AMP-activated protein kinase and ATP-citrate lyase are two distinct molecular targets for ETC-1002, a novel small molecule regulator of lipid and carbohydrate metabolism.

ETC-1002 (8-hydroxy-2,2,14,14-tetramethylpentadecanedioic acid) is a novel investigational drug being developed for the treatment of dyslipidemia and other cardio-metabolic risk factors. The hypolipidemic, anti-atherosclerotic, anti-obesity, and glucose-lowering properties of ETC-1002, characterized in preclinical disease models, are believed to be due to dual inhibition of sterol and fatty acid synthesis and enhanced mitochondrial long-chain fatty acid β-oxidation. However, the molecular mechanism(s) mediating these activities remained undefined. Studies described here show that ETC-1002 free acid activates AMP-activated protein kinase in a Ca2+/calmodulin-dependent kinase β-independent and liver kinase β 1-dependent manner, without detectable changes in adenylate energy charge. Furthermore, ETC-1002 is shown to rapidly form a CoA thioester in liver, which directly inhibits ATP-citrate lyase. These distinct molecular mechanisms are complementary in their beneficial effects on lipid and carbohydrate metabolism in vitro and in vivo. Consistent with these mechanisms, ETC-1002 treatment reduced circulating proatherogenic lipoproteins, hepatic lipids, and body weight in a hamster model of hyperlipidemia, and it reduced body weight and improved glycemic control in a mouse model of diet-induced obesity. ETC-1002 offers promise as a novel therapeutic approach to improve multiple risk factors associated with metabolic syndrome and benefit patients with cardiovascular disease.

and stored in sterile microcentrifuge tubes at 4°C for up to four weeks (stability was assessed). Working solutions of ETC-1002 were prepared in serum-free DMEM containing 12 mM HEPES, 10,000 U/ml penicillin, and 100 g/ml streptomycin. ETC-1002-CoA was synthesized using rat liver microsomes essentially as described by Cramer et al. ( 12 ). For in vivo experiments, ETC-1002 dosing solutions were formulated by preparing a disodium salt aqueous solution using 2:1 molar ratio of NaOH to ETC-1002 in water. Carboxymethyl cellulose (CMC) and Tween-20 were added to make a fi nal solution containing 0.5% CMC and 0.025% Tween, with a fi nal pH 7-8 . Compound concentrations in dosing solutions were based upon a 10 ml/kg body weight dosing volume.

Cell culture and siRNA transfection
HepG2 cells were grown and treated in DMEM containing 1 g/L D-glucose supplemented with 10% FBS. Reverse transfections were performed in 6-well culture plates at 2.5 × 10 5 cells/well using Lipofectamine 2000. Cells were incubated for 48 h with 10 nM silencer siRNA for liver kinase ␤ (LKB)1 or negative control prior to compound treatment.

Nucleotide measurements
Cells were placed on ice, deproteinized with ice-cold 6% perchloric acid, scraped, neutralized with 10 M NaOH, and buffered with 1 M K 2 HPO 4 to precipitate potassium perchlorate. Solutions were transferred to microcentrifuge tubes and centrifuged for 5 min. Supernatant (20 l) was diluted in cold HPLC-grade water and maintained at approximately 4°C until injection. Diluted sample (20 µl) was injected into the LC-MS/MS system and three m/z transitions were monitored ( m/z 348.2 → 136.5 for AMP, m/z 428.1 → 136.5 for ADP, and m/z 508.2 → 136.5 for ATP) on an API-4000 triple-quadrupole mass spectrometer (AB Sciex, Framingham, MA). The relative amounts of AMP, ADP, and ATP were determined for each sample by normalizing triplicate measures of test conditions to vehicle treatment from the same plate. To determine the adenine nucleotide levels in freeze-clamped liver, approximately 500 mg of frozen liver was homogenized in ice-cold methanol and diluted in cold HPLC water before injecting 30 l into the LC-MS/MS system. AMP, ADP, and ATP concentrations in liver were determined by comparing the sample peak area to the peak area of known calibration standard samples prepared in methanol.

De novo lipid synthesis assay
Rates of lipid synthesis were assessed in cultured primary rat hepatocytes using [ 14 C]acetate or [ 14 C]citrate. Experiments were performed in DMEM with 4.5 g/l glucose. Cells were treated with compound or vehicle (0.1% DMSO) for up to 4 h followed by lipid isolation. After metabolic labeling, saponifi ed and nonsaponifi ed lipids were extracted from cells essentially as described by Slayback et al. ( 17 ). benefi cially affect other common cardiometabolic risk factors associated with atherosclerosis and MetS.
ETC-1002 (8-hydroxy-2,2,14,14-tetramethylpentadecanedioic acid), also known as ESP55016, is a novel investigational drug being developed for the treatment of dyslipidemia and other cardio-metabolic risk factors. ETC-1002 favorably changes lipid profi les in preclinical models of dyslipidemia ( 12 ), benefi ts glucose homeostasis in mouse models of impaired glycemic control ( 12,13 ), and decreases atherosclerosis in LDL-receptor-defi cient mice ( 14 ). Importantly, clinical studies have shown that ETC-1002 reduces LDL-C levels in subjects with mild dyslipidemia and has benefi cial effects on other relevant cardio-metabolic risk factors, including insulin levels, hsCRP, and blood pressure ( 15 ).
In the current study, we demonstrate that these benefi cial effects of ETC-1002 on lipid and carbohydrate metabolism are tightly linked to activation of hepatic AMPactivated protein kinase (AMPK), a master kinase controlling whole-body energy homeostasis. Additionally, the CoA thioester of ETC-1002 revealed potent inhibitory activity against hepatic ATP-citrate lyase (ACL), another central enzyme coordinating extra-mitochondrial carbon fl ux into the synthesis of lipids. The combination of these two distinct molecular mechanisms not only may regulate LDL-C but also may exhibit additional benefi cial attributes for the treatment of CVD and provide clinically meaningful effi cacy for other risk factors associated with MetS.

In vivo studies
Wistar rats.

Glucose production assay
Glucose production was measured in primary rat hepatocyte cultures. Cells were cultured in glucose-and phenol red-free DMEM, containing 10 mM lactate, 1 mM pyruvate, and nonessential amino acids (glucose production buffer, GPB). To assess the effects of ETC-1002 on glucagon-stimulated glucose production, cells were incubated with and without 0.3 M glucagon (Sigma, St. Louis, MO) with various concentrations of ETC-1002 (0.1 to 100 M). Media was sampled over time. Following specifi ed treatments, cells were washed twice in GPB. Cells were then incubated for an additional hour to assess glucose production by adding GPB containing equivalent glucagon concentrations without ETC-1002. Cells were incubated for 1 h, and the concentration of glucose in the media was determined using a glucose oxidase assay kit (catalog #GAGO20-1KT; Sigma Chemicals).

ATP-citrate lyase enzyme activity assay
The activity of recombinant human ACL was carried out essentially as described in ( 18 ). Briefl y, 7.

Short-chain acyl-CoA and citrate measurements
For in vitro studies, cell culture samples were prepared as previously described for nucleotide measurements. For in vivo studies, freeze-clamped liver was homogenized in ice-cold methanol, and then samples were prepared by liquid-liquid extraction using chloroform and 0.1% formic acid in water. A portion of the aqueous phase was transferred and injected into the LC-MS ETC-1002-CoA concentrations were determined by comparing the sample peak area to the peak area of a ETC-1002-CoA calibration standard.

RESULTS
ETC-1002 has been previously shown to inhibit de novo sterol and fatty acid synthesis in primary rat hepatocytes in vitro and in vivo, with equal potency ( 12 ). In these studies, ETC-1002-CoA thioester was identifi ed as the primary active form of ETC-1002 and was shown to inhibit partially purifi ed ACC (IC 50 = 29 M) without activating the AMPK pathway in vitro ( 12 ). In a follow-up study reported here, we unexpectedly found a marked and sustained increase in AMPK (T172) (358.3% ± 48.14; P = 0.0007) and ACC (S79) phosphorylation (164.7% ± 12.39; P = 0.001) ( Fig. 1 ) in rat livers following two weeks of treatment with ETC-1002. Interestingly, in liver extracts, the ETC-1002 free acid concentration was approximately 110:1 molar ratio compared with the CoA thioester indicating that previously uncharacterized free acid may be involved in regulating ETC-1002-mediated metabolic activities. Furthermore, while ACC inhibition has been attributed to ETC-1002-CoA, this only explains the inhibition of fatty acid synthesis, leaving the mechanism for the equipotent inhibition of sterol synthesis unidentifi ed.
To obtain better insight into the molecular targets for ETC-1002 free acid and the CoA thioester, we fi rst characterized the temporal nature of ETC-1002 uptake and CoA New Brunswick, NJ) for 12 weeks. Mice were randomized into two treatment arms at 20 weeks of age based on 4 h fasted blood glucose and body weight and received oral dosing of either CMC/ Tween vehicle or 30 mg/kg/day ETC-1002 q.d in the morning for an additional two weeks. Body weight and food consumption were monitored throughout the study. Following the two-week dosing period, food was removed at 8 AM, and bedding was changed 2 h prior to oral administration of ETC-1002. Two hours post dose, fasting samples were collected. Fasting blood glucose levels were measured immediately prior to anesthesia using a hand-held Alphatrak glucometer (Abbott, Chicago, IL), with blood collected by unrestrained tail snip. For insulin determinations, blood was collected under isofl urane anesthesia via retroorbital sinus into EDTA-coated tubes, and plasma was isolated by centrifugation. Plasma insulin levels were measured with a commercially available ELISA (Crystal Chem Associates, Downers Grove, IL).

Lipoprotein profi les and size exclusion chromatography
Plasma samples were transferred to autosampler vials and maintained at 4°C until injection onto the FPLC system (Waters Alliance 2695 Separations Module) utilizing size-exclusion chromatography with a Superose 6 10/300GL column (GE Healthcare Biosciences, Uppsala, Sweden) and 0.9% sodium chloride/0.02% sodium azide in water. Postcolumn effl uent and CHOL CHOD-PAP cholesterol reagent (Roche Diagnostics, Indianapolis, IN) were mixed in-line and reacted in a 37°C heated knitted coil prior to monitor at 490 nm (Waters 2996) PDA. Very low density lipoprotein (VLDL), low density lipoprotein (LDL), and high density lipoprotein (HDL) ratios were determined by calculation of the peak area for each protein as a percentage of the total peak area of all proteins detected in the sample.

HPLC-UV for ETC-1002-CoA determination
A 15 l aliquot of the aqueous layer (top, approximately 0.8 ml) from the extraction procedure described above was injected into the HPLC system utilizing an Alltima C8 5 , 250 × 4.6 mm ID HPLC column (Alltech Associates, Deerfield, IL) running a 15-40% acetonitrile in 25 mM potassium hydrogen phosphate (pH 7.0) gradient before being UV detection at 254 nm on a G1314A detector (Agilent Technologies, Santa Clara, CA). malonyl-CoA, HMG-CoA, acetyl-CoA, and citrate were selected to provide greater insight into ETC-1002-related effects occurring between mitochondrial citrate production and the entry of acetyl-CoA into the fatty acid and sterol synthesis pathways. The treatment of primary rat hepatocytes with ETC-1002 resulted in concentrationdependent reductions in acetyl-CoA, malonyl-CoA, and HMG-CoA, with concomitant increases in citrate ( Fig. 3A and Table 1 ) and ETC-1002-CoA ( Fig. 3B and Table 1 ), whereas no signifi cant effect on CoASH concentration was observed ( Fig. 3A and Table 1 ). Remarkably, these effects occurred within 5 min of treatment (data not shown). Conversely, the human hepatoma cell line HepG2 served as a negative control, showing the absence of ETC-1002-CoA formation ( Fig. 3B ). Importantly, Triacsin C, an inhibitor of multiple long-chain acyl-CoA synthetase (ACS) isoforms, not only reduced intracellular concentrations of ETC-1002-CoA but also attenuated the effects of ETC-1002 on metabolic intermediates ( Table 2 ) and de novo lipid synthesis ( Fig. 3C ). These data demonstrate that the conversion of ETC-1002 to a CoA thioester is required for equipotent inhibition of de novo sterol and fatty acid synthesis and that inhibition occurs after citrate formation, at or before ACL-dependent acetyl-CoA production. Indeed, hepatic knockdown of ACL expression thioesterifi cation in primary rat hepatocytes. Treatment with ETC-1002 resulted in rapid uptake and CoA thioesterifi cation ( Fig. 2A ), which was associated with immediate inhibition ( р 5 min) of de novo lipid synthesis ( Fig. 2B ) and transient increases in phosphorylation of AMPK (T172), ACC (S79) and HMGR (S182) ( Fig. 2C ). These data revealed that ETC-1002 uptake in primary rat hepatocytes is closely linked to CoA thioesterifi cation and, unlike in vivo, results in an approximately 1:1 to 2:1 molar ratio (ETC-1002 free acid:ETC-1002-CoA) and only transient AMPK activation ( Fig. 2C ). The identifi cation of relatively low intracellular ETC-1002 free acid concentrations and lack of sustained AMPK activation in primary rat hepatocytes further suggested that, in vivo, the ETC-1002 free acid may indeed be linked to the AMPK activation, while ETC-1002-CoA may mediate its effects through a mechanism distinct from AMPK.

ETC-1002-CoA inhibits de novo sterol and fatty acid synthesis via direct inhibition of ACL
To characterize the link between ETC-1002-CoA thioesterifi cation and equipotent inhibition of sterol and fatty acid synthesis, mass changes in intermediates of lipid synthesis were analyzed with a radioisotope tracer-independent LC-MS/MS method. Free CoA (CoASH), sterol regulatory element binding proteins (SREBP) activity, resulting in the upregulation of key enzymes involved in fatty acid synthesis. Consistent with in vitro data, in nutritionally staged (lipogenic) rats treated with a single oral dose of 30 mg/kg of ETC-1002, the CoA thioester was detected in liver extracts within 2 h after dosing. Furthermore, ETC-1002 treatment was associated with a reduction in hepatic intermediates of lipid synthesis, including acetyl-CoA, malonyl-CoA, and HMG-CoA, and an increase in citrate levels ( Fig. 4B ). The effects on these intermediates of lipid synthesis were sustained when assessed 8 h after dosing, whereas citrate returned to levels comparable to vehicle-treated rats (data not shown).
While the ability of ETC-1002-CoA to directly inhibit ACL may explain, at least in part, the inhibition of de novo lipid synthesis both in vitro and in vivo, further understanding of the molecular mechanism for the ETC-1002 free acid should provide crucial information as to how these two active forms might cooperate to mediate multiple benefi cial effects in preclinical disease models.
has been shown to reduce liver acetyl-CoA and malonyl-CoA concentrations in a manner consistent with that observed with ETC-1002 in vitro ( 19,20 ).
To assess the effects of ETC-1002 and ETC-1002-CoA on ACL directly, a cell free activity assay was adapted from methods by Ma et al. ( 18 ) using recombinant human ACL. ETC-1002-CoA, but not the free acid, markedly inhibited ACL activity in a concentration-dependent manner (IC 50 = 17.8 M) under conditions saturated for ATP, citrate, and CoASH ( Fig. 4A ). When the assay was performed with CoASH concentration comparable to the reported apparent K m for ACL, the potency increased approximately 4.5-fold (IC 50 = 4.01 M). These data support that ETC-1002-CoA directly inhibits ACL at concentrations consistent with in vitro inhibition of lipid synthesis and that it is competitive for CoASH. Hydroxycitrate (HCA), a natural citrate analog, was used as a reference inhibitor. To confi rm the physiological relevance of ACL inhibition by ETC-1002, Wistar rats were nutritionally staged by fasting and refeeding a high-carbohydrate diet, which increases lipogenesis through the induction of potential AMPK-activating properties of ETC-1002 free acid. Consistent with the apparent association between ETC-1002 uptake and CoA thioesterifi cation, HepG2 cells required higher media concentrations of the compound to achieve intracellular levels comparable to those measured

ETC-1002 activates the AMPK pathway via a calcium-and energy-independent mechanism
Unlike primary rat hepatocytes, HepG2 cells fail to generate measurable amounts of ETC-1002-CoA thioester ( Fig. 3B ), making this cell line a viable model for characterizing the   and AMPK signaling), hormone signaling (e.g., insulin and glucagon), and substrate fl ux. As ETC-1002 can potentially mediate its effects on lipid metabolism through AMPK-and ACL-related pathways, glucagon-stimulated glucose production was assessed in primary rat hepatocytes exposed to ETC-1002 or insulin. Glucagon treatment alone induced an approximately 2-fold increase in glucose levels in media conditioned by hepatocytes overnight, which was reduced by ‫ف‬ 50% in cells treated with 10 M ETC-1002 or 100 nM insulin ( Fig. 6A ). Additionally, inhibition of glucose production by ETC-1002 was concentration-dependent, with IC 50 = 3.6 M ( Fig. 6B ). Further investigation into the underlying mechanism(s) mediating these effects showed that ETC-1002 decreased glucagon-induced PEPCK and G6Pase protein expression ( Fig. 6C ). Furthermore, unstimulated hepatocytes treated overnight with metformin or ETC-1002 showed reductions in total FOXO1 levels by ‫ف‬ 30% and ‫ف‬ 40%, respectively ( Fig. 6D ), consistent with its role in hepatic glucose production. These data demonstrate that ETC-1002 reduces hepatocyte glucose production in primary rat hepatocytes through a mechanism associated with reductions in FOXO1, PEPCK, and G6Pase protein levels.

LKB1 is required for ETC-1002-mediated activation of the AMPK pathway
In liver, physiological activation of AMPK is considered to be mediated primarily through LKB1-dependent phosphorylation. To determine whether the activation of AMPK by ETC-1002 was LKB1-dependent, we used small inhibitory RNA (siRNA) interference to reduce endogenous LKB1 protein levels in HepG2 cells and assessed the effects of ETC-1002 on ACC (S79) phosphorylation along with AEC, intracellular lipids, and the effectors of gluconeogenesis, FOXO1, and HNF-4 ␣ .
HepG2 cells were transfected with negative control "mock" or LKB1 siRNA and treated for 24 h with vehicle, ETC-1002 (100 M), or metformin (1,000 M). HepG2 cells transfected with LKB1 siRNA resulted in a 75% ( P = 0.0119) reduction in LKB1 protein compared with cells transfected with mock siRNA ( Fig. 7A ). While mocktransfected cells showed an ETC-1002-and metformindependent increase (ETC-1002 = +780% ± 35; P = 0.0374; metformin = +730% ± 80; P = 0.0187) in ACC (S79) phosphorylation and reductions in HNF-4 ␣ protein levels ( Fig. 7B, C ), these effects were abolished in cells transfected with LKB1 siRNA, demonstrating that both ETC-1002 and metformin activate AMPK in an LKB1-dependent fashion. Interestingly, FOXO1 protein levels were not changed with ETC-1002 treatment, metformin treatment, or LKB1 knockdown, suggesting the LKB1-AMPK axis does not control FOXO1 expression under these conditions (data not shown). Consistent with inhibition of mitochondrial respiration, the treatment of HepG2 cells with metformin resulted in a reduction in AEC, which was further decreased by LKB1 knockdown ( Fig. 7D and supplementary Table I). Contrary to the activities of metformin and consistent with energy-independent nature of AMPK activation by ETC-1002, no changes in in primary rat hepatocytes and rat liver. No signifi cant increases in markers of cell viability including LDH release and caspase 3/7 activity could be linked to compound exposure (data not shown).
Consistent with ETC-1002-induced AMPK activation observed in rat liver ( Fig. 1 ), HepG2 cells treated with ETC-1002 revealed a sustained and concentration-dependent increase in AMPK (T172) and ACC (S79) phosphorylation comparable to the AMPK-activating effect of metformin (1,000 M) ( Fig. 5A ). To further characterize the mechanism leading to AMPK activation by ETC-1002, HepG2 cells were pretreated with STO-609, an AMPK kinase Ca 2+ / calmodulin-dependent kinase ␤ (CaMKK ␤ )-specifi c inhibitor. STO-609 did not signifi cantly reduce AMPK or ACC phosphorylation in ETC-1002-or metformin-treated cells, indicating that AMPK activation is not dependent on intracellular Ca 2+ signaling ( Fig. 5B ). Intriguingly, while the ATP analog and AMPK inhibitor, compound C, signifi cantly reduced AMPK and ACC phosphorylation by metformin, it did not inhibit ETC-1002-dependent AMPK activation ( Fig. 5B ). To determine whether ETC-1002-dependent AMPK activation is associated with reductions in AEC, intracellular ATP, ADP, and AMP concentrations were measured in HepG2 cells treated with vehicle, rotenone (10 M), or ETC-1002 (100 M). Treatment with rotenone (complex I inhibitor) resulted in increased AMP and ADP levels and in reduced ATP levels and AEC compared with vehicle treatment, while ETC-1002 had no effect ( Fig. 5C ). These data suggest that the activation of the AMPK pathway by ETC-1002 may be independent of reductions in energy production.
Many immortalized and highly proliferative cell lines, such as HepG2, are known to synthesize the majority of their ATP from glycolysis despite fully functional mitochondria and suffi cient oxygen availability ( 21 ). Therefore, to ensure that the apparent energy-independence of AMPK activation by ETC-1002 was not an artifact of compensatory anaerobic metabolism, HepG2 cells were grown in normal glucose-containing media or glucosefree media supplemented with galactose ( 22 ). Intracellular ATP was measured in cells treated with ETC-1002, rotenone, or CCCP (mitochondrial uncoupler). While HepG2 cells grown in the presence of glucose showed an approximately 50% reduction in ATP content when exposed to rotenone or CCCP, ATP levels dropped signifi cantly further when galactose was substituted for glucose ( Fig. 5D ). Remarkably, treatment with up to 300 M ETC-1002 in either growth medium did not alter ATP content, further indicating that AMPK activation by ETC-1002 was independent of effects on mitochondrial-dependent oxidative energy production. These intriguing fi ndings in HepG2 cells suggest that ETC-1002 does not likely affect AMPK through the disruption of calcium homeostasis or energy production but through a distinct mechanism.

ETC-1002 decreases glucagon-dependent glucose production in hepatocytes in vitro
Hepatic glucose production is regulated by multiple factors, including nutrients, energy availability (ATP availability and TG levels ( Fig. 7E, F ). These data demonstrate that LKB1 activity is required for ETC-1002-mediated AMPK activation via a pathway that does not alter energy production in HepG2 cells. AEC in either mock or LKB1 siRNA-treated cells were detected in cells treated with ETC-1002. Importantly, LKB1 knockdown also prevented ETC-1002-and metformin-induced reduction in intracellular total cholesterol aminotransferase (AST) ( Fig. 8A ). Activation of hepatic AMPK with ETC-1002 treatment from the two-week Wistar rat study was not associated with altered hepatic AEC ( Fig. 8B and Table 3 ); however, ETC-1002 treatment coincided with a 70% reduction in hepatic TG, while low basal levels of CE and FC remained insensitive to treatment ( Fig. 8B ). Importantly, reductions in liver TG content was accompanied by 51% increase in plasma ␤ -HBA, indicating enhanced liver fatty acid ␤ -oxidation (supplementary Table II). Likewise, reductions in plasma TG ( Ϫ 34%) and non-HDL-C ( Ϫ 21%), along with a greater than 2-fold increase in plasma HDL-C (supplementary Table I), were consistent with previously published effects of pharmacological

ETC-1002 modulates plasma and tissue functional biomarkers of AMPK and ACL activity in vivo
To determine whether modulation of AMPK and ACL activities by ETC-1002 could be linked to physiological responses in vivo, plasma and liver samples from chowfed Wistar rats treated with ETC-1002 for two weeks at 30 mg/kg were examined for functional biomarkers of alterations in lipid and carbohydrate metabolism. Safety studies in the Wistar rat orally dosed with ETC-1002 at 30 mg/kg for four weeks showed plasma exposures equivalent to clinical exposures with no meaningful drugrelated effect on safety parameters, including liver markers of injury alanine aminotransferase (ALT) and aspartate

ETC-1002 corrects dyslipidemia in golden Syrian hamsters and improves glycemic control in a mouse model of diet-induced obesity
To evaluate the therapeutic effects of ETC-1002 on metabolic disease in vivo, a series of pharmacology studies were undertaken in dyslipidemic golden Syrian hamsters and in the mouse model of diet-induced obesity (DIO). The human-like lipoprotein metabolism of the golden Syrian hamster characterized by comparable LDL receptor regulation ( 24 ), plasma cholesteryl ester transfer protein activity ( 25 ), and hepatic secretion of full-length apoB (apoB100) containing VLDL particles ( 26 ) make it an attractive model for preclinical evaluation of hypolipidemic agents. Furthermore, feeding hamsters a HFHC diet increases weight gain and induces dearrangements in lipoprotein metabolism, resulting in hypertriglyceridemia and hypercholesterolemia with a human-like dyslipidemic lipoprotein profi le and activation of AMPK in rats ( 23 ). Furthermore, ETC-1002 treatment increased levels of hepatic PGC-1 ␣ (68.4% ± 25.6; P = 0.0240) and mitochondrial citrate synthase (105.6% ± 25.2; P = 0.0030), indicating the upregulation of hepatic oxidative metabolism ( Fig. 8C ). Although the chow-fed rat is not a model for hyperglycemia or dyslipidemia, the reduced basal levels of liver FOXO1 and HNF-4 ␣ (61.4% ± 10.0; P = 0.0003 and 54.7% ± 12.2; P = 0.002, respectively) associated with ETC-1002 treatment further supports a potential mechanistic link to the regulation of carbohydrate metabolism.
This mechanistic translation into a normal chow-fed rat model provides further evidence for the AMPK-activating properties of ETC-1002 observed in vitro. Importantly, these data support the AEC-independent activation of AMPK by ETC-1002, which was associated with multiple markers of AMPK activation at the levels of signal transduction, transcription factors, hepatic lipid mass, and lipid biomarkers. An excess of visceral adipose tissue coupled with hyperglycemia and hyperinsulinemia makes the mouse model of DIO an attractive in vivo tool for studying mechanisms of impaired glycemic control and for the characterization of novel therapeutic agents (33)(34)(35)(36)(37)(38). To determine whether ETC-1002 improves glucose homeostasis in vivo, mice were kept on HFD for 12 weeks prior to ETC-1002 treatment. Administration of ETC-1002 at 30 mg/kg for 14 days resulted in a 9% ( P < 0.05) reduction in body weight ( Fig. 10 ) while daily food consumption remained unchanged. Consistent with decreased glucose production by hepatocytes in vitro ( Fig. 6 ), two weeks of treatment with ETC-1002 was suffi cient to signifi cantly reduce fasting plasma glucose levels by 13% ( P < 0.05). Likewise, plasma insulin levels in animals treated with ETC-1002 were reduced by 42% ( P < 0.05) ( Fig. 10 ) further supporting the benefi cial effects of ETC-1002 on regulation of glucose homeostasis in vivo.
Based on these data generated in a hamster model of dyslipidemia and a mouse model of DIO, ETC-1002 treatment has a broad range of favorable metabolic effects, consistent with the activation of AMPK and inhibition of hepatic ACL. While the relative contributions of AMPK activation and ACL inhibition in vivo are not addressed in these studies, the robust effects of ETC-1002 hepatic steatosis (27)(28)(29)(30)(31)(32). To determine whether ETC-1002 benefi cially affects tissue and plasma parameters associated with dyslipidemia, HFHC-fed hamsters were treated for three weeks with ETC-1002 (30 mg/kg/day). ETC-1002 administration was associated with detectable levels ETC-1002-CoA in liver extracts (data not shown) and a 14.4% ( P < 0.01) decrease in body weight gain, while no signifi cant changes in food consumption or plasma ALT or AST were observed (data not shown). Consistent with enhanced fatty acid ␤ -oxidation, ETC-1002 treatment increased plasma ␤ -HBA (20%; P < 0.05) and reduced plasma NEFA and epididymal fat mass by 34% and 35%, respectively ( P < 0.05) ( Fig. 9A ). Remarkably, ETC-1002 treatment resulted in dramatic reductions in hepatic TG ( Ϫ 64%, P < 0.001), CE ( Ϫ 67%, P < 0.05), and FC ( Ϫ 31%, P < 0.05) content ( Fig. 9B ), along with reductions in plasma TG ( Ϫ 41%, P < 0.05) and total cholesterol ( Ϫ 41%, P < 0.05) ( Fig. 9C ). Further insight into the hypolipidemic effects of ETC-1002 was obtained by measuring the distribution of cholesterol among the proatherogenic lipoprotein particles LDL and VLDL. Hamsters treated with ETC-1002 showed a 64% and 62% reduction in in LDL-C and VLDL-C, respectively ( P < 0.05) ( Fig. 9C ), while HDL-C levels remained unchanged (data not shown).  Fig. 1 were analyzed for AEC, TG, CE, and FC content, and expressed as mg/g liver wet weight. (C) Protein levels of PGC-1 ␣ , citrate synthase, FOXO1, and HNF-4 ␣ were measured by Western blot, normalized to ␤ -actin, and expressed as mean percentage of vehicle control ± SEM, n = 5. Comparisons were made using unpaired Student t -test; * P < 0.05. ii ) direct inhibition of hepatic ACL by the ETC-1002-CoA thioester ( Fig. 11 ). The elucidation of these distinct molecular mechanisms was contingent on the identifi cation of a key difference that exists between our in vitro rat hepatocyte model and intact rat liver. In our in vitro model, primary rat hepatocytes predominantly convert ETC-1002 to a CoA thioester, limiting intracellular exposure to the free acid. In vivo, we show that ETC-1002 free acid is >100fold more prevalent than the CoA thioester in rat liver and is associated with AMPK activation. This difference between in vitro and in vivo results may explain the absence of AMPK activation observed in previous studies ( 12 ). The rapid formation of ETC-1002-CoA and subsequent transient increases in citrate with concomitant reductions in acetyl-CoA, malonyl-CoA, and HMG-CoA in rat hepatocytes is consistent with ACL inhibition ( Fig. 10 ) and dual inhibition of fatty acid and sterol synthesis ( Fig. 10 ). The role of ETC-1002-CoA in inhibition of lipid synthesis was further supported by exploiting the ACS inhibitory activity of Triacsin C, which reduced ETC-1002-CoA formation, and ETC-1002-dependent reductions in acetyl-CoA, malonyl-CoA, and HMG-CoA. Interestingly, cotreatment of hepatocytes with ETC-1002 and Triacsin C trended toward increased citrate levels, which may appear inconsistent with ACL deinhibition. This result could be due to inhibition of fatty acid oxidation by Triacsin C, resulting in a dependence on glucose oxidation with increased acetyl-CoA levels supporting additional citrate production. Inhibition on hepatic and plasma lipid parameters along with improved glucose homeostasis suggest that these two activities may work cooperatively.

DISCUSSION
These in vitro and in vivo studies support that ETC-1002 works through two distinct mechanisms: i ) activation of the hepatic AMPK pathway by the ETC-1002 free acid and  acetyl-CoA and malonyl-CoA, plasma triglycerides and fatty acids ( 19 ), and hepatic lipids ( 20 ).
Intriguingly, recent work has identifi ed a novel regulatory role for ACL-derived acetyl-CoA levels, which links energy substrate availability with gene expression through controlling substrate availability for the acetylation of key proteins (49)(50)(51). These fi ndings have raised the possibility that extra-mitochondrial acetyl-CoA concentrations may also regulate PGC-1 ␣ acetylation through controlling activity of the acetyltransferase GCN5 at the substrate level. Until recently, the regulatory roles of acetyl-CoA have been believed to be primarily mediated within mitochondria through allosteric means, such as controlling the fate of pyruvate toward carboxylation or dehydrogenation. Recently, many extra-mitochondrial regulatory acetylations have been characterized and shown to be dependent on ACL-derived acetyl-CoA pools ( 50 ). These fi ndings support a link between energy excess and gene regulation, as sensed through glucose-derived acetyl-CoAdependent cytosolic/nuclear acetylation reactions ( 49,50 ). For many years, the position of ACL in lipid biosynthesis has made it an attractive target for pharmacological of ACL was supported in vivo by demonstrating that hepatic ETC-1002-CoA formation was associated with decreases in the ACL product acetyl-CoA, which is the fi nal common substrate for both fatty acid and sterol synthesis. Furthermore, we show that ETC-1002 -CoA inhibits recombinant human ACL (rhuACL) directly in a cell-free system.
ACL is a key cytosolic enzyme, which precedes HMGR and ACC in the lipid biosynthesis pathways and catalyzes the cleavage of mitochondrial-derived citrate to cytosolic oxaloacetate and acetyl-CoA (39)(40)(41). ACL is highly expressed in lipogenic tissues, such as liver and adipose ( 42 ), and its products are acetyl-CoA and oxaloacetate. In liver, oxaloacetate can serve as the rate-limiting substrate for gluconeogenesis. It is interesting to speculate that ETC-1002-CoA-mediated ACL inhibition may contribute to reduced rates of gluconeogenesis.
The effects of ETC-1002 on hepatic PGC-1 ␣ protein levels are also interesting. In liver, PGC-1 ␣ is known to be an important transcriptional coactivator that mediates many metabolic adaptations associated with the fasting and exercise phenotype. PGC-1 ␣ activity increases in response to glucagon and low energy signals, resulting in increasing nuclear receptors' (e.g., PPAR ␣ ) target gene transcription and increasing fatty acid catabolism ( 101 ). Additionally, activation of AMPK in liver has been shown to result in the upregulation of PGC-1 ␣ activity and protein levels, which was associated with increased mitochondrial content ( 102 ). Although mitochondrial number was not measured in the studies presented here, the increases in PGC-1 ␣ and citrate synthase associated with ETC-1002 treatment suggest mitochondrial adaptations consistent with AMPK activation.
PGC-1 ␣ also coordinates the hepatic fasting gluconeogenic response through its interactions with transcription factors FOXO1 and HNF-4 ␣ . The coordination of these interactions maintains hormone sensitivity during the fasted and fed conditions, ensuring appropriate PEPCK and G6Pase expression levels ( 76,103 ). Interestingly, the activation of AMPK has been shown to directly phosphorylate HNF-4 ␣ , resulting in altered nuclear localization, reduced DNA binding, and increased degradation ( 77,104 ). Consistently, ETC-1002 treatment was associated with reduced hepatic HNF-4 ␣ protein levels, an activity that was shown in HepG2 cells to be dependent on the LKB1/ AMPK axis. Furthermore, pathophysiological conditions associated with hepatic insulin resistance results in the deregulation of posttranscriptional modifi cation (phosphorylation and acetylation) of FOXO1, leading to hyperactive transcriptional activity, which is believed to contribute to hyperglycemia ( 105 ). Although reductions in FOXO1 protein levels were observed in rat liver and primary rat hepatocytes, protein levels did not change in response to ETC-1002 treatment in HepG2 cells. Zatara et al. ( 106 ) showed that the LCFA analogs of the AMPK-activating MEDICA series reduced liver FOXO1 protein levels in Sprague-Dawley rats and suppressed the nuclear:cytosolic ratio in HepG2 cells in an AMPK-dependent manner. Additional investigations are required to determine whether ETC-1002 modulates FOXO1 nuclear localization in a similar manner or through a distinct mechanism linked to ACL inhibition.
In the present studies, we have described two molecular targets for ETC-1002 and found that the free acid activates AMPK, whereas the ETC-1002-CoA thioester directly inhibits hepatic ACL. We show that these distinct molecular mechanisms are complementary in their benefi cial effects on the regulation of lipid and carbohydrate metabolism in vitro and in vivo. Furthermore, we have demonstrated therapeutic relevance by showing that ETC-1002 reduces circulating proatherogenic LDL-and VLDL-cholesterol concentrations, as well as body weight gain and hepatic lipids in a human-like hyperlipidemic hamster model. These unique properties of ETC-1002 can be potentially inhibition. However, the emerging role of ACL-derived acetyl-CoA level in linking energy substrate availability with gene expression heightens the relevance of ACL in regulating energy homeostasis.
The underlying mechanism mediating the in vitro and in vivo activation of AMPK by ETC-1002 is particularly intriguing. We show that the ETC-1002 free acid activates AMPK in HepG2 cells independently of changes in the AEC and the calcium-sensing calcium CaMKK ␤ pathway. Furthermore, compound C, an ATP analog, failed to inhibit activation of AMPK by ETC-1002, consistent with a mechanism independent of AEC. Long-chain fatty acids (LCFA) and their modifi ed analogs have been shown to regulate AMPK activity through direct and indirect mechanisms (52)(53)(54)(55)(56)(57)(58)(59)(60)(61)(62)(63)(64). The mechanism most consistent with the activities described for ETC-1002 is that shown by Watt et al., which demonstrates that natural LCFA directly interacts with AMPK ␣ ␤ ␥ and enhances LKB1-dependent AMPK activation in L6 myotubes independently of AEC ( 57 ). This leaves open the possibility that ETC-1002 may increase AMPK activity through a similar direct mechanism. Importantly, the activation of AMPK by natural LCFAs is unlikely to yield therapeutic benefi t as they also provide substrates for energy production and the formation of proinfl ammatory-signaling lipids such as ceramides and DAG ( 65,66 ). Therefore, fatty acid mimetics and/or modifi cations to LCFAs that limit their metabolism may prove to be a viable therapeutic strategy if they retain their AMPK-activating properties.
The AMPK activator metformin is currently the fi rst-line oral therapy for hyperglycemia in individuals with T2D ( 67,68 ). However, monotherapy is capable of maintaining target glycemic control for only a short time ( 69 ), and low bioavailability and GI symptoms can limit the treatment dosage. Metformin decreases hepatic glucose production, in part, through inhibition of complex I of the mitochondrial respiratory chain ( 70,71 ), resulting in the activation of AMPK ( 72 ). The antidiabetic and antisteatotic effects of AMPK activation are primarily mediated through downregulation of the gluconeogenic genes pepck and g6pase ( 73 ) and modulation of transcription factors, such as SREBP1, ChREBP, TORC2, HNF4 ␣ , FOXO1, and PGC-1 ␣ ( 72-78 ). Furthermore, AMPK acutely inhibits fatty acid and sterol de novo synthesis through inhibitory phosphorylation of ACC ( 79,80 ) and HMGR ( 81 ), respectively. Interestingly, unlike ETC-1002, metformin does not consistently reduce LDL-C in dyslipidemic and hyperglycemic subjects, which may highlight the benefi cial effects of ACL inhibition mediated by ETC-1002-CoA.
The combination of AMPK activation by ETC-1002 and inhibition of ACL by the ETC-1002-CoA thioester would be expected to complement one another, as they affect hepatic lipid synthesis at the signal transduction and substrate-levels, respectively. This unique activity may provide benefi t over other AMPK activators (e.g., metformin), which have been shown to provide little benefi t for normalizing plasma LDL-C. Furthermore, this dual activity may, in part, account for the pleotropic nature of nonclinical therapeutic benefi ts observed with structurally related by guest, on  www.jlr.org Downloaded from .html http://www.jlr.org/content/suppl/2012/11/01/jlr.M030528.DC1 Supplemental Material can be found at: utilized for therapeutic interventions designed not only to produce statin-like effects for lowering LDL-C in dyslipidemic subjects but also to reduce other risk factors associated with CVD and MetS.