Insulin resistance is a cellular antioxidant defense mechanism

We have previously described a reproducible system for studying IR in myotubes and adipocytes in culture relying on the translocation of the facilitative glucose transporter GLUT4 to the plasma membrane (5). This process represents one of the earliest defects in IR (2) and so provides a robust measure of this defect. Insulin acutely increased surface levels of GLUT4 by approximately 4-fold in adipocytes and myotubes and this was impaired by approximately 50% in response to hyperinsulinemia (chronic insulin, 10 nM for 24 h; CI), inflammation (2 ng/mL TNFα for 72 h; TNF), steroids (1 μM dexamethasone for 24 h; DEX), and hyperlipidemia (palmitate, 0.15 mM for 18 h; PALM) defining four separate models of IR (Fig. 1 A and C).

Oxidative stress has been implicated in IR, but the precise source of these reactive species is not known. Since mitochondria are also implicated in IR and are a rich source of oxidative end products, we set out to test the hypothesis that mitochondrial oxidative stress may link these two processes as a major cause of IR. Mitochondrial oxidative species are primarily formed by the escape of high energy electrons from complex I and/or complex III of the electron transport chain (ETC). These electrons reduce molecular oxygen creating O₂^•−. Therefore, we examined the level of mitochondrial O₂^•− in each of our IR models using the mitochondria-targeted O₂^•−-sensitive fluorophore MitoSOX Red. MitoSOX Red is hydroethidine (HE) coupled to a mitochondrial targeting lipophilic cation triphenylphosphine (TPP) (10). With one unpaired electron, the O₂^•− radical selectively oxidizes the nonfluorescent HE moiety to form the fluorescent product 2-hydroxyethidium (2-OH-E). This increase in fluorescence was used as an index of relative mitochondrial O₂^•− production. Increased mitochondrial O₂^•− production was observed in each of our models of IR (Fig. 1 B and D) signifying an association between IR and mitochondrial O₂^•−.

To determine if mitochondrial O₂^•− is upstream of IR we used a pharmacologic approach to decrease mitochondrial O₂^•− focusing on palmitate-induced IR in L6 muscle cells. After induction of IR with 12 h of PALM treatment (Fig. 2A) we added drugs aimed at decreasing mitochondrial O₂^•− for an additional 6 h (Fig. 2B). Since the ETC is a major site of O₂^•− production we first sought to determine if this was the source linked to IR. We used the mitochondrial uncoupler carbonyl cyanide p-trifluoromethoxyphenylhydrazone (FCCP) and the respiratory chain complex I inhibitor rotenone. FCCP allows protons to leak back into the mitochondria independent of ATP synthase increasing the speed and efficiency of electron flow and oxygen consumption (11). This results in a more oxidized state of the ETC with less O₂^•− production (12–14). Rotenone also reduces O₂^•− production and electron flow by partially inhibiting complex I (13, 15). FCCP and rotenone reversed PALM-induced IR in these cells (Fig. 2C) concomitant with reduced mitochondrial reactive oxygen species (ROS) production (Fig. 2D).

Next we tested the effects of a series of O₂^•− scavengers on IR. The mitochondria-targeted superoxide dismutase (SOD) mimetic MitoTEMPO (2-(2,2,6,6-tetramethyl-piperidin-1-oxyl-4-ylamino)-2-oxoethyl)-TPP) reversed IR in a dose dependent manner whereas the TPP moiety or TEMPOL had no effect (Fig. 2E). Two other mitochondria-penetrating SOD mimetics, manganese 5,10,15,20-tetrakis (4-benzoic acid) porphyrin (MnTBAP, 300 μM) and manganese (III) tetrakis(1-methyl-4-pyridyl)porphyrin (MnTMPyP, 100 μM) (16, 17), also reversed PALM-induced IR (Fig. 2E). Since MnTBAP can substitute for the endogenous mitochondrial MnSOD in vivo (18) we tested this compound against a range of IR models including CI, TNF, and DEX in both 3T3-L1 adipocytes and L6 myotubes and found that MnTBAP reversed all models of IR (Fig. 2 F and G).

These data suggest that mitochondrial O₂^•− is a common feature across multiple models of IR. To confirm the findings obtained with these drugs and antioxidants we used a more specific genetic strategy by overexpressing MnSOD. Overexpression of MnSOD by 2.5-fold had no significant effect on insulin-stimulated GLUT4 translocation in control L6 cells but overcame the ability of CI, TNF, DEX, and PALM to impair insulin-stimulated GLUT4 translocation (Fig. 2H). In insulin resistant cells but not in control cells, overexpression of MnSOD led to an overshoot in insulin-stimulated GLUT4 translocation (Fig. 2H). One possibility is that these IR models still induce mitochondrial O₂^•−, however its rapid conversion to hydrogen peroxide (H₂O₂) by the overexpressed MnSOD may somehow potentiate insulin action. Such a model has been described for growth factor-induced H₂O₂ production, which inhibits tyrosine phosphatases to enhance signal transduction (19). However, overexpression of MnSOD overcame IR without significantly modulating insulin signaling (Fig. 2I), suggesting that if H₂O₂ is responsible for the overshoot it must be mediating its effects via an alternate mechanism.

The above findings show that many models of IR are associated with increased mitochondrial O₂^•−, which if reversed may restore insulin sensitivity. We next investigated whether the induction of mitochondrial O₂^•− within the ETC under defined experimental conditions is sufficient to drive IR. Complex III (cytochrome bc₁) of the ETC is a well established source of mitochondrial O₂^•− so we selectively induced O₂^•− formation with the Complex III Q_i site inhibitor antimycin A (AntA) (20, 21). Low doses of AntA (50 nM) caused a rapid (30 min) decrease in insulin-stimulated GLUT4 translocation in L6 myotubes (Fig. 3A) concomitant with increased O₂^•− production from mitochondria (Fig. 3B). To exclude the possibility that partial inhibition of mitochondrial electron transfer rather than O₂^•− production explained the effects of AntA, we performed three further experiments. First, stigmatellin, which blocks AntA-induced O₂^•− production by inhibiting the transfer of electrons to the AntA-binding site of complex III (21), was shown to prevent AntA-induced IR (Fig. 3C). Second, the mitochondrial SOD mimetics MnTBAP and MitoTEMPO were protective against AntA-induced IR (Fig. 3D), and third, overexpression of MnSOD in L6 cells prevented AntA-induced IR (Fig. 3E). Again, MnSOD overexpression led to supercompensation of insulin-stimulated GLUT4 translocation only in cells treated with an insult (Fig. 3E), similar to the effect seen with other IR models (Fig. 2H). Notably, the inhibitory effect of AntA on insulin-stimulated GLUT4 was not due to inhibition of signal transduction through Akt as evidenced by normal Akt phosphorylation at both maximal (Fig. 3F) and submaximal insulin concentrations (Fig. S1) in the presence of AntA.

The studies described above provide strong evidence for the involvement of mitochondrial O₂^•− production in multiple models of IR in muscle and fat cells in vitro. To confirm these findings in a physiological setting we used three approaches. First, we showed that acute administration of MnTBAP to high fat fed mice caused a significant improvement in glucose tolerance (Fig. 4 A and B) principally due to increased peripheral insulin sensitivity in muscle and fat (Fig. 4 C and D). Next, we examined the effects of MnSOD overexpression using transgenic (TG) mice expressing this enzyme under the control of the endogenous mouse promoter. The MnSOD-TG mice displayed a 2- to 2.5-fold increase in MnSOD expression in muscle and fat (Fig. S2A), normal body weight (Fig. S2B), and similar glucose tolerance vs. wild-type (WT) mice fed a standard low fat diet (LFD, Fig. 4E). We then challenged these same mice with a high fat diet (HFD) for one week. WT animals had impaired glucose tolerance under these conditions; however, this was not the case for MnSOD-TG mice (Fig. 4F). Importantly, there was no change in food intake (Fig. S2C), body weight gain (Fig. S2B), or insulin secretion (Fig. 4G) during the glucose tolerance test. To determine if MnSOD overexpression also protected against the effects of long-term high-fat feeding, we subjected another cohort of mice to a standard LFD or HFD for 12 weeks. Both WT and MnSOD-TG mice accrued the same amount of body mass (Fig. S2D), however, the MnSOD-TG mice were partially protected against HFD-induced glucose intolerance compared to WT animals (Fig. 4H). Accordingly, insulin tolerance tests showed that MnSOD-TG remained more insulin responsive than WT controls when fed a HFD for up to 24 weeks (Fig. 4I). These effects occurred despite no significant change in circulating triglyceride or free fatty acid levels between WT and MnSOD-TG mice fed chow or HFD (Fig. S3). These data confirm our in vitro studies where L6 muscle cells overexpressing MnSOD were also protected from IR.

To examine whether reduced mitochondrial O₂^•− scavenging capacity might be sufficient to cause IR in vivo, we tested glucose tolerance in MnSOD deficient mice. Homozygous deletion of MnSOD is lethal, whereas MnSOD^−/+ mice have no significant change in longevity or body weight regulation and breed normally (22). MnSOD^−/+ mice displayed a 70% reduction in MnSOD protein in muscle and fat compared to WT littermates (Fig. S1E), that was associated with a significant impairment in glucose tolerance on standard LFD (Fig. 4J) despite similar insulin levels to control mice (Fig. 4K).

3T3-L1 adipocytes and L6 myotubes overexpressing HA-GLUT4 were used for all in vitro experiments. HA-GLUT4 is essential for the measurement of GLUT4 at the plasma membrane (PM) as described in ref. 5. Models of IR were performed as described here. Chronic insulin (CI) treated cells were cultured in serum free DMEM media with 0.2% BSA (stepdown media) and given doses of 10 nM insulin at 12 PM, 4 PM, 8 PM, and 8 AM the following day. At 12 PM the cells were washed three times with PBS and incubated in stepdown media for 90 min before experimentation. Palmitate (PALM) treatment was performed essentially as described in ref. 5, with the exception that the cells were serum free for the entire 18 h of treatment. Dexamethasone (DEX) treatment consisted of 24 h incubation with 1 μM dexamethasone in stepdown media before experimentation. Finally, inflammation was mimicked with 2 ng/mL tumor necrosis factor α (TNF). TNF was added at 1:30 PM each day for 4 days with the final day in SF media. In experiments where drugs were added to the IR models the compounds remained on the cells with the insults for the entire duration of the final 6 h, except in the case of CI where the drugs were washed out during the PBS washes but replaced for the final 90 min. MnSOD cDNA was obtained from JA Melendez (Albany Medical College, NY) and cloned into the retroviral vector pBABE-puro. MnSOD and empty vector control retroviruses were produced in Platinum-E cells and used to infect L6 myoblasts overexpressing HA-GLUT4 driven from a pWZL-derived retrovirus. Cells were stably selected with G418 (400 μg/mL) and puromycin (2 μg/mL) for >3 passages before differentiation and experimentation. TNF and DEX treatment increased cellular GLUT4 levels by 22 and 94%, as described in ref. 5, however in all experiments cell surface GLUT4 was normalized to total cellular GLUT4 thus taking into account changes in GLUT4 expression.

MitoSOX Red was administered essentially as described by the manufacturer (Molecular Probes); however, the two cell lines required different concentrations and incubation times. L6 myotubes were incubated with 5 μM MitoSOX Red for the final hour of IR treatment while 3T3-L1 adipocytes were incubated with 1 μM MitoSOX Red for the final 30 min of treatment. Cells were cultured in low-absorbance, black-walled 96-well plates. After MitoSOX treatment cells were quickly washed with PBS and fluorescence was detected on a Fluo-Star plate reader with excitation at 390 or 485 nm and emission settings at 590 nm.

ROS production by isolated mitochondria was measured by monitoring Amplex Red oxidation to Resorufin in the presence of HRP (ex/em at 540/590 nm). Mitochondria were isolated from L6 cells using a cell culture mitochondrial isolation kit (Pierce). Isolated mitochondria were resuspended in KCl buffer (125 mM KCl, 20 mM HEPES, 2 mM MgCl₂, 2 mM KH₂PO₄, and 40 μM EGTA, pH 7.2) containing 1 mM pyruvate and 7 mM succinate and diluted to 0.5 mg/mL. ROS production was measured by adding 50 μM Amplex Red and 0.1 U/mL HRP and monitoring Resorufin fluorescence over time.

Antibodies against p-Akt and p-GSK3 were from Cell Signaling. pT642-AS160 antibody was from Symansis. MitoSOX Red was from Invitrogen. MitoTEMPO, MnTBAP, and MnTMPyP were purchased from Alexis Biochemicals. Rotenone, Stigmatellin, AntA, FCCP, and TEMPOL were purchased from Sigma. MnTBAP was prepared in haemaccel for use in mice and water for use in cell culture.

MnTBAP experiments were performed in male C57BL/6 mice at the Garvan Institute and experiments in MnSOD-TG and MnSOD^−/+ mice were performed at the University of Texas Health Science Center at San Antonio (UTHSCSA). Food and water were provided ad libitum until the date of study and all animal care was in compliance with the US National Institutes of Health and the Australian National Health and Medical Research Council guidelines, as well as institutional guidelines at both Garvan and UTHSCSA. i.p. glucose tolerance tests were performed after a 6 h fast (8 AM–2 PM) and insulin was measured using an ultrasensitive mouse insulin ELISA from Crystal Chem. The HFD used at the Garvan (45% kcal from fat) was made in house as described in ref. 33 and the Garvan LFD (8% calories from fat) was from Gordon's Specialty Stock Feeds. The MnSOD-Tg mice and WT controls were fed HFD (45% kcal from fat, 58V8) or LFD (10% kcal from fat, 58Y2) from Purina Test diets. The MnSOD^−/+ and WT controls were fed Harlan Teklad 7912.

Data are expressed as means ± standard error. For GLUT4 translocation analysis each experiment was normalized to maximal insulin stimulation of control cells and this value was set to 100%. P values were calculated by two-tailed Student's t test using Microsoft excel (Microsoft Corp) or Graph Pad Prism. Statistical significance was set at P < 0.05.