Thiazolidinedione Treatment Decreases Oxidative Stress in Spontaneously Hypertensive Heart Failure Rats Through Attenuation of Inducible Nitric Oxide Synthase–Mediated Lipid Radical Formation

α-(4-Pyridyl-1-oxide)-N-t-butylnitrone (POBN) and rosiglitazone were obtained from Alexis Biochemicals (San Diego, CA). The 1400 W was from Calbiochem-Novabiochem (La Jolla, CA). Other materials were all obtained from Sigma-Aldrich (St. Louis, MO).

Ten-week-old Wistar-Kyoto (WKY) male rats (Charles River Laboratories) were used in all experiments as controls; age-matched male spontaneously hypertensive heart failure (SHHF) rats were used as obese animals. Rats were housed in a room with air conditioning and a 12/12 h light/dark cycle, were fed a standard rat chow (NIH open formula; Zeigler Brothers, Gardners, PA), and had access to water ad libitum. All studies were approved by the institutional review boards at the National Institute of Environmental Health Sciences and the Pennington Biomedical Research Center and adhered to National Institutes of Health guidelines for the care and handling of experimental animals.

Body composition was characterized by a Bruker MiniScope NMR (Bruker, Billerica, MA) at the Pennington Biomedical Research Center. WKY and SHHF rats were placed in a plastic tube and their lean mass, fat mass, and free fluid amounts were compared.

For the EPR experiments, POBN was dissolved in saline and administered at 1 g/kg body wt i.p. and bile cannulation was performed. Bile samples (∼300 μL) were collected every 30 min for 2.5 h into plastic Eppendorf tubes containing a 50-μL solution of bathocuproine disulfonic acid and 2,2’-dipyridyl (30 mmol/L, respectively, to prevent ex vivo free radical production). Samples were frozen in dry ice immediately after collection and stored at −80°C until the EPR measurements.

In the inhibitor studies, WKY or SHHF rats received the iNOS inhibitor 1400 W (15 mg/kg i.p.) 1 h before bile cannulation and spin trapping. Each group contained at least six animals. A different group of animals received l-arginine (100 mg/kg i.p.) 30 min before spin trapping experiments. The rosiglitazone-treated group received rosiglitazone for 10 days (4 mg/kg/day i.p.). Another group received the cell permeable free radical scavenger 4-hydroxy-2,2,6,6-tetramethylpiperidine-N-oxyl (TEMPOL) for 10 days (30 mg/kg/day i.p). Control groups received vehicle only.

For the various experiments with the insulin signaling pathway proteins, liver and skeletal muscle samples were taken from WKY and SHHF rats after insulin stimuli. Anesthetized animals received human insulin (50 IU/kg) via the portal vein; liver and skeletal muscle samples were taken 30 and 90 s, respectively, after the stimuli and frozen immediately in liquid nitrogen.

For oral glucose tolerance test (OGTT), WKY and SHHF rats received 300 mg/100 g body wt glucose in solution via oral gavage. Blood glucose levels were tested at 0, 30, 60, 90, and 120 min after OGTT with a OneTouch Mini apparatus.

All EPR spectra were recorded at room temperature in a quartz flat cell on a Bruker EMX EPR spectrometer equipped with a super high-Q cavity. Spectra were recorded using an IBM-compatible computer interfaced with the spectrometer with the following instrument settings and conditions: 20.2 mW microwave power, 100 kHz modulation frequency, 1 G modulation amplitude, 1,300 ms time constant, 655 ms conversion time, and a single scan of 80 G. Bile samples were treated with ascorbate oxidase strips to reduce the interfering background signal of ascorbate radical that is abundant in bile.

For immunoprecipitation studies, polyclonal IR-β antibody was obtained from Santa Cruz Biotechnology (San Diego, CA). Polyclonal IRS-1 antibody from Millipore was used to immunoprecipitate IRS-1. Polyclonal rabbit Akt as well as p-Ser473-Akt antibodies were purchased from Cell Signaling Technology (Danvers, MA) to detect or immunoprecipitate these proteins. Phospho-IR-β (Tyr1162), phosphatidylinositol 3-kinase (PI3K)-p85 subunit, and pIRS (Tyr989) antibodies were purchased from Santa Cruz Biotechnology. Liver and skeletal muscle tissue samples were homogenized with a Polytron homogenizer in lysis buffer containing a mixture of protease and phosphatase inhibitors, 2% Triton X-100, 300 mmol/L NaCl, 20 mmol/L Tris, 2 mmol/L EDTA, and 1% Nonidet P-40. Samples were centrifuged at 10,000 rpm for 5 min at 4°C. Protein concentrations were quantitated with a BCA protein assay kit (Pierce). Supernatants containing 1,500 μg protein were incubated with various polyclonal antibodies according to the manufacturer’s recommendation (typically 4 μg antibody) overnight in 4°C. These samples were then incubated with 40–100 μL Protein A Plus agarose beads (Pierce) for 2 h at 4°C. The beads were prewashed three times with cold PBS before use. Complexes were centrifuged at 3,000 rpm for 1 min between exhaustive washing steps with cold lysis buffer (at least five washes). For Western blot analysis, beads containing the immunoprecipitated protein were incubated in 50 μL elution buffer for 10 min and then 30 μL SDS sample buffer was added. In some experiments, eluates were boiled for 10 min. Supernatants were separated on reducing NuPAGE 4–12% Bis-Tris gels (Invitrogen) and transferred to a nitrocellulose membrane.

Liver or skeletal muscle homogenate supernatants containing equal amounts of protein were reduced similarly to the previous procedure, and 20–40 µg protein per lane was loaded for electrophoresis and transferred as described above. For nuclear fractionation, liver samples were homogenized in a buffer containing 0.32 mol/L sucrose, 20 mmol/L Tris-HCl (pH 7.4), 2 mmol/L EDTA, 1 mmol/L dithiothreitol, and protease inhibitor cocktail at 4°C. The homogenates were centrifuged (1,000g, 25 min, 4°C) and the pellets were washed once with this buffer (1,000g, 10 min, 4°C), suspended in Triton buffer (1% Triton X-100, 20 mmol/L Tris-HCl [pH 7.4], 150 mmol/L NaCl, 200 mmol/L EDTA, plus protease inhibitor cocktail), kept on ice for 30 min, and centrifuged (15,000g, 30 min, 4°C) to obtain the nuclear fraction.

After blocking (1% BSA in 0.1 mol/L PBS, pH = 7.4), the membrane was probed with the corresponding antibody against the protein of interest in the immunoprecipitation studies (IR-β, pIR-β, IRS-1, p85, pIRS-1, Akt, or p-Ser473-Akt). In the case of whole homogenates, monoclonal anti-iNOS antibody (Sigma-Aldrich), polyclonal IRS-1 primary (Millipore), goat primary anti–peroxisome proliferator–activated receptor (PPAR)-γ coactivator 1-α (anti–PGC1-α; Novus Biologicals), rabbit primary anti-PEPCK (Cell Signaling), or anti-G6Pase (Santa Cruz Biotechnology) were used, followed by the appropriate secondary antibody conjugated with horseradish peroxidase (1:10,000) and development by enhanced chemiluminescent substrate (Pierce). Western blot band intensities were quantified using the ImageJ program (download available at http://rsb.info.nih.gov/ij/).

For confocal studies, liver and skeletal muscle samples from WKY and SHHF rats were fixed in 4% buffered formaldehyde for 24 h then placed into 30% sucrose for 24 h. Tissues were then embedded into optimal cutting temperature media, sliced on a cryocut microtome into 30-μm sections, and mounted on frosted glass slides. Samples were permeabilized with 0.1% Surfact-Amps-X-100 for 30 min. After blocking with 1% BSA in PBS for 30 min, staining of 4-HNE was performed using polyclonal anti–HNE-adduct primary antibody (Axxora, San Diego, CA) (1:500) and an Alexa Fluor 488 anti-mouse fluorescent secondary antibody (1:1,000). Staining for nitrotyrosine was made using Abcam anti-nitrotyrosine primary (1:500) and Alexa Fluor 568 secondary (1:1,000) antibodies. Secondary controls were done by omitting the primary antibody but applying the secondary antibody. Sections were observed under an LSM510 confocal laser microscope (Zeiss).

Liver and skeletal muscle nitrite levels were determined with the Griess method using a nitrite measurement kit obtained from Invitrogen.

Data are expressed as mean ± SD. Statistical significance between groups was determined by ANOVA and Student t test as appropriate. P < 0.05 was considered to be statistically significant.

Compared with their age-matched controls, SHHF rats showed increase in their fat mass and free fluid content (Fig. 1A–C). WKY and obese SHHF rats were subjected to an oral glucose challenge after overnight fasting to confirm the presence of impaired oral glucose tolerance. As shown in Fig. 1D, SHHF rats were unable to recover their blood glucose levels in time compared with age-matched WKY. This result confirms that SHHF rats are glucose intolerant, a characteristic feature of metabolic syndrome.

SHHF rats showed significant lipid peroxidation compared with age-matched WKY rats, which was confirmed by EPR spectroscopy and in vivo spin trapping experiments. These experiments led to the detection of strong six-line EPR signals of a POBN radical adduct that were reproducibly observed in the bile of SHHF rats 2 h after spin trap administration (Fig. 2A–B). In WKY, only residual signals of POBN radical adducts were recorded, confirming increased free radical formation in the obese animals. EPR is the unambiguous way to detect radical species; because of their paramagnetic nature, they will give rise to a specific signal. As a result of the radical species’ short half-life, it is usually necessary to use spin traps to trap these radicals as adducts with a longer half-life. POBN is specific for carbon-centered radicals, and the whole six-line of the trace indicates a lipid-derived radical species. This is verified from previous simulations using an EPR database. On the basis of the distances between the peaks (hyperfine splitting constants), the detected species is identified unequivocally. The free radicals trapped were identified through their EPR parameters, a^N = 15.75 ± 0.06 G and a_β^H = 2.77 ± 0.07 G, corresponding to those reported previously for the POBN radical adduct of a carbon-centered, lipid-derived radical (34) (see also Supplementary Fig. 1).

A significant decrease in the formation of lipid free radical adducts was found in SHHF rats when rats were injected with 1400 W (15 mg/kg i.p.), a specific inhibitor of iNOS (35), 1 h before spin trap administration (Fig. 2A–B). The inhibitory effect remained for >2 h. Western blot studies also confirmed elevated iNOS protein levels in the liver of SHHF rats (Fig. 2C–D). We were not able to detect iNOS protein in the skeletal muscle. These results suggest a contribution of hepatic iNOS to the radical generation process in these obese rats. Although a specific mechanism is hard, if not impossible, to prove in vivo, this may imply peroxynitrite and its decomposition or reaction with CO₂ in the tissues leading to radical intermediates.

Because SHHF rats showed increased lipid radical production (Fig. 2), which was attenuated by specific iNOS inhibition, we investigated whether iNOS levels and the accumulation of lipid peroxidation or protein oxidation end products, such as 4-HNE and nitrotyrosine, are elevated in liver and skeletal muscle tissues. It is interesting that these two tissues showed different characteristics. Confocal studies colocalized significant 4-HNE and nitrotyrosine accumulation in the liver of obese SHHF rats, but only slight changes were notable in the skeletal muscle, compared with age-matched WKY rats (Fig. 3A–B). Consistent with iNOS and nitrotyrosine levels, liver nitrite levels were significantly higher in the obese SHHF liver but only slightly changed in skeletal muscle (Fig. 3C). Since iNOS levels were elevated in the liver but not in skeletal muscle, this result suggests that hepatic iNOS overexpression may drive the reactions leading to the accumulation of 4-HNE, 3-nitrotyrosine, and nitrite in the liver, but this process is absent in skeletal muscle.

To gain a better understanding about the possible perturbations of insulin signaling in SHHF rats, we first examined the basal levels of IRS-1 protein and the transcriptional regulators of gluconeogenesis, PGC1-α, PEPCK, and G6Pase. To determine the degree of change in the insulin signaling cascade, IR-β, IRS-1, and Akt proteins were immunoprecipitated from WKY and SHHF rat liver and skeletal muscle tissues after insulin injection. We investigated Tyr phosphorylation on IR-β, the attachment of the p85 subunit of the PI3K protein to IRS-1, and Ser473 phosphorylation on Akt from these samples. SHHF rats did not show decreased levels of IRS-1 in the liver or skeletal muscle; however, expression of the gluconeogenetic master regulators were elevated in the liver, which is typical of many diabetic animal models (Fig. 4A–B). Tyr phosphorylation of IR-β was significantly reduced in SHHF rat liver and skeletal muscle, together with impaired p85 attachment to IRS-1 (Fig. 5A–D). Akt levels were unchanged, but Ser473 Akt phosphorylation was reduced in both tissues (Fig. 5E–F).

As the commonly used diabetes drug rosiglitazone has also been suggested to affect iNOS levels, we asked whether rosiglitazone treatment is able to reduce iNOS levels and lipid radical formation in obese rats, while in parallel improving their condition. Ten days of intraperitoneal rosiglitazone treatment significantly reduced the detectable lipid radical formation in the bile and almost completely attenuated the increased iNOS protein levels in the liver (Fig. 6A–C).

The treatment also improved OGTT response and corrected the insulin signaling pathway as shown by enhanced phosphorylation and p85 attachment in the liver (Fig. 6D–E). Of interest, a 10-day treatment with a general scavenger TEMPOL (30 mg/kg/day) resulted only in a mild improvement of OGTT response (Supplementary Fig. 2), indicating that thiazolidinedione is necessary to achieve both the redox and the insulin-sensitizing effects.

Increasing evidence indicates that during the development of insulin resistance, toxic lipid metabolites accumulate in tissues, contributing to the overall reduction of insulin sensitivity (26,36). The possible underlying mechanisms are currently under investigation, but incomplete fatty acid oxidation and the accumulation of lipotoxic metabolites compromise insulin signaling. Here, we studied a lipid peroxidation mechanism and impairment of the insulin signaling cascade of an obese, heart-failure prone, insulin-resistant animal model. We asked whether toxic lipid metabolites accumulate in insulin-sensitive tissues, whether iNOS is a contributing source, and whether thiazolidinedione treatment is able to attenuate lipid radical formation.

The unique methodology of EPR spectroscopy combined with spin trapping allowed us to characterize a lipid radical–driven process in the liver of SHHF rats. We provide in vivo evidence that a lipid radical–generating mechanism contributes to toxic lipid end product accumulation. In spin trapping, short-lived free radical intermediates react with the spin trapping agent, producing stable free radical adducts that are mainly excreted in the bile. These can then be detected and characterized through their unique signature EPR spectra (37). Lipid radical formation was increased in the SHHF rat and was significantly reduced with a specific iNOS inhibitor (1400 W), therefore suggesting the active participation of iNOS in the initial free radical–generating process. POBN gets distributed through the tissues, and POBN radical adducts are excreted in the bile. Although this is a convenient way to detect lipid peroxidation, the method probably reflects hepatic processes only. In contrast to our previous results obtained in a type 1 diabetes model (25), the lipid radical formation was unchanged by l-arginine addition (Supplementary Fig. 1). If iNOS is uncoupled, generating both superoxide and NO, the substrate l-arginine should abrogate the process and restore iNOS coupling. Our data indicate that iNOS is in its coupled state in this model. Furthermore, the addition of DMSO did not improve the EPR signals (Supplementary Fig. 1). DMSO specifically reacts with ^•OH radicals if present, and ^•OH is then converted into a methyl radical (^•CH₃) through its diffusion-limited reaction with DMSO (k = 7 × 10⁹ ⋅ mol/L⁻¹ ⋅ s⁻¹) (38). The ^•CH₃ radicals are then trapped and detected by EPR as POBN radical adducts, resulting in an additional increase of the initial POBN lipid radical adduct signal. Our data suggests the lack of the ^•OH radical in the process. The initiation of lipid peroxide generation depends on other contributing mechanisms.

As a result of increased lipid peroxidation, 4-HNE accumulation was also observed in the SHHF liver (Fig. 3A). Reactive oxygen species generation stimulates the peroxidation of polyunsaturated fatty acids resulting in lipid hydroperoxide production. Our EPR experiments demonstrated the presence of increased lipid radical formation, which is an intermediate step leading to the generation of primary products of lipid peroxidation, lipid hydroperoxides. Given their unstable nature, the decomposition of these hydroperoxides leads to various reactive aldehydes, including 4-HNE (28). 4-HNE is a well-characterized toxic aldehyde end product of lipid peroxidation that reacts with amine groups to produce covalent modification of proteins and enzymes via nonenzymatic Michael addition (29). We found a strong colocalization of HNE and 3-nitrotyrosine in the liver (Fig. 3B), together with increased nitrite formation (Fig. 3C). This observation is particularly interesting because it indicates a correlation of lipid radical production and nitration in vivo and allows a speculation that lipid radical formation can lead to nitration as well through Tyr radical formation. This was recently suggested via in vitro model systems in a hydrophobic environment (32,33).

To further link lipid radical formation and the possible role of proinflammation and iNOS activation, we treated SHHF rats with the diabetes drug rosiglitazone for 10 days. Thiazolidinediones are known to act on not only PPAR-γ but also iNOS and cyclooxygenase; therefore, they may have the potential to inhibit an iNOS-driven lipid radical production. Rosiglitazone treatment not only improved the OGTT response and insulin signaling in obese rats but also led to significantly attenuated lipid radical formation and reduced iNOS protein levels, thus presenting a strong mechanistic link between these oxidative processes (Fig. 6). Here, we show that in addition to the known positive and negative effects, a PPAR-γ agonist can reduce oxidative stress in this model through reducing lipid radical formation possibly with an iNOS-mediated origin. This effect can certainly contribute to a better redox status in insulin-sensitive tissues. Rosiglitazone may have other potential redox effects that can contribute to the overall mechanism. It has been implied before that rosiglitazone has anti-inflammatory potential through blocking iNOS and cyclooxygenase, whereas a PPAR-γ antagonist blunted these effects (39). To test whether the PPAR-γ agonist potential is important in our model, we next treated SHHF rats with the general scavenger TEMPOL (30 mg/kg) for 10 days and performed an OGTT after treatment. TEMPOL had only a mild effect and was not able to reduce blood glucose levels significantly (Supplementary Fig. 2). This result indicates that the PPAR-γ agonist activity of rosiglitazone may be important to modulate iNOS levels and subsequent lipid peroxidation, and it is necessary to achieve both the redox and the insulin-sensitizing effects.

Charbonneau and Marette (40) recently showed that lipid infusion in mice induces nitrotyrosine formation, which damages all the major insulin signaling proteins in the IR-β/IRS/PI3K/Akt cascade. An in vitro model from Bartesaghi et al. (33) shows that lipid peroxides are mediating Tyr oxidation processes in hydrophobic compartments. Our studies are in agreement with those mentioned above in an in vivo model in SHHF rats and further tie the role of iNOS and the subsequent lipid peroxidation process together. While improving glucose tolerance and signaling, treatment with thiazolidinediones can also attenuate oxidative stress through inhibiting lipid radical formation. The detrimental effect of HNE in SHHF rats may occur via a direct mechanism partially disabling protein function. Another possibility is that, as we indicated before, lipid peroxidation itself can trigger the formation of Tyr^⋅, which in turn can lead to nitration on functionally important residues. This warrants further studies of possible modifications on insulin signaling proteins and can provide details about the interference of these metabolites. Besides these mechanisms, others have also shown S-nitrosation of insulin signaling proteins (41) in various type 2 diabetes or obesity models.

The lipid peroxidation process was accompanied by impaired insulin signaling in SHHF rat liver and skeletal muscle. Previous studies show similar patterns in the aorta of SH (spontaneously hypertensive) rats (42) and that oxidative stress can be a significant factor in activating signaling cascades and hypertension. Because SHHF rats are insulin resistant and glucose intolerant, but are only slightly hyperglycemic or normoglycemic on fasting, our study shows that indeed, oxidative stress and the accumulation of toxic lipid metabolites in peripheral tissues may play a significant role in the development of metabolic syndrome and insulin resistance.

In conclusion, our findings demonstrate that hepatic iNOS induction contributes to lipid radical formation and, therefore, to redox imbalance and to the accumulation of toxic metabolites in obesity. Rosiglitazone treatment not only promotes improved insulin signaling but also attenuates iNOS-mediated lipid radical formation through the reduction of iNOS protein levels. This effect seems to be connected to its PPAR-γ agonist activity. These interactions could be important factors that may connect proinflammation, oxidative stress, and lipotoxicity together in obesity. Future therapies could focus to act on multiple targets, such as iNOS, peroxynitrite, and lipid peroxides, to improve the pathogenesis of insulin resistance and metabolic syndrome.

This work was supported by grants from the National Institutes of Health (R00-DK-083615 to K.S. and T32-AT004094 to S.W.), the Pennington Foundation, the Louisiana Board of Regents (to K.S.), and the American Heart Association (AHA-09SDG2250933 to M.G.B.).

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

M.B.K. researched data, contributed to discussion, and wrote the manuscript. M.G.B. and S.W. researched data and contributed to discussion. C.R. and E.C. researched data. K.S. designed the concept, directed the experiments, researched data, contributed to discussion, and wrote the majority of the manuscript. K.S. is the guarantor of this manuscript and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

The authors are very grateful to Drs. Andre Marette and Alexandre Charbonneau (both from Laval University, Quebec City, Quebec, Canada) and Dr. Jacqueline Stephens (Louisiana State University, Baton Rouge, LA) for helpful discussions.