Inhibition of Inducible Nitric-oxide Synthase by Activators of AMP-activated Protein Kinase A NEW MECHANISM OF ACTION OF INSULIN-SENSITIZING DRUGS*

AMP-activated protein kinase (AMPK), an energy-sensing enzyme that is activated in response to cellular stress, is a critical signaling molecule for the regulation of multiple metabolic processes. AMPK has recently emerged as an attractive novel target for the treatment of obesity and type 2 diabetes because its activation increases fatty acid oxidation and improves glucose homeostasis. Here we show that pharmacological activation of AMPK by insulin-sensitizing drugs markedly inhibits inducible nitric-oxide synthase (iNOS), a proinflammatory mediator in endotoxic shock and in chronic inflammatory states including obesity-linked diabetes. AMPK-mediated iNOS inhibition was observed in several cell types (myo-cytes, adipocytes, macrophages) and primarily resulted from post-transcriptional regulation of the iNOS protein. AMPK activation in vivo also blunted iNOS induction in muscle and adipose tissues of endotoxin-challenged rats. Reduction of AMPK expression by small interfering RNA reversed the inhibitory effects of AMPK activators on iNOS expression and nitric oxide production in myocytes. These results indicate that AMPK is a novel anti-inflammatory signaling pathway and thus represents a promising therapeutic target for immune-inflammatory disorders. AMP-activated protein kinase (AMPK) 1 emerging energy-sensing/signaling metabolite-sensing

acts to switch off ATP-consuming pathways and switch on alternative pathways for ATP regeneration. AMPK is a heterotrimer consisting of a catalytic ␣-subunit and two regulatory subunits, ␤ and ␥. In response to cellular energy depletion, as reflected by an increase in the AMP/ATP ratio, AMPK is phosphorylated and activated by a still uncharacterized upstream AMPK kinase (2). It also can be activated allosterically by increases in the AMP/ATP and creatine/creatine-P ratios. The metabolic function of AMPK perhaps has been documented best in exercising skeletal muscle, where its activation seems to contribute to increased glucose transport and fatty acid oxidation (3,4). AMPK can be activated chemically with 5-aminoimidazole-4-carboxamide riboside (AICAR), which is taken up by cells and phosphorylated by adenosine kinase to form 5-aminoimidazole-4-carboxamide ribonucleoside, a nucleotide that mimics the effect of AMP (5). More recent studies also have identified AMPK as the mediator of the metabolic effects of the adipose-derived peptidic hormones leptin and adiponectin in skeletal muscle and liver (6 -8).
Chronic treatment of animal models of type 2 diabetes with the AMPK activator AICAR improves glucose homeostasis and insulin sensitivity (9 -11). These beneficial effects are thought to be explained mainly by the well known actions of AMPK on glucose metabolism and lipid oxidation in muscle and liver. Accordingly, the anti-diabetic drugs metformin and rosiglitazone also activate AMPK, and this is believed to contribute to their insulin-sensitizing actions in diabetic subjects (12)(13)(14)(15). However, the potential effects of AMPK activation on proinflammatory mediators of insulin resistance have not been investigated yet. Indeed, an inflammatory component also is present in obesity-linked diabetes (16,17) as reflected by increased systemic and tissue concentrations of the proinflammatory cytokines TNF-␣ and interleukin-6 in obese human subjects (18,19) and several animal models of obesity (20 -22). Furthermore, we recently have reported that inducible nitricoxide synthase (iNOS), a cytokine-inducible proinflammatory mediator in several pathological conditions, is overexpressed in muscle and fat of genetic and dietary models of obesity and type 2 diabetes (22). Targeted disruption of the iNOS gene was found to protect high fat-fed obese mice from developing insulin resistance and to significantly improve glucose tolerance (22). A role for iNOS in mediating insulin resistance also is supported by the finding that pharmacological inhibition of iNOS reverses impaired insulin action in cultured myocytes exposed to cytokines and the endotoxin LPS (23). Because an inflammatory component that includes iNOS is involved in the pathogenesis of muscle insulin resistance in obesity-linked type 2 diabetes, we asked whether AMPK activation might improve insulin action in muscle cells through inhibition of iNOS, and if so, whether AMPK is a key iNOS inhibitory pathway in other cell types where increased iNOS expression is pathogenic.

EXPERIMENTAL PROCEDURES
Materials-AICAR was purchased from Toronto Research Chemicals (Toronto, Canada). Troglitazone and 15-deoxy PGJ 2 were from Cayman Chemical Co. (Ann Arbor, MI). Interferon-␥ and TNF-␣ were from Research Diagnostics (Flanders, NJ) and RD Systems (Minneapolis, MN), respectively. 1400W was from Biomol Research Laboratories (Plymouth, PA), and Oligofectamine was from Invitrogen. ADP-Sepharose 4B beads were from Amersham Biosciences, and the Renaissance enhanced chemiluminescence kit was from PerkinElmer Life Sciences (Boston, MA). All other chemicals were from Sigma. Monoclonal antibodies against iNOS and eNOS and a polyclonal antibody against nNOS were obtained from Transduction Laboratories (Mississauga, Canada). Polyclonal antibodies against AMPK (which recognizes both ␣1and ␣2-AMPK), phospho-AMPK, and phospho-ACC (acetyl-CoA carboxylase) were purchased from New England Biolabs (Beverly, MA). IRS-1 antibody (C-20) was obtained from Santa Cruz Biotechnology (Santa Cruz, CA).
Cell Culture-L6 myoblasts and 3T3-L1 fibroblasts were grown in ␣-minimum Eagle's medium (10% fetal bovine serum) or ␣-Dulbecco's modified Eagle's medium (20% calf serum) containing 1% antibiotic/ antimycotic solution and differentiated into myotubes and adipocytes as described previously (23,24). Bone marrow-derived macrophages were a kind gift of Dr. M. Olivier (CHUL Research Center, Quebec, Canada). Generation and immortalization of the macrophages from mouse femurs was performed as described previously (25). Induction of iNOS by cytokines and/or LPS in L6 myotubes, 3T3-L1 adipocytes, and macrophages and the effects of AMPK-activating agents were carried out as described in the figure legends. The accumulation of nitrite in the incubation medium was used as an index of NO production. Nitrite was determined spectrophotometrically as described previously (23).
Animals-This study was approved by the Animal Care and Handling Committee of Laval University. Male Sprague-Dawley rats (200 -250 g) purchased from Charles River (Montréal, Canada) were used in these studies. Rats were pretreated with a single intraperitoneal injection of AICAR (1 g/kg, BW), metformin (100 mg/kg, BW) or saline (vehicle) 30 min prior to LPS (20 mg/kg, BW, intraperitoneally) or saline treatment. Animals were sacrificed 4 h later, and hind limb muscles and white adipose tissues were removed and stored at Ϫ80°C until further processing.
Measurement of Plasma and Tissue Nitrite and Nitrate-Nitrite and nitrate levels in plasma, skeletal muscles, and adipose tissue were measured by fluorometry (26). Briefly, blood was collected in tubes containing EDTA and centrifuged for 10 min at 3200 ϫ g to obtain plasma. Tissues were grounded in liquid nitrogen using a mortar. The tissue powder was resuspended in 5 volumes of Tris-EDTA buffer (20 mM Tris, pH 7.5, 10 mM EDTA) containing a protease inhibitor mixture. Tissue lysates were centrifuged at 2800 ϫ g for 20 min, and protein concentration of the supernatant was measured by the BCA protein assay. Plasma and tissue lysates then were centrifuged at 5000 ϫ g (4°C) overnight in an Ultrafree®-Microcentrifuge 10,000 NMWL filter unit. Nitrate was reduced to nitrite using nitrate reductase and the NADPH regenerating system as described previously (26). The fluorescence was measured at ex of 360 nm and em of 450 nm.
Determination of Nitric-oxide Synthase Protein Expression and Western Blot Analysis-Muscle and adipose tissue lysates (1 mg of protein) were used to purify nitric-oxide synthase enzymes using 2Ј,5Ј-ADP-Sepharose resin as described previously (22). Samples were subjected to SDS-PAGE, and immunoblotting was performed as described previously (27). Immunoreactive bands were detected by the enhanced chemiluminescence method.
RNA Analysis-Total cellular RNA was isolated using guanidium thiocyanate-phenol-chloroform extraction, and iNOS mRNA was measured by semiquantitative reverse transcriptase-polymerase chain reaction as described previously (23,24).
PI 3-Kinase Activity-Cell lysates (500 g) were immunoprecipitated with 2 g of anti-IRS-1 coupled to protein A-Sepharose overnight at 4°C. PI 3-kinase activity was determined as described previously (22).
Data Analysis-All data are presented as means Ϯ S.E. The effects of the treatments were compared by an analysis of variance followed by Fisher's post hoc test. Differences were considered to be statistically significant at p Ͻ 0.05.

AMPK Activators Inhibit iNOS and Reverse Insulin Resistance in Skeletal
Muscle Cells-We first tested the effect of the AMPK activator AICAR on iNOS induction in L6 myocytes chronically exposed to cytokines and LPS, which is a model of chronic inflammation associated with insulin resistance (23,27). As shown in Fig. 1A, treatment of muscle cells with stimulated AMPK activity as reflected by increased phosphorylation of AMPK on Thr-172 (a site known to activate the enzyme (28)) as well as by its ability to enhance phosphorylation of ACC, a downstream target of AMPK and a good correlate of its activation (29). The anti-diabetic drug metformin and the PPAR␥ agonists troglitazone and 15dPGJ2 also activated AMPK in these cells (Fig. 1A). Chronic exposure of L6 myocytes to cytokines (TNF-␣ and IFN-␥) and LPS markedly induced NO production as detected by the accumulation of the stable bioproduct nitrite in the culture medium (Fig. 1B). This effect was explained by a robust induction of iNOS mRNA and protein levels ( Fig. 1, C and D). Neither eNOS nor nNOS is expressed in control or cytokine/LPS-treated L6 myocytes, and NO production was completely inhibited by co-incubation with the specific iNOS inhibitor 1400W (30) (data not shown).
Addition of AICAR caused a dose-dependent reduction of cytokine/LPS-induced NO production, reaching a maximal 80% inhibition at 2.5 mM (Fig. 1B). Treatment with either metformin or PPAR␥ agonists also decreased (by 50 -90%) the effect of cytokine/LPS on NO production. All AMPK activators inhibited NO production through the reduction of cellular iNOS protein levels (Fig. 1C). AMPK activators failed to change the cellular expression of desmin, the main intermediate filament protein in myocytes, here used as an internal control (data not shown). Reduction of iNOS protein expression was not linked to suppression of iNOS transcription because iNOS mRNA levels were not affected by AICAR or the anti-diabetic drugs metformin and troglitazone (Fig. 1D). However, high concentrations of 15dPGJ2 did reduce iNOS mRNA levels (see 25 M 15dPGJ2 in Fig. 1D). In additional studies, we also tested whether AICAR could inhibit iNOS-mediated NO production either through reduction in substrate (L-arginine) uptake or by direct inhibition of iNOS enzymatic activity. L-Arginine uptake by muscle cells was not found to be affected by AICAR (2.5 mM) (1297 Ϯ 209 versus 1369 Ϯ 90 pmol/min/mg for control and AICAR-treated cells, respectively). Furthermore, AICAR did not affect iNOS subunit dimerization and failed to reduce iNOS catalytic activity in cytokine/LPS-treated myocytes, as assessed by the formation of L- activity in L6 myocytes, and the inhibitory effect of AICAR on NO production was not decreased by blocking adenosine receptors with 8-(p-sulfophenyl) theophylline (see Figs. C and D in supplemental material), ruling out any role for adenosine release in AMPK-dependent iNOS inhibition.
To test whether AMPK-induced iNOS inhibition reverses the insulin resistance-promoting effects of cytokine/LPS, next we measured the effects of AMPK activators on insulin action in L6 myocytes. We chose to assess insulin-induced activation of PI 3-kinase, an essential mediator of the metabolic actions of insulin, rather than glucose uptake because the latter is increased by AMPK through insulin-independent pathways. As shown in Fig. 2, A and B, cytokines and LPS treatment caused insulin resistance as revealed by a 60% reduction in the ability of insulin to stimulate IRS-1-associated PI 3-kinase activity as compared with control cells. However, addition of either AICAR, metformin, or troglitazone to cytokine/LPS-treated cells fully restored insulin action on PI 3-kinase. AMPK activating agents failed to increase PI 3-kinase in the absence of insulin ( Fig. 2A). The restoration of insulin action by AMPKactivating agents was similar to that obtained with the specific iNOS inhibitor 1400W (Fig. 2B).
AMPK Activators Inhibit iNOS in Adipocytes and Macrophages-Next we tested whether AMPK activation inhibits iNOS in other cell types. We have reported that adipose tissue is a major site of iNOS expression in endotoxemia and that chronic treatment of adipocytes with cytokines and endotoxins reproduces this inflammatory response in vitro (24). As shown in Fig. 3A, addition of AICAR or anti-diabetic drugs to cultured 3T3-L1 adipocytes activated AMPK as reflected by enhanced ACC phosphorylation. Activation of AMPK reduced cytokine/ LPS-mediated NO production (Fig. 3B) and cellular iNOS protein content (Fig. 3C) but not iNOS mRNA levels (Fig. 3D). Monocytes and macrophages are major sites of iNOS expression, particularly in infectious or inflammatory diseases. Sustained production of NO endows macrophages with cytostatic or cytotoxic activity against several pathogens and tumor cells, but it also can exert deleterious proinflammatory effects on host cells (32). Activation of AMPK by AICAR and metformin inhibited iNOS induction upon exposure of macrophages to interferon-␥ (Fig. 3, E and F). The PPAR␥ ligand troglitazone also increased AMPK and caused a dose-dependent inhibition of NO production in these cells (Fig. 3, E and F). The inhibitory effects of AMPK activators also were linked to reduction in iNOS protein induction through a post-transcriptional mechanism (Fig. 3, G and H). However, at higher concentrations (Ͼ 10 M) of troglitazone, a reduction in iNOS mRNA levels was also observed.

Activation of AMPK by in Vivo Injection of AICAR and Metformin Inhibits iNOS in Muscle and Adipose Tissues of LPStreated Rats-To test whether AMPK also blunts iNOS induction in vivo,
we assessed the ability of AICAR and metformin to inhibit iNOS in skeletal muscle and adipose tissue of rats injected with the bacterial endotoxin LPS, a model of septic shock associated with insulin resistance (27). Pretreatment of LPS-challenged rats with AICAR or metformin increased ACC phosphorylation (Fig. 4A) and significantly reduced NO production in the plasma as well as in muscle (gastrocnemius and tibialis) and adipose tissues (Fig 4B). AICAR and metformin treatments prevented induction of iNOS protein expression in muscle and fat tissues of septic rats (Fig. 4C). In marked contrast, neither eNOS nor nNOS enzymes were affected by LPS or AMPK activators in muscle and fat (nNOS was not detectable in fat), highlighting the isoform-specific effect of AMPK on iNOS protein. As observed for cell lines, the inhibitory effects of AMPK activators on iNOS-mediated NO production was explained by post-transcriptional mechanisms because LPS-induced iNOS mRNA expression was not affected by AICAR or metformin treatments (Fig. 4D). However, one notable exception is metformin, which caused a small but reproducible reduction in iNOS mRNA only in adipose tissue.
RNA Interference-mediated Depletion of ␣-AMPK Reverses the Inhibitory Effects of AMPK Activators on iNOS Protein and NO Production-To further establish whether AICAR and antidiabetic drugs inhibit iNOS through AMPK activation, we next used RNA interference technology to "knock down" endogenous AMPK ␣-subunits. Immunoblotting experiments confirmed that transfection of siRNA specific for the ␣1 and ␣2 catalytic subunits reduced ␣-AMPK expression to 53 Ϯ 5% of that of cells mock transfected with unrelated siRNA (Fig. 5A). Moreover, combined siRNA treatment resulted in a comparable inhibition of AMPK and ACC phosphorylation by AMPK activators (Fig.  5A). In AMPK-depleted myocytes, the ability of AICAR, metformin, and troglitazone to inhibit cytokine/LPS-induced NO production and iNOS protein expression was reduced significantly as compared with cells treated with unrelated siRNA (Fig. 5B). These results confirm that these agents blunt iNOS expression and NO production at least in part through activation of AMPK. DISCUSSION Elucidation of the multiple roles of NO in health and disease states has uncovered many areas where manipulation of NO production would be of therapeutic benefit. Thus, whereas focal iNOS induction and NO production is an important anti-infectious and anti-tumor mechanism of innate immunity (33), high amounts of iNOS-derived NO also can lead to self-tissue destruction in autoimmune diseases, allograft rejection, sepsis, rheumatoid arthritis, chronic heart failure, and other disorders accompanied by excessive activation of the immune system (see Refs. 34 -37). iNOS induction also is implicated in chronic metabolic disorders such as atherosclerosis (38, 39) and obesity-linked diabetes (22,40) in which an inflammatory condition also is believed to play a pathogenic role. Such wide implication has produced an intense interest in understanding the modulation of iNOS expression and activity, with the goal of finding novel mechanisms of control of this high output NO pathway. For these reasons, the design of selective iNOS inhibitors and the identification of iNOS inhibitory pathways have received much attention in recent years. In the present study, we show for the first time that pharmacological activation of AMPK by AICAR and two classes of anti-diabetic drugs inhibit iNOS induction in cells and tissues exposed to inflammatory mediators. Gene silencing experiments confirmed that AMPK-activating agents blunt iNOS-mediated NO production at least in part via activation of AMPK. Indeed, depletion of ␣-AMPK in L6 myocytes by using the combination of ␣1and ␣2-AMPK siRNA species led to a significant loss in the ability of AICAR, metformin, and troglitazone to reduce cytokine/LPS-induced iNOS protein expression and NO production.
Metformin and the PPAR␥ agonist rosiglitazone activate AMPK, and this is believed to contribute to their insulin-sensitizing actions in diabetic subjects (12)(13)(14)(15)41). These anti-diabetic effects mainly have been attributed to the lipid-and glucose-related metabolic effects of AMPK activation. However, our data strongly suggest that inhibition of iNOS induction and reduced NO formation represent a novel mechanism by which AMPK improves insulin action in muscle. Indeed, we show here that AMPK activation by several drugs reduces iNOS induction in L6 myocytes chronically exposed to cytokines and LPS, an in vitro model of insulin resistance for glucose transport (23). Accordingly, we also found that iNOS-mediated impairment in insulin-induced PI 3-kinase activation can be reversed by AMPK activators. Furthermore, the inhibitory effect of AMPK activation on iNOS was confirmed in vivo in a model of septic shock associated with insulin resistance (27). Further studies will be needed to test whether long term AMPK activation also can prevent or reduce iNOS induction in lower grade but chronic inflammatory disorders associated with insulin resistance such as atherosclerosis and obesity-linked diabetes.
AMPK activation also suppressed iNOS protein expression and NO production in cytokine/LPS-treated adipocytes and in adipose tissue of LPS-challenged rats. Whereas the precise role of iNOS-derived NO in adipose cells still is not fully understood, it is thought to be involved in the modulation of lipid metabolism in endotoxemia (24) and may be a paracrine mediator of skeletal muscle insulin resistance in obesity-linked diabetes (22). Activation of AMPK in macrophages, an important site of iNOS expression, also caused inhibition of interferon-␥induced NO production. The latter finding is of important clinical relevance because macrophage/monocyte iNOS induction has been implicated in the pathogenesis or promotion of several inflammatory/immune disorders such as septic and hemorrhagic shock (42,43), multiple sclerosis (44), myocardial infarction (45,46), inflammatory bowel disease (47,48), and pancreatic ␤-cell dysfunction (49).
It will be important to dissect out the molecular mechanisms by which AMPK activators cause iNOS inhibition. It is well established that inflammatory cytokines and LPS trigger iNOS transcription through a complex network of intracellular pathways including NF-B, Janus kinase/signal transducers and activators of transcription, and mitogen-activated protein kinase (50,51). Moreover, PPAR␥ ligands rosiglitazone and 15dPGJ 2 were previously shown to reduce iNOS expression in several cell types (49,52) through transcriptional repression (52). Therefore, we were expecting that PPAR␥-dependent activation of AMPK would inhibit iNOS by reducing the transcription of iNOS. Surprisingly, we found that all AMPK activating agents, including the PPAR␥ agonists troglitazone and 15dPGJ 2 , decreased iNOS protein but barely affected mRNA expression. Only when higher concentrations of troglitazone and 15dPGJ 2 were used did we observe a reduction in iNOS transcript levels. It should be noted that recent studies have shown that inhibition of iNOS transcription by rosiglitazone may result from activation of both PPAR␥ and PPAR␦ (53). Thus, high concentrations of rosiglitazone inhibit iNOS in PPAR␥ Ϫ/Ϫ macrophages at least in part by activating PPAR␦. Troglitazone is 100-fold less potent than rosiglitazone for PPAR␥, but its affinity for PPAR␦ has not been established. Therefore, it cannot be excluded that its inhibitory action on iNOS may be mediated in part through PPAR␦. Studies of cells lacking both PPAR␥ and PPAR␦ will be needed to answer this important issue.
The inhibitory effects of AMPK activators on NO production were highly correlated with the reduction in iNOS protein content, suggesting that AMPK modulates iNOS protein turnover. In this regard, recent reports suggest that AMPK switches off protein synthesis either through suppression of the mTOR-p70S6 kinase pathway (54) or by direct activation of eukaryotic elongation factor 2 kinase, resulting in the phosphorylation and inactivation of eukaryotic elongation factor 2 (55). It also is possible that AMPK reduces iNOS protein content by promoting its ubiquitination, a process required for targeting iNOS through the proteasome proteolysis pathway (56). However, ubiquitination also has been reported to be involved in the proteolytic degradation of nNOS (57) whereas the inhibitory effects of AMPK on NO production appear to be specific for the iNOS isoform. Indeed, neither nNOS nor eNOS proteins were affected by AICAR or metformin treatments in muscle and adipose tissue of LPS-challenged rats, and AMPK activators blunted LPS-induced NO production but did not reduce basal NO formation by constitutive nitric-oxide synthase enzymes. This argues for inhibition of protein synthesis rather than ubiquitin-mediated proteasomal degradation as the principal mechanism by which AMPK regulates iNOS protein turnover.
The identification of AMPK as a potent inhibitory pathway for iNOS may provide a molecular basis for the recent observations that exercise training, which increases AMPK expression (9), improved exercise capacity in association with reduced skeletal muscle iNOS expression in patients with chronic heart failure (37). Our data further suggest that AMPK represents a promising therapeutic target for immune-inflammatory disorders in which iNOS induction is pathogenic but not critical for health (i.e. non-infectious diseases). These include metabolic disorders such as atherosclerosis and obesity-linked diabetes. However, even for those inflammatory conditions, caution must be exerted because iNOS also can play more subtle yet important physiological roles with potential adverse consequences attendant on its inhibition in diseased states. For example, iNOS plays a protective role in the development of transplant arteriosclerosis (58) and also is critical for optimal wound healing (59), which calls into question the use of iNOS inhibitory agents to treat diseases such as atherosclerosis and diabetes. Bearing these cautions in mind, it is crucial to elucidate the molecular mechanisms by which AMPK inhibits iNOS because this may help design novel strategies for an effective and tissue-specific inhibition of iNOS in inflammatory settings while limiting adverse physiological consequences.