Control of Muscle Mitochondria by Insulin Entails Activation of Akt2-mtNOS Pathway: Implications for the Metabolic Syndrome

Background In the metabolic syndrome with hyperinsulinemia, mitochondrial inhibition facilitates muscle fat and glycogen accumulation and accelerates its progression. In the last decade, nitric oxide (NO) emerged as a typical mitochondrial modulator by reversibly inhibiting citochrome oxidase and oxygen utilization. We wondered whether insulin-operated signaling pathways modulate mitochondrial respiration via NO, to alternatively release complete glucose oxidation to CO2 and H2O or to drive glucose storage to glycogen. Methodology/Principal Findings We illustrate here that NO produced by translocated nNOS (mtNOS) is the insulin-signaling molecule that controls mitochondrial oxygen utilization. We evoke a hyperinsulinemic-normoglycemic non-invasive clamp by subcutaneously injecting adult male rats with long-lasting human insulin glargine that remains stable in plasma by several hours. At a precise concentration, insulin increased phospho-Akt2 that translocates to mitochondria and determines in situ phosphorylation and substantial cooperative mtNOS activation (+4–8 fold, P<.05), high NO, and a lowering of mitochondrial oxygen uptake and resting metabolic rate (−25 to −60%, P<.05). Comparing in vivo insulin metabolic effects on gastrocnemius muscles by direct electroporation of siRNA nNOS or empty vector in the two legs of the same animal, confirmed that in the silenced muscles disrupted mtNOS allows higher oxygen uptake and complete (U-14C)-glucose utilization respect to normal mtNOS in the vector-treated ones (respectively 37±3 vs 10±1 µmolO2/h.g tissue and 13±1 vs 7.2±1 µmol 3H2O/h.g tissue, P<.05), which reciprocally restricted glycogen-synthesis by a half. Conclusions/Significance These evidences show that after energy replenishment, insulin depresses mitochondrial respiration in skeletal muscle via NO which permits substrates to be deposited as macromolecules; at discrete hyperinsulinemia, persistent mtNOS activation could contribute to mitochondrial dysfunction with insulin resistance and obesity and therefore, to the progression of the metabolic syndrome.


Introduction
The powerhouse of the cell, mitochondria are responsible for sustaining energy levels. However, a number of the new functions discovered for mitochondria in the past decade depend not on energy demand but on adjustments to respiration [1][2]. Critical reduction of respiration takes part as well in the mechanisms of prevalent illnesses; a reduction of mitochondrial activity and the decrease in energy expenditure contribute substantially to metabolic dysfunction in aging, insulin resistance and diabetes and conducts to lipid accumulation [3]. Of the many studies focused on insulin resistance and mitochondrial dysfunction in the last decade [4], few have critically examined mitochondrial activity in terms of alternation between complete substrate oxidation and storage as macromolecular deposits [glycogen or fat stores]. A mutual connection between decreased respiratory rate and the synthesis pathways was reported by Petersen et al who found that 40% reduction in mitochondrial oxidative and phosphorylation activity increased intramyocellular lipid content in elderly volunteers with severe muscle insulin resistance, compared with matched young control subjects [5]. Although lipid and glycogen accumulation further increases insulin resistance [6], a clear mechanism for reduction of mitochondrial oxidations with displacement of substrates to storage was not defined in this context yet.
With high affinity for O 2 , cytochrome c oxidase is responsible for the final transference of electrons to oxygen for reduction to water in mitochondria. In the last years, cumulative evidence shows that mitochondrial NO is a typical regulator that binds cytochrome oxidase at subunit II [7], competes with high-affinity for oxygen occupancy, and thus reversibly inhibits cytochrome oxidase and reduces respiration at nanomolar concentrations; at equal demands, physiologic in vivo oxygen utilization by mitochondria depends on the matrix NO/O 2 ratio [8].
Although the diffusible NO yielded by cytosol nitric oxide synthase (NOS) isoforms modulates respiration in vivo [9], an advantage comes from the mitochondrial compartmentalization of NOS (mitochondrial nitric oxide synthase; mtNOS) [10] that directs NO flux vectorially to the matrical space. In particular, neuronal nitric oxide synthase [nNOSa] binds complexes I [11] and IV [12] through specific domains; most of the mitochondrial NO yield is abolished by targeted disruption of the nNOSa gene [13]. Likewise, activation of mtNOS and enhanced mitochondrial NO levels were previously demonstrated in normal rat development, hypothyroidism, cold acclimation, and hypoxia [14][15][16]. At reduced mitochondrial respiratory activity, mtNOS modulation should therefore be equally important for insulin effects. Matrix NO reduces oxygen uptake and ATP levels [17], thereby increasing the level of reduced equivalents and acetyl-S-coenzyme A (acetyl-CoA), the substrates for anaplerotic reactions and synthesis pathways stimulated by insulin. Moreover, in rat mitochondria, mtNOS is phosphorylated in an Akt-sensitive domain at Ser 1412 . The Akt/protein kinase B [PKB], a serine/ threonine protein kinase with high homology for protein kinases A and C, was identified as the cellular homologue of the viral oncoprotein v-Akt (herein referred to as Akt). It is now clear that Akt1 and Akt2 have distinct functions, that both kinases are effectors of the insulin-PI3K pathway and that, Akt2 deficiency associates to a diabetes mellitus-like syndrome [18].
We therefore link here for the first time the mtNOS activity that results from nNOS confinement to muscle mitochondria with the respective rates of oxidative utilization of glucose or of glycogen synthesis induced by insulin. We postulate that insulin signaling activates Akt2, which not only activates the GLUT4 recycling pathway but also increases NOS activity in the organelles in an attempt to restrain mitochondria, thus favoring the replenishment of glycogen and fat energy stores. To that, we used a rat model of hyperinsulinemia with normal glycemia that allowed testing lasting effects of insulin on mitochondria. We succeeded in creating an in vivo comparative metabolic condition in which nNOS-and mtNOS-deficient muscles increase glucose and oxygen utilization and reciprocally restrict glycogen-synthesis providing molecular basis for understanding a NO-dependent progression of metabolic syndrome and type 2 diabetes.

Insulin increases p-Akt2 in skeletal muscle mitochondria
Under fairly constant stimulation, insulin early increased the expression of Akt2 and phospho-Akt (p-Akt) in cytosol and mitochondria from skeletal muscle; thereafter, only mitochondrial Akt2 and p-Akt remained very high up to twelve hours after insulin (by about 4 and 6 folds, respectively). In contrast, insulin effects on Akt1 were poor and non-significant in the different fractions ( Figure 1A). Flow cytometry of the isolated and purified mitochondria confirmed a net increase of p-Akt fluorescence, with a similar temporal kinetics to that detected by western blotting. (Figure1B). To test p-Akt2 activity under these experimental conditions, we measured the activation of the natural target of Akt, glycogen synthase kinase (GSK-3a/b), to its phosphorylated form (p-GSK-3a/b). Early in the experiment, p-GSK-3a/b peaked up in cytosol and mitochondria and those levels declined for the remainder of the experiment ( Figure 1C). Submitochondrial fractionation demonstrated that p-Akt is localized at the inner mitochondrial membrane ( Figure 1D). Selective increase of muscle Akt2 suggests the in vivo accumulation of mitochondrial p-Akt2 in the insulin-stimulated skeletal muscle.

Insulin boosts nNOS activity in mitochondria
We next examine the relationship between the enhanced insulin-p-Akt2 pathway and the mitochondrial nNOS activity. At maximal mitochondrial p-Akt2, mtNOS activity proportionally increased by about 8 fold and remained high up to 24 hours ( Figure 2A). Accordingly, high mitochondrial NO concentration was revealed by flow cytometry of isolated organelles with 4amino-5-methyl-amino-29, 79-difluorescein diacetate (DAF-FM) ( Figure 2B). To link NO to metabolic signaling, the muscle was challenged with insulin and the phosphatidyl-inositide 3 -kinase (PI 3 K) inhibitor LY294002, that resulted in a considerable 75% reduction in mitochondrial NOS activity and in blocking p-Akt accumulation in the organelles ( Figure 2C and 2D). Taking into consideration that insulin elicits antagonistic responses, we searched for the optimal dose to obtain maximal NOS activity in mitochondria ( Figure S1). Interestingly, remarkable effects on mtNOS were precisely achieved at half-dose insulin while they were less significant at the highest dose. Increase of NOS activity was not associated with changes in protein expression ( Figure 2E), and other isoforms were not found in the organelles (not shown).
To quantify the inhibitory effects of NO on the electron transfer chain, we measured O 2 uptake in muscle slices and mitochondria. Muscle O 2 uptake peaked 3 hours after insulin administration (P,.05 respect control), but significantly decreased at 12-24 h (P,.05 either respect control or 3 h) ( Figure 2F). In accord, at 12-24 h insulin-administered animals with maximal mtNOS activity had only 40-50% of the mitochondrial respiratory rate of control non-treated animals (P,.05) ( Figure 2G). Likewise, in those periods O 2 utilization was maximally restricted to 20-40% of the control values by supplementation of organelles with L-arginine. Dependence of NO was as well revealed by the fact that O 2 inhibition was significantly relieved by the NOS inhibitor Nmonomethyl-L-arginine (L-NMMA) (P,.05) ( Figure 2G). To discern the effect of mtNOS in the different periods, we calculated the sum of opposite effects of NOS substrate and inhibitor on basal O 2 utilization that determines the mitochondrial NOS functional activity on respiration (11): In accord to Figure 2G, mtNOS functional activity on respiration was maximal at 12 (91%) and 24 h (46%) post insulin treatment while at 3 h, it did not change respect to the control values (19%).
To test whether these effects actually depend on p-Akt2 and whether p-Akt2 acts directly on NOS within mitochondria, we next incubated organelles from control muscle ex vivo with recombinant active Akt1 and Akt2 phosphorylated at Ser 473 and Thr 308 . In the presence of ATP and substrate, both phosphorylated Akt1 and Akt2 translocated to energized mitochondria and were detected even when incubated with proteinase K, whereas dephosphorylated Akt1 and Akt2 remained in the supernatant ( Figure 3A); proteinase K removed non-translocated proteins that associate to the cytosol face of the outer mitochondrial membrane. In addition, when mitochondria charged with active Akt1 or Akt2 were exposed to DAF-FM, a peak of fluorescence was detected only in those treated with phosphorylated Akt2; this effect was not observed in mitochondria supplemented with dephosphorylated Akt and dissapeared by co-incubation with NOS inhibitor L-NMMA ( Figure 3B). These results indicate that, at constant nNOS expression, subcellular distribution of p-Akt2 accounts for the temporal kinetics of nNOS activation in cytosol and mitochondria.

Mitochondrial nNOS is phosphorylated by translocated p-Akt2: a cooperative kinetics
To confirm that the increase in mtNOS activity was a product of nNOS phosphorylation by translocated p-Akt, mitochondria from insulin-administered animals were supplemented with cold ATP and 15 mC c 32 P-ATP and additional recombinant Akt2 before being subjected to autoradiography. In these conditions, a spontaneous signal of phospho-mtNOS was detected in skeletal muscle mitochondria with a high p-Akt2 charge ( Figure 4A), that was further increased by exogenous recombinant p-Akt2. To confirm specificity, mitochondria were also isolated from nNOSdeficient muscle directly electroporated with nNOS siRNA. The right legs of rats were injected with nNOS siRNA and the left ones with an empty vector, and both were subjected to direct electroporation in vivo. These mitochondria exhibited low mtNOS content and therefore the Akt2-induced phosphorylation signal substantially decreased ( Figure 4B). In accord, direct injection of Akt2 siRNA produced a dose-dependent inhibitory effect on mtNOS activity (60% at the highest dose), similar to the reduction of p-Akt2 and p-GSK3 ( Figure 5A).
This effect suggests that the insulin-dependent increase in NOS activity is based upon the kinetic variations induced by Akt2related enzyme phosphorylation in mitochondria ( Figure 5A). To test this hypothesis, His-tagged nNOS was purified from bacteria transfected with cDNA in pCWori+ vector, and enzyme activity was assayed for phosphorylation in the presence of p-Akt2 and c 32 PATP. The pioneer work by Fulton et al reported that Akt activates eNOS by phosphorylating Ser 1179 in transfected COS cells [19]. In their study, analogous phosphorylation of nNOS was observed only when the transfected construct had a myristoylation  consensus site, indicating that attachment to membranes is also required for enzyme phosphorylation. Elfering et al reported that mtNOS is myristoylated and has only one Akt-sensitive domain (RXRXXS/T) at Ser 1412 , which has been reported to be phosphorylated in the Maldi/MS analysis [20]. We therefore measured K' for L-arginine in phosphorylated or non-phosphorylated recombinant nNOS in the presence of BH 4 and cholesterol, as a lipid phase representative of membrane insertion ( Figure 5B). The S1412G nNOS mutation, which simulates phosphorylation, increases the heme reduction rate and decreases NO yield and [21] electron transfer in the calmodulin-bound state. Here, a sigmoid kinetic curve was noted for both the phosphorylated and non-phosphorylated nNOS, but with different Hill slopes (3.7 and 1.9), findings that suggest a cooperative behavior (r 2 = 0.99 and 0.97); in the lipid phase, p-Akt2 reduced K' for L-Arg from 19 to 11 mM. mtNOS is the final insulin/Akt effector that regulates muscle utilization of available glucose A cornerstone of this study is that, after energy replenishment, mitochondrial oxidations ought to be depressed to allow substrates to be deposited as macromolecules. We tested the contribution of mtNOS to the balance maintained between mitochondrial and intermediary metabolism, by injecting siRNA nNOS into rat gastrocnemius muscles with high nNOS content. This procedure reduced nNOS mRNA levels as well as protein expression and activity in mitochondria by approximately 50-77% in comparison to constant complex IV ( Figure 6A). Under these experimental conditions and because of the decreased levels of matrix NO ( Figure 2B), oxygen uptake became insensitive to insulin administration and remained high in the nNOS-silenced muscles. The increase of glucose uptake elicited by insulin administration was not modified by siRNA nNOS. Instead, siRNA nNOS markedly increased glucose utilization evidenced by complete oxidation to CO 2 and H 2 O. Thus, at fixed insulin with similar glucose uptake and glycogen synthase activity, high oxygen utilization with siRNA nNOS was in line with high 3 H 2 O and 14 CO 2 production rates, i.e., 50-100% higher in the right siRNAtreated muscle than in the left control ( Figure 6B). No significant variations of metabolic rate were detected at low mtNOS activity. Interestingly, under these conditions the glucose used for glycogen synthesis was halved in the siRNA nNOS-treated muscle; at normal mtNOS content similar effects were observed when the Akt2 gene was silenced, reducing protein expression and phosphorylation of GSK-3a/b by 56% and 75%, respectively ( Figure 6C). In accordance with previous reports, direct administration of siRNA Akt2 did not result in significant variations of glucose uptake or glycogen synthase activity, although the peak of glucose uptake decreased at 24 h in the deficient muscles ( Figure 6C).

Discussion
We herein demonstrate the modulatory connection between insulin signaling and mitochondrial function: insulin decreases the muscle oxidative rate via mitochondrial NO. We confirmed this finding using a number of different experimental approaches, showing that the effect relies on p-Akt2-selective phosphorylation of mitochondrial nNOS, after kinase translocation to the organelles.
A rapid translocation of p-Akt to the mitochondria and the phosphorylation of target GSK-3a/b by insulin-PI3K were previously reported by Bijur and Jope in a variety of cells [22]. It is confirmed here that phosphorylation is absolutely required for Akt translocation to muscle mitochondria and that, in this framework, both p-Akt1 and p-Akt2 can be translocated to the organelles ex vivo ( Figure 3A). In vivo however, insulin only increases p-Akt2 probably because of high expression of SH2-phophatase that selectively dephosphorylates p-Akt1, but not p-Akt2 in skeletal muscle [23]. Moreover, solely translocated p-Akt2 and not p-Akt1 induced a robust generation of NO by mtNOS in the isolated mitochondria ( Figure 3B). Considering that eNOS is activated in the caveolae by insulin/PI3K via p-Akt1, specific functional and topographic connections between NOS and Akt isoforms become evident. In this way, a recent study reported p-Akt2 phosphorylation of nNOS at Ser 1412 in synaptosomes, suggesting the functioning of this mechanism in different organs and subcellular fractions [24].
The phosphorylation of nNOS/mtNOS by p-Akt2 involves cooperative effects centered at the C-terminal tail that contains the Akt motif (Ser 1412 ; Figures 4B and 7A). Although structural changes of phosphorylated nNOS are not defined yet, it is worth noticed that C-terminal tail participates in the regulation of nNOS activity. Allosteric inhibition of electron transfer from NADPH to FAD and poor NO release at low Ca 2+ -calmodulin level result from electrostatic interactions between acidic negative charges of 29phosphate of NADP and positive charges of basic Arg 1400 [25]. On this basis, we surmise that p-Akt2 activation of nNOS/mtNOS (Figures 4 and 5) promote opposed electrostatic interactions between Ser 1412 and Arg 1410 that, in the lipid phase, equilibrates the oscillation of the C-terminal tail to cooperatively increase the NO production rate ( Figure 7A).
Different approaches confirmed in this study that phosphorylation of muscle mtNOS increases mitochondrial NO yield by several folds and reduces systemic O 2 utilization. We show here as well that the temporal activation of mtNOS and the resulting decline of mitochondrial O 2 uptake via NO drive the muscle insulin-transition from glucose oxidation to glycogen deposition. In absence of NO or, in nNOS-silenced muscles, mitochondrial O 2 uptake is completely released leading to a preferential oxidation of glucose to CO 2 and H 2 O; in these conditions, prevalent oxidation of glucose reflects the respective Km of regulatory enzymes like cytochrome oxidase for O 2 and pyruvate dehydrogenase complex for acetyl-CoA, at least one order of magnitude lower (10 27 -10 28 M, ref 26), than that of glycogen synthase for glucose (10 25 M, ref 27). In contrast, in the presence of NO, glucose utilization is delivered to glycogen synthesis. These results therefore support the notion that a discrete inhibition of the respiratory chain as produced by NO is required to reduce oxidative rates and to allow glycogen or fat deposition ( Figure 7B). At ten-fold higher insulin dose, maximal mtNOS activation was reduced by a half ( Figure S1). This fact reminds that insulin activates other mitochondrial reactions as well; i.e., early activation of pyruvate dehydrogenase complex through insulin-inhibition of pyruvate dehydrogenase kinase (PDK4) could transiently compete with NADH/NADPH availability for NOS in skeletal muscle mitochondria [28]. In this study, effects at low and high insulin concentrations thus explain: a) lack of stimulation and null mtNOS activity in muscle at initial blood insulin peak (3 hours); b) high respiratory rate at peak insulin level with null mtNOS ( Figure 2G); c) decreased respiratory rate at stable insulin concentration with full mtNOS and, d) physiological alternation of insulin effects promoting complete glucose oxidation or glycogen synthesis (Figures 2 and 6). The increase or decrease of oxygen consumption was previously reported to occur in muscle at respectively high or low insulin concentration [29].
The presented data confirm insulin as a normal regulator of oxidative phosphorylation rate in skeletal muscle but as well they could explain cardinal features of the progression of metabolic syndrome and obesity to type 2 diabetes. Zucker Fatty (ZF) rats with a leptin receptor defect exhibit obesity, insulin resistance with hyperinsulinemia and normal glycemia that progress to type 2  diabetes in 6-12 weeks (4); ZF rats have severe mitochondrial inhibition with low oxidation of palmitic acid to CO 2 [30]. Moreover, leptin-deficient ob 2 /ob 2 obese mice exhibit insulin resistance and obesity associated to hyperinsulinemia with mitochondrial inhibition [31]. Preliminary results of our group demonstrated that ob 2 /ob 2 mice with metabolic syndrome have increased expression of mtNOS and high mitochondrial NO yield and that mitochondrial inhibition is released by NOS inhibitors or by leptin infusion, both decreasing NO yield in the organelles [32]. As in genetic experimental models of metabolic syndrome, the term mitochondrial dysfunction has been referred in humans as to a condition with decreased oxidation rate and ATP synthesis [33] and, in this context, reduced respiration was reported to contribute to glycogen or fat deposition [3] and can be independent of insulin resistance [34]. Considering the metabolic syndrome as accompanied by insulin resistance with prolonged hyperinsulinemia, such a mitochondrial dysfunction appears to be analogue to the persistent inhibitory effects on mitochondrial respiration demonstrated by the presented data at intermediate insulin concentration.
Our findings therefore suggest that the mechanisms for progression to severe mitochondrial dysfunction in type 2 diabetes are contributed by NO-dependent mechanisms. We reported that NO markedly increases superoxide anion yield in rat skeletal muscle mitochondria by inhibiting electron transfer at Complexes I and III [35], finally damaging Complex I [36]. Increased mitochondrial superoxide anion was related to hyperglycemia and insulin resistance, as well as to persistent mitochondrial dysfunction [37]. In the presence of superoxide anion plus nitric oxide, 3-nitrotyrosine was found as a marker of nitration of mitochondrial proteins in the ob 2 /ob 2 metabolic syndrome [31] and in experimental and clinical diabetes [38]. Excessive NO leads both to nitrosylation and inactivation of the insulin receptor and Akt, increasing insulin resistance [39]. A number of different NO-based mitochondrial mechanisms can contribute to insulin resistance as well. First, an increase of ADP/ATP ratio due to inhibition of electron transfer [17] can restrict the phosphorylation of insulin receptor substances (IRSs) and the translocation of GLUT4 to the cell membrane. Second, accumulation of NADH due to NO-dependent complex I inhibition can negatively modulate glycolisis and the Krebs cycle [40]. Finally, nitric oxide stimulates mitochondrial biogenesis by increasing PGC 1a coactivator and thus could compensate mitochondrial dysfunction in the early metabolic syndrome, while reduced biogenesis with further decay of mitochondrial population has been reported in already established type 2 diabetes [41]. From these perspectives, an additional explanation is required for progression of diabetes and tissue damage in hypoinsulinemic status, like type I diabetes, Akt2 deficiency or nNOS knocked out mice. In contrast to activation of insulin-NOS pathway in mitochondria, in these models progressive tissue and mitochondrial damage should not depend on an increased ADP/ATP ratio; it is worth noticed that Akt2 has extramitochondrial effects on GLUT4 and facilitates per se the mechanism of glucose uptake. However, it is apparent as well that weight loss in low-insulin diabetes or severe lypoatrophy in hypoinsulinemic Akt2-deficient mice could occur because of the lack of NOS activation, leading to uncontrolled mitochondrial respiration and to high oxidative stress in the organelles with a marked reduction of glycogen content [18].
We conclude that increase of NO in the mitochondrial compartment by insulin is a powerful physiological resource to properly adjust muscle O 2 utilization, while persistence of this mechanism in hyperinsulinemic states is harmful and could contribute to diabetic mitochondrial dysfunction.

Animals and treatments
Animals were housed in a temperature-controlled room, provided food, and subjected to a 12:12 dark/light cycle. National Institutes of Health criteria for animal research were followed after approval by the University Hospital.
To achieve a non-invasive hyperinsulinemic-normoglycemic clamp, male Wistar rats [250-300g body wt] were subcutaneously inoculated with long-acting human analogue insulin glargine (Supporting Figure S2, [42][43][44] or NaCl 0.9% and skeletal muscle was excised at 3, 12 and 24 hours. Insulin glargine, with its three modifications to human insulin (Gly A21 , ,Arg B31 , Arg B32 ) is a stable molecule that is soluble in slightly acidic conditions and precipitates in the neutral pH of subcutaneous tissue. Because of these properties, absorption of insulin glargine is delayed and the analogue provides a fairly constant, basal insulin supply without peaks in plasma insulin levels for approximately 24 hours, similar to that achieved by a continuous subcutaneous insulin infusion ( Figure S2). During the 24-hour period, active metabolites sustain insulin effects after the decay of the hormone. Animals were allowed to drink water with 5% sucrose ad libitum to avoid hypoglycemia ( Figure S2). Without glucose intake, insulin glargine results in hypoglycemia 2-4 h after administration (215 to 30% of basal values; 45). Plasma insulin levels were followed independently of endogenous hormone with a specific human immunoassay (IMX Microparticle Enzyme Immunoassay, Abbott Labs, Ramsey, MN). Because the cross reactivity between insulin glargine and human insulin is approximately 60%, thus resulting in underestimation of its level in the immune assay, obtained values were multiplied by 1.8 [43]. All experiments were carried out in rat extensor digitorum longus and gastrocnemius red-mixed muscles with appropriate mass (,0.6 g per muscle); adult rat muscle is comprised of approximately 4-50% slow oxidative fibers [I] and 20-40% fast oxidative fibers [IIa] with a large mitochondrial population, high glycogen content, and concomitantly, a high level of nNOS activity [46].

Akt/insulin Signaling Pathway and Nitric Oxide in Muscle Mitochondria
Muscle mitochondria were isolated from homogenized muscles by differential centrifugation and further purified with Percoll gradients. After purification, the mitochondrial fraction has no more than 10% activity of cytosol lactic dehydrogenase activity or 5% of calreticulin from SRE (western blot) while cytosolic fraction did not express complex I proteins or 39 kDa subunit I of cytochrome oxidase (western blot). Cytosol and mitochondrial expression of nNOS, p-Akt, Akt1, Akt2 and p-GSK-3 a/b was assessed by immunoblotting with specific antibodies and by detecting variations of NO and p-Akt, by flow-cytometry in freshly isolated mitochondria.
To analyze effects of PI3K pathway on mtNOS, muscles were dissected out and rapidly cut into 20 to 30 mg strips and incubated in a shaking water bath at 30uC for 30 min into a 25 ml flask containing 3 ml of oxygenated Robinson buffer supplemented with glucose and 0.1% BSA; strips were homogenized and mitochondria were obtained. Flasks were gassed continuously with 95% O 2 -5% CO 2 throughout the experiment.
NOS activity was followed in cytosol and purified mitochondria by conversion of  [11]. Ca 2+ -dependent specific activity was calculated by subtracting the remaining activity in the presence of the NOS inhibitor L-N G -methyl-L-arginine (5 mM of L-NMMA) or 2 mM of EGTA 3 H arginine assay. NOS activity was measured as the number of mitochondria isolated from rat muscle injected with insulin and from muscle incubated with insulin or LY29400 (PI3K-specific inhibitor). Muscles were dissected out and rapidly cut into 20-to 30-mg strips and incubated in a shaking water bath at 30uC for 30 min before being transferred into a 25-ml flask containing 3 ml of oxygenated Krebs-Ringer bicarbonate buffer (KRB) supplemented with 8 mM of glucose, 32 mM of mannitol and 0.1% BSA. Flasks were gassed continuously with 95% O 2 -5% CO 2 throughout the experiment. To measure the mtNOS activity, muscle mitochondria were isolated and purified after incubation, as described previously.
Total muscle RNA was extracted with Trizol, and RT-PCR was performed under standard conditions.

Translocation of Akts to mitochondria
Isolated organelles at 1 mg/ml were purified with percoll gradients and incubated ex vivo with pure recombinant active Akt1-GST or Akt2-GST (100 units/ml) (Cell Signaling) dually phosphorylated at Thr-308 and Ser-473 or with isoforms dephosphorylated with acid phosphatase and co-incubated with the NOS inhibitor L-NMMA. Mitochondria were then centrifuged and separated from supernatant, co-incubated with 50 mg/ml proteinase K and DAF-FM, and subjected to flow cytometry to assess fluorescence intensity. To allow translocation of Akt2 to the inner mitochondrial membrane [47], mitochondrial membrane potential was sustained by coincubating the isolated organelles with 2 mM NADH and 2 mM ATP without Mg 2 Cl.

Oxygen uptake
Oxygen uptake was measured at 220 mM of O 2 in a standard polarograph (IQUIFYB-MADEIC, Buenos Aires, Argentina) in sliced muscle with Robinson buffer or in isolated mitochondria with MSHE buffer with 6 mM of malate plus 6 mM of glutamate, with or without 0.2 mM of ADP, 0.3 mM of L-arginine or 3 mM of L-NMMA.

In vivo muscle siRNA electroporation
Animals were briefly anesthetized with sodium thiopental [40 mg/kg weight, intraperitoneal], and gastrocnemius muscles were percutaneously treated with one proximal and one distal injection of siRNA nNOS (10 mg/per muscle] [GenScript, New-Jersey, N.Y.) (right leg) or empty pRNAT-U6.1/Neo vector (left leg) or with Akt2 siRNA from Santa Cruz Biotechnology (Santa Cruz, CA) (right leg) or vehicle (left leg), in parallel. Each injection was performed in 5 ml free-RNAase solution, slowly in 15-20 s with a Hamilton syringe with a 22 G needle. Around 30 s after injection, electric pulsation was applied. For pulse delivery to muscle, two stainless steel connectors were applied on the needles implanted in muscle. Therefore, the needles were used as electrodes and connected to the generator; in these conditions the depth of insertion was about 0.5 cm and the gap between the two needles was 0.9 cm. The muscle were then electropulsed with selected parameters using the Gene Pulser II Electropulsator (Bio-Rad Laboratories,USA).Voltage (120V), pulse duration (20 ms) and frequency of pulses (1Hz) were all preset on the electropulsator. A train of eight pulses was delivered. [46]. The sequences of nNOS siRNAs were designed on the structure of rat nNOS g.i.: 16258811 and are provided in Figure S3.

Metabolic studies
Thirty six hours after electroporation, insulin was administered at appropriate times and finally, animals were anesthetized and gastrocnemius muscles were removed, weighed, and washed with NaCl (0.9%). Finally, muscles were excised and minced in 0. Determinations were done after 12 hours of fasting; to maintain both appropriate light-dark cycle and fasting, samples for glycogen synthesis after 12 hours of insulin administration were obtained at 6:00 PM and for glucose oxidation at 3:00 and 11:00 AM.
To measure glucose uptake, muscle was placed separately in the appropriate buffers and supplemented with radioactive deoxyglucose [48]. Glycogen synthase activity was measured using the method developed by Leloir and Goldemberg [49].

Statistical analysis
Data were compared with ANOVA and the Dunnett post hoc test or the two-tailed Student's t test as appropriate; significance was accepted at P,.05. Figure S1 Muscle mtNOS activity at different doses of insulin glargine. Each value represents separate experiments from animals with insulin (closed circles) and without insulin (closed squares) (n = 6, *P,.05 by ANOVA and Dunnett test). Strong activation of mtNOS selectively occurs at half-maximal insulin dose. Found at: doi:10.1371/journal.pone.0001749.s001 (0.17 MB EPS) Figure S2 Effects of long-lasting insulin in rats. (A) A noninvasive hyperinsulinemic normoglycemic clamp was achieved in rats by subcutaneous administration of 0.1 U/Kg of insulin glargine and 5% oral sucrose ad libitum; after an initial peak, level of insulin+metabolites remains remarkably stable by 20 hours. (B). The rate of sucrose intake of animals with or without (C) insulin (white bar), at the same dose but at different times (dark bars); *P,.05 respect control by ANOVA and Dunnett test). Found at: doi:10.1371/journal.pone.0001749.s002 (0.24 MB EPS) Figure S3 Sequence of hairpin siRNAs designed for rat nNOS (gi: 16258810). siRNAs were cloned in vector pRNAT-U6.1/Neo with BamH I and Hind III restriction enzymes; in red, poly A tail. siRNA # 1 was preferentially used in the in vivo studies because of its maximal inhibitory effect. Found at: doi:10.1371/journal.pone.0001749.s003 (0.33 MB EPS) Figure 6. mtNOS Is a final insulin/Akt effector regulating muscle utilization of available glucose. (A) Representative western blot of disrupted and normal mtNOS, densitometries, and mtNOS activities of mitochondria isolated from right and left gastrocnemius of the same animal, 36 h after direct electroporation of respectively 10 mg of siRNA nNOS or empty pRNAT-U6.1/Neo vector to muscle, as described in Experimental Procedures. (n = 6; densitometry: t = 6.55, DF = 10, P = 0.000; activity at 12 h: t = 6.07, DF = 10, P = 0.000; activity at 24 h: t = 2.22, DF = 10, P = 0.05) (B) To test the interdependence of oxidative and intermediary metabolism as related to activation of mtNOS, metabolic studies including oxygen uptake, glycogen synthesis, complete glucose oxidation to CO 2 and H 2 O, glucose uptake and glycogen synthase activity were performed in nNOS-silenced right gastrocnemius-and vector-administered left muscles of the same animal in parallel under the different conditions by radioactive methods (n = 6; vector: O 2 uptake, F = 10.47, DF = 23, P = 0.000; 14  The threshold for significance is the same as that described in Figure 4.*P,.05 vs controls or between nNOS or Akt2 silenced and not-silenced muscle samples; ** P,.05 between different times of insulin administration. doi:10.1371/journal.pone.0001749.g006

Acknowledgments
We are grateful to Dennis Stuehr (Ph.D) for providing us with cloned nNOS, to Daniel De Seta for insuline glargine and to Natalia Riobó (Ph.D) and Juan J. Gagliardino (Ph.D) for their helpful comments.