M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 1 6 7 9 – 1 6 9 1

Ava i l ab l e on l i ne a t www.sc i enced i r ec t . com

Metabolismwww.metabo l i sm jou rna l . com

Activation of the AMPK/Sirt1 pathway by a

leucine–metformin combination increases insulinsensitivity in skeletal muscle, and stimulatesglucose and lipid metabolism and increases lifespan in Caenorhabditis elegans

Jheelam Banerjee⁎, Antje Bruckbauer, Michael B. ZemelNuSirt BioPharma Inc., 11020 Solway School Road, Knoxville, TN 37931, USA

A R T I C L E I N F O

Abbreviations: ACC, acetyl-CoA carboxmonophosphate-activated protein kinase;coactivator 2; ECAR, extracellular acidificatisynthase kinase; IR, insulin receptor; IRS1, inMet, metformin; NAD, nicotinamide adeninecarboxykinase; pGS, phosphorylated glycogenSirt1, Sirtuin 1; TNF-α, tumor necrosis factor a⁎ Corresponding author.

E-mail addresses: jbanerjee@nusirt.com,

http://dx.doi.org/10.1016/j.metabol.2016.06.010026-0495/© 2016 Elsevier Inc. All rights rese

A B S T R A C T

Article history:Received 28 January 2016Accepted 29 June 2016

Background. Wehave previously shown leucine (Leu) to activate Sirt1 by lowering its KM forNAD+, thereby amplifying the effects of other sirtuin activators and improving insulinsensitivity. Metformin (Met) converges on this pathway both indirectly (via AMPK) and bydirect activation of Sirt1, and we recently found Leu to synergize with Met to improve insulinsensitivity and glycemic control while achieving ~80% dose-reduction in diet-induced obesemice. Accordingly, we sought here to define the mechanism of this interaction.

Methods.Muscle cells C2C12 and liver cells HepG2wereused to test the effect ofMet–Leu onSirt1 activation. Caenorhabditis elegans was used for glucose utilization and life span studies.

Results. Leu (0.5 mmol/L) + Met (50–100 μmol/L) synergistically activated Sirt1 (p < 0.001) atlow (≤100 μmol/L) NAD+ levels while Met exerted no independent effect. This was associatedwith an increase in AMPK andACC, phosphorylation, and increased fatty acid oxidation, whichwas prevented by AMPK or Sirt inhibition or silencing. Met–Leu also increased P-IRS1/IRS1 andP-AKT/AKT and in insulin-independent glucose disposal in myotubes (~50%, p < 0.002) evidentwithin 30min aswell as a 60% reduction in insulin EC50. In addition, in HepG2 liver cells nuclearCREB regulated transcription coactivator 2 (CRTC2) protein expression and phosphorylation ofglycogen synthase was decreased, while glycogen synthase kinase phosphorylation wasincreased indicating decreased gluconeogenesis and glycogen synthesis. We utilized C. elegansto assess the metabolic consequences of this interaction. Exposure to high glucose impairedglucose utilization and shortened life span by ~25%,while addition of Leu + Met to high glucoseworms increasedmedian andmaximal life spanby 29 and15%, respectively (p = 0.023), restorednormal glucose utilization and increased fat oxidation ~two-fold (p < 0.005), while metforminexerted no independent effect at any concentration (0.1–0.5 mmol/L).

Keywords:AMPKSirt1Insulin sensitivityLeucineMetformin

ylase; AICAR, 5-aminoimidazole-4-carboxamide ribonucleotide; AMPK, 5′ adenosineCREB, cAMP-responsive element binding protein; CRTC2, CREB regulated transcriptionon rate; ERK, extracellular signal regulated kinase; GS, glycogen synthase; GSK, glycogensulin receptor substrate 1; JNK, c-Jun N-terminal kinases; Leu, leucine; LKB1, liver kinase B1;dinucleotide; NAMPT, nicotinamide phosphoribosyltransferase; PEPCK, phosphoenolpyruvatesynthase; pGSK, phosphorylated glycogen synthase kinase; PKC-ZETA, protein kinase C zeta;lpha.

jheelambanerjee@yahoo.com (J. Banerjee).

1rved.

1680 M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 1 6 7 9 – 1 6 9 1

Conclusion. Thus, LeuandMet synergize to enable Sirt1 activationat lowNAD+concentrations(typical of energy replete states). Sirt1 andAMPKactivations are required forMet–Leu's full action,which result in improvements in energy metabolism and insulin sensitivity.

© 2016 Elsevier Inc. All rights reserved.

1. Introduction

Metformin is considered the initial drug of choice for treatingtype 2 diabetes, as it is highly efficacious, exhibits an excellentsafety profile, does not promote weight gain, does not increasethe risk for hypoglycemia and has been shown to reduce the riskof diabetes-related comorbidities and death [1]. However,metforminmonotherapy often fails to achieve optimal glycemiccontrol due to inter-individual variability in response to druginitiation and maintenance, failure to achieve optimal dosetitration secondary to drug-induced gastrointestinal discomfort,or drug discontinuation due to the development of adverseeffects [2]. Although there are multiple other antihyperglycemicdrug classes which can be used as second choice monotherapyor add-on treatments, they still show some disadvantagesin comparison to successful metformin monotherapy [3,4]Accordingly, there is still a need to optimize metformin efficacyat lower doses [5].

The AMP-activated protein kinase (AMPK) and the NAD+-dependent sirtuins (Sirt1 and Sirt6) are key sensors of energystatus and regulators of glucose and lipid metabolism, activatingeach other in a finely tuned network [6,7].

While insulin resistance and diabetes are associated withimpairment of this pathway, activation of the AMPK-Sirt1axis improves hyperglycemia and insulin sensitivity [8,9].Accordingly, Sirt1 transgenic mice exhibit reduced levels offasting blood glucose and insulin as well as improved glycemiccontrol, showing anti-diabetic effects, during the glucosetolerance test [10].

Wepreviously found the branched-chain amino acid leucine(Leu) to activate Sirt1 by lowering the activation energy forNAD+, thus mimicking the effects of caloric restriction, andenabling co-activation and amplification of the effects of otherAMPK/sirtuin activators at low concentrations [11]. As a resultof metformin and leucine converging on this AMPK-Sirt1 axis[1], we found leucine and metformin to exhibit synergisticeffects to improve insulin sensitivity and glycemic controlin diet-induced obese mice while achieving a ~80% dose-reduction ofmetformin [12]. Therefore, this studywas designedto further define the mechanism of the interaction of leucineand metformin on AMPK/Sirt1 activation, and to assess themetabolic consequences on glucose utilization and life span inCaenorhabditis elegans.

2. Materials and Methods

2.1. Cell Culture

Murine C2C12muscle cells were grown in the absence of insulinin Dulbecco's modified Eagle's medium (DMEM, 25 mmol/Lglucose) containing 10% fetal bovine serum (FBS) and antibiotics(1% penicillin–streptomycin) at 37 °C in 5% CO2 in air. They were

then seeded into plates for the final treatments and theninduced to differentiate with a DMEM containing 2% horseserum and 1% Pen-Strep. The cells were maintained in thisdifferentiationmedium for 5 days before treatment. HepG2 cellswere either grown in low glucose DMEM (5 mmol/L glucose) orhigh glucose DMEM (25 mmol/L glucose) containing 10% fetalbovine serum (FBS) and antibiotics (1% penicillin–streptomycin)at 37 °C in 5% CO2 in air. C2C12 cells were treated withmetformin (0.1 mmol/L), leucine (0.5 mmol/L) and a combina-tion of both drugs for 2 h, followed by 20 min insulin (10 nmol/L)stimulation. Proteins were then extracted using NP40 buffer.HepG2 cells were treatedwith a combination ofMet–Leu for 24 hto 48 h and then cells were harvested. Protein concentrationswere quantitated with the BCA kit (Thermo Scientific).

2.2. siRNA Transfection

C2C12 muscle cells were seeded with 104 cells/well on 96-wellplate or 24-well Seahorse plate (confluence, ~50 to 60%). ~24 hlater, cells were transfected with siRNA against Sirt1 (AmbionID# 174219) or AMPK (Ambion ID# 221537) complexed toLipofectamine RNAiMAX reagent (ThermoFisher Scientific,Cat# 13778-030) according to manufacturer's instructions for24 to 48 h. Then they were differentiated in 2% horse serumfor an additional 4 days prior to treatment.

2.3. Gene Expression Analysis

Cells were grown in a 96-well plate. Cell Lysis, reverse transcrip-tion and RT-PCR were performed using the TaqMan® GeneExpression Cells-to CT™ Kit (Life Technologies, Cat #4399002)according to manufacturer's instructions. Gene expression wasassessed by RT-PCR using StepOnePlus™ PCR system (ThermoFisher Scientific) and TaqMan® Gene expression assays forAMPK (Life Technologies, Cat #Mm01264789) and Sirt1 (LifeTechnologies, Cat #Mm01168521).

2.4. Nuclear and Cytosolic Extraction

Treated HepG2 cells were scraped from culture flasks andwashed with cold PBS twice. Cell pellet was re-suspendedin 500 μL of hypotonic buffer (20 mmol/L Tris–HCl, pH 7.4;10 mmol/L NaCl; 3 mmol/L MgCl2) and incubated on ice for15min. Then 25 μL detergent (10%NP40)was added and vortexedfor 10 s. The homogenatewas centrifuged for 10min at 3000 rpmat 4 °C, and the supernatant containing the cytoplasmic fractionwas then aliquoted and stored at −80 °C for further experiments.The pellet, containing the nuclear fraction, was re-suspended in50 μL of Cell Extraction buffer (Life Technologies, Grand Island,NY; Cat #FNN0011) supplemented with 1 mmol/L PMSF andProtease Inhibitor Cocktail (Sigma, St. Louis, MO; Cat #P-2714,1:100 dilution) for 30 min on ice with vortexing at 10 minintervals. Then the homogenate was centrifuged for 30 min at

1681M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 1 6 7 9 – 1 6 9 1

14,000g at 4 °C. The supernatant was aliquoted and stored at−80 °C for further experiments. Protein content of the cyto-plasmic and nuclear fraction was measured with the BCA kit(Thermo Scientific, Grand Island, NY).

2.5. ELISA

Differentiated C2C12 muscle cells were treated for 2 h with Met(0.1 mmol/L)–Leu (0.5 mmol/L). 20 min before harvest they werestimulated with 100 nmol/L insulin. The ELISA assays (PathScanP-IRS1 (S307) and PathScan Total IRS1, Cell Signaling (Danvers,MA), cat #7287and#7328, respectively)wereperformedaccordingto the manufacturer's instructions.

2.6. Western Blot

Phospho-AMPK (Thr172), AMPK, pACC (Ser79), ACC, pIRS1 (S318),IRS1, p-AKT (Thr308), AKT, pGSK, GSK, pGS, GS, TORC2/CRTC2and β-actin antibodies were obtained from Cell Signaling(Danvers, MA). Protein levels of cell extracts were measured byBCA kit (Thermo Scientific, Pittsburgh, PA). For Western blot,10–50 μg protein was resolved on 4–15% gradient polyacrylamidegels (Criterion precast gel, Bio-Rad Laboratories, Hercules, CA),transferred to either PVDF or nitrocellulosemembranes, incubat-ed in blocking buffer (3% BSA or 5% non-fat dry milk in TBS) andthen incubated with primary antibody (1:1000 dilution), washedand incubated with horseradish peroxidase- or fluorescence-conjugated secondary antibody (1:10,000 dilution). Visualizationwas conducted using BioRad ChemiDoc instrumentation andsoftware (Bio-Rad Laboratories, Hercules, CA) and band intensitywas assessed using Image Lab 4.0 with correction for backgroundand loading controls.

Fig. 1 – Met–Leu effect on Sirt1 activation. Recombinant human Smetformin (0.1 mmol/L) for 45 min under different NAD+ concensubstrate was measured. (A and B) Representative fluorescencerespectively. Data are expressed as means ± SEM (n = 5 to 6, p ≤ 0activity ± SEM (n = 5 to 6, p ≤ 0.05). * indicates significantly differ

2.7. Glucose Utilization in C2C12 Muscle Cells

In the absence of a fatty acid source and oxidative metabo-lism, glycolysis and subsequent lactate production result inextracellular acidification. Extracellular acidification rate(ECAR) was measured using a Seahorse Bioscience XF24analyzer (Seahorse Bioscience, Billerica, MA) in 24-well plates at37 °C. C2C12 cells were seeded at 40,000 cells per well,differentiated as described above, treated for 24 h with orwithout metformin (0.1 mmol/L)–leucine (0.5 mmol/L). ForSH-experiment, media was changed to 11 mmol/L glucose.After three baseline readings, insulin (0 to 500 nmol/L) wasinjected. After 20 min incubation, 14 mmol/L glucose wasinjected and extracellular acidification rate (ECAR)wasmeasuredover a two-hour period.

2.8. Sirt1 Activity

The Sirt1 FRET-based screening assay kit (Cayman ChemicalCompany, Ann Arbor, MI) was used for the measurement ofSirt1 activity in a cell-free system as well as in C2C12 musclecells. For the cell-free experiment, Leu (0.5 mmol/L), Met(0.1 mmol/L) or combination was incubated with recombinantSirt1 enzyme under different NAD+ concentrations (500 μmol/L,100 μmol/L, 50 μmol/L and 10 μmol/L) for 30 min according tothe manufacturer's protocol. The protocol was modified forthe measurements in cells as follows; cell lysate of differen-tiated C2C12 muscle cells treated with Leu (0.5 mmol/L), Met(0.1 mmol/L) or combination for 24 h was incubated withpeptide substrate under low NAD+ (100 μmol/L) concentra-tions. The fluorescence was measured with excitation andemission wavelengths of 360 nm and 450 nm, respectively.

irt1 enzyme was incubated with Leucine (0.5 mmol/L) and/ortrations (10 to 500 μmol/L) and fluorescence of deacetylateddata with incubation of 50 μmol/L and 500 μmol/L NAD+,.05). (C) Summary of results represented as % change of Sirt1ent from control (p < 0.05).

1682 M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 1 6 7 9 – 1 6 9 1

The increase in fluorescence is proportional to the amountof deacetylated substrate and thus Sirt1 activity.

2.9. C. elegans Maintenance

Worms (N2 Bristol wild-type) were obtained from theCaenorhabditis Genetics Center (CGC) at the University ofMinnesota and grown on standard NGM plates with E. coli(OP50) as food source at 21 °C. For treatments, adult wormswere bleached to obtain eggs, which were allowed to hatch

overnight inM9 buffer. Synchronized L1 larvaewere transferredto Escherichia coli fed NGM plates containing indicated treat-ments for about 35 h to reach L4/young adult stage. Alltreatments were added to the agar in indicated concentrations.

2.10. Glucose Content in C. elegans

Glucose content of worms was measured using the Amplex®Red Glucose/Glucose Oxidase Assay Kit (Invitrogen, nowThermoFisher Scientific, Grand Island, NY).

1683M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 1 6 7 9 – 1 6 9 1

Synchronized L1 worms were maintained in liquid media,as described in Ref. [13], containing 40 mmol/L glucose andindicated treatmentsuntil adulthood.Adultwormswerewashedoff the plate. The supernatant was removed after centrifugationat 500g for 1 min. The worms were washed three times with M9buffer. After the lastwash, supernatantwas removed except that~150 μL and 50 μL of Reagent buffer from assay kit were added.Then worm solution was homogenized and lysate was used forassay according to the manufacturer's instructions.

2.11. Life Span Study Protocol

50 synchronized stage L4 worms per group were placed on NGMagar plates seeded with E. coli strain OP50 (= day 1 of study). Alltreatments were addedwith the indicated concentrations to theagar plates. The worms were maintained at 21 °C throughoutthe duration of the study. Livewormswere placed on newplatesevery day to eliminate progeny. Worms were scored as dead ifthey did not respond to repeated toucheswith the platinumpickand scored as censored if they crawled off the plate.

2.12. Palmitate-Induced Fatty Acid Oxidation

Cellular oxygen consumption was measured using a SeahorseBioscience XF24 analyzer (Seahorse Bioscience, Billerica, MA) in24-well plates with appropriate modifications for C. elegans orfor cells, as follows. For themeasurement of fatty acid oxidationin worms, treated L4/adult worms were washed off the platewithM9 buffer, and thenwashed (centrifuged at 500g for 1 min)three times. After final wash, theworm suspensionwas dilutedto aworm count of 2worms/20 μLworm suspension drop. Then500 μL of theworm suspension was added to each well to reacha final worm count of ~50 worms/well with at least 5 replicatesper group in each experiment. The Seahorse program was setup at 27 °C. Five successive baselinemeasures were taken priorto injection of palmitate (200 μmol/L final concentration). Threesuccessive measurements of O2 consumption were then con-ducted, each followed by a 10-minwaiting period. Thismeasure-ment pattern was then repeated over a 2-h period. For fatty acidoxidationmeasurement in C2C12muscle cells, cells were seededwith 10,000 cells perwell, then treatedwith Lipofectamine/siRNAagainst Sirt1 orAMPKor lipofectamine vehicle for 24h. Then theywere differentiated with 2% horse serum for 4 days, and thentreated for 24 h with the indicated treatments. To measure fatoxidation, theywerewashed twicewith non-buffered carbonate-

Fig. 2 – Met–Leu effects on Sirt1 and AMPK signaling in C2C12 muindicated treatments for 2 h and stimulated with insulin (10 nmolphospho-AMPK (Thr172) and total AMPK (B) were measured. Quanrepresented as mean ± SEM (n = 3 to 4). (C) Sirt1 activity was mea(0.1 mmol/L), Leu (0.5 mmol/L) or the combination for 24 h. Fluores(n = 4 to 6). (D) Protein expressionof phospho (Thr172)-AMPKand tMet–Leu in presence or absence of the Sirt1 inhibitor EX527. Bar graof two different blots (n = 3 to 6) and representative blots are showinsert (E) Protein expression of Sirt1: C2C12 muscle cells were infethen treated after differentiation with Met–Leu for 24 h. Bar graphshown (n = 2). (F) Protein expression of phospho (Ser79)-ACC anMet–Leu in presence or absence of the Sirt1 inhibitor EX527 andmeasurement and representative blot are shown (n = 2).

free pH 7.4 low glucose (2.5 mmol/L) DMEM containing carnitine(0.5 mmol/L), equilibrated with 600 mL of the same media in anon-CO2 incubator for 45 min, and then inserted into theinstrument for 15 min of further equilibration, followed by O2

consumption measurement. Three successive baseline mea-sures at five-minute intervals were taken prior to injection ofpalmitate (200 μmol/L final concentration). Three successive 5-min measurements of O2 consumption were then conducted,followed by 10 min waiting period. This measurement patternwas then repeated over a 2-h period.

2.13. Statistical Analysis

All data are expressed as mean ± SEM, with the exception ofSeahorse data, which are shown as mean ± SD. Data wereanalyzed by one-way ANOVA, and significantly different groupmeans (p < 0.05)were separatedby the least significantdifferencetest using GraphPad Prism version 6 (GraphPad Software, LaJolla, CA;www.graphpad.com). For the statistical analysis of thelife span study with C. elegans, data were analyzed via Kaplan–Meier survival curves and statistical significance betweencurves was determined with the Gehan–Breslow–Wilcoxontest using Prism 6 (GraphPad Software, La Jolla, CA; www.graphpad.com).

3. Results

First we tested the effects of the Met–Leu combination on Sirt1activation in a cell-free system (Fig. 1). Met–Leu synergisticallyactivated human recombinant Sirt1 enzyme under low NAD+

conditions (50 μmol/L, Fig. 1A) but not under high NAD+

conditions (500 μmol/L, Fig. 1B), shifting the activation curve tothe left and lowering the Michaelis constant (KM) value for NAD+

from 55.05 to 21.45 (Fig. 1C); the individual compounds exertedlittle or no significant independent effect. There was also asignificant increase in the protein expression and activity of Sirt1and of the ratio of Phospho-AMPK (Thr172)/AMPK in C2C12muscle cells treated with the Met–Leu combination (Fig. 2A–C).Treatment of C2C12 muscle cells with the Sirt1 inhibitor Ex527prevented the effects of Met–Leu on AMPK phosphorylation.However, the ratio of P-AMPK/AMPK was not changed withEX-527 compared to Met–Leu since there were small butstatistically insignificant differences in total AMPK between thegroups. AMPK knockdown blunted the effects of Met–Leu on

scle cells. Differentiated C2C12 muscle cells were treated with/L) 20 min prior to harvest. Protein expression of Sirt1 (A) andtitative data and representative blots are shown. Data are

sured in differentiated C2C12 muscle cells treated with Metcence values were normalized to protein amount of cell lysateotal AMPK: differentiated C2C12muscle cellswere treatedwithphswere calculated frompooled densitometrymeasurementsn. The bar graph of the ratio of P-AMPK/AMPK is shown as ancted with siRNA against AMPK as described under methods,of densitometry measurement and representative blots ared ACC: differentiated C2C12 muscle cells were treated withAMPK inhibitor Compound C. Bar graphs of densitometry

Fig. 3 – AMPK or Sirt1 knockdown blocks the effects of Met–Leu on fatty acid oxidation. C2C12 muscle cells were infected withsiRNA against AMPK or Sirt1 as described under methods, then treated after differentiation with Met–Leu for 24 h. (A)Representative figure for AMPK knockdown experiment: oxygen consumption rate (OCR) was measured after 200 μmol/Lpalmitate injection (time points A and C). The area under the curve (AUC) of the OCR for (B) AMPK knockdown and (C) Sirt1knockdown experiment. Data are represented as mean ± SEM (n = 4 to 5). Validation of successful knockdown was measuredby RT-PCR for (D) AMPK and (E) Sirt1. siRNA (Sirt1) was used as positive control for AMPK knockdown and siRNA (AMPK) aspositive control for Sirt1 knockdown.

1684 M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 1 6 7 9 – 1 6 9 1

Sirt1 protein expression, indicating the interactive effects of Sirt1and AMPK (Fig. 2D and E). To further explore whether AMPK andSirt1 are both required for the full effect of the Met–Leucombination, we measured ACC phosphorylation as a ratio tototal ACC and fatty acid oxidation as downstream targets ofAMPK activation. Both, ACC phosphorylation and palmitate-induced oxygen consumption ratewere increased by ~100% and~20%, respectively, by Met–Leu and totally prevented by either

AMPK and Sirt1 knockdown or inhibition (Figs. 2F and 3A to C).The level of AMPK or Sirt1 knockdown is shown in Fig. 3D and E.

Wenext examined the effects of theMet–Leu combination oninsulin signaling. Stimulation with 10 nmol/L insulin 20 minprior to harvest produced the expected increase in phosphory-lated (Ser318)-IRS in C2C12 muscle cells. Met–Leu treatment for2 h in the absence of insulin resulted in a comparable (~25%)increase in IRS phosphorylation, and further significantly

1685M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 1 6 7 9 – 1 6 9 1

enhanced IRS phosphorylation by an additional ~15% duringinsulin stimulation compared to insulin-treated controls(Fig. 4A). Similarly, Met–Leu treatment in the absence ofinsulin resulted in a significant increase in phosphorylatedAKT expression in muscle cells comparable to that achievedwith insulin, and Met–Leu further augmented insulin stimula-tion of P-AKT (Fig. 4B). However, IRS phosphorylation on Ser307was not further stimulated by Met–Leu treatment whencompared to insulin-stimulated control (Fig. 4C).

Next we measured ECAR, which is an indicator of the cellularglucose utilization, in C2C12 cells treated with Met–Leu for 2 h inresponse to different insulin concentrations via the SeahorseBioscience XF24 analyzer. Fig. 5A shows a representative exper-iment with A, the time point of insulin injection (0.5 nmol/L,5 nmol/L, 50 nmol/L or 500 nmol/L), and C, (20 min after A) thetime point of 14 mmol/L glucose injection, to reach a final wellconcentration of 25 mmol/L glucose. Fig. 5B shows the calcu-lated area under the curve of that experiment and Fig. 5C thesummarized results of all four experiments. Therewas a shift inglucose utilization at the tested insulin concentrations withMet–Leu treatment. Met–Leu treatment increased glucose utiliza-tion comparable to 500 nmol/L insulin stimulation, andadditionalinsulin to the Met–Leu treatment further enhanced this effect.

Fig. 4 – Met Leu effects on insulin signaling in C2C12 muscle celindicated treatments for 2 h and stimulated with insulin (10 nm(Ser318) and IRS (A), and P-AKT and AKT (B) were measured. Barmeasurements of two different blots and representative blots ar(C) P-IRS (Ser307) and IRS were measured in cell extract via ELIS

In liver HepG2 cells we found a trend (p = 0.07) for reductionin CRTC2 in the nucleus with Met–Leu treatment, as well as asignificant increase in glycogen synthase kinase (GSK)-3 βphosphorylation and an associated decrease in glycogensynthase (GS) phosphorylation treatment (Fig. 6).

Next we tested the effects on Met–Leu on glucose and fatmetabolism in a simple model organism, C. elegans. Fig. 7Ashows the oxygen consumption rate and the calculated areaunder the curve in worms treated with Met–Leu. There was asignificant ~50% increase in fatty acid oxidation in C. elegans.Next, we measured the glucose content of worms kept eitherunder normal (low) glucose conditions or under high (40 mmol/Lglucose added to liquid media) glucose conditions. The highglucose conditions significantly increased the glucose content inthe worms, which was prevented by simultaneous treatmentwith Met–Leu, indicative of higher glucose utilization in theworms (Fig. 7B). To test whether these effectsmodulate survival,we conducted life span studies under normal and high glucoseconditions (2%glucose added to agar plates)withorwithoutMet,Leu and the combinations (Fig. 8). High glucose shortenedmedian survival by 22% compared to low glucose conditions.Leu combined with Met (0.1 mmol/L (Fig. 8A) and 0.5 mmol/L(Fig. 8B)) prolonged median and maximum life span under

ls. Differentiated C2C12 muscle cells were treated withol/L) 20 min prior to harvest. Protein expressions of P-IRSgraphs were calculated from pooled densitometry

e shown. Data are represented as mean ± SEM (n = 3 to 5).A. Data are represented as mean ± SEM (n = 8).

Fig. 5 – Met–Leu effects on glucose utilization in C2C12 muscle cells. The extracellular acidification rate (ECAR) was measuredafter different insulin (5 to 500 nmol/L) and glucose (14 mmol/L) injection (points A and C) in differentiated C2C12muscle cells.(A) Representative experiment after 500 nmol/L insulin injection. (B) The area under the curve (AUC) of this experiment. Dataare represented as mean ± SD (n = 5). ECAR response to four different insulin concentrations represented as group means ofeach individual experiment. * indicates significant difference between Met–Leu curve to control insulin curve (p < 0.03).

1686 M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 1 6 7 9 – 1 6 9 1

these high glucose conditions. The individual compoundsexerted small but statistically insignificant effect (Fig. 8C).

4. Discussion

The major findings of this paper are that the Met–Leucombination increases insulin-independent stimulation ofthe insulin signaling pathway by activating the AMPK/Sirt1pathway and increases insulin sensitivity and overall glucoseand lipid metabolism.

Metabolic sensors such as AMPK and Sirt1 are crucial to theregulatory network for metabolic homeostasis by modulatingpleiotropic effects in key metabolically relevant tissues such asliver, muscle and adipose tissue. SIRT1, a NAD+-dependentregulator of energy metabolism, has been reported to beactivated through AMPK-mediated induction of nicotinamidephosphoribosyltransferase (NAMPT), the rate-limiting enzymefor NAD+ biosynthesis or changes in the NAD+/NADH ratio [14].On the other hand, Sirt1-induced deacetylation of LKB1 causesits movement from the nucleus to the cytosol and subsequentactivation ofAMPK [14], suggesting reciprocal activation of thesetwo energy sensing systems. Both systems represent targets fortherapeutic intervention, as they are down-regulated duringperiods of over-nutrition and in obesity, diabetes andmetabolicsyndrome [15–18].

Metformin, acts in part via AMPK activation in liver to reducehepatic gluconeogenesis as well as in muscle to increase glucose

uptake [4,19]. In addition, metformin appears to increase bothSirt1 activity and protein expression via AMPK-dependent andindependentpathways [20,21].Wehavepreviouslydemonstratedthat leucine amplifies the effects of metformin on AMPK andSirt1 activation and enables a substantial dose reduction invitro and in vivo [12,22]. Our previous data suggest that some ofthe effects of leucine are mediated by acting as a directallosteric activator of Sirt1, reducing the KM for NAD+ andthereby allowing Sirt1 activation at lowerNAD+ concentrations,which are characteristic for a metabolic replete state. [11].Moreover, leucine facilitates the interaction of Sirt1 with otherSirt1 activating compounds such as resveratrol, which resultedin a further left-ward shift of the activation curve [11]. Here wedemonstrate that this ability of leucine can be extrapolatedto metformin, which also has some direct effects on Sirt1activation [23], as the Met–Leu combination further increasedSirt1 activity at low NAD+ concentrations (Fig. 1). Therefore, theactivation of Sirt1 by the combinationmay be an early effect andAMPK independent, which leads to subsequent AMPK activationand further Sirt1 activation at later time points, as suggested byLiang et al. [24]. Moreover, it was demonstrated that Sirt1 actsupstream of AMPK to regulate hepatocellular lipid metabolism[25]. Consistent with this concept, the Sirt1 inhibitor EX527blocked the effects of Met–Leu on AMPK phosphorylation,demonstrating that Sirt1 activation is necessary for AMPKactivation (Fig. 2D). In contrast, AMPK knockdown only partiallyblunted the Met–Leu effect on Sirt1, suggesting both AMPK-dependent and independent activation of Sirt1 (Fig. 2E). However,

Fig. 6 – Met–Leu effects on enzyme of hepatic glucose metabolism in HepG2 liver cells. HepG2 liver cells were treated withindicated treatments for 24 h. Protein expression of P-GSK and GSK (A), and P-GS and GS (B) were measured. Quantitative dataand representative blots are shown. Data are represented as mean ± SEM (n = 3 to 4). (C) Nuclear and cytosolic fraction of cellextract was separated as described under Materials and Methods. CRTC2 protein expression was measured in the cytosolicand nuclear fraction and normalized to β-actin. Data are represented as mean ± SEM (n = 3).

1687M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 1 6 7 9 – 1 6 9 1

Fig. 7 – Met–Leu effects on glucose and fat metabolism in C. elegans. C. elegans L1 larvae were maintained on agar plates orliquid media with indicated treatments until adulthood. (A) Oxygen consumption rate (OCR) in adult worms was measuredafter 200 μmol/L palmitate injection (time points A and C) and the area under the curve (AUC) was calculated. Data arerepresented as mean ± SD (n = 10). * indicates significantly different from control (p < 0.05). (B) Glucose content of worms wasmeasured as described under Methods and normalized to protein content. Different letter superscripts indicate statisticallysignificant difference between groups (p < 0.05). Treatments: control low gluc (no glucose or treatments added), control highgluc (40 mmol/L glucose), Met (metformin 0.1 mmol/L)–Leu (leucine 0.5 mmol/L).

1688 M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 1 6 7 9 – 1 6 9 1

knockdown of either Sirt1 or AMPK prevented the effects of Met–Leu on downstream events such as ACC phosphorylation andpalmitate-induced fatty acid oxidation (Fig. 2F and 3), implyingthat both are necessary for Met–Leu's full action.

Since the Met–Leu combination increased AMPK and Sirt1protein expression in C2C12 muscle cells (Fig. 2), we wanted todetermine if this results in an insulin-independent increase inglucosemuscle uptake. IRS-1 protein is a key player in the insulinsignal transduction pathway to stimulate glucose uptake and isregulated by various kinases in response to metabolic stimuli atmultiple serine, threonine and tyrosine phosphorylation sites[26]. AMPK and Sirt1 also regulate IRS-1 phosphorylation andmuscle glucose uptake, although this may be a secondaryresponse to inhibition of inflammatory mediators. For example,knockdown of Sirt1 led to increased inflammatory responseaccompanied by impaired IRS-1 signaling and decreased insulin-stimulated glucose transport in 3T3-L1 adipocytes whereas Sirt1activators suppressed the inflammatory pathway and increasedinsulin sensitivity [27]. Similarly, treatment of 3T3-L1 adipocyteswith AICAR, an AMPK activator, reduced TNF-α induced serinephosphorylation of IRS-1 through ERK and JNK and improvedinsulin resistance [28]. We found an upregulation of Ser 318phosphorylation of IRS-1 with Met–Leu treatment comparableto insulin alone, and a further augmentation of insulin-inducedIRS-1phosphorylation.Although thisphosphorylationsite canbeactivated by PKC-zeta, JNK and other kinases which presumablydisrupt the interaction of IR and IRS-1 and provide a negative

feedback mechanism [26], there is also evidence that Ser 318phosphorylation also occurs in the early phase of insulinsignaling increasing AKT phosphorylation and glucose uptakeand is required for an enhanced induction of insulin action [29].However, in the late phase of insulin action, this phosphorylationsite attenuates insulin-stimulated signal transduction and sub-sequently decreases glucose uptake in skeletal muscle. There-fore, this Ser318 phosphorylation may be necessary to maintainthe physiological balance of cellular energy homeostasis [29]. Incontrast, the Ser307 phosphorylation site is assumed to be aninhibitory phosphorylation site stimulated by factors such as freefatty acids, TNF-α and hyperinsulinemia and is involved in thedevelopment of insulin resistance [26,30,31]. In agreement withthis concept, our data indicate that Met–Leu treatment did notaffect the inhibitory Ser307 phosphorylation site. However, theincreased Ser318 phosphorylationwas associatedwith increasedAKT phosphorylation and an increased glucose utilization inmuscle (Figs. 4 and 5).

Diabetes and hyperglycemia are associated with increasedgluconeogenesis and glycogen synthesis in the liver. Phospho-enolpyruvate carboxykinase (PEPCK) and glycogen synthase,key rate limiting enzymes of gluconeogenesis and glycogensynthesis, respectively, are tightly controlled by transcriptionalregulation, phosphorylationandallosteric effectors in response toenergetic signals [32,33]. Glycogen synthase kinase-3-β activatesGS and is also negatively regulated by phosphorylation. AMPKactivation reduces hepatic glucose production and glycogen

Fig. 8 – Met–Leu effects on survival in C. elegans. (A and B) Survival curves of synchronized L4 worms maintained on NGM agarplates with indicated treatments under high glucose conditions (2% glucose added to agar plates) until death. (C) Life span tablesummarizingmedian survival,maximum life span and statistical significance. Treatments: control low (no glucose or treatmentsadded to NGM agar plates), control high (2% glucose added to NGM agar plates), Met (metformin 0.1 or 0.5 mmol/L added to NGMagar plates), Leu (leucine 0.25 mmol/L added to NGM agar plates).

1689M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 1 6 7 9 – 1 6 9 1

synthesis through phosphorylation of GSK-3β and of TORC2(transducer of regulated CREB activity 2 or CRTC2), a co-activatorof cAMP-responsive element binding protein (CREB) [34]. Phos-phorylation of CRTC2 leads to its translocation from the nucleusto the cytosol and consequently to a reduced transcriptionalactivity of CREB, an important transcriptional regulator of hepaticPEPCK. CREB, in turn, is also directly phosphorylated by GSK-3β[34,35]. Therefore, AMPK-induced reduction of gluconeogenesis ismediated both by phosphorylation of GSK-3β and consequentreduction in CREB activity, and by decreased nuclear presence ofTORC2 (CRTC2), which together reduce transcription of PEPCK.Sincewehave recently demonstrated inmice treatedwith aMet–Leu combination that hepatic AMPK was activated associatedwith increased fat oxidation and decreased liver triglycerideaccumulation [36], we wanted to explore the effects of AMPKactivation with our treatment on hepatic glucose metabolism.Consistent with downstream effects of AMPK activation, wefound trend for reduction (20%) ofCRTC2 in thenucleuswithMet–Leu associated with a significant increase in GSK-3β phosphory-lation and an accompanying decrease in GS phosphorylation inliver HepG2 cells (Fig. 6).

The nematode C. elegans conserves 65% of the humandisease-related genes, including pathways for fat and glucosemetabolism, and is used as awell-defined simplemodel systemfor obesity and insulin resistance [37]. Worms kept on highglucose (40 mmol/L) agar plates reached an internal glucoseconcentration of 14 mmol/L, comparable to plasma glucoseconcentrations under poorly controlled diabetic conditions.

This concentration shortened both median and maximum lifespan [38,39]. In another study [40], 2% glucose added to the agarplates shortened life span of C. elegans by ~20%, similar to datafrom the present study. Addition of theMet–Leu combination tothe agar plates reversed the high glucose-induced impairmentin fat and glucose metabolism in the worms and produced acorresponding increase in life span (Figs. 7 and 8).

The concentrations of metformin and leucine, used in thecell studies, were derived from our previous work for optimalSirt1 activation and are pharmacologically relevant, asfollows [11,12,36]. The fasting plasma concentration of leucinein humans is ~0.1 mmol/L and a plasma concentrationof 0.5 mmol/L can be reached after ingestion of ~300 mgleucine/kg/day [41]. The peak systemic plasma concentrationreached after a therapeutic dose of metformin (1000 mg) is~2.4 mg/L (=18 μmol/L). However, it is higher in the portalvein where it reaches between 40 and 70 μmol/L [42,43].Therefore, the 0.1 mmol/L concentration used in our cell studiesis close to a pharmacologically achievable concentration, and isabout 10 to 100 times lower than the metformin concentrationused in typical cell studies, which usually range from 1 to10 mmol/L [44–48]. In fact, it was demonstrated that metforminat low concentration (≤80 μmol/L) increases AMPK activityand suppresses dibutyryl-cAMP-stimulated glucose productionwhile the inhibition of the respiratory chain complex 1 happensonly at high concentrations (~5 mmol/L) [43].

There are some limitations to this study. While we mainlyfocus on the downstream targets of Sirt1 and AMPK, as they

1690 M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 1 6 7 9 – 1 6 9 1

aremaster regulators for energymetabolism,wedid not furtherexplore the bi-directional interaction of Sirt1 and AMPK.Previous reports demonstrate that Sirt1 can activate LKB1, theupstream kinase of AMPK, by lysine deactylation and thuspromoting its translocation to the cytoplasm where it caninteract with AMPK [6,14,21]. Although activation of AMPKdepends primarily on phosphorylation of the alpha subunit onThr172 by LKB1 [49], there is evidence that the AMPK enzymaticactivity may also be stimulated without changes in Thr172phosphorylation [50], However, we did not measure AMPKactivity directly. Instead we measured ACC (Ser79) phosphory-lation, a direct target of AMPK, as a measure of AMPK activity.ACC phosphorylation was increased two-fold by Met–Leu, andthis effect was prevented by either Sirt1 or AMPK inhibition.

5. Conclusion

In summary, these data demonstrate that leucine combinedwith a low dose of metformin stimulates the AMPK/Sirt1pathway, especially under the low NAD+ concentrations charac-teristic of energy-replete conditions. This study also providesevidence for the interacting network of AMPK and Sirt1, withboth being required forMet–Leu's overall action. This AMPK/Sirt1activation increases insulin-dependent and -independent glu-cose utilization in muscle, decreases hepatic glucose output andincreases fat oxidation. These effects together restored highglucose induced impairments in fat and glucose metabolism inC. elegans and increased median and maximum survival in theworms. These data provide mechanistic support for the thera-peutic benefit of metformin–leucine combinations suggested byour previous preclinical studies [12,36].

Author Contributions

J.B. and A.B. were responsible for design and execution of thecell experiments. All authorswere responsible for experimentaldesign, data analysis, interpretation, writing and editing themanuscript. All authors read and approved the manuscript.

Funding

Financial support for this study was provided by NuSirtBiopharma, Nashville, TN.

Disclosure Statement

The authors A.B., J.B., and M.B.Z. are employees of NuSirtBiopharma. A.B. and M.B.Z. are stockholder of NuSirt Biopharmaand have patents related to the reported work.

Acknowledgments

C. elegans strains were provided by the CGC, which is funded byNIH Office of Research Infrastructure Programs (P40 OD010440).

R E F E R E N C E S

[1] Viollet B, Guigas B, Sanz Garcia N, Leclerc J, Foretz M, AndreelliF. Cellular and molecular mechanisms of metformin: anoverview. Clin Sci (Lond) 2012;122:253–70.

[2] Pawlyk AC, Giacomini KM, McKeon C, Shuldiner AR, Florez JC.Metformin pharmacogenomics: current status and futuredirections. Diabetes 2014;63:2590–9.

[3] Nathan DM. Diabetes advances in diagnosis and treatment.JAMA 2015;314:1052.

[4] Brietzke SA. Oral antihyperglycemic treatment options fortype 2 diabetes mellitus. Med Clin North Am 2015;99:87–106.

[5] Lim PC, Chong CP. What’s next after metformin? Focus onsulphonylurea : add-on or combination therapy. Pharm Pract(Granada) 2015;13:606.

[6] Lan F, Cacicedo JM, Ruderman N, Ido Y. SIRT1 modulation ofthe acetylation status, cytosolic localization, and activity ofLKB1: possible role in AMP-activated protein kinase activation.J Biol Chem 2008;283:27628–35.

[7] Cantó C, Gerhart-Hines Z, Feige JN, Lagouge M, Noriega L,Milne JC, et al. AMPK regulates energy expenditure bymodulating NAD+ metabolism and SIRT1 activity. Nature2009;458:1056–60.

[8] Sun C, Zhang F, Ge X, Yan T, Chen X, Shi X, et al. SIRT1improves insulin sensitivity under insulin-resistant conditionsby repressing PTP1B. Cell Metab 2007;6:307–19.

[9] Ruderman NB, Carling D, Prentki M, Cacicedo JM. AMPK,insulin resistance, and the metabolic syndrome. J Clin Invest2013;123:2764–72.

[10] Banks AS, Kon N, Knight C, Matsumoto M, Gutiérrez-Juárez R,Rossetti L, et al. SirT1 gain of function increases energy efficiencyand prevents diabetes in mice. Cell Metab 2008;8:333–41.

[11] Bruckbauer A, Zemel MB. Synergistic effects of polyphenols andmethylxanthines with leucine on AMPK/Sirtuin-mediatedmetabolism in muscle cells and adipocytes. PLoS One 2014;9:e89166.

[12] Fu L, Bruckbauer A, Li F, Cao Q, Cui X, Wu R, et al. Leucineamplifies the effects of metformin on insulin sensitivity andglycemic control in diet-induced obese mice. Metabolism2015;64:845–56.

[13] Stiernagle T. Maintenance of C. elegans. WormBook 2006;1–11.[14] Ruderman NB, Xu XJ, Nelson L, Cacicedo JM, Saha AK, Lan F,

et al. AMPK and SIRT1: a long-standing partnership? Am JPhysiol Endocrinol Metab 2010;298:E751–60.

[15] Pfluger PT, Herranz D, Velasco-miguel S, Serrano M, TschoMH. Sirt1 protects against high-fat diet-induced metabolicdamage. PNAS 2008;105:9793–8.

[16] Gauthier M-S, O'Brien EL, Bigornia S, Mott M, Cacicedo JM, XuJX, et al. Decreased AMP-activated protein kinase activity isassociated with increased inflammation in visceral adiposetissue and with whole-body insulin resistance in morbidlyobese humans. Biochem Biophys Res Commun 2011;404:382–7.

[17] LiuY,WanQ,GuanQ,GaoL, Zhao J.High-fat diet feeding impairsboth the expression and activity of AMPKa in rats' skeletalmuscle. Biochem Biophys Res Commun 2006;339:701–7.

[18] Caton PW, Kieswich J, Yaqoob MM, Holness MJ, Sugden MC.Metformin opposes impaired AMPK and SIRT1 function anddeleterious changes in core clock protein expression in whiteadipose tissue of genetically-obese db/db mice. DiabetesObes Metab 2011;13:1097–104.

[19] Bosi E. Metformin–the gold standard in type 2 diabetes: whatdoes the evidence tell us? DiabetesObesMetab 2009;11(Suppl. 2):3–8.

[20] Caton PW, Nayuni NK, Kieswich J, Khan NQ, Yaqoob MM,Corder R. Metformin suppresses hepatic gluconeogenesisthrough induction of SIRT1 and GCN5. J Endocrinol 2010;205:97–106.

1691M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 1 6 7 9 – 1 6 9 1

[21] Zheng Z, Chen H, Li J, Li T, Zheng B, Zheng Y, et al. Sirtuin1-mediated cellular metabolic memory of high glucose viathe LKB1/AMPK/ROS pathway and therapeutic effects ofmetformin. Diabetes 2012;61:217–28.

[22] Bruckbauer A, Zemel MB. Synergistic effects of metformin,resveratrol, and hydroxymethylbutyrate on insulin sensitivity.Diabetes Metab Syndr Obes 2013;6:93–102.

[23] Foretz M, Hébrard S, Leclerc J, Zarrinpashneh E, Soty M,Mithieux G, et al. Metformin inhibits hepatic gluconeogenesisin mice independently of the LKB1/AMPK pathway via adecrease in hepatic energy state. J Clin Invest 2010;120:2355–69.

[24] Liang C, Curry BJ, Brown PL, Zemel MB. Leucine modulatesmitochondrial biogenesis and SIRT1-AMPK signaling inC2C12 myotubes. J Nutr Metab 2014;2014:239750.

[25] Hou X, Xu S, Maitland-Toolan KA, Sato K, Jiang B, Ido Y, et al.SIRT1 regulates hepatocyte lipid metabolism through activatingAMP-activated protein kinase. J Biol Chem 2008;283:20015–26.

[26] Boura-Halfon S, Zick Y. Phosphorylation of IRS proteins,insulin action, and insulin resistance. Am J PhysiolEndocrinol Metab 2009;296:E581–91.

[27] Yoshizaki T, Milne JC, Imamura T, Schenk S, Sonoda N,Babendure JL, et al. SIRT1 exerts anti-inflammatory effectsand improves insulin sensitivity in adipocytes. Mol Cell Biol2009;29:1363–74.

[28] Shibata T, Takaguri A, Ichihara K, Satoh K. Inhibition of theTNF-α-induced serine phosphorylation of IRS-1 at 636/639 byAICAR. J Pharmacol Sci 2013;122:93–102.

[29] Weigert C, Hennige AM, Brischmann T, Beck A, Moeschel K,SchaM, et al. The phosphorylation of Ser 318 of insulin receptorsubstrate 1 is not per se inhibitory in skeletalmuscle cells but isnecessary to trigger the attenuation of the insulin-stimulatedsignal *. J Biol Chem 2005;280:37393–9.

[30] D'Alessandris C, Lauro R, Presta I, Sesti G. C-reactive proteininduces phosphorylation of insulin receptor substrate-1 onSer307 and Ser 612 in L6 myocytes, thereby impairing theinsulin signalling pathway that promotes glucose transport.Diabetologia 2007;50:840–9.

[31] Prada PO, ZecchinHG, Gasparetti AL, TorsoniMA, UenoM, HirataAE, et al. Western diet modulates insulin signaling, c-JunN-terminal kinaseactivity, and insulin receptor substrate-1ser307phosphorylation in a tissue-specific fashion. Endocrinology 2005;146:1576–87.

[32] VonWilamowitz-Moellendorff A, Hunter RW, García-Rocha M,Kang L, López-Soldado I, Lantier L, et al. Glucose-6-phosphate-mediated activation of liver glycogen synthase plays a key role inhepatic glycogen synthesis. Diabetes 2013;62:4070–82.

[33] Sharabi K, Tavares CDJ, Rines AK, Puigserver P. Molecularpathophysiology of hepatic glucose production. Mol AspectsMed 2015;46:21–33.

[34] Horike N, Sakoda H, Kushiyama A, Ono H, Fujishiro M, KamataH, et al. AMP-activated protein kinase activation increasesphosphorylation of glycogen synthase kinase 3beta andthereby reduces cAMP-responsive element transcriptionalactivity and phosphoenolpyruvate carboxykinase C geneexpression in the liver. J Biol Chem 2008;283:33902–10.

[35] Zhang BB, Zhou G, Li C. AMPK: an emerging drug target fordiabetes and the metabolic syndrome. Cell Metab 2009;9:407–16.

[36] Fu L, Bruckbauer A, Li F, Cao Q, Cui X, Wu R, et al. Interactionbetween metformin and leucine in reducing hyperlipidemiaand hepatic lipid accumulation in diet-induced obese mice.Metabolism 2015;64:1426–34.

[37] Zheng J, Greenway FL. Caenorhabditis elegans as a model forobesity research. Int J Obes (Lond) 2012;36:186–94.

[38] Choi SS. High glucose diets shorten lifespan of Caenorhabditiselegans via ectopic apoptosis induction. Nutr Res Pract 2011;5:214.

[39] Schlotterer A, Kukudov G, Bozorgmehr F, Hutter H, Du X,Oikonomou D, et al. C elegans as model for the study of highglucose-mediated life span reduction. Diabetes 2009;58:2450–6.

[40] Lee S-J, Murphy CT, Kenyon C. Glucose shortens the life spanof C. elegans by downregulating DAF-16/FOXO activity andaquaporin gene expression. Cell Metab 2009;10:379–91.

[41] Elango R, Chapman K, Rafii M, Ball RO, Pencharz PB.Determination of the tolerable upper intake level of leucinein acute dietary studies in youngmen. Am J Clin Nutr 2012;96:759–67.

[42] Gabr RQ, Padwal RS, Brocks DR. Determination ofmetformin inhuman plasma and urine by high-performance liquidchromatography using small sample volume and conventionaloctadecyl silane column. J Pharm Pharm Sci 2010;13:486–94.

[43] He L, Wondisford FE. Metformin action: concentrationsmatter. Cell Metab 2015;21:159–62.

[44] Ouyang J, Parakhia RA, Ochs RS. Metformin activates AMPkinase through inhibition of AMP deaminase. J Biol Chem2011;286:1–11.

[45] Vytla VS, Ochs RS. Metformin increases mitochondrialenergy formation in L6 muscle cell cultures. J Biol Chem 2013;288:20369–77.

[46] Zang M, Zuccollo A, Hou X, Nagata D, Walsh K, Herscovitz H,et al. AMP-activated protein kinase is required for thelipid-lowering effect of metformin in insulin-resistanthuman HepG2 cells. J Biol Chem 2004;279:47898–905.

[47] Lee SK, Lee JO, Kim JH, Kim SJ, You GY, Moon JW, et al.Metformin sensitizes insulin signaling throughAMPK-mediatedPTEN down-regulation in preadipocyte 3T3-L1 cells. J CellBiochem 2011;112:1259–67.

[48] Owen MR, Doran E, Halestrap AP. Evidence that metforminexerts its anti-diabetic effects through inhibition of complex1 of the mitochondrial respiratory chain. Biochem J 2000;348(Pt 3):607–14.

[49] Woods A, Johnstone SR, Dickerson K, Leiper FC, Fryer LGD,Neumann D, et al. LKB1 is the upstream kinase in theAMP-activated protein kinase cascade. Curr Biol 2003;13:2004–8.

[50] Gowans GJ, Hawley SA, Ross FA, Hardie DG. AMP is a truephysiological regulator of AMP-activated protein kinase byboth allosteric activation and enhancing net phosphorylation.Cell Metab 2013;18:556–66.