Thioesterase superfamily member 2/acyl-CoA thioesterase 13 (Them2/Acot13) regulates hepatic lipid and glucose metabolism

Abstract

Members of the acyl-CoA thioesterase (Acot) gene family catalyze the hydrolysis of fatty acyl-CoAs, but their biological functions remain unknown. Thioesterase superfamily member 2 (Them2; synonym Acot13) is a broadly expressed mitochondria-associated Acot. Them2 was previously identified as an interacting protein of phosphatidylcholine transfer protein (PC-TP). Pctp^−/− mice exhibit altered fatty acid metabolism that is accompanied by reduced hepatic glucose production. To examine the role of Them2 in regulating hepatic lipid and glucose homeostasis, we generated Them2^−/− mice. In livers of Them2^−/− mice compared with Them2^+/+ controls, a 1.9-fold increase in the K_m of mitochondrial thioesterase activity was accompanied by a 28% increase in fatty acyl-CoA concentration. A reciprocal 23% decrease in free fatty acid concentration was associated with reduced activation of peroxisome proliferator-activated receptor α. However, fatty acid oxidation rates were preserved in livers of Them2^−/− mice, suggesting that Them2 functions to limit β-oxidation. Hepatic glucose production was also decreased by 45% in the setting of reduced hepatocyte nuclear factor 4α (HNF4α) expression. When fed a high-fat diet, Them2^−/− mice were resistant to increases in hepatic glucose production and steatosis. These findings reveal key roles for Them2 in the regulation of hepatic metabolism, which are potentially mediated by PC-TP-Them2 interactions.—Kang, H. W., Niepel, M. W., Han, S., Kawano, Y., Cohen, D. E. Thioesterase superfamily member 2/acyl-CoA thioesterase 13 (Them2/Acot13) regulates hepatic lipid and glucose metabolism.

The cellular balance between free fatty acids (FFAs) and fatty acyl-CoAs regulates fatty acid oxidation, the biosynthesis of complex lipids, signal transduction events, and gene transcription (1). The formation of fatty acyl-CoAs is catalyzed by members of the long-chain acyl-CoA synthetase (ACSL) gene family. Among the functions of these enzymes is to promote the synthesis of complex lipids (2) and β-oxidation (3). Although the hydrolysis of fatty acyl-CoAs to form FFA and coenzyme A (CoASH) can be catalyzed by any of 13 members of the acyl-CoA thioesterase (Acot) gene family, the metabolic functions of these enzymes have yet to be elucidated.

Apparently due to convergent evolution, 2 distinct types of Acot structures lead to similar enzymatic activities (4). Type 1 Acots (Acots 1–6) contain N-terminal β-sandwich and C-terminal α/β hydrolase domains. Type 2 Acots (Acots 7–13) utilize N-terminal hotdog-fold thioesterase domains (5). With the exception of Acot13 [synonym: thioesterase superfamily member 2 (Them2)], type 2 Acots contain tandem ∼140-aa thioesterase domains. Acot13/Them2 (hereafter referred to as Them2) is a single-hotdog-fold thioesterase domain, which in solution forms an enzymatically active homotetramer with substrate specificity for long-chain fatty acyl-CoAs (6, 7).

We identified Them2 as an interacting partner for phosphatidylcholine transfer protein [PC-TP; synonym: steroidogenic acute regulatory protein-related lipid transfer domain 2 (StarD2); ref. 8], which is a highly specific intracellular lipid binding protein. Consistent with a functional interaction, the fatty acyl-CoA thioesterase activity of purified recombinant Them2 in vitro was increased by the addition of purified recombinant PC-TP (6, 8). Both proteins are enriched in liver and oxidative tissues and are up-regulated by peroxisome proliferator-activated receptor α (PPARα; refs. 6, 9), suggesting roles in nutrient metabolism. Them2 is primarily associated with mitochondria (6) but is also present in cytosol. The subcellular distribution of PC-TP is well suited for interacting with Them2: it is enriched in cytosol, with a fraction that associates with mitochondria (6, 10).

Pctp^−/− mice exhibit increased hepatic insulin sensitivity and utilize fatty acids as a preferred substrate for oxidative metabolism (11). Pctp^−/− mice are also resistant to high-fat-diet-induced increases in hepatic glucose production but not to obesity (12). To gain insights into the biological function of Them2, as well as its potential role in the metabolic regulation by PC-TP, we created Them2^−/− mice. The phenotypes of Them2^−/− mice strongly support roles for Them2 in the control of lipid and glucose homeostasis. Although these mice did not phenocopy Pctp^−/− mice, key similarities suggest that Them2-PC-TP interactions contribute to the regulation of fatty acid oxidation and glucose production by the liver.

MATERIALS AND METHODS

Generation of Them2^−/− mice

The Them2 gene, located on mouse chromosome 13, is 13.5 kb in length and contains 3 exons. The translation start codon is located in the first exon, translation stop codon in the third exon. With the use of bacterial artificial chromosome (BAC) recombineering techniques (13), a gene-targeting vector was constructed by replacing exon2 through exon3 (2974 bp) of Them2 with a Loxp-Neo-Loxp (LNL) cassette followed by retrieval of a 5-kb homology arm (5′ to LNL cassette), LNL cassette, and 2-kb short homology arm (end of LNL cassette to 3′) onto a pPTL-DTA vector (Precision Targeting Lab LLC, Tenafly, NJ, USA). Several targeted ES cells were identified by PCR and injected into C57BL/6 blastocysts to generate chimeric mice. Male chimeras were bred to C57BL/6 females to achieve germline transmission of the Them2-null allele. Heterozygous Them2-null allele mice (Them2^+/−) were then intercrossed to generate homozygous (Them2^−/−) mice. These offspring were genotyped by PCR analysis from tail genomic DNA using the following primers: forward 5′-ATCAGATCCCATTACCGATGGTTGT-3′ and reverse 5′-CTAGCTTGGCTGGACGTAAACTCCT-3′ for Them2^−/− mice; forward 5′-AAACCCATGGATAAATGAAGGATGG-3′ and reverse 5′-ACTGAGGCAAATGCAAGTGTCTTTC-3′ for wild-type mice. Mice were maintained on the mixed 129/B6 genetic background.

Animals and diets

Male mice were housed in a 12-h light-dark cycle barrier facility with free access to drinking water and were fed a standard chow diet, (PicoLab Rodent Diet 20, 5053; LabDiets, St. Louis, MO, USA). For some experiments, mice at 5 wk of age were begun on a high-fat diet (60% kcal from fat; D12492; Research Diets, New Brunswick, NJ, USA) for 13 wk. Before death, mice were unfed for 16 h unless otherwise stated. Plasma samples were collected by cardiac puncture, and tissue samples were utilized immediately or snap-frozen in liquid nitrogen and then stored at −80°C. Protocols for animal use and euthanasia were approved by the Institutional Committee of Harvard Medical School.

Body composition

Micro–computed tomography (micro-CT) scans were performed to measure body composition at the Longwood Small Animal Imaging Facility [Beth Israel Deaconess Medical Center, Boston, MA, USA; supported in part by U.S. National Institutes of Health (NIH)/National Center for Research Resources (NCRR) shared instrumentation grant S10-RR-023010]. Briefly, mice were anesthetized with isoflurane, and imaging was performed using the CT component of a NanoSpect/CT scanner (Bioscan, Washington, DC, USA) equipped with an 8-W X-ray source running at 45 kVp (177 mA) and a 48-mm-pitch CMOS-CCD X-ray detector. Continuous helical micro-CT scanning was performed with the following parameters: 1 s exposure, 240 angles, 1.3 magnification, 37 mm pitch (1 field of view), and a 512- × 256-pixel frame size (0.192-mm pixels). Images were reconstructed as 170- × 170-pixel transverse matrices with varying axial length and slice thickness of 0.1 mm (isotropic voxel size 0.1 mm) using filtered back-projection (Shepp-Logan filtering). Quantification of body fat was performed using InVivoScope software (Bioscan) employing a Hounsfield unit windowing technique.

Analytical techniques

Cholesterol and triglyceride concentrations in liver and plasma were measured enzymatically using reagent kits (Sigma-Aldrich, St. Louis, MO, USA) as described previously (14). Plasma glucose concentrations were determined using a OneTouch Ultra glucose monitor (LifeScan, Milpitas, CA, USA). Plasma and hepatic concentrations of FFA were measured using a reagent kit (Roche Applied Sciences, Indianapolis, IN, USA). Tissue concentrations of long-chain fatty acyl-CoA esters in liver were determined by ESI/MS/MS analysis as a service of the Mouse Metabolic Phenotype Center (Yale University School of Medicine, New Haven, CT, USA; supported in part by NIH grant U24 DK76169). Plasma concentrations of insulin, leptin, and adiponectin were determined by ELISA, and β-hydroxybutyrate concentrations were determined enzymatically (Joslin Diabetes and Endocrinology Research Center Specialized Assay Core, Joslin Diabetes Center, Boston, MA, USA; supported in part by NIH grant 5P30 DK36836). Freshly harvested tissues were flash-frozen or stored in Bouin's fixative solution and then processed for hematoxylin and eosin or oil red O staining by the Rodent Histophathology Core Facility (Dana Farber/Harvard Cancer Center, Boston, MA, USA; supported in part by National Cancer Institute Cancer Center Support grant NIH 5 P30 CA06516). Immunoblot analyses were performed as described by Kang et al. (15) using polyclonal antibodies to Them2 (8), as well as a monoclonal antibody to β-actin (Sigma-Aldrich).

Real-time quantitative PCR (qPCR)

Total RNA was extracted from tissues using a TRIzol reagent (Invitrogen, Carlsbad, CA, USA). cDNA was synthesized using a reagent kit according to the manufacturer's protocol (Invitrogen). qPCR measurements were performed using a Roche 480 LightCycler (Roche Applied Sciences). Most of the primer sequences have been described previously (9, 11, 16–20). Other primer sequences used in this study were as follows: Acot2 (forward 5′-CTGAAGTCAACGACGCAAAA-3′, reverse 5′-CGGCGGAGGTACAAACAG-3′), Acots 9 and 10 (forward 5′-TGCTACATGCACAACCACAA-3′, reverse 5′-AAGGTTGCATCCAAAACAGG-3′), and Acot11 (forward 5′-AGATCATGGCTTGGATGGAG-3′, reverse 5′-AAAGGCGTTATTCACGATGG-3′). Because the nucleotide sequences of mRNAs that encode Acots 9 and 10 share 95% nucleic and amino acid identity (21), we were unable to design oligonucleotide primers that discriminated between the two. Therefore, these genes were quantified together and referred to as Acot9/10. Ribosomal protein L 32 (RPL32) was used as an invariant gene (15).

Fatty acyl-CoA thioesterase activity

Fatty acyl-CoA thioesterase activity was measured in liver homogenates and in isolated mitochondria using myristoyl (C14:0)-CoA as an exogenous substrate. Briefly, liver samples were homogenized in 20 mM Tris (pH 8.0), 137 mM NaCl, 1 mM EDTA, 10% glycerol, and 0.5% Triton X-100; sonicated; and then centrifuged at 16,000 g for 10 min at 4°C. Myristoyl-CoA (Avanti Polar Lipids, Alabaster, AL, USA) was dissolved in H₂O. Reactions were performed in 96-well plates using final volumes of 200 μl containing 50 mM KCl, 10 mM HEPES (pH 7.5), 0.3 mM 5,5′-dithiobis(nitrobenzoic acid), myristoyl-CoA, and 50 μg protein of liver homogenates. Reactions were initiated by the addition of the liver homogenates and then monitored at 412 nm for 1 h at 37°C in a SpectraMax M5 microplate reader (Molecular Devices, Sunnyvale, CA, USA). To measure mitochondrial thioesterase activity, mitochondria were isolated (22), resuspended in lysis buffer (20 mM Tris, pH 8.0; 137 mM NaCl; 1 mM EDTA; 10% glycerol; and 0.5% Triton X-100), sonicated, and then incubated at 4°C for 30 min. After centrifugation, supernatants (50 μg protein) were used for thioesterase activities. Values of initial rate of reaction (V_o) were used to calculate values of maximum rate of reaction (V_max) and Michealis constant (K_m; substrate concentration at which V_o = V_max/2) according to the Michaelis-Menten equation, as described previously (6).

Fatty acid oxidation rates

Rates of fatty oxidation were determined in liver homogenates (3, 23). Liver tissue was homogenized in ice-cold buffer (250 mM sucrose, 1 mM EDTA, and 10 mM Tris-HCl, pH 7.4). After centrifugation at 320 g for 5 min, supernatants were resuspended in reaction buffer (62.5 mM sucrose, 12.5 mM K₂HPO₄, 100 mM KCl, 1.25 mM MgCl₂, 1.25 mM l-carnitine, 0.125 mM malic acid, 10 mM Tris-HCl, 1.25 mM DTT, 0.125 mM NAD⁺, 2.0 mM ATP, and 0.063 mM CoA, pH 7.4) containing the substrates 200 μM palmitic acid plus [1-¹⁴C] palmitic acid (specific activity 55 mCi/mmol; American Radiolabeled Chemicals, St. Louis, MO, USA) bound to BSA (0.25 μCi/reaction) and then incubated at 37°C for 30 min. After additional incubation with concentrated perchloric acid at 4°C overnight, acid soluble metabolites were collected by centrifugation at 16,000 g for 20 min and quantified by scintillation counting.

Indirect calorimetry and physical activity

Measurements were performed using an Oxymax/Comprehensive Laboratory Animal Monitoring System (Columbus Instruments, Columbus, OH, USA). Mice were housed in individual metabolic cages with free access to food and water. Following a 24 h acclimation period, rates (ml/kg/h) of O₂ consumption (Vo₂) and CO₂ production (Vco₂), food consumption, and activity were measured over a 24-h period. Respiratory exchange rate (RER) was calculated as the ratio of Vco₂ to Vo₂. Values of heat production (kcal/h) were obtained from RER values as heat = (3.815 + 1.232 × RER) × Vo₂ (24). Calorimetry data were analyzed by the percentage relative cumulative frequency (PRCF) method as described previously (11).

Fecal fat excretion and absorption

Fecal fat excretion and absorption measurements were performed as described previously (25). Briefly, mice were acclimated for 4 d in individual cages on metal mesh racks that were inserted over bedding. Feces were then collected for 2 d and dried overnight at 60°C. Dried feces were pulverized and soaked in water for 2 h. Total lipids were extracted twice in chloroform/methanol (2:1), dried under N₂, and reconstituted in 2% Triton X-100. Fecal triglyceride, cholesterol, and fatty acid contents were measured enzymatically using a reagent kit (Wako Diagnostics, Richmond, VA, USA). Fecal fat was expressed by a percentage of total lipid mass extracted from 0.2 g of feces. Percentage fat absorption was calculated as (fat consumed − fecal fat)/fat consumed × 100.

Hepatic triglyceride secretion rates

Rates of triglyceride secretion from the liver were measured as described previously (11). Briefly, mice were unfed for 4 h before measurements of baseline plasma triglyceride concentrations. Following intravenous injection of 10% w/v of the lipoprotein lipase inhibitor Triton WR1339 (Sigma-Aldrich) dissolved in PBS, plasma was collected at intervals up to 4 h. Plasma triglyceride concentrations were used to calculate hepatic triglyceride secretion rates (11).

Glucose, pyruvate, and insulin tolerance tests

Tolerance tests to glucose, pyruvate, and insulin were each performed after overnight food withdrawal. Following the measurement of baseline glucose concentrations, mice were injected intraperitoneally with 1–2 mg/g bw d-glucose (20% w/v) for glucose tolerance tests, 4 mg/g bw sodium pyruvate (40% w/v in sterile PBS) for pyruvate tolerance tests, or 1 U/kg bw insulin (HumulinR; Eli Lilly, Indianapolis, IN, USA) for insulin tolerance tests. Plasma glucose concentrations were then monitored periodically for up to 180 min. Mice were allowed to recover for 1 wk between tolerance tests. Values for area under the curve (AUC) were calculated using baseline values (26).

Transcriptional activity of PPARα and hepatocyte nuclear factor 4α (HNF4α)

Human embryonic kidney (HEK) 293T cells were maintained and transfected as described previously (9). Endogenous Them2 expression was knocked down using 20 nM siRNA (5′-CCUGGGAAACUGAGAGAAC-dTdT-3′), and a scrambled Negative siRNA Control 1 (Applied Biosystems/Ambion, Austin, TX, USA) served as the control. After 6 h of transfection, fresh medium was added for 48 h. Transcriptional activity of PPARα and HNF4α was measured in promoter-reporter assays as described previously (9). In some experiments, docosahexaenoic acid (DHA) complexed with BSA (9) were added to the media for 4 h before transcriptional activity was measured.

Statistics

Data are reported as means ± se. Differences between groups were assessed using a 2-tailed unpaired Student's t test using a Bonferroni correction for multiple testing (Prism 5; GraphPad Software Inc., La Jolla, CA, USA).

RESULTS

Them2^−/− mice developed normally but exhibited modest decreases in body weight

Them2^−/− mice were generated by deleting exons 2 and 3 (Supplemental Fig. S1A, B). These mice, which lacked detectable Them2 expression (Supplemental Fig. S1C), were fertile and without evident physical defects. A comprehensive tissue survey did not reveal any histological abnormalities. Compared with chow-fed Them2^+/+ mice, body weights of chow-fed Them2^−/− mice were reduced by up to 5% beginning at 7 wk of age (Fig. 1A). When fed the high-fat diet, body weights of Them2^−/− mice were reduced by up to 10% beginning at 8 wk of age compared with high-fat-fed Them2^+/+ mice. Whereas absolute rates of chow or high-fat diet consumption (g/d) did not differ between Them2^−/− and Them2^+/+ mice when fed either chow (Them2^+/+: 4.3±0.1; Them2^−/−: 4.3±0.3) or the high-fat diet (Them2^+/+: 14.3±1.9; Them2^−/−: 18.1±2.0), food consumption normalized to body weight (Fig. 1B) was increased in chow and tended to increase in high-fat-fed animals lacking Them2 expression due to their decreased sizes. When compared with Them2^+/+ mice, chow-fed Them2^−/− mice exhibited modestly increased lean body mass and reduced white adipose tissue (WAT; Fig. 1C). The high-fat diet increased the proportion of WAT in both genotypes and eliminated differences between genotypes. Consistent with these observations, epidydymal fat pad weights were lower in chow-fed Them2^−/− mice, but there was no difference after high-fat feeding, which increased fat pad weights to similar values (Fig. 1D).

Figure 1.

Influence of Them2 expression on body weight and composition. A) Mice were fed chow (C: Them2^+/+, n=8; Them2^−/−, n=25) or high-fat diet (H: Them2^+/+, n=11; Them2^−/−, n=14) for up to 13 wk. B) Food consumption was measured for 7 d at 13 wk for mice in A. C) Percentages of lean body (lean) and WAT mass were determined for 12-wk-old chow or high-fat-diet-fed mice by computed tomography (n=3/group). D) Epidydymal fat pad weights were measured on the death of mice (C: Them2^+/+, n=7; Them2^−/−, n=9; H: Them2^+/+, n=4; Them2^−/−, n=6). *P < 0.025 vs. Them2^+/+; ^†P < 0.025 vs. C.

Reduced activity and energy expenditure in high-fat-fed Them2^−/− mice

To determine whether reduced body weights in Them2^−/− mice might be attributable to increased energy consumption, mice were subjected to activity monitoring and indirect calorimetry in metabolic cages (Fig. 2 and Supplemental Fig. S2). For both genotypes, total and ambulatory activities were higher during the dark phase than during the light phase (Fig. 2A and Supplemental Fig. S2A, B). In Them2^+/+ mice, the high-fat diet decreased total and ambulatory activities during the dark phase but only ambulatory activity during the light phase. Total activity tended to be lower in chow- and high-fat-fed Them2^−/− mice during both dark and light phases, although these differences did not achieve statistical significance. Chow-fed Them2^−/− mice exhibited less ambulatory activity in the light but also tended to be less active in the dark.

Figure 2.

Influence of Them2 expression on energy expenditure. Mice (C: Them2^+/+, n=9; Them2^−/−, n=9; H: Them2^+/+, n=6; Them2^−/−, n=5) were subjected to comprehensive monitoring of physical activity and indirect calorimetry measurements over a 24 h period. A) Total and ambulatory activities were determined during 12-h dark and light periods for mice fed chow or high-fat diet. B) Values of Vo₂ were determined every 60 min for each mouse. Not shown here are corresponding values of CO₂, heat production, and RER, which are presented in Supplemental Fig. S2. C–F) PRCF analyses were performed for values of Vo₂ (C), Vco₂ (D), heat production (E), and RER (F). Nonlinear least square analyses of the data points are displayed, and values of EC₅₀ and R² are provided for each. Individual data points are not shown for clarity. *P < 0.025 vs. Them2^+/+; ^†P < 0.025 vs. C.

By indirect calorimetry, Vo₂ values for chow-fed mice were not influenced by Them2 expression (Fig. 2B) and were reduced in high-fat-fed Them2^−/− compared with Them2^+/+ mice. This was confirmed by PRCF analysis, which revealed similar half maximal effective concentration (EC₅₀; PRCF=50) values in chow-fed Them2^−/− and Them2^+/+ mice (Fig. 2C, top panel). Compared with chow feeding, the high-fat diet reduced EC₅₀ values for Vo₂ in both genotypes (Fig. 2C, bottom panel), but this effect was greater in Them2^−/− mice, so that the EC₅₀ was 15% lower than observed for Them2^+/+ mice. The calculation of heat production was based on concurrent measurement of Vco₂ values (Supplemental Fig. S2C and Fig. 2D), and it exhibited similar trends as observed for Vo₂ (Fig. 2B). As a result, there were no changes in heat production attributable to Them2 expression in chow-fed mice (Supplemental Fig. S2D and Fig. 2E, top panel), whereas the EC₅₀ value for high-fat-fed mice was reduced by 17% (Fig. 2E, bottom panel). Indirect calorimetry also yields values of RER, which were unchanged in chow-fed Them2^−/− compared with Them2^+/+ mice (Supplemental Fig. S2E and Fig. 2F, top panel). EC₅₀ values were reduced by the high-fat diet in both genotypes (Fig. 2F, bottom panel) but were 6% higher in Them2^−/− mice. Taken together, data for activity and indirect calorimetry did not provide an explanation for decreased body weights of Them2^−/− mice.

Reduced intestinal lipid absorption in chow-fed Them2^−/− mice

Considering that rates of food consumption were increased in Them2^−/− mice in the absence of increased activity or energy expenditure, we next examined the possibility that the modest weight reductions observed in the absence of Them2 expression might be attributable to reduced intestinal lipid absorption. Consistent with a report of Them2 mRNA expression in mouse small intestine and colon (27), Them2 protein expression was robustly expressed in all 3 segments of the small intestine as well as in the colon of Them2^+/+ but not Them2^−/− mice (Supplemental Fig. S3A). Although the absence of Them2 expression did not alter the gross appearance of feces on either diet, the total lipid content in feces from chow-fed Them2^−/− mice was increased (Supplemental Fig. S3B) and absorption decreased (Supplemental Fig. S3C). This reflected an increase in fecal contents of triglycerides (Supplemental Fig. S3D) and a trend (P=0.043) toward increased fecal FFA (Supplemental Fig. S3E), and cholesterol (Supplemental Fig. S3F). However, the same effect of Them2 expression was not apparent in high-fat-fed mice, in which there were no differences in fecal lipid contents or percentage of lipid absorption. Histological examination of the small and large intestine in chow- and high-fat-fed mice did not show any evident changes due to Them2 expression nor were any differences noted in amounts or patterns of oil red O staining. These findings suggest that reduced body weights in chow-fed Them2^−/− mice were, at least in part, attributable to modestly reduced intestinal lipid absorption.

Because an important objective of this study was to explore the influence of Them2 expression on hepatic lipid and glucose metabolism, we sought to minimize the potentially confounding influence of body weight per se. This was accomplished in subsequent experiments by choosing individual chow or high-fat-fed mice such that there were no differences in body weight between genotypes for mice fed the same diet.

Reduced mitochondrial fatty acyl-CoA thioesterase activity in livers of Them2^−/− mice

We next determined whether loss of Them2 expression influenced the fatty acyl-CoA thioesterase activity in liver tissue. Liver homogenates from Them2^−/− mice exhibited increased values of K_m for the exogenous substrate myristoyl-CoA, with no effect on V_max (Fig. 3A). Consistent with its enrichment in mitochondria, a similar effect was observed for the fatty acyl-CoA thioesterase activity associated with purified mitochondria. Them2^−/− mice exhibited a 28% increase in total hepatic acyl-CoA concentrations (Fig. 3B). This was mainly attributable to increases in C16:0-CoA, C18:0-CoA and C18:3-CoA by 79, 47, and 80%, respectively, in livers of Them2^−/− mice, with other acyl-CoAs species tending to be increased. In keeping with reduced hydrolysis of fatty acyl-CoAs, hepatic FFA concentrations were decreased by 23% in the absence of Them2 expression (Fig. 3C). To explore whether deletion of the Them2 gene resulted in compensatory changes, we measured mRNA expression levels of the other Acot genes (5). There were significant decreases in expression of Acots 1, 2, 5, 7, and 9/10 (Fig. 3D). The mRNA expression of ACSL1, which promotes fatty acyl-CoA formation within the liver (28), was unaffected by the loss of Them2 expression.

Figure 3.

Reduced fatty acyl-CoA thioesterase activity in livers of Them2^−/− mice. A) Acyl-CoA thioesterase activity was determined using homogenates and purified mitochondria from mouse livers (n=3/group). Values of K_m and V_max were determined as described in the text. Where not visible, error bars are contained within the symbol sizes. *P < 0.05 vs. Them2^+/+. B, C) Hepatic concentrations of fatty acyl-CoAs (B; n=3/group) and FFA (C; Them2^+/+, n=7; Them2^−/−, n=9) were determined for chow-fed mice. *P < 0.05 vs. Them2^+/+. D) Influence of Them2 expression on the hepatic mRNA expression of each Acot gene was determined in chow-fed mice (n=7/group). *P < 0.025 vs. Them2^+/+.

Them2^−/− mice were resistant to high-fat-diet-induced hepatic steatosis

In response to high-fat feeding, livers of Them2^+/+ mice exhibited a modest increase in Them2 mRNA expression but a similarly modest decrease in steady-state protein levels (Fig. 4A). Whereas the lack of Them2 expression did not alter liver histology, the high-fat-diet-induced hepatic steatosis was markedly attenuated in Them2^−/− mice (Fig. 4B). In keeping with these observations, the high-fat diet promoted an almost 400% increase in hepatic triglyceride concentrations of Them2^+/+ mice, but only a 60% increase was observed for Them2^−/− mice compared with their chow-fed counterparts (Fig. 4C). Whereas FFA concentrations were reduced in chow-fed Them2^−/− mice (Fig. 3C), values increased to similar levels (nmol/mg protein) in high-fat-fed mice (Them2^+/+: 14.1±1.7; Them2^−/−: 12.9±0.9). Hepatic cholesterol concentrations in liver were similarly influenced by Them2 expression and high-fat feeding, such that high-fat-fed Them2^+/+ and Them2^−/− mice exhibited 250 and 110% increases in hepatic cholesterol concentrations, respectively, compared with their chow-fed controls (Fig. 4D). As a potential explanation for reduced accumulation of hepatic triglycerides in chow and high-fat Them2^−/− mice, we measured rates of hepatic triglyceride secretion, which tended to be increased in chow-fed mice (Fig. 4E). The high-fat diet increased hepatic triglyceride secretion rates in both genotypes, and values were slightly higher in the absence of Them2 expression. In chow-fed mice, rates of fatty acid oxidation (nmol/mg/min) were the same in Them2^+/+ and Them2^−/− mice following 4 h of fasting (Them2^+/+: 2.23±0.12; Them2^−/−: 2.29±0.19). Whereas fatty acid oxidation rates in Them2^+/+ mice tended to increase following 16 h (P=0.031), this effect was attenuated in Them2^−/− mice (Fig. 4F).

Figure 4.

Influence of Them2 expression on hepatic lipid metabolism. A) Expression levels of Them2 mRNA and protein were measured in livers of mice fed chow (n=5–7/group) or high-fat diet (n=5–7/group) after 16 h of food withdrawal. Inset: representative immunoblots. B) Hematoxylin and eosin staining of liver sections from mice fed high-fat diet. C, D) Hepatic concentrations of triglycerides (C) and cholesterol (D) were measured following 16 h of food withdrawal (C: 16 h of food withdrawal (C: Them2^+/+, n=7; Them2^−/−, n=9; H: Them2^+/+, n=5; Them2^−/−, n=6). E) Hepatic triglyceride secretion rates were determined after 16 h of food withdrawal (C: Them2^+/+, n=6; Them2^−/−, n=5; H: Them2^+/+, n=4; Them2^−/−, n=4). F) Rates of fatty acid oxidation were determined in liver homogenates after 16 h of food withdrawal (C: Them2^+/+, n=4; Them2^−/−, n=3; H: Them2^+/+, n=3; Them2^−/−, n=3). G–J) Plasma concentrations of triglycerides (G), cholesterol (H), FFA (I), and β-hydroxybutyrate (J) were measured after 16 h of food withdrawal (C: Them2^+/+, n=7; Them2^−/−, n=9; H: Them2^+/+, n=5; Them2^−/−, n=6). K–M) Expression of genes that mediate synthesis (K), uptake (L), and oxidation (M) of fatty acids was measured in livers of mice after 16 h of food withdrawal (C: Them2^+/+, n=7; Them2^−/−, n=7; H: Them2^+/+, n=5; Them2^−/−, n=6). *P < 0.025 vs. Them2^+/+; ^†P < 0.025 vs. C.

The absence of Them2 did not influence plasma triglyceride concentrations (Fig. 4G), which were modestly increased by high-fat feeding in both genotypes. Plasma cholesterol concentrations were not affected by the absence of Them2 in either chow or high-fat-fed mice (Fig. 4H). The absence of Them2 expression had no effect on plasma concentrations of FFA (Fig. 4I) or β-hydroxybutyrate in chow-fed mice (Fig. 4J), but it did prevent the increases in both that occurred in Them2^+/+ mice in response to high-fat feeding.

To gain additional insights into regulation of hepatic lipid metabolism by Them2, we measured expression levels of genes that mediate fatty acid synthesis, uptake and oxidation. Them2 expression did not influence hepatic mRNA expression levels of the lipogenic transcription factor sterol regulatory element binding protein 1c (SREBP1c) or its target genes in livers of mice fed chow or the high-fat diet (Fig. 4K). Expression of the fatty acid transporter CD36 was up-regulated in high-fat-fed in Them2^+/+, but not Them2^−/− mice (Fig. 4L), and there was no major effect of Them2 expression or diet on other fatty acid transporters [i.e., FATP, fatty acid transport protein 2 or 5 (FATP2 or FATP5)]. Hepatic expression of PPARα, which promotes fatty acid oxidation, tended to be reduced in chow-fed Them2^−/− mice (P=0.038), and there were associated decreases in its target genes fatty acid binding protein 1 (FABP1) and fibroblast growth factor 21 (FGF21) (Fig. 4M). CPT1α, another PPARα target gene that plays a key role in fatty acid oxidation, also tended to be decreased (P = 0.033). In high-fat-fed Them2^−/− mice, only FGF21 was decreased. PC-TP, which is also under the transcriptional control of PPARα, was down-regulated in livers chow, but not high-fat-fed Them2^−/− mice.

Improved glucose homeostasis in mice lacking Them2

Fasting plasma concentrations of glucose (mg/dl) did not differ in mice fed chow (Them2^+/+: 129.6±12.6; Them2^−/−: 117.1±8.8) or the high-fat diet (Them2^+/+: 149.5±12.4; Them2^−/−: 133.8±8.1), nor did plasma insulin concentrations (ng/ml) differ in chow (Them2^+/+: 0.20±0.03; Them2^−/−: 0.33±0.05) or high-fat-fed (Them2^+/+: 0.23±0.03; Them2^−/−: 0.20±0.02) mice. However, glucose tolerance tests in both chow (Fig. 5A) and high-fat-fed (Fig. 5B) mice demonstrated that the absence of Them2 expression reduced time-dependent plasma glucose concentrations. Increased glucose clearance in Them2^−/− mice was evidenced by a 27% reduction in AUC for chow-fed mice (Fig. 5A, inset) and a trend toward reduced AUC in high-fat-fed mice (Fig. 5B, inset).

Figure 5.

Improved glucose tolerance and reduced hepatic production in Them2^−/− mice. A–F) After 16 h of food withdrawal, chow-fed mice (A, C, E; Them2^+/+, n=5–7; Them2^−/−, n=9) and high-fat-fed mice (B, D, F; Them2^+/+, n=6; Them2^−/−, n=4–5) were subjected to glucose tolerance tests (A, B), pyruvate tolerance tests (C, D), and insulin tolerance tests (E, F). Glucose tolerance tests were optimized by injecting chow- and high-fat-fed mice intraperitoneally with 2 mg/g and 1 mg/g of d-glucose, respectively. Insets: bar plots provide AUC values. ^‡P < 0.05 vs. Them2^+/+. G, H) Plasma concentrations of leptin (G) and adiponectin (H) were measured for chow-fed (C: Them2^+/+, n=7; Them2^−/−, n=7) and high-fat-fed (H: Them2^+/+, n=5; Them2^−/−, n=6) fed mice. I) mRNA expression levels were determined in livers of chow-fed (Them2^+/+, n=7; Them2^−/−, n=7) and high-fat-fed (Them2^+/+, n=5; Them2^−/−, n=6) mice after 16 h of food withdrawal. *P < 0.025 vs. Them2^+/+; ^†P < 0.025 vs. C.

We utilized pyruvate tolerance tests to determine whether Them2 expression might influence hepatic glucose production. Following pyruvate challenge, plasma glucose concentrations peaked earlier in chow-fed Them2^−/− compared with Them2^+/+ mice and then decreased more rapidly toward baseline values (Fig. 5C), so that AUC values were decreased by 45% (Fig. 5C, inset). In Them2^+/+ mice, high-fat feeding led to a delayed peak of plasma glucose concentrations (Fig. 5D) compared with chow-fed counterparts (Fig. 5C). By contrast, high-fat-fed Them2^−/− mice exhibited an earlier peak in plasma glucose concentrations, followed by a rapid decline to near basal values (Fig. 5D). These differences were reflected by a 73% decrease in AUC for high-fat-fed Them2^−/− compared with Them2^+/+ mice (Fig. 5D, inset).

Insulin tolerance tests were performed to determine the influence of Them2 expression on sensitivity to exogenous insulin. In chow-fed Them2^−/− mice, plasma glucose concentrations were lower than in Them2^+/+ mice only at 15 min following insulin administration compared with wild-type mice (Fig. 5E), but there were no differences in AUC values (Fig. 5E, inset). Following high-fat feeding, there was no influence of Them2 expression on the response to exogenously administered insulin (Fig. 5F, inset).

To determine whether reduced rates of hepatic glucose production in Them2^−/− mice might have been attributable to altered concentrations of adipokines that modulate hepatic insulin sensitivity, we next measured plasma concentrations of leptin and adiponectin. Consistent with the reduction in body fat, plasma leptin concentrations tended to be decreased in chow-fed Them2^−/− mice (Fig. 5G). The high-fat diet increased plasma leptin concentrations in both genotypes, and there was a trend toward lower concentrations in Them2^−/− mice. Plasma adiponectin concentrations were not influenced by either genotype or high-fat feeding (Fig. 5H).

Them2 expression did influence expression of transcription factors that regulate hepatic gluconeogenesis (Fig. 5I). Hepatic mRNA levels of HNF4α were reduced in chow-fedThem2^−/− mice and tended to decrease following high-fat feeding (P=0.045). Forkhead box protein O1 (FOXO1) levels were decreased only in chow-fed Them2^−/− mice, and peroxisome proliferator-activated receptor-γ coactivator 1α (PGC1α) expression did not differ in livers of either chow or high-fat-fed Them2^−/− mice.

Influence of Them2 expression of PPARα and HNF4α transcriptional activities

When taken together with reduced fatty acid oxidation and decreased hepatic glucose production, reduced expression of PPARα and HNF4α suggests the possibility that Them2 could regulate these transcription factors via the intracellular balance of FFA and fatty acyl-CoAs (29). To test this possibility, we explored the influence of Them2 expression on the transcriptional activities of PPARα and HNF4α. HEK 293T cells express abundant Them2 (8), which could be knocked down using siRNA (Fig. 6A). Them2 knockdown did not alter the endogenous PPARα transcriptional activity. However, the addition of the PPARα ligand DHA (9) led to greater induction of PPARα activity in the absence of Them2 than the presence of Them2 (Fig. 6B). In keeping with decreased expression of HNF4α gene, the absence of Them2 reduced the transcriptional activity of HNF4α (Fig. 6C).

Figure 6.

Influence of Them2 expression in PPARα and HNF4α transcriptional activities in HEK 293T cells. A) Representative immunoblot demonstrating knockdown of endogenous Them2 in HEK 293T cells. B, C) Following treatment of cells with scrambled control or Them2 siRNAs, transcriptional activity of PPARα (B) was determined in the absence (−) or presence (+) of PPARα expression, as well as exposure to BSA or DHA complexed to BSA (DHA/BSA), and HNF4α (C) was determined following expression of HNF4α. *P < 0.05 vs. scrambled.

DISCUSSION

This study was designed to gain insights into the metabolic functions of Them2, which is a broadly expressed, mitochondria-associated long chain acyl-CoA thioesterase. We also sought to determine whether Them2^−/− mice exhibit similar phenotypes as observed in Pctp^−/− mice, which would support the hypothesis that PC-TP-Them2 interactions regulate hepatic lipid and glucose metabolism (30). Homozygous disruption of Them2 expression in mice reduced hepatic fatty acyl-CoA thioesterase activity, leading to alterations in fatty acid metabolism and improved glucose homeostasis. In chow-fed mice, these changes paralleled the transcriptional activities of PPARα and HNF4α in liver, suggesting that Them2 regulates the intracellular balance of hepatic FFA and fatty acyl-CoAs under physiological conditions. On challenge with a high-fat diet, livers of Them2^−/− mice were protected against developing steatosis or increased hepatic glucose production by complementary compensatory mechanisms. Although not a phenocopy of Pctp^−/− mice, Them2^−/− mice exhibited changes in hepatic lipid and glucose metabolism that were consistent with regulatory mechanisms that involve interactions between Them2 and PC-TP.

The absence of Them2 decreased hepatic acyl-CoA thioesterase activities, as evidenced by increased K_m values for hydrolysis of the exogenous substrate myristoyl-CoA in liver homogenates and in mitochondria. When taken together with increases in fatty acyl-CoA concentrations and decreases in concentrations of FFA, this suggested that Them2 contributes significantly to the total acyl-CoA thioesterase activity in the liver. However, the loss of Them2 expression was accompanied by decreases in mRNAs that encode Acots 1, 2, 5, 7, and 9/10. Therefore, it was also possible that reduced expression of these enzymes contributed to the observed reduction in thioesterase enzymatic activity. Like Them2, Acots 2 and 9/10 are expressed in liver and localized to mitochondria (5, 21, 31, 32). Although the relative mitochondrial concentrations of Acots 2 and 9/10 and Them2 in mouse liver are not known, our data nevertheless suggested that Them2 made a substantial contribution to the mitochondrial acyl-CoA thioesterase activity. This is because the values of K_m values we measured for Them2^+/+ mice (24.9 μM for homogenate; 27.5 μM for mitochondria) were close to those previously reported for purified recombinant Them2 either alone (K_m=15.5 μM) or in the presence of model membranes (K_m=22.8 μM; ref. 6). By contrast, the K_m value of purified recombinant Acot2 was considerably lower (K_m=2.8 μM; ref. 31). Whereas the enzymatic activities of Acots 9/10 have not been well characterized, the hepatic expression of these genes is low compared with other tissues (21). Finally, because Acots 1 and 7 are cytosolic and Acot5 is peroxisomal (5), it was less likely that changes in expression of these genes accounted for the reductions in Acot activity observed in Them2^−/− mice. Taken together, these findings support a function for Them2 as mitochondria-associated long chain acyl-CoA thioesterase in mouse liver.

The mitochondrial localization of Them2 suggests its potential role regulating the β-oxidation of fatty acids in the liver. Because the conversion of FFA to fatty acyl-CoAs is required for entry into mitochondria, we have speculated that the activity of Them2 could limit access of fatty acids and thereby reduce rates of β-oxidation (30). Our data support this possibility, even though rates of fatty acid oxidation and plasma concentrations of β-hydroxybutyrate were both unchanged in chow-fed Them2^−/− mice: the absence of Them2 expression led to decreased PPARα activation, as evidenced by reduced hepatic mRNA levels of PPARα and its target genes, as well as decreased hepatic concentrations of FFA, which serve as endogenous ligands for this nuclear hormone receptor (33). Reduced PPARα activation would have been expected to decrease β-oxidation rates. Therefore, the absence of a decrease in rates of fatty acid oxidation in livers of Them2^−/− mice suggested that a function of Them2 was to reduce β-oxidation rates.

The high-fat diet, which decreased Them2 expression in Them2^+/+ mice, also up-regulated CD36. CD36 promotes the hepatic uptake of FFA and contributes to fatty liver and insulin resistance under conditions of nutritional excess (34, 35). The marked reduction in hepatic triglyceride concentrations in the absence of Them2 expression was most likely attributable to the reduction in CD36 expression combined with a modest increase in hepatic triglyceride secretion rates. The rates of β oxidation were reduced in livers of high-fat-fed Them2^−/− mice, as evidenced by direct measurements and by reduced plasma concentrations of β-hydroxybutyrate. In contrast to chow-fed Them2^−/− mice, there was no evidence for a generalized decrease in activation of PPARα in the absence of Them2 expression. However, FGF21 mRNA was still reduced in livers of high-fat-fed mice lacking Them2, and this presumably contributed to the observed decrease in fatty oxidation rates (36, 37).

The absence of Them2 also improved glucose tolerance in both chow and high-fat-fed mice, principally by decreasing hepatic glucose production. In chow-fed mice, reduced hepatic glucose production was in keeping with decreased expression of HNF4α and FOXO1, which are transcription factors that promote gluconeogenesis (38, 39). Because they are also ligands for HNF4α (40), reduced hepatic FFA concentrations in Them2^−/− mice may have accounted for this effect. In livers of high-fat Them2^−/− mice, we did not observe down-regulation of the same transcription factors. Under these conditions, the decrease in hepatic glucose production was most likely explained by decreased hepatic triglyceride concentrations together with reduced expression of FGF21, which also promotes gluconeogenesis (36, 37).

We previously proposed a model in which PC-TP stimulates Them2 activity at the mitochondrial membrane to reduce hepatic fatty acid oxidation and to activate PPARα and HNF4α via increased concentrations of FFA within liver cells (30). This model was based on observations that PC-TP interacts with and stimulates the enzymatic activity of Them2 in vitro (6, 8); PC-TP and Them2 each associate with mitochondria (6); knockdown of PC-TP in HEK 293 cells leads to activation of the transcriptional activities of PPARα and HNF4α (9); and livers of chow-fed Pctp^−/− mice exhibit reduced hepatic glucose production (11). Our current findings in chow-fed Them2^−/− mice largely support this model. In HEK 293T cells, transcriptional activities of PPARα and HNF4α following siRNA-mediated knockdown of Them2 demonstrated the same responses as we observed following knockdown of PC-TP (9). In livers of Them2^−/− mice, reduced fatty mitochondrial acyl-CoA thioesterase activity, whether due to the absence of Them2 or down-regulation of other Acot genes, led to a decrease in hepatic concentrations of FFA and to decreased hepatic gluconeogenesis.

As observed in high-fat-fed Them2^−/− mice, Pctp^−/− mice exhibited reduced hepatic glucose production under the same experimental conditions when compared with wild-type controls (12). However, unlike Them2^−/− mice, which were resistant to diet-induced hepatic steatosis, mice lacking PC-TP accumulated triglycerides in liver to the same extent as Pctp^+/+ controls. This difference could reflect a scenario in which PC-TP functions to enhance Them2 activity: it would be expected that absence of PC-TP yields an intermediate phenotype due to a relative reduction of Them2 activity, whereas the absence of Them2 produces a more striking phenotype that is attributable to the complete lack of enzyme activity. Alternatively, there could be separate mechanisms whereby PC-TP and Them2 attenuate high-fat diet-induced increases in hepatic glucose production.

Although there were important similarities noted in hepatic phenotypes of Them2^−/− and Pctp^−/− mice, the extrahepatic phenotypes differed in several respects. There was a reduction in body weights of chow-fed Them2^−/− mice, which was at least partially explained by modest decreases in intestinal lipid absorption. This was not observed in Pctp^−/− mice, which exhibited similar body weights as Pctp^+/+ controls and did not malabsorb lipids (11). Them2 expression had no impact on physical activity or heat production in chow-fed mice (11). By contrast, chow-fed Pctp^−/− mice exhibited increased heat production despite a trend toward decreased activity (11). In the absence of appreciable changes in intestinal lipid absorption, we did not identify a mechanism to explain the modest decrease in body weights of high-fat-fed mice that lacked Them2. This phenotype also differed because high-fat-fed Pctp^−/− mice gained the same amount of weight as Pctp^+/+ mice (12). For high-fat-fed Them2^−/− mice, we observed reduced heat production without changes in physical activity, as well as increased RER values. Although RER was increased in high-fat-fed Pctp^−/− mice, heat production and physical activity were unchanged (unpublished results). When taken together, these observations do not support our previously proposed model by which a PC-TP and Them2 interaction reduces energy expenditure in oxidative tissues by limiting mitochondrial oxidation of fatty acids (30). Additional studies will be required to better define the individual roles of both Them2 and PC-TP in the regulation of energy homeostasis.

In this study, we undertook a global knockout strategy to probe the biological functions of Them2, particularly in the liver. Notwithstanding the use of mice with equal body weights in our studies of hepatic metabolism, we cannot discount the possibility that extrahepatic effects of Them2 expression may have influenced the results. In future studies, this may be addressed using a conditional-knockout strategy that facilitates more precise definition of tissue-specific roles for Them2. Nevertheless, we have demonstrated important in vivo functions of a member of the Acot enzyme family in metabolic regulation and have determined that Them2^−/− mice share key phenotypes with Pctp^−/− mice. These findings suggest that the reduction of hepatic glucose production that was achieved in high-fat-fed mice by small-molecule inhibition of PC-TP (12) may have been attributable to disruption of PC-TP-Them2 interactions. Finally, considering that Them2^−/− mice are themselves protected against diet-induced hepatic steatosis and increases in hepatic glucose production, the enzymatic activity of Them2 could represent an attractive target in the management of obesity-related metabolic diseases.