Deficiency in hepatic ATP-citrate lyase affects VLDL-triglyceride mobilization and liver fatty acid composition in mice.

ATP-citrate lyase (ACL) is a key lipogenic enzyme that converts citrate in the cytoplasm to acetyl-CoA, the initial precursor that yields malonyl-CoA for fatty acid biosynthesis. As cytosolic citrate is derived from the tricarboxylic acid cycle in the mitochondrion, ACL catalyzes a critical reaction linking cellular glucose catabolism and lipid synthesis. To investigate the metabolic action of ACL in lipid homeostasis, we specifically knocked down hepatic ACL expression by adenovirus-mediated RNA interference in mice maintained on a low-fat or high-fat diet. Hepatic ACL abrogation markedly reduced the liver abundance of both acetyl-CoA and malonyl-CoA regardless of dietary fat intake, which was paralleled with decreases in circulating levels of triglycerides and free fatty acids. Moreover, hepatic ACL knockdown resulted in diet-dependent changes in the expression of other lipogenic enzymes, accompanied by altered fatty acid compositions in the liver. Interestingly, ACL deficiency led to reduced serum VLDL-triglyceride levels but increased hepatic triglyceride content, resulting at least partially from decreased hepatic secretion of VLDL-containing apolipoprotein B-48. Together, these results demonstrate that hepatic ACL suppression exerts profound effects on triglyceride mobilization as well as fatty acid compositions in the liver, suggesting an important role for ACL in lipid metabolism.

Previously we have demonstrated that ACL in the liver, which is found to be markedly suppressed in mice fed a high-fat diet ( 13 ), acts as a critical component in mediating leptin's metabolic functions in regulating lipid and glucose metabolism ( 32 ). In leptin receptor-defi cient db/ db mice where hepatic ACL is abnormally elevated, targeted suppression of ACL in the liver can prominently correct obesity-associated fatty liver and hyperglycemia ( 32 ). To further understand the physiological contribution of hepatic ACL to lipid homeostasis, we investigated the metabolic consequences of hepatic ACL defi ciency in mice under different nutritional conditions.

EXPERIMENTAL PROCEDURES
Animal studies C57BL/6 male mice (Shanghai Laboratory Animal Co. Ltd) were housed in laboratory cages at a temperature of 23 ± 3°C and a humidity of 35 ± 5% under a 12 h dark/light cycle (lights on at 6:30 a.m.) in accredited animal facilities at the Shanghai Institutes for Biological Sciences, CAS. Mice maintained on normal chow diet were infected with the adenoviruses and then fed ad libitum a low-fat diet (LFD, 10 kcal% fat) or high-fat diet (HFD, 60 kcal% fat) (Research Diets, Inc., New Brunswick, NJ). Total body fat content was measured by nuclear magnetic resonance (NMR) using the Minispec Mq7.5 (Bruker, Germany). Oxygen consumption and motility were determined for mice fed ad libitum in the comprehensive laboratory animal monitoring system (CLAMS, Columbus Instruments) according to the manufacturers' instructions. After acclimation to the system for 6 h, O 2 and CO 2 were measured for the following 24 h (through a 12 h light/ dark cycle). Oxygen consumption was normalized to lean mass, and respiratory exchange ratio (RER) was calculated. Voluntary activity was determined by monitoring the X-axis beam breaks every 15 min. All animals were euthanized under anesthetic conditions. Livers were snap-frozen in liquid nitrogen immediately after resection and stored at Ϫ 80°C. All experimental procedures and protocols were approved by the Institutional Animal Care and Use Committee of the Institute for Nutritional Sciences, CAS.

Generation and administration of recombinant adenoviruses
Recombinant adenoviruses Ad-shACL and Ad-shLacZ, which express shRNAs directed against ACL and LacZ genes, respectively, were generated using the BLOCK-iT™ Adenoviral RNAi Expression System (Invitrogen, Carlsbad, CA) as previously described ( 32 ). The two adenoviruses used for ACL knockdown contain the following target sequences: 5 ′ -GCTGAATACCGAG-GACATTAA-3 ′ for Ad-shACL1# and 5 ′ -GCATTAAGCCTGGAT-GCTTTA-3 ′ for Ad-shACL4#, respectively. High-titer stocks of amplifi ed recombinant adenoviruses were purifi ed, and viral titers were determined by the tissue culture infectious dose 50 (TCID50) method. Viruses were diluted in PBS and administration was performed through tail-vein injection at approximately 5 × 10 8 pfu/mouse.

Blood and liver measurements
Hepatic triglyceride levels were measured by using 40-50 mg of liver tissue homogenized in 1.5 ml of a mixture of CHCl 3 -CH 3 OH (2:1, v/v), followed by shaking at room temperature for 2 h. After addition of 0.5 ml of 0.1 M NaCl, the suspension was centrifuged at 3,700 rpm for 10 min, and the lower organic phase Lipogenesis is a metabolic pathway that is coordinately regulated in responding to nutritional and hormonal stimuli. Dysregulated lipogenesis has been shown to contribute signifi cantly to the occurrence of dyslipidemia and the progression of metabolic disorders (7)(8)(9). Although gene expression profi ling studies in animal models have revealed obesity-associated dysregulation of lipogenic gene expression (10)(11)(12)(13), the metabolic contribution in the liver of each individual lipogenic enzyme to lipid metabolism as well as whole-body energy homeostasis has not been fully addressed until recently ( 11,(14)(15)(16)(17)(18)(19)(20). For instance, it has been reported that liver-specifi c ACC1 knockout mice (LACC1KO), while showing decreased liver TG contents with unchanged serum TG levels under fed states, exhibit no alterations in glucose homeostasis but much lower accumulation of lipids in the liver when fed a fat-free diet for 28 days ( 21 ). Unlike LACC1KO mice, however, mice with ablation of FAS in the liver (FASKOL mice) fed a zero-fat diet for 28 days develop fatty liver and hypoglycemia with decreased ketone bodies ( 22 ). Whereas hepatic FAS ablation dramatically increases malonyl-CoA accumulation, ACC1 defi ciency in the liver results in decreased production of malonyl-CoA, which is known as a crucial physiological inhibitor of lipid ␤ -oxidation ( 21 ). In addition, liver-specifi c knockout of SCD-1 (LSCD1KO mice), which catalyzes the production of various unsaturated fatty acids, leads to protection of mice from adiposity and hepatic steatosis induced by a high-sucrose, very low-fat diet, but not by a high-fat diet ( 19 ). These studies all point to distinct metabolic consequences resulting from defi ciency in each of the lipogenic enzymes, especially under different nutritional conditions. As a key lipogenic enzyme, ACL is most abundantly expressed in the liver and white adipose tissues, while exhibiting low expression levels in the brain, heart, small intestine, and muscles ( 23 ). Global ACL ablation in mice has been shown to result in embryonic lethality, indicative of its essential role in embryonic development ( 24 ). With respect to ACL's cellular functions, several groups have recently demonstrated that knockdown of ACL expression can suppress the proliferation and survival of several tumor cell lines with high glycolytic activities, suggesting that ACL, as an essential molecule in utilizing carbohydrates to provide carbon sources for cellular lipid synthesis required for tumor cell growth, represents a promising target for developing antitumor therapeutics (25)(26)(27). Animal studies have also shown that pharmacological inhibition of ACL at the whole-body level results in hypolipidemia and leads to prevention of high-fat diet-induced weight gain and hyperlipidemia (28)(29)(30). Similar to other lipogenic enzymes, ACL is also nutritionally regulated in response to changes in body's energy status. It has been reported that hepatic ACL in rodents displays relatively low expression levels in the state of starvation, whereas it is dramatically upregulated upon refeeding, with most prominent induction seen when fed a high-carbohydrate, low-fat diet ( 31 ). Currently, it remains largely unclear to what extent hepatic ACL affects lipid metabolism in response to changes in body's nutritional states.

Quantifi cation of acetyl-CoA and malonyl-CoA
Homogenized liver tissues (20-25 mg) were added to the extraction solution (5% 5-sulfosalicylic acid containing 50 M 1,4-dithioerythritol) at a ratio of 1:10 (w/v). After centrifugation at 14,000 g for 15 min at 4°C, the tissue homogenates were subjected to analysis by HPLC (Agilent 1200) and tandem mass spectrometry using negative-ion mode electrospray ionization with a 4000 Q-Trap system (Applied Biosystems, Foster City, CA) as previously described ( 32
Total FAMEs were analyzed by gas chromatography-mass spectrometry (GC-MS) with a fl ame ionization detector (Agilent 5975B Inert XL MSD with 6890N GC; SP-2560 capillary column: 100 m × 0.25 mm I.D. × 0.20 m fi lm, Supelco, Inc., PA). Quantitative profi les were calculated using methyl-17:0 as the internal standard and an equal weight FAME mixture 68A (Nuchek Prep, Elysian, MN) as the response factor for each FAME. The amount of each fatty acid was calculated as relative to the wet weight of liver.

Hepatic triglyceride secretion rate and lipoprotein fractionation
Hepatic triglyceride secretion rates were determined by measuring the increases in serum triglycerides after inhibition of lipoprotein lipase via tail-vein injection of tyloxapol (Sigma-Aldrich, St. Louis, MO) at 600 mg/kg, a dose that was supposed to completely inhibit triglyceride clearance during VLDL secretion experiments ( 34 ). Serum samples were taken from the tail vein every h for triglyceride analysis.
For lipoprotein fractionation analysis, equal volumes of serum samples were pooled from mice for each group in the fed states (a total volume of 400 l). Lipoproteins were fractionated using a Superose 6 10/300 GL FPLC column (Amersham Biosciences, Piscataway, NJ). Fractions (500 l) were collected and used for triglyceride and cholesterol analysis. For detection of apoB100 and apoB48, aliquots of the FPLC elution fractions were mixed with protein loading buffer and heated at 95°C for 6 min before Western immunoblot analysis.

Analysis of apoB mRNA editing
For the measurement of C to U editing effi ciency of apoB mRNA in the liver, a gel-purifi ed 537-bp RT-PCR product harboring the editing site was subjected to direct sequencing. The primers used for PCR amplifi cation of the editing region of apoB mRNA were: 5 ′ -GC-CCTGAGCAGACTTCCT-3 ′ and 5 ′ -AATA GCGTCCATCTGTCG-3 ′ . Editing effi ciency was determined based on the T/C nucleotide signal ratio from the sequencing eletropherograms.

Histology
Liver tissue specimens were fi xed in 10% neutral buffered formalin, and then paraffi n-embedded sections were subjected to standard hematoxylin-eosin (H and E) staining. was transferred and air-dried in a chemical hood overnight. The residual liquid was resuspended in 400 µl of 1% Triton X-100 in absolute ethanol, and the concentrations of triglyceride and cholesterol were analyzed using the commercial kits for blood measurements. Blood glucose was measured by a glucometer (FreeStyle, Alameda, CA). Serum and liver levels of triglycerides, FFA, ␤ -hydroxybutyrate, and alanine transaminase (ALT) were determined using the Serum Triglyceride Determination Kit (Sigma, St. Louis, MO), FFA Half-micro Test (Roche Applied Science, Penzberg, Germany), ␤ -Hydroxybutyrate LiquiColor (Stanbio, Boerne, TX) and Alanine Transaminase Determination Kit (ShenSuoYouFu, Shanghai, China), respectively.

Determination of ACL activity
ACL enzyme activity was determined using the malate dehydrogenase (MDH)-coupled method as described ( 32 ). Briefl y, liver extracts were incubated in the reaction buffer containing 20 mM citrate, 10 mM MgCl 2 , 10 mM DTT, 0.5 U/ml malic dehydrogenase, 0.33 mM CoASH, 0.14 mM NADH, and 100 mM Tris (pH 8.7), with or without 5 mM ATP. The yield of oxaloacetate generated by ACL catalysis was measured as the change in absorbance at 340 nm resulting from the consumption of NADH by the MDH-catalyzed reaction at 37°C. After subtracting the background changes, relative ACL activities were calculated by normalization to the total protein abundance of the extracts.

Antibodies and Western immunoblot analysis
ACL and ACC antibodies were from Cell Signaling, Boston, MA; Monoclonal FAS antibody was purchased from BD Biosciences, San Jose, CA; GAPDH antibody was from KangChen, Shanghai, China; apoB antibody was from RayBiotech, Norcross, GA. For Western immunoblot analysis, tissue extracts were prepared by lysis with CelLytic™ MT (Sigma, St. Louis, MO) and centrifuged for 20 min at 20,000 g to remove the debris. Proteins (20-40 g) from liver extracts or fast-protein liquid chromatography (FPLC) fractionation samples were separated by SDS-PAGE and transferred to PVDF fi lter membrane (Amersham Biosciences, Piscataway, NJ), which was subsequently subjected to immunoblotting with the desired antibodies.

Diet-dependent effects of hepatic ACL knockdown on the expression of ACC and FAS and the production of lipogenic precursors
We next sought to determine the metabolic effects of targeted suppression of ACL on lipid metabolism under different nutritional states. Male C57BL/6 mice at 8 weeks of age were infected and maintained on an LFD versus HFD (60 kcal% fat). At 25 days post infection, Ad-shACL1# (indicated as Ad-shACL throughout the subsequent experiments) did not cause signifi cant increases in functional liver damage or infl ammation relative to Ad-shLacZ control virus, as evaluated by the serum levels of ALT ( Table 1 ). Ad-shACL effectively diminished ACL protein expression as well as ACL enzyme activity (by >85% for LFD-fed mice and by >88% for HFD-fed mice) in the livers of both LFD-and HFD-fed mice ( Fig. 1A , B ). Curiously, in LFD-fed mice, hepatic ACL knockdown led to marked increases in the protein expression levels of ACC and FAS. By contrast and consistent with previously reported observations ( 13 ), ACL, ACC, and FAS were prominently suppressed in HFD-fed mice infected with Ad-shLacZ control virus. Further knockdown of ACL in HFD-fed mice led to signifi cantly decreased FAS expression but unaltered expression of ACC ( Fig. 1A ). Therefore, hepatic ACL abrogation resulted in distinct dietdependent alterations in the expression of other downstream lipogenic enzymes.
To examine whether ACL defi ciency exerts diet-dependent effects on the production of precursor molecules for fatty acid synthesis, we measured the hepatic contents of both acetyl-CoA and malonyl-CoA by LC-MS-MS. As a result of HFD-induced suppression of hepatic ACL expression and activity (>70% decrease in ACL enzyme activities; Fig.  1B ), a signifi cant reduction in acetyl-CoA was observed in Ad-shLacZ-infected HFD-fed mice ( Fig. 1C ). Interestingly, in comparison with that of Ad-shLacZ-infected control mice, hepatic ACL knockdown resulted in ‫ف‬ 25% reduction of the hepatic acetyl-CoA levels in Ad-shACL-infected mice fed LFD or HFD ( Fig. 1C ), although >85% decrease in ACL enzyme activities were observed ( Fig. 1B ). This indicates the

Statistical analysis
Data are presented as means ± SEM. Differences were analyzed by unpaired two-tailed t -test between two groups or otherwise by one-way ANOVA.

Hepatic ACL suppression results in hypotriglyceridemia with elevated liver triglyceride contents
Previously we found that targeted suppression of dysregulated ACL in the liver of db/db mice dramatically reduced lipogenesis and led to markedly decreased levels of hepatic but not serum triglycerides ( 32 ). To further investigate the physiological importance of hepatic ACL in lipid homeostasis, we took the RNAi approach using two recombinant adenoviruses (Ad-shACL1# and 4#) that expressed shRNAs directed against two different coding regions of ACL. First we tested the effects of ACL knockdown in C57BL/6 mice maintained on a low-fat diet (10 kcal% fat). At 23 days post infection, both Ad-shACL1# and 4# effi ciently knocked down the expression of ACL protein (supplementary Fig. I-A ), resulting in a ‫ف‬ 77% reduction in the hepatic ACL enzyme activity compared with mice infected with the control virus Ad-shLacZ (supplementary Fig. I-B ). In contrast to our observations in db/db mice ( 32 ), hepatic ACL suppression by both viruses led to marked reductions in the serum levels of TG  dyslipidemia. Compared with Ad-shLacZ-infected mice, comparable increases were observed in both body weight gain and fat mass accumulation in Ad-shACL-infected mice fed HFD for 25 days ( Table 1 ), and no appreciable changes were found in their metabolic rate, respiratory exchange ratio (RER), or physical activities (supplementary Fig. II ). Moreover, serum levels of glucose were not significantly altered in the fasted state. Therefore, in the settings of the moderate adiposity induced by short-term HFD feeding (25 days), ACL abrogation did not cause significant changes in diet-dependent fat mass accumulation or glucose metabolism. Interestingly, ACL knockdown caused signifi cantly elevated ketone bodies ( ␤ -hydroxybutyrate) in the fasted state after HFD feeding ( Table 1 ), possibly refl ecting increased fatty acid ␤ -oxidation caused by reduced malonyl-CoA. On the other hand, compared with Ad-shLacZ-infected mice, hepatic ACL knockdown resulted in similar decreases in serum TG levels (by ‫ف‬ 42%; likely involvement of other compensatory sources of hepatic acetyl-CoA production as a result of ACL defi ciency. In addition, despite of distinct ACC expression patterns, ‫ف‬ 50% and ‫ف‬ 28% decreases in hepatic malonyl-CoA levels, respectively, were detected in LFD-and HFD-fed mice infected with Ad-shACL ( Fig. 1D ), indicating the likely presence of additional factors that infl uence ACC activities (e.g., ACC phosphorylation) under these conditions. These results suggest that hepatic ACL defi ciency can affect the lipogenic program in a diet-dependent manner, and the resulting alterations in the production of lipogenic precursors as well as in downstream lipogenic enzymes may lead to changes in fatty acid compositions in the liver.

Hepatic ACL abrogation affects circulating TG levels and hepatic TG storage regardless of diet
We then examined the metabolic effects of hepatic ACL knockdown in the context of HFD-induced obesity and Fig. 1. Effects of hepatic ACL knockdown on the expression of lipogenic enzymes and the production of lipogenic precursors in mice fed LFD versus HFD. Male C57BL/6J mice (n = 7-8/group) at 8 weeks of age were infected through tail-vein injection with the control adenovirus Ad-shLacZ or the targeting adenovirus Ad-shACL1# (Ad-shACL) and then maintained on LFD or HFD for 25 days. Animals were euthanized under the fed state. A: Western immunoblots showing ACL, ACC, and FAS protein levels from the liver extracts. Each lane represents pooled samples from 2-3 mice. The bar graph indicates the relative protein abundance as quantifi ed by densitometry from the immunoblots after normalization to GAPDH. B: ACL enzyme activities were measured from the liver extracts. C and D: Determination of acetyl-CoA and malonyl-CoA levels from the liver extracts by LC-MS/MS using the ABI 4000 Q-Trap system. Shown are representative analyses by tandem mass spectrometry of (C) acetyl-CoA (ion pair at 808.3/408.2) and (D) malonyl-CoA (at 852.3/408.2) from each individual mouse infected with Ad-shLacZ versus Ad-shACL, with n-propionyl-CoA (at 822.4/408.3) used as an internal reference. The bar graphs indicate the average levels calculated according to their standard curves (n = 7-8/group). Data are shown as means ± SEM. * P < 0.05 and ** P < 0.01 versus Ad-shLacZ-infected mice fed the same diet by ANOVA; ## P < 0.01 versus LFD-fed mice infected with the same adenovirus by ANOVA. ACC, acetyl-coenzyme A carboxylase; ACL, ATP-citrate lyase; HFD, high-fat diet; LFD, low-fat diet.

Knockdown of hepatic ACL reduces the circulating levels of VLDL-containing apolipoprotein B48
Fatty acids are transported from the liver to adipose tissues and other sites in the form of VLDL in circulation, and the major lipid constituents of VLDL are triglycerides ( 6 ). To determine whether decreased serum TG levels in ACL knockdown mice refl ect reduced amounts of VLDL particles, we measured the TG contents from fractionated serum lipoproteins. In contrast to the unaltered levels of cholesterol associated with HDL, marked reductions in the TG contents (by ‫ف‬ 47% in LFD-fed mice and ‫ف‬ 57% in HFD-fed mice) were detected from the serum VLDL fractions from Ad-shACL-infected mice compared with Ad-sh-LacZ-infected control mice, either fed LFD or HFD ( Fig.  3A , B ). Moreover, decreased levels of TG associated with intermediate-density lipoprotein or low-density lipoprotein (IDL/LDL) were also observed (by ‫ف‬ 38% and ‫ف‬ 16% in LFD-and HFD-fed mice, respectively), presumably stemming from reductions in VLDL-TG ( Fig. 3A ). These data further indicate the presence of defective production of VLDL as a result of hepatic ACL defi ciency.
As the major structural component of VLDL, two forms of apolipoprotein B, apoB100 and apoB48, exist as the products of a single gene with the same N-terminal region, arising from a C to U RNA editing event that introduces a translational stop codon in the apoB RNA transcript ( 35,36 ). To test whether ACL knockdown-induced serum VLDL-TG reductions were associated with alterations of a particular apoB protein component, we analyzed the FPLC-fractionated VLDL lipoprotein profi les from the serum of LFD-fed mice infected with Ad-shLacZ or Ad-shACL. Surprisingly, as shown by Western immunoblot analysis, marked decreases in the amount of apoB48 but not apoB100 were detected in Ad-shACL-infected mice compared with control mice ( Fig. 3C ), with obviously reduced apoB48/aopB100 ratios (by >50%) observed in major VLDL fractions. These data indicate that hepatic ACL defi ciency caused a selective decrease in the abundance of apoB48-containing VLDL in the circulation.

Hepatic ACL defi ciency reduces hepatic secretion of VLDL-containing apoB48
Given the fact that changes in the circulating VLDL-TG levels can refl ect its altered clearance from circulation or the impact of its hepatic secretion, we next analyzed the rate of hepatic lipid export in LFD-fed mice treated with tyloxapol, an inhibitor of lipoprotein lipase which could block VLDL hydrolysis by the peripheral tissues, thereby permitting the measurement of hepatic VLDL-TG production. Indeed, ACL abrogation led to a 23% decrease in the rate of hepatic VLDL-TG production in Ad-shACL-infected mice fed LFD relative to Ad-shLacZ-treated control mice ( Fig. 4A ), suggesting that ACL defi ciency leads to decreased TG output from the liver.
We next performed FPLC fractionation of lipoproteins in the serum from mice similarly treated with tyloxapol for 4 h. In parallel with marked reductions in VLDL-TG, apparently decreased abundance of apoB48 protein was detected in the VLDL fractions in Ad-shACL-infected mice Fig. 2A and Table 1 ) and FFA levels (by 25-30%; Fig. 2B and Table 1 ) in Ad-shACL-infected mice fed LFD or HFD, completely correcting HFD-induced hypertriglyceridemia. However, whether maintained on LFD or HFD, hepatic ACL knockdown resulted in signifi cantly increased TG levels in the liver (by ‫ف‬ 45% in LFD-fed mice and by ‫ف‬ 29% in HFD-fed mice; Fig. 2C ), even though Ad-shLacZ-infected mice fed HFD already had a ‫ف‬ 50% elevation in hepatic TG content relative to LFD-fed mice. Histological examination of liver sections revealed that in HFD-fed mice, hepatic ACL defi ciency resulted in obvious appearance of lipid droplets ( Fig. 2D ). Thus, ACL knockdown in the liver exerted a prominent hypolipidemic effect independent of dietary fat intake that might result from decreased hepatic output or increased uptake of TG by the liver. (supplementary. Fig. III-B ), which would presumably result in decreased hepatic translation production of apoB48 protein. Thus, these results further support that hepatic ACL knockdown caused decreased export of apoB48containing VLDL from the liver, indicating a possible connection between ACL defi ciency and the assembly or maturation of selective lipoprotein particles.

Impact of hepatic ACL defi ciency upon lipogenic gene expression and fatty acid composition
To further determine the extent of the effects of hepatic ACL abrogation on lipid metabolism in the liver, we examined the expression profi les of genes related to triglyceride production as well as liver fatty acid composition. Given glucose as the major carbon source for acetyl-CoA production, the majority of cytosolic acetyl-CoA is thought to emanate from citrate by ACL-catalyzed reaction. In mammals, however, there are two acetyl-CoA synthetases (AceCS), AceCS1 and AceCS2, which are localized, respectively, in the cytoplasm and mitochondria to catalyze the formation of acetyl-CoA from acetate, thus offering alternative sources of acetyl-CoA. Consistent with the reduced ( Fig. 4B ), while discernable increases were observed for the apoB100 protein whose abundance was much lower than that of apoB48 upon tyloxapol treatment. Similarly, prominently reduced apoB48:apoB100 ratios were seen from major VLDL fractions in ACL knockdown mice. Consistent with a selective decrease in the output of apoB48containing VLDL, elevated apoB48 protein abundance was found in the liver of Ad-shACL-infected mice compared with that of Ad-shLacZ-infected control mice ( Fig.  4C ). To determine whether this increase in hepatic apoB48 protein level resulted primarily from reduced secretion or from changes in apoB RNA editing, we measured the efficiency of C to U editing by directly sequencing the RT-PCR products derived from apoB mRNA that harbor the editing site. In the liver of ACL knockdown mice, no signifi cant alterations (supplementary Fig. III-A ) were detected in the abundance of mRNAs encoding either apoB or apobec-1, the catalytic subunit within the multicomponent enzyme complex responsible for apoB RNA editing ( 36,37 ). Curiously, a signifi cant reduction in apoB RNA editing effi ciency (by ‫ف‬ 20% when expressed as the T/C ratio) was observed in the liver of Ad-shACL-infected mice Fig. 3. Hepatic ACL defi ciency results in reduced circulating VLDL containing apoB48. Lipoprotein profi les were determined using pooled serum samples (n = 5) by FPLC fractionation from adenovirus-infected mice fed LFD or HFD for 25 days (n = 7-8/group). A and B: TG and cholesterol concentrations were measured in each indicated fraction corresponding to VLDL, IDL/LDL, and HDL, respectively. The relative TG content in VLDL fractions and cholesterol content in HDL are also shown using the areas under curve from the FPLC profi les. C: Mice likewise infected were fed LFD for 28 days (n = 8). Serum samples were pooled (n = 5) and fractionated by FPLC, and the TG concentration in each fraction was measured (left). Equal volumes of the indicated VLDL fractions were subjected to Western immunoblot analysis using the apoB antibody. The bar graph indicates the relative apoB48/apoB100 ratio as quantifi ed by densitometry from the immunoblots. ACL, ATP-citrate lyase; Apo, apolipoprotein; FPLC, fast-protein liquid chromatography; HFD, high-fat diet; LFD, low-fat diet; TG, triglyceride. and ELOVL6 was signifi cantly suppressed, and the most marked downregulation was observed for SCD1. Notably, ACL knockdown led to a ‫ف‬ 38% increase in the mRNA expression of SCD1 in LFD-fed mice but a ‫ف‬ 70% decrease in HFD-fed mice ( Fig. 5A ), suggesting that the SCD1 expression is under complex control in response to nutritional changes.
Given the observed impact of ACL defi ciency on the expression patterns of genes related to lipid metabolism, we attempted to determine how ACL knockdown might alter the entire fatty acid composition in total liver lipids upon different dietary fat intake using GC-MS ( Fig. 5B ). This fatty acid profi ling analysis revealed that HFD feeding did not elicit dramatic alterations in the fatty acids in high abundances (e.g., the saturated fatty acids palmitate and stearate) and caused merely slight increases in the fatty acid 18:2n6. However, in both Ad-shLacZ-and protein levels observed ( Fig. 1 ) in LFD-and HFD-fed mice infected by Ad-shACL, quantitative RT-PCR analysis showed markedly decreased ACL mRNA levels in the liver ( Fig. 5A ). Interestingly, while AceCS2, but not AceCS1, was downregulated by HFD feeding, hepatic ACL knockdown resulted in signifi cant increases of the mRNA expression of AceCS2, but not AceCS1, in the liver of Ad-shACL-infected mice fed LFD and HFD (by ‫ف‬ 73% and ‫ف‬ 29%, respectively; Fig. 5A ). This suggests that the mitochondrion-localized AceCS2 may be transcriptionally responsive to the cytoplasmic event of ACL defi ciency. Somewhat surprisingly, the mRNA levels of FAS or ACC were not signifi cantly altered in the liver of ACL knockdown mice, which contrasted with the changes observed in their protein levels ( Fig. 1A ), indicating that ACL deficiency did not infl uence their expression at the transcriptional level. Upon HFD feeding, the expression of FAS Fig. 4. Hepatic ACL defi ciency suppresses the secretion of apoB48-containing VLDL. A: Male C57BL/6J mice at 8 weeks of age were infected with Ad-shLacZ or Ad-shACL and maintained on LFD for 18 days (n = 8/group). After a 4 h fast, tyloxapol was administered through tail-vein injection at 600 mg/kg to block the LPL activity. Blood triglycerides were measured at the time points as indicated, and the VLDL secretion rate was derived from the slope of the line using least squares regression. Data are shown as means ± SEM. * P < 0.05 and ** P < 0.01 versus Ad-shLacZ-infected mice by t-Test. B: Mice likewise infected were treated with tyloxapol for 4 h, and serum samples were collected and pooled for FPLC fractionation analysis (n = 4). The TG concentration was determined from each of the indicated fractions, and equal volumes of the VLDL fractions were subjected to Western immunoblot analysis using the apoB antibody. The bar graph indicates the relative apoB48:apoB100 ratio after quantifi cation by densitometry from the immunoblots. C: Western immunoblot analysis of the liver ACL and apoB48 protein abundance from the mice treated for 4 h with tyloxapol as in (B). Each lane represents analyzed sample from an individual mouse. The bar graphs indicate the relative abundance of ACL and apoB48 protein as quantifi ed by densitometry from the immunoblots after normalization to GAPDH (n = 4/group). Data are shown as means ± SEM. ** P < 0.01 versus Ad-shLacZinfected mice. ACL, ATP-citrate lyase; Apo, apolipoprotein; FPLC, fast-protein liquid chromatography; HFD, high-fat diet; LFD, low-fat diet; TG, triglyceride.
We found that abrogation of hepatic ACL resulted in hypotriglyceridemia, even in the face of a high-fat diet challenge, with alterations in several aspects of lipid metabolism in the liver that ranged from fatty acid composition to triglyceride mobilization and VLDL secretion. These results reveal the metabolic importance of hepatic ACL in lipid homeostasis.
Previously we have shown that in db/db mice, a genetic obesity mouse model with hepatic steatosis and dyslipidemia, ACL expression is dramatically elevated in the liver; knockdown of the abnormally upregulated ACL leads to subsequent suppression of the key lipogenic regulator PPAR ␥ as well as the entire lipogenic program, preventing the development of fatty liver but not hypertriglyceridemia ( 32 ). In contrast, here we observed different metabolic consequences resulting from hepatic ACL abrogation in normal mice, most notably hypotriglyceridemia along with increased hepatic TG content. Given the profound metabolic derangements caused by leptin receptor defi ciency in db/db mice, it is conceivable that in the presence versus Ad-shACL-infected mice fed HFD, signifi cant reductions were detected in the amount of monounsaturated fatty acids (MUFA) 16:1n7-palmitoleate and 18:1n7-oleate ( Fig. 5B ), possibly as a result of HFD-induced SCD1 suppression ( Fig. 5A ). Moreover, among other changes in low-abundance fatty acids, only in LFD-fed mice did hepatic ACL knockdown lead to signifi cantly increased contents of MUFAs, most notably C16:1n7-palmitoleate (by 67%), C16:1n9-palmitoleate (by 70%), C18:1n7-oleate (by 32%) and C18:1n9-oleate (by 30%). Together, these results indicate that hepatic ACL defi ciency can bring about profound effects on the fatty acid composition, particularly affecting the production of MUFAs in the liver.

DISCUSSION
This study aimed to explore the physiological actions of hepatic ACL in lipid homeostasis in response to nutritional changes using mice fed a low-fat diet versus a high-fat diet.  5. Impact of ACL knockdown on the expression of genes related to TG metabolism and total hepatic fatty acid composition. Mice at 8 weeks of age were infected with Ad-shLacZ or Ad-shACL and fed LFD or HFD for 25 days. Liver RNAs or total fatty acids were prepared from mice under the fed state. A: The mRNA expression level was determined for each indicated gene involved in acetyl-CoA generation and TG synthesis by real-time quantitative RT-PCR and expressed as relative to that of the Ad-shLacZ-infected control group fed LFD (set as 1.0) with GAPDH used as the internal control. Data are shown as means ± SEM (n = 7-8/ group). B: Profi ling analysis by GC-MS of liver fatty acid species ranging from 16 to 22 carbons. The relative abundance of each indicated fatty acid is expressed as the percentage of the total fatty acids determined for each sample. Data are shown as means ± SEM (n = 5-6/group). * P < 0.05 and ** P < 0.01 versus Ad-shLacZinfected mice fed the same diet by ANOVA; # P < 0.05, ## P < 0.01 versus LFD-fed mice infected with the same adenovirus by ANOVA. ACL, ATP-citrate lyase; GC-MS, gas-chromatography mass-spectroscopy; HFD, highfat diet; LFD, low-fat diet; TG, triglyceride.
inhibition of the hepatic SCD1 activity could mediate the central glucose's action in downregulating triglyceride synthesis and VLDL secretion from the liver ( 34,39 ). In addition, we observed that ACL knockdown caused marked reductions in the hepatic output of apoB48containing VLDL molecules, accompanied by increased abundance of apoB48 protein in the liver. Interestingly, without infl uencing the levels of mRNA transcripts encoding apoB or apobec-1, ACL knockdown led to considerable reduction in apoB RNA editing. Given the existence of multiple regulatory factors involved in the assembly of a multiprotein editosome responsible for apoB RNA editing ( 40 ), how ACL defi ciency infl uenced this editing process remains currently unknown. It is worth noting that a recent study demonstrated a crucial role of ACL in regulation of histone acetylation through controlling the nuclear levels of acetyl-CoA ( 41 ), linking cellular ACL activities to the metabolic control of histone acetylation and, consequently, gene expression programs. In this regard, ACL defi ciency may exert a po tential impact on the extent of protein acetylation modifi cations and thus the activation status of key metabolic regulators such as PGC1 ␣ and LXR (42)(43)(44). Therefore, it is not surprising that hepatic ACL suppression could have various effects on a range of lipid metabolism-related events from lipogenic gene expression, to apoB RNA editing, to VLDL secretion. While the detailed molecular nature of these ACL-dependent changes has yet to be further elucidated, our fi ndings provide insights for fully understanding the metabolic importance of hepatic ACL in the face of changing nutritional conditions, especially in the maintenance of lipid homeostasis.
The authors thank Jun Shen for assistance with CLAMS.
absence of the functional leptin receptor, ACL abrogation could exert distinct effects on lipid metabolism through affecting different sets of crucial regulatory molecules, as exemplifi ed by hepatic PPAR ␥ , whose expression is prominently increased in db/db mice ( 32 ).
The lipogenic pathway is a highly regulated process that plays an important part in coordinating body's partitioning of energy fuels. Perturbation at different steps of lipogenic pathway brings about distinct metabolic phenotypes under different conditions of dietary fat intake, as demonstrated by the studies of mouse models with liver-specifi c knockout of ACC1 ( 21 ), FAS ( 22 ), or SCD1 ( 19 ). The results from the current study showed that ACL defi ciency could infl uence the expression of downstream lipogenic genes in a diet-dependent fashion (e.g., triggering increased expression levels of ACC and FAS proteins as well as SCD1 mRNA when fed LFD). It is notable that, whether maintained on LFD or HFD, liver ACL defi ciency resulted in lower amount of malonyl-CoA and increased TG storage in the liver, which is different from the metabolic phenotypes of hepatic ACC1 knockout mice exhibiting reduced malonyl-CoA levels but dramatically decreased accumulation of lipids in the liver when fed a fat-free diet. Of additional note, in the face of HFD feeding that led to suppressed expression of lipogenic enzymes including FAS, ACC, and SCD1 in the liver, ACL defi ciency failed to trigger the induction of these lipogenic genes as seen when fed LFD. Whereas it is currently unclear what dietdependent mechanisms are involved in mediating the distinct effects of ACL defi ciency on the expression levels of these lipogenic enzymes, this may at least partially account for the observation that ACL knockdown caused extensive diet-responsive changes in the total fatty acid composition in the liver, particularly the levels of MUFAs, such as C16:1n7-palmitoleate, C16:1n9-palmitoleate, C18:1n7-oleate, and C18:1n9-oleate. The alterations in the fatty acid profi les could refl ect part of the ACL deficiency-induced metabolomic changes, which could in turn infl uence not only lipid metabolism but also glucose homeostasis. For instance, it was recently reported that C16:1n7-palmitoleate could act as a so-called "lipokine" playing a key role in regulation of the lipogenic pathway and systemic glucose metabolism ( 38 ). On the other hand, we did not observe a signifi cant impact of hepatic ACL suppression on HFD-induced accumulation of body fat mass or glucose metabolism during the relatively short period (25 days) of the adenoviral knockdown experiment. While this is consistent with our earlier fi ndings ( 13 ) that the moderate adiposity levels were not associated with overt insulin resistance in mice upon short-term HFD feeding, it remains to be deciphered whether chronic defi ciency in hepatic ACL can affect glucose homeostasis under long-term HFD feeding conditions.
Our results also showed that the hypotriglyceridemic effect of hepatic ACL defi ciency might largely arise from decreased VLDL secretion, in parallel with increased TG content in the liver. This provides another example of the lipogenic enzyme in affecting hepatic VLDL secretion and triglyceride output, as it was documented that