and tubulointerstitial fibrosis in the Akita mice at 26 weeks of age, providing a relevant model for testing the protective properties of MitoQ on mitochondrial redox signalling.

Diabetic kidney disease is a chronic pathological condition that has a limited number of effective therapies [³⁶,³⁷]. Indirect approaches of strict glycaemic control and maintenance of normal blood pressure delay the progression of the disease [³⁶,³⁷]. Despite these measures, a significant number of cases progress to diabetic nephropathy, which is characterized histologically by tubulointerstitial fibrosis, diffuse or nodular glomerulosclerosis and variable degrees of hyaline arteriolosclerosis. Mechanisms including changes in glomerular haemodynamics due to altered vascular resistance, inflammation and augmented formation of AGEs (advanced glycation end-products) have been suggested in the pathogenesis [³⁵]. Increased synthesis of TGFβ and an abnormal renin–angiotensin system have also been proposed to be involved in the progression of diabetic nephropathy [³⁶]. Changes in the β-catenin signalling pathway are now also thought to play a pro-fibrotic role in a number of pathologies including diabetes [³⁸]. Several studies have shown that altered mitochondrial function occurs in the development of diabetic kidney disease, but the contribution to the pathophysiological changes, particularly fibrosis, associated with diabetes is not clear [³⁹].

In vitro and in vivo studies have shown that overexpression of mitochondrial antioxidant enzymes in renal proximal tubular cells and in the retina protect against oxidant-induced damage in experimental hyperglycaemia [⁴⁰,⁴¹]. Evidence from the cancer and cardiovascular fields suggest that oxidant production from the mitochondria appears to play a role in the cell signalling associated with tissue remodelling, proliferation and angiogenesis [⁹,⁴²]. For these reasons we hypothesized that a mitochondrially targeted antioxidant that can modulate the production of ROS from enzymes associated with CoQ may have beneficial effects in diabetic nephropathy. Of the potential compounds available the ubiquinone analogue MitoQ has attracted particular interest since it is orally bioavailable, has low toxicity and reaches concentrations of 200–700 pmol per g of wet weight in the tissue of a number of organs, including the kidney [⁴³]. For example, in animal models MitoQ treatment has been shown to prevent ischaemia/reperfusion-induced cardiac dysfunction [¹⁷,⁴⁴], decrease inflammation in a LPS (lipopolysaccharide)–peptidoglycan-induced rat sepsis model [¹⁴] and to protect the heart against doxorubicin toxicity [¹⁵]. It has been shown that MitoQ treatment was also protective against endothelial cell dysfunction in vitro and in vivo [¹⁹,⁴⁵]. Importantly, MitoQ has been used in Phase I and Phase II clinical trials for Parkinson's disease and hepatitis-C-induced liver damage, and is well-tolerated in human and animal subjects [⁴⁶,⁴⁷].

In the present study the Ins2^+/−AkitaJ mice were selected as a model of Type 1 diabetes to test the potential benefits of MitoQ therapy. This model develops insulin resistance over time and has many of the characteristics of chronic hyperglycaemia associated with both Type 1 and 2 diabetes [²²,²⁵]. Ins2^+/−AkitaJ mice have been used in several studies and are reported to have increased oxidative stress in the kidney [³⁰,⁴⁸] and abnormal mitochondrial morphology in the diabetic kidney, but no change in volume density and number [²⁴]. Decreased levels of SOD1 and SOD3, but not SOD2 exacerbate diabetic nephropathy in these mice and the disease progression was prevented using the SOD mimetic tempol, supporting a role for the dysregulation of ROS in diabetic nephropathy [⁴⁹]. Recent studies have reported the attenuation of diabetic kidney disease using recombinant angiotensin-converting enzyme treatment [⁵⁰] and cyclo-oxygenase 2 inhibitor treatment, which are both pathways known to generate increased levels of ROS [⁵¹].

Oral MitoQ treatment was sustained over a period of 11–12 weeks in wild-type control and Ins2^+/−AkitaJ mice, after which a number of end-points relevant to diabetes were determined. The pharmacological delivery of MitoQ was adjusted based on the water intake of the animals for a body-weight-independent dose, due to substantially increased water consumption by the Ins2^+/−AkitaJ mice. However, when corrected for body weight the control mice were heavier, resulting in a slightly lower dose of MitoQ for the controls. MitoQ treatment resulted in lower HbA1c levels in the Ins2^+/−AkitaJ mice, however the changes were modest and unlikely to account for the prevention of renal pathology with MitoQ treatment.

Histochemical staining of the kidney tissue sections revealed mesangial expansion and tubulointerstitial fibrosis. MitoQ treatment significantly prevented these tissue pathologies associated with diabetic nephropathy. Assessment of kidney function required the use of sensitive non-toxic and minimally invasive techniques, and for this we used 99mTc–MAG3 and 99mTc–DTPA [³²]. Increased clearance of these tracers was evident for MitoQ-treated Ins2^+/−AkitaJ mice, consistent with MitoQ-mediated protection of renal tubular and glomerular function. As demonstrated in the present study, and by others [³⁴], Ins2^+/−AkitaJ mice have pronounced GBM thickening in the peripheral capillary loops, which was prevented by MitoQ (Figure 6). Prevention of GBM thickening with MitoQ treatment is consistent with the improvement in glomerular clearance function as assessed with 99mTc–DTPA. Although plasma creatinine remained unchanged, the decreased urine albumin in Ins2^+/−AkitaJ mice with MitoQ treatment also indicates that renal function was improved. This can be explained on the basis of reports in animal models of nephropathy that proteinuria is a more sensitive marker of kidney disease compared with plasma creatinine measurements [⁵²]. Ins2^+/−AkitaJ mice demonstrated mesangial expansion, tubulointerstitial fibrosis and decreased renal function, which closely reflects human diabetic nephropathy [³⁵]. The decreased mesangial expansion, interstitial fibrosis and GBM thickening observed in the MitoQ-treated Ins2^+/−AkitaJ animals compared with the untreated diabetic animals suggests that the MitoQ treatment modulates the profibrotic signalling pathways in the diabetic kidney.

Several pathophysiological characteristics of diabetes including hyperglycaemia, hypertension, mechanical stress and ROS production can activate the expression of the profibrotic cytokine TGFβ. The signalling mediated by TGFβ occurs by phosphorylation and nuclear translocation of Smad2/3, resulting in the synthesis of proteins involved in extracellular matrix accumulation [⁵³]. Several studies have demonstrated the stimulation of TGFβ-mediated signalling results in increased production and accumulation of extracellular matrix proteins, and activation of protease inhibitors. In addition, TGFβ induces its own synthesis and stimulates the proliferation of tubulointerstitial fibroblasts, resulting in the amplification of the fibrotic response. Demonstration of tubulointerstitial fibrosis (Figure 7), and increased phosphorylation and nuclear translocation of Smad2/3 (Figure 8A and 8B), confirms the activation of TGFβ-mediated profibrotic signalling in the diabetic animals. Interestingly, MitoQ inhibited the nuclear translocation of phospho-Smad2/3. In addition, the Wnt/β-catenin signalling has been shown to play a crucial role as a profibrotic signal in a number of pathologies, and recently has been implicated in diabetes-induced retinal fibrosis [³⁸]. The majority of Wnt/β-catenin studies have been in the fields of embryogenesis and cancer, and it has shown to have an important role in the EMT (epithelial-to-mesenchymal transition) [⁵⁴]. A number of studies have implicated EMT as a mechanism of fibrotic deposition in a range of insults including diabetes [⁵⁴]. Wnt/β-catenin has been shown to exert different effects in the kidney, e.g. protective effects on mesangial cells and yet profibrotic effects in tubular cells [⁵⁵,⁵⁶]. The Wnt/β-catenin pathway is redox-sensitive and is therefore potentially sensitive to mitochondrial oxidants [⁵⁷]. In support of this argument we found that MitoQ suppressed β-catenin translocation to the nucleus (Figure 9).

Since MitoQ was administered orally it is possible it may exert its effects in several different cell types and for this reason the mechanism(s) of action will require further investigation. There are, however, several possibilities which can be suggested on the basis of previous studies with this compound. In isolated mitochondria it has been shown that MitoQ can be converted into mitoubiquinol by complex II, but is not readily oxidized by complex III or reduced by complex I [¹³]. This suggests it would potentially change the redox balance and change mitochondrial redox signalling, since this is an important site of superoxide and hydrogen peroxide formation in the mitochondrion. Since mitochondrial respiratory chain proteins are not significantly altered in the kidney mitochondria from the Akita mice [²⁴], and in the present study MitoQ treatment did not alter the respiratory capacity of mitochondria, it is unlikely that protection of these proteins is important in the mechanism leading to prevention of diabetic nephropathy.

In conclusion, we report that the Type 1 diabetic model of Ins2^+/−AkitaJ mice display both renal tubular and glomerular defects associated with diabetes, which are attenuated by oral administration of MitoQ. Histological evidence suggests that the improvement of renal function in MitoQ-treated Ins2^+/−AkitaJ mice correlates with attenuation of tubulointerstitial fibrosis. Diabetes-induced alterations in the cellular redox status could have an impact upon the redox signalling pathways that control tissue remodelling, which maybe ameliorated by MitoQ. Taken together these findings support the hypothesis that dysregulation of mitochondrial signalling plays a key role in diabetic kidney disease.

Balu Chacko, Colin Reily, Anup Srivastava, Michelle Johnson, Elena Ulasova, Kurt Zinn, Anupam Agarwal and Victor Darley-Usmar designed the research. Balu Chacko, Colin Reily, Anup Srivastava, Michelle Johnson, Yaoza Ye and Elena Ulasova performed the research and analysed the data. Balu Chacko, Colin Reily and Victor Darley-Usmar wrote the paper. Michael Murphy and Balaraman Kalyanaraman provided MitoQ for the study and reviewed the manuscript.

The authors thank Jeff Dubuisson for excellent technical assistance.

This study was supported by the National Institutes of Health [grant number DK075865]; the UAB-UCSD O'Brien Core Center for Acute Kidney Injury [grant number NIH/NIDDK 1P30 DK 079337]; and the UAB Small Animal Imaging shared facility [grant number P30CA013148].