Noninvasive In Vivo Imaging of Diabetes-Induced Renal Oxidative Stress and Response to Therapy Using Hyperpolarized ¹³C Dehydroascorbate Magnetic Resonance

Diabetes (type 1 and type 2) currently affects 8.3% of the population in the U.S. (1). Diabetic nephropathy develops in about one-third of diabetic patients and is one of the most devastating complications from diabetes (2). In diabetes, chronic hyperglycemia leads to excessive production of reactive oxygen species (ROS), exceeding local antioxidant capacity and leading to oxidative stress. Oxidative stress has been proposed to be a unifying cause for the onset and progression of diabetic nephropathy, and a target for novel therapies (3–7).

The link among diabetes, oxidative stress, and renal injury has been illustrated in many studies. A primary source of ROS and oxidative stress in renal cells is NAPDH oxidase 4 (Nox4; EC 1.6.3.1), which is activated by chronic hyperglycemia (8–13). Nox4 belongs to a family of enzymes responsible for the production of ROS (superoxide) by transferring electrons across the membrane from NADPH to molecular oxygen. Other isoforms of the Nox family expressed in the kidneys include Nox1 and Nox2, but these are much less abundant in comparison with Nox4. Nox4-dependent superoxide generation and oxidative stress have been shown to mediate glomerular hypertrophy and mesangial matrix accumulation (8,14), which are key histological features of diabetic nephropathy. Nox4-induced oxidative stress also contributes to tubulointerstitial fibrosis seen in diabetic nephropathy (15,16). ACE inhibitors and angiotensin receptor blockers, commonly used classes of drugs used to treat diabetic kidney disease, are thought to protect the kidneys in part by suppressing Nox4-mediated superoxide generation (17,18). Given the major role of oxidative stress in diabetes, a noninvasive strategy to interrogate oxidative stress in vivo may further enhance our understanding of diabetic renal injury. Such a strategy may also improve the monitoring of the onset and progression of diabetic nephropathy and, importantly, provide biomarkers for response to new therapies.

Hyperpolarized (HP) ¹³C magnetic resonance (MR) spectroscopy is a new molecular imaging technique that allows noninvasive investigation of dynamic metabolic and physiological processes in real time (19). Hyperpolarization, achieved through the dynamic nuclear polarization technique (20), can provide dramatic enhancement of the ¹³C nuclear MR signals (>50,000-fold at 3 T) of the substrates as well as subsequent metabolic products. Recently, we developed HP ¹³C dehydroascorbate (DHA) as an endogenous reduction-oxidation (redox) sensor (21). DHA is an oxidized form of vitamin C (VitC); it is rapidly taken up by cells via the GLUT1, GLUT3, and GLUT4 transporters (22) and is reduced to VitC via a glutathione-dependent mechanism, with coupled reactions with NADPH (Fig. 1). Glutathione, which usually exists in its reduced form (GSH), functions as a key antioxidant by scavenging ROS. It follows that the rate of DHA reduction to VitC reflects cellular redox capacity (23), and serves as an indicator for oxidative stress. We have previously observed rapid in vivo conversion of HP ¹³C DHA to ¹³C VitC in tissues expected to be rich in GSH, such as in the kidneys, liver, and brain, and in prostate cancer in a TRAMP (transgenic adenocarcinoma of the mouse prostate) model (21,24).

In this study, we apply HP ¹³C DHA to interrogate the renal redox capacity in a mouse model of type 2 diabetes and diabetic nephropathy. We show lower HP ¹³C DHA reduction to ¹³C VitC during the development of diabetic kidney injury, and its normalization after treatment with an ACE inhibitor (ramipril). HP ¹³C DHA reduction to VitC reflects the intracellular GSH concentrations and provides a noninvasive probe for monitoring redox alterations in vivo.

All procedures were approved by the Institutional Animal Care and Use Committee. Male mice homozygous for the leptin receptor mutation (BKS.Cg-lepr^db/lepr^db [db/db mice]) and their age-matched control littermates (BKS.Cg-lepr^db/+ [db/m mice]) were purchased from The Jackson Laboratory (Bar Harbor, ME). The db/db mice are one of the most commonly used murine models of human type 2 diabetes and diabetic nephropathy. The animals were allowed free access to standard chow and drinking water ad libitum. The db/db mice at 8, 12, and 16 weeks of age (n = 5 for each age group), and their age-matched db/m mice (n = 4 for each age group) underwent MRI studies, and subsequent histological and biochemical assays described below. In addition, a separate group of db/db mice (n = 5) were treated with an ACE inhibitor, ramipril (Sigma-Aldrich; 10 mg/kg daily via drinking water) from 8 to 12 weeks of age, followed by MRI and tissue analysis. Contemporaneous 24-h urine samples were collected from the mice for determination of urine albumin excretion by ELISA (Albuwell; Exocell, Philadelphia, PA). Urine 8-hydroxy-2′-deoxyguanosine (8-OHdG) concentrations were measured using a competitive ELISA kit (OxiSelect Oxidative DNA Damage ELISA Kit; Cell Biolabs, San Diego, CA) as previously described (14). Urinary 8-OHdG excretion was expressed as a total amount in nanograms over 24 h. Fasting blood glucose levels of the mice were also measured at the time of the MRI (AlphaTRAK; Abbott Laboratories).

The db/db mice and their control littermates db/m mice were fasted for 6 h prior to image acquisition. A tail vein catheter was placed for intravascular access. A 2.2 mol/L solution of [1-¹³C]DHA in dimethyacetamide containing 15 mmol/L OX063 trityl radical (Oxford Instruments) was HP on a HyperSense dynamic nuclear polarization instrument (Oxford Instruments). The frozen sample was dissolved in distilled water containing 0.3 mmol/L EDTA. Imaging was performed using a 3-T MRI scanner (GE Healthcare, Waukesha, WI) equipped with a multinuclear spectroscopy hardware package. The radiofrequency coil used in these experiments was a dual-tuned ¹H-¹³C coil with a quadrature ¹³C and ¹H channels. Prior to ¹³C studies, three-plane T₂-weighted images were acquired for anatomic localization (echo time 100 ms; repetition time 4 s; six averages) using a standard fast spin echo sequence. ¹³C MR spectroscopic imaging studies were then acquired 25 s postinjection of 15 mmol/L HP ¹³C DHA, at 5-mm isotropic resolution, as previously published (21,24).

Two-micrometer-thick renal sections were cut from 10% formalin-fixed, paraffin-embedded kidney samples, and stained with periodic acid Schiff (PAS). Each section was scanned at high magnification (×400) to produce digitized whole slide images and was loaded into the ImageJ software (National Institutes of Health). Ten randomly selected glomeruli in the outer cortex were selected. The percent glomerular mesangial area was calculated as the fraction of the total glomerular tuft cross-sectional area, as previously described (25). 8-OHdG immunostaining of the renal slices was performed using anti–8-OHdG mouse monoclonal antibody (Santa Cruz Biotechnology). In brief, ethanol-fixed sections were prepared, and antigen retrieval was performed using a microwave. Antibodies against 8-OHdG were used for the primary reactions followed by secondary reactions with biotin-labeled anti-mouse goat IgG.

Renal tissues were homogenized in PBS with EDTA. Protein was precipitated with 10% metaphosphoric acid, and glutathione concentrations were assayed spectrophotometrically using a commercially available 5,5′-dithio-bis-2-(nitrobenzoic acid)–based absorbance assay (Cayman Chemical).

Total RNA was isolated from renal tissues from db/db (n = 4) and db/m (n = 4) mice using the Qiagen RNeasy Kit, and was reverse transcribed using SuperScript III Reverse Transcriptase (Invitrogen) according to the manufacturer’s instructions. Mouse Nox1, Nox2, and Nox4 Invitrogen primers were obtained from Life Technologies, and real-time PCR determination of cDNA amounts was performed. Relative expression to control gene hypoxanthine phosphoribosyltranserase (HPRT) was determined using the ΔCt method.

Student t tests and one-way ANOVA with Tukey-Kramer post hoc tests were used to assess the difference between relevant groups using the statistical software package STATA version 8.0. All values are reported as the mean ± SE. A P value of <0.05 was considered statistically significant.

We performed HP ¹³C DHA MR spectroscopic imaging in db/db mice and age-matched db/m mice at 8, 12, and 16 weeks. The db/db mice have been reported to reliably develop frank hyperglycemia by 8 weeks of age (26). Table 1 summarizes the weight, blood glucose level, and urine albumin excretion in the db/db and db/m mice in this study. Figure 2A shows representative MR spectra for voxels corresponding to kidneys in a db/m and a db/db mouse at 12 weeks of age after the injection of HP ¹³C DHA, demonstrating a lower reduction of DHA to VitC in the db/db mouse. Figure 2B shows VitC/(VitC + DHA) ratios in the kidneys of db/m and db/db mice at 8, 12, and 16 weeks of age. For db/m mice, the VitC/(VitC + DHA) ratios did not change significantly over time. For db/db mice, the VitC/(VitC + DHA) ratios were 13%, 35%, and 33% lower compared with the age-matched db/m mice at 8, 12, and 16 weeks, respectively (P = 0.02, 0.03, and 0.02, respectively). A significant difference in the VitC/(VitC + DHA) ratios was found among the db/db mice of different ages (P = 0.01). Tukey-Kramer post hoc analysis showed the VitC/(VitC + DHA) ratios were significantly lower in the 12- and 16-week-old db/db mice compared with those in the 8-week-old db/db mice (P < 0.05 for both), but were not significantly different between the 12- and 16-week-old db/db mice. PAS stains of the renal slices demonstrated that the percent mesangial matrix area increased over time in the db/db mice, at 13.9% for the 8-week-old db/db mice, 17.7% for the 12-week-old db/db mice, and 20.4% for the 16-week-old db/db mice (P = 0.02, one-way ANOVA). The percent mesangial matrix area was significantly higher in 12- and 16-week-old db/db mice compared with 8-week-old db/db mice (P < 0.05 for both).

Since there is a redox coupling mechanism between glutathione and VitC, we hypothesized that the observed in vivo reduction of HP ¹³C DHA reflects changes in the concentration of GSH. To test this hypothesis, we assayed the kidneys of db/db and db/m mice for both GSH and oxidized glutathione dimer (GSSG) concentrations using an enzymatic recycling method (27). The results are summarized in Fig. 3. For db/m mice, the GSH concentrations did not change significantly among the three time points. For db/db mice, the mean renal GSH concentrations were 18%, 26%, and 29% lower compared with age-matched db/m mice at 8, 12, and 16 weeks of age, respectively (P = 0.04, 0.003, and 0.02, respectively). The renal GSH/GSSG ratios were 27%, 42%, and 38% lower in the db/db mice compared with the db/m mice at 8, 12, and 16 weeks, respectively (P = 0.04, 0.002, and 0.02, respectively). The total glutathione (GSH + GSSG) concentrations in the db/db mice were slightly lower than those in the age-matched db/m mice, although they did not reach statistical significance (P = 0.12, 0.06, and 0.08, respectively). We additionally assayed the mRNA expression of Nox4, which has been shown to be a major source of renal superoxide generation in diabetes (8–13), as well as the expression of Nox1 and Nox2. The renal Nox4 expression was significantly increased in the db/db mice compared with db/m mice at all three time points (P = 0.04, 0.03, and 0.03 for 8, 12, and 16 weeks, respectively) (Fig. 3). The Nox2 expression was much lower than that of Nox4, and was not significantly different between the db/db and db/m mice (Supplementary Fig. 1). The Nox1 expression was very low (<1% of the control gene HPRT) in the db/db and db/m mice (results not shown).

Hyperglycemia has been shown to increases cellular angiotensin II (Ang II) production (28), which in turn activates Nox4-mediated superoxide production and oxidative stress (18). We treated a group of db/db mice with an ACE inhibitor, ramipril, from 8 to 12 weeks of age and measured the in vivo HP ¹³C DHA reduction. We observed that HP ¹³C DHA reduction to VitC in the kidneys of treated 12-week-old db/db mice was restored to a level similar to that found in age-matched db/m mice (Fig. 4A and B). One-way ANOVA with post hoc analysis showed that the VitC/(VitC + DHA) ratios were significantly lower in the 12-week-old untreated db/db mice compared with the 12-week-old db/m mice (P < 0.05) and the 12-week-old treated db/db mice (P < 0.05). The VitC/(VitC + DHA) ratios were not significantly different between the treated db/db mice and the age-matched db/m mice. Similar analyses also showed that there were no significant differences between the treated db/db mice and the age-matched db/m mice with regard to the renal GSH and total glutathione concentrations, GSH/GSSG ratios, and Nox4 mRNA expression level (Fig. 4C and D). Immunohistochemical stains of 8-OHdG in renal slices demonstrate increased staining, indicating increased oxidative DNA damage, in the untreated db/db mice, which is diminished with ramipril treatment (Fig. 4E). Similarly, 24-h urine 8-OHdG concentrations were increased in the untreated db/db mice, and were normalized to the level in the db/m mice after ramipril treatment (Fig. 4F). PAS stains of renal slices demonstrate an increased mesangial matrix area in the untreated db/db mice, which is diminished with treatment (Fig. 4G).

Diabetic nephropathy is the major cause of end-stage renal disease. Increased oxidative stress has been proposed to be both a key initiator and a downstream effect of a number of pathways involved in the pathogenesis of diabetic nephropathy. In this study, we monitored, in real time, the renal redox alteration associated with diabetes and after treatment using HP ¹³C DHA MR. We showed that the db/db mice have lower renal HP ¹³C DHA reduction to VitC compared with age-matched control mice. After 4 weeks of treatment with an ACE inhibitor, ramipril, the db/db mice show normalization of the HP ¹³C DHA reduction to VitC.

The observed alteration in the HP ¹³C DHA reduction to VitC appears to reflect renal GSH concentration. GSH is the most abundant intracellular thiol-based antioxidant. It protects the cells from oxidative stress by preventing the accumulation of ROS. GSSG, the oxidized form of glutathione, is formed as a product of this detoxification. GSH is regenerated by the action of glutathione reductase (EC 1.8.1.7), an NADPH-dependent enzyme, thus completing the redox cycle. GSH is also a cofactor for the enzyme DHA reductase (EC 1.8.5.1), which recycles DHA back to reduced ascorbic acid (VitC), therefore linking the redox couple between glutathione and VitC. GSH alterations have been observed in diabetes and diabetic nephropathy. For example, renal mesangial cells exposed to chronic hyperglycemia have significantly decreased GSH concentrations (29). Diabetic patients with microalbuminuria were noted to have lower GSH levels in red blood cells than diabetic subjects without microalbuminuria (30). In agreement with these previous studies, we found significantly lower concentrations of GSH and lower GSH/GSSG ratios in the kidneys of db/db mice compared with those of age-matched control mice at all three time points. The observed changes in renal HP ¹³C DHA reduction to VitC appeared to reflect the changes in GSH concentration. The mean VitC/(VitC + DHA) ratios in the db/db mice were 13%, 35%, and 33% lower, respectively, compared with the age-matched db/m mice at 8, 12, and 16 weeks of age. Correspondingly, the measured renal GSH concentrations were 18%, 26%, and 29% lower, respectively, in the db/db mice compared with age-matched db/m mice at 8, 12, and 16 weeks of age.

We noted an early decrease in the renal VitC/(VitC + DHA) ratios in the 8-week-old db/db mice. Phenotypically, the 8-week-old db/db mice in our study demonstrated increased albuminuria, but no significant glomerular histological changes compared with control mice. As the duration of diabetes increased, the glomerular histological changes, as assessed by mesangial matrix area, became evident, although they remained relatively mild even in the 16-week-old db/db mice. Taken together, these findings support the notion that alterations in oxidative stress/redox capacity are an early event and may play a primary role in the development of diabetic nephropathy (6,15,31). Noninvasive techniques such as HP ¹³C DHA MR may facilitate future work to better understand diabetic renal injury, such as the role that oxidative stress plays in determining susceptibility to diabetic nephropathy, as suggested by a previous study (32), and the temporal relationship between oxidative stress and the development of kidney injury.

Nox4 is an enzyme that has been reported to be a major source of renal ROS in diabetes. Upregulation of Nox4 together with increased superoxide generation has been shown in response to high glucose concentration in renal cells and in experimental models of diabetes (8–13). In our study, the observed lower HP ¹³C DHA reduction to VitC in the kidney of the db/db mice corresponded to an increase in Nox4 expression. We postulate that this may be due to Nox4-induced increase in superoxide generation with consumption of GSH (Fig. 5). It should be noted that other sources or pathways are also involved in the increased renal oxidative stress in diabetes. For example, hyperglycemia leads to increased superoxide production by the mitochondrial electron transport chain. This process can in turn activate other superoxide production pathways, such as via Nox4, that may amplify the original damaging effect of hyperglycemia (33). Therefore, other sources of oxidative stress may also contribute to the observed alterations in renal GSH in the diabetic mice. These additional sources likely contribute to the metabolic phenotype reflected in lower HP ¹³C DHA reduction to VitC in the diabetic kidneys.

It is well recognized that hyperglycemia increases cellular Ang II production (28), which in turn activates Nox4-mediated superoxide production and oxidative stress (18,34). ACE inhibitors, which decrease the production of Ang II, and Ang II receptor blockers are commonly used drugs for the treatment of diabetic nephropathy. They are thought to protect the kidneys in part through the reduction of oxidative stress (35,36) (Fig. 5). In this study, we demonstrated that HP ¹³C DHA can monitor the targeted effect of an ACE inhibitor (ramipril) on oxidative stress in vivo by observing the HP ¹³C DHA reduction to VitC. After 4 weeks of ramipril treatment, HP ¹³C DHA reduction to VitC in the kidneys of 12-week-old db/db mice was restored to a level similar to that found in age-matched db/m mice. The MR findings parallel the normalization of renal GSH concentration and Nox4 expression. Because of the important role of oxidative stress in the development and progression of diabetic nephropathy as well as other forms of chronic kidney disease (37), new pharmacological strategies that directly or indirectly target oxidative stress are being explored (7,14,38). Noninvasive techniques such as HP ¹³C DHA MR may provide companion biomarkers that can provide better information on drug targeting and enhance treatment monitoring.

It should be pointed out that, in general, cellular DHA uptake may be facilitated by several factors such as insulin, cyclic adenosine monophosphate, and colony-stimulating factors (39–41) and suppressed by homocysteine (42). Such factors may potentially influence DHA uptake into the renal cells of db/db mice with hyperinsulinemia and insulin resistance. Additionally, different cell types within the renal glomeruli and tubules may differentially reduce DHA to VitC due to different insulin sensitivity or resistance. Therefore, future in vitro cell studies using HP ¹³C DHA are warranted to elucidate the uptake and reduction of DHA in different renal cells in a diabetic milieu. It should also be noted that DHA reduction to VitC can be affected by enzymes such as glutaredoxin (Glrx) and glutathione S-transferase ω (GSTO) that have DHA reductase activity (43,44). The mRNA expressions of Glrx-1 and GSTO-1, were noted to be higher in the db/db mice compared with the db/m mice (Supplementary Fig. 2). Therefore, the observed lower renal HP ¹³C DHA reduction to VitC in the db/db mice compared with the db/m mice cannot be explained by the alterations of expression of these enzymes. Future studies are needed to investigate the role that other enzymes play in the HP ¹³C DHA reduction to VitC.

There are presently limited noninvasive in vivo methods for measuring oxidative stress in the kidneys. Urine markers of oxidative stress, such as 8-OHdG (a marker of DNA oxidative damage), have been used to study diabetes-associated oxidative stress (45). Ex vivo measurement of such markers, however, can be adversely affected by variables, such as how the samples are collected and stored (46), and may not accurately reflect oxidative stress in real time at the tissue level. Electron spin resonance (ESR) imaging has been applied to detect free radicals in the kidneys in vivo (47). However, the sensitivity of the ESR instruments for in vivo studies and the specificity and stability of the probes are presently insufficient (48). A recent study reported in vivo imaging of immunospin-trapped radicals with MRI in a mouse diabetes model (49). A drawback of this technique is that it required multiple injections to trap radicals over the course of days (49). Both the ESR imaging and immunospin-trapped radical MRI assess oxidative stress by measuring radical levels.

HP ¹³C DHA MR provides a different approach of determination of oxidative stress via real-time evaluation of the redox capacity in vivo. In this preclinical study, because of the small size of the mouse kidneys, the spatial resolution of the HP ¹³C DHA MR does not currently allow for separate evaluation of the renal cortex and medulla. Several approaches will address this problem, including better substrate polarization and coil design, and more efficient sampling of k-space. These advances will allow specific interrogation of the various renal compartments in preclinical models, as well as in potential future clinical studies. An important consideration in the clinical translation of HP ¹³C probes is that injection of probes at doses in the micromolar-to-millimolar range is necessary to achieve sufficient MR signals. While probes such as ¹³C DHA have the advantage of being an endogenous compound, the metabolic effects of the injected dose and the safety in humans will need to be established. Notably, the clinical translation of HP ¹³C MR technology has been recently achieved with the successful completion of the phase I clinical trial of HP ¹³C pyruvate in prostate cancer patients (50), which opens doors for potential clinical translation of other endogenous HP probes such as DHA.

In conclusion, our study has shown that HP ¹³C DHA reduction to VitC correlates to the GSH component of the redox couple, and likely reflects the impact of diabetes-induced alterations in renal superoxide generation and oxidative stress. HP ¹³C DHA can, in real time, noninvasively probe the redox changes associated with diabetic renal injury and after successful treatment. Such an imaging approach may potentially enhance the prediction and early detection of diabetic nephropathy and provide companion biomarkers that can better inform on the response to new therapies targeting oxidative stress in patients. More broadly, such imaging strategies can be extended to the noninvasive in vivo evaluation of other complications from diabetes as well as other oxidative stress-related diseases.

Acknowledgments. The authors thank Romelyn Delos Santos from the University of California, San Francisco, for excellent histological assistance.

Funding. This research was supported by National Institutes of Health grants R00-EB-014328 (K.R.K.), R01-CA-166766 (D.M.W.), R00-EB-012064 (P.L.), P41-EB-013598 (J.K.), and R01-DK-097357 (Z.J.W.); a Society of Abdominal Radiology Morton A. Bosniak Research Grant (Z.J.W.); and INSERM grant P41EB013598.

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

Author Contributions. K.R.K. conceived the study, designed and performed the experiments, and wrote the manuscript. D.M.W. designed the experiments and contributed to the writing of the manuscript. V.S., R.B., K.-Y.J., P.L., and M.V.C. performed the experiments. J.K. designed the experiments and reviewed the manuscript. Z.J.W. conceived the study, designed and performed the experiments, and wrote the manuscript. K.R.K. and Z.J.W. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the Society of Nuclear Medicine and Molecular Imaging 2013 Annual Meeting, Vancouver, BC, Canada, 8–12 June 2013.