ACCS: Acetyl-CoA Synthetase; ACSL: Long-chain acyl-CoA Synthetase; AICAR: 5-aminoimidazole-4-carboxamide riboside; BAT: Brown Adipose Tissue; GFP: Green Fluorescent Protein; MOI: Multiplicity of Infection; SIRT: Sirtuin Deacetylase

Acyl-CoA synthetases (ACS) catalyze the ATP-dependent ligation of fatty acids with coenzyme A and comprise a 26 member mammalian family whose members are named for fatty acid chain length preference and identified by homologies in two major amino acid regions [1]. The five isoforms of long chain acyl-CoA synthetase (ACSL) are each encoded by a separate gene and catalyze the activation of saturated and unsaturated fatty acids of 8-22 carbon chains [2]. Most of their acyl-CoA products are directed towards either oxidation pathways or towards the synthetic pathways for triacylglycerol, phospholipids and cholesterol esters. Long chain acyl-CoAs are also important signaling molecules in lipid metabolism, insulin secretion, and glucose uptake, storage, and metabolism [3,4].

ACSL1, the major mRNA isoform in liver, adipose tissue, and heart [5], contributes 50% of total ACSL activity in liver and 80 to 90% in adipose tissue and heart [6-8]. ACSL1 is regulated transcriptionally by PPARα [9] and by nutrient and hormonal signals in a tissue specific manner [10]. For example, obesity increases ACSL1 mRNA in adipose, but changes in ACSL1 mRNA expression in the liver vary, depending on the obesity model examined [10]. Exercise decreases ACSL1 mRNA and activity in adipose, while increasing both ACSL1 mRNA and activity in gastrocnemius muscle [10].

Indirect evidence suggests that ACSL1 is also regulated posttranslationally. For example, mice that have been either fasted or fasted then refed show disparate changes in ACSL1 mRNA, protein expression, and activity [5]. Thus after fasting, Acsl1 mRNA expression increases in gastrocnemius, inguinal adipose, and liver, while ACSL1 protein levels increase in gastrocnemius, but decrease in inguinal adipose and liver. mRNA expression is unchanged in gonadal adipose tissue after a 48 h fast, but adipose protein levels and ACSL specific activity decrease [5]. Acute regulation of ACSL1 is also suggested by decreased enzyme activity in rat adipocytes after they are incubated with the membrane-permeable cAMP analog, dibutyryl-cAMP [11].

In order to investigate the post-translational modifications of ACSL1 under different physiological conditions, we used mass spectrometry to analyze overexpressed ACSL1 in hepatocytes, brown adipocytes, and differentiated 3T3-L1 adipocytes after treatment with different hormones and drug activators of protein kinase A or AMPactivated kinase (AMPK). Because Acsl1 mRNA is regulated by PPARa [9] and is required for fatty acid ß-oxidation [6-8], we also analyzed endogenous ACSL1 in liver and brown adipose tissue from mice under conditions in which fatty acid oxidation would be upregulated, i.e. fasting or exposure to a cold environment. We found universal N-terminal acetylation, and identified 15 acetylated lysines and 25 phosphorylation sites on ACSL1. Thirteen of the acetylated lysines had not previously been identified, and none of the phosphorylation sites had been previously identified. Several modifications were unique to a specific tissue or physiological condition and many modified critical functional regions of the protein. This study indicates a remarkable array of post-translational modifications that may control the activity of ACSL1.

Animals and tissue fractionation

Animal protocols were approved by the UNC Institutional Animal Care and Use Committee. C57Bl/6 mice were maintained on a 12-h light/dark cycle with free access to water and low fat diet (Prolab 5P76 Isopro 3000; 5.4% fat by weight). Six to 11 week old mice were euthanized by cervical dislocation after each of four treatments: fed, fasted for 20 h, fasted for 9 h at room temperature, or fasted for 4 h and then exposed to 4°C for 5 h while continuing to fast. Liver and brown adipose tissues (BAT) were immediately removed and all subsequent steps were performed at 4°C. Unless stated otherwise, buffers contained proteinase inhibitor (Roche) and phosphatase inhibitor cocktail I and II (Sigma, St. Louis MO). For crude mitochondria collection, liver was rinsed with ice cold phosphate buffered saline and homogenized with 10 up-and-down strokes in a Teflon-glass motorized homogenizing vessel (setting 6) in cold buffer H (10 mM HEPES-KOH, pH 7.4, 0.25 M sucrose, 1 mM EDTA and 1 mM dithiothreitol). Crude mitochondria were collected as previously described [12], except that the pellet was resuspended in gentle lysis buffer (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% NP-40,5 mM EDTA), and homogenized in a Dounce tissue grinder with a glass pestle. For lysate collection, liver or BAT was homogenized in gentle lysis buffer and the lysate was collected after centrifuging at 600 x g.

Cell culture

Cells were maintained at 37°C in 5% CO₂/air. Lysates were collected in cold RIPA buffer (25 mM Tris-HCl pH 7.6, 150 mM NaCl, 1% NP-40, 1% sodium deoxycholate, 0.1% SDS) or gentle lysis buffer supplemented with proteinase inhibitor (Roche) and phosphatase inhibitor cocktail I and II (Sigma, St. Louis MO).

Hepatocyte isolation and adenovirus infection

Primary hepatocytes were isolated from male Wistar rats (250-300 g) by the UNC Cellular Metabolism and Transport Core. Hepatocytes were seeded at a density of 4.5 × 10⁶ cells per 100-mm collagen-coated dish in Minimal Essential Media supplemented with 10% Fetal Bovine Serum (v/v) and 50 units/ml penicillin/streptomycin. For adenovirus infection, recombinant adenoviruses Ad-Acsl1 that carries a C-terminal Flag tag or Ad-GFP (control) [13] were added at an MOI of 20 for 24 h. Cells were either untreated or treated for 20-30 min with one of the following hormones: 200 nM insulin, 1 mM 5-aminoimidazole- 4-carboxamide riboside (AICAR), 2 mM dibutyryl-cAMP, or 10 µM isoproterenol.

Brown adipocyte differentiation and adenovirus infectiona

Interscapular BAT from wild-type mice (postnatal day 5) was removed, minced, and digested for 30 min with collagenase (CSL1, 0.2%, Worthington) in isolation buffer (123 mM NaCl, 5 mM KCl, 1.3 mM CaCl₂, 5 mM glucose, 100 mM HEPES, 4% bovine serum albumin) [14]. Precursor cells were collected and maintained in growth medium (Dulbecco's Modified Eagle's Medium with 10 mM HEPES, 25 mM glucose, 10% fetal bovine serum, 100 units/ml penicillin/ streptomycin, 17 µM D-pantothenic acid, 33 µM D-biotin, 100 µM ascorbic acid, 8.3 mM L-glutamine and 4 nM insulin) [15]. Confluent cells (post-confluence day 2) were differentiated with growth medium supplemented with 50 nM insulin, 50 nM T3 and 1 µM rosiglitazone. Four days later, differentiated brown adipocytes were infected with adenovirus (Ad-GFP or Ad-Acsl1) using Lipofectamine (Invitrogen, Carlsbad, CA) at an MOI of 100 [16]. Four days after infection, adipocytes were preincubated in serum-free Dulbecco's Modified Eagle's Medium with 0.5% fatty acid-free bovine serum albumin for 4 h before hormone treatment for 30 min. Cells were treated with either 10 µM of the ß3-adrenergic receptor agonist CL-316243, 200 nM insulin, or 20 µM forskolin.

3T3-L1CARΔ1 Adipocyte Differentiation and Adenovirus Infection

A 3T3-L1 adipocyte cell line stably expressing the truncated receptor for coxsackievirus and adenovirus receptor (3T3-L1CARΔ1) was a gift from Dr. D.J. Orlicky (University of Colorado). 3T3- L1CARΔ1 preadipocytes were cultured in Dulbecco's Modified Eagle's Medium containing 25 mM glucose, 10% fetal bovine serum and 100 U/ml penicillin/streptomycin [17]. Differentiation was induced in 2-d post-confluent cells with 0.25 µM dexamethasone, 0.5 mM isobutylmethylxanthine, and 100 nM insulin for 48 h [18]. Cells were then grown supplemented with 100 nM insulin, which was replaced every 2 d thereafter until analysis. After 6 days, the adipocytes were infected with adenovirus (Ad-GFP or Ad-Acsl1) at an MOI 200 for 16 h. Adipocytes were serum starved for 3 h before hormone treatment with 500 nM insulin for 20 min, 2 mM dibutyryl-cAMP for 30 min, or 200 µM 1,2-dioctanoyl-sn-glycerol for 30 min.

Immunopurification

Immunoprecipitation was performed in RIPA buffer or gentle lysis buffer at 4°C. ACSL1-Flag was immunoprecipitated overnight from Ad-Acsl1 infected cells using anti-Flag M2 affinity agarose gel (Sigma, St. Louis MO). Ad-GFP infected cells were used as a negative control. Endogenous ACSL1 was immunoprecipitated overnight with rabbit anti-ACSL1 (21^st Century Biochemicals, Marlborough, MA; target sequence Ac-EVHELFRYFRMPELIDIC-amide) from crude liver mitochondria, liver lysate, or BAT lysate. Crude mitochondria from liver-specific Acsl1^L-/- liver or lysates from adipose-specific Acsl1^A-/- BAT were used as negative controls. The immune complex was incubated with protein A/G agarose beads for 1 h (Santa Cruz Biotechnology, Santa Cruz, CA). The immune complex beads were treated with 4X LDS buffer (Invitrogen, Carlsbad, CA) supplemented with 1 mM dithiothreitol at 70°C for 10 min before alkylation by 2 mM iodoacetamide [DOI: 10.1038/nprot.2008.123]. The supernatant was subjected to electrophoresis in a NuPAGE 4-12% Bis-Tris gel, and the gel was stained for total protein using GelCode Blue Stain (Invitrogen, Carlsbad, CA).

LC-MS/MS

The 78 kDa bands were excised, cut into 1 mm³ cubes, destained, and incubated with trypsin, and peptides were extracted [DOI: 10.1038/ nprot.2008.123]. Reversed-phase nanoflow liquid chromatographytandem mass spectrometry (nLC-MS/MS) was performed on an ABI TEMPO capillary LC coupled to an Applied Biosystems 4000QTrap mass spectrometer. Peptides were reconstituted in 15 µl of 0.5% formic acid and incubated for 10 min at 25°C with shaking. The pH of the solution was neutralized with 7.5 µl of 0.1 M NH₄HCO₃ and incubated for 10 min at 25°C with shaking. A 10 µl aliquot was loaded and desalted on a C18 trap at 20 µl/min with 95% A (95% water/5% acetonitrile with 0.1% formic acid) and 5% B (5% water/95% acetonitrile with 0.1% formic acid) and separated on a 75 µm inner diameter × 15 cm Acclaim PepMap 100 C18 nanoLC column (LC Packings). The gradient flowed at 300 nl/min and started at 95% A and 5% B, increased to 40% B from 15 to 140 min, and was again increased to 90% B from 140 to 150 min. The gradient was held at 90% B for 10 min before the original conditions were resumed. An acetonitrile blank was run after each sample; the gradient started at 95% A and 5% B, increasing to 90% B at 5 min, and held for 40 min before returning to baseline conditions. Mass spectra were acquired for the samples and the blanks with Analyst Software v1.4.1 (AB SCIEX, Framingham, MA) scanning the mass range of 400- 1800 m/z and the 3 most abundant multiply-charged peaks were selected for MS/MS. MS/MS centroid peak lists were created from precursors of all charge states using Mascot.dll script. Proteome Discoverer v1.3 (Thermo Fisher Scientific) was used for data analysis. MS/MS spectra were first searched against the Swiss-Prot (November 8, 2010 release) and reversed databases using a Mus or Rattus taxonomy filter with MASCOT v2.2.1 and a Mus musculus or Rattus norvegicus database subset for SEQUEST v1.2 with up to 1 missed cleavage allowed and carbidomethyl (C) static modification with a precursor tolerance of ± 1.8 Da and a MS/MS tolerance of ± 0.8 Da. The search results filtered for medium and high confidence peptides (defined according to the Percolator algorithm [19] with a q-value threshold of 0.05) and for proteins with more than 2 peptides (only rank 1 peptides were counted, and if a peptide belonged to more than one protein, it was only counted in the top scored protein) with protein grouping enabled are shown in Supplementary Table 1. All of the search results files belonging to each species, rat and mouse, respectively, were opened into a multiconsensus report. We filtered for high-confidence peptides (defined according to the Percolator algorithm [19] with a q-value threshold of 0.01) and for proteins with more than 5 peptides (only rank 1 peptides were counted, and if a peptide belonged to more than one protein it, was only counted in the top scored protein). The remaining proteins, including all proteins in all protein groups, were exported as a FASTA file to use in a second search for modifications. The second search was performed in Proteome Discoverer v1.3 (Thermo Fisher Scientific) using SEQUEST v1.2. MS/MS spectra were searched against the rat or mouse subset and corresponding decoy database created in the first pass search with up to 2 missed cleavages allowed and acetylation (K, N-terminal), carbamidomethyl (C), deamidation (NQ), oxidation (M), and phosphorylation (S,T, Y) variable modifications with a peptide tolerance of ± 1.8 Da and MS/MS tolerance of ± 0.8 Da. The smaller subset database was used to search for modifications in order to reduce the false discovery rate. The search results were filtered for high confidence peptides (defined according to the Percolator algorithm [19] with a q-value threshold of 0.01) and for proteins with more than 2 peptides (only rank 1 peptides were counted, and if a peptide belonged to more than one protein, it was only counted in the top scored protein). The modification search results for ACSL1 are shown in Supplementary Table 3. Replicates are biological replicates.

Site-directed mutagenesis

Full-length rat ACSL1 cDNA with a C-terminal Flag epitope was subcloned from a previously constructed plasmid to mammalian expression vector pcDNA3.1. PCR-based site-directed mutagenesis was performed to mutate single amino acids according to Stratagene's Quickchange Site-Directed Mutagenesis Manual. The primers (Supplemental Table 2) were synthesized by the UNC Nucleic Acids Core Facility. The full length ACSL1 with mutations was confirmed by sequencing by the UNC Genome Analysis Facility.

Transient transfection

CHO cells were routinely cultured in Modified Eagle's Medium containing 10% fetal bovine serum and 100 U/ml penicillin/ streptomycin. CHO cells were plated a cell density of 2.0 x 10⁵ per well in 6-well dishes and transfected 18 h after plating with 2 µg of plasmid carrying rat ACSL1 or mutants (Fugene 6, Roche, Florence, SC). Cell homogenates were collected 24 h post-transfection in ice-cold Medium A (10 mM Tris, pH 7.4, 250 mM sucrose, 1 mM EDTA, 1 mM dithiothreitol, and Protease Inhibitor Mixture (Sigma, St. Louis MO)). Aliquots were stored at -80°C until use.

ACSL assay

ACSL initial rates were measured with 50 µM [1-¹⁴C] palmitic acid (Perkin Elmer, Waltham, MA), 10 mM ATP, and 0.25 mM CoA in total membrane fractions (0.5-4.0 µg) [20].

Western blot

Homogenates from transfected CHO cells (30 µg) were separated by electrophoresis on a 10% polyacrylamide gel and transferred to nitrocellulose membranes (Bio-Rad). Protein expression of ACSL1 with and without mutations was detected by anti-ACSL1 polyclonal antibody (Cell Signaling 4047). GAPDH was used as the loading control (Abcam, Cambridge, MA; ab36840).

Acetylated lysines in overexpressed and endogenous ACSL1

To investigate the post-translational regulation of ACSL1 in different cell types and after different hormonal treatments, we overexpressed ACSL1 in rat hepatocytes, mouse brown adipocytes, and differentiated 3T3-L1 adipocytes. As has been reported for 80% of mammalian proteins, including in rat liver mitochondrial ACSL1 [21], the N-terminal methionine was acetylated under all conditions (Tables 1, 2). Although the function of N-terminal acetylation is unknown [22], in yeast an N-terminal acetylation appears to act as a degradation signal [23].

Eleven lysines were acetylated within the over-expressed protein (Table 1). Nine of these acetylated lysines had not been previously reported. Lys217 and Lys223 were acetylated in all rat hepatocytes and mouse brown adipocytes and in untreated and insulin-treated 3T3-L1 cells. In hepatocytes, Lys562 was also acetylated in all treatment groups, except in untreated hepatocytes. In contrast, Lys562 was acetylated in both untreated mouse brown adipocytes and untreated 3T3-L1 cells. Treatment of mouse brown adipocytes and 3T3-L1 cells with CL- 316243, a ß3-adrenergic agonist that increases fatty acid oxidation and thermogenesis [24], and 1,2-dioctanoyl-sn-glycerol (di8), respectively, also resulted in Lys562 acetylation. Five acetylated lysines (K49, K341, K504, K533, and K544) were unique to cell types or hormone treatment (Table 1). Acetylated Lys544 was previously identified in mouse liver after a calorie-restricted diet and in the livers of both fasted and fed mice [25,26].

Because post-translational modifications could be an artifact of overexpression, we also analyzed endogenous ACSL1 from mouse liver and brown adipose tissue (Table 2). None of the lysine acetylations, except Lys387, identified in the endogenous ACSL1, had been previously identified. Acetylated Lys387 was identified in liver after mice were fed a calorie restricted diet [25]. Although fewer total modifications were present in the endogenous ACSL1 compared to the overexpressed protein, three of the nine acetylated lysines identified in the endogenous samples were at the same sites. Thus, the overexpression and endogenous data complemented each other. Also congruent with the data from overexpressed ACSL1, Lys562 was the most prevalent acetylation in the endogenous samples in all treatment groups. Six of the acetylated lysines found endogenously (K52, K241, K250, K387, K641, and K649) were unique to a particular treatment or tissue.

Phosphorylated Residues in Overexpressed and Endogenous ACSL1

In overexpressed ACSL1 in rat hepatocytes, mouse brown adipocytes, and differentiated 3T3-L1 adipocytes that were either untreated or treated with different hormones, we identified 20 different phosphorylated residues (Table 3). None of these phosphorylation sites had been previously identified. Unlike the identified acetylated lysines, no specific residues were phosphorylated in all cell types or under all hormone treatments. However, Ser58 was consistently phosphorylated in hepatocytes in all treatment groups, and thirteen phosphorylation sites were unique either to a specific cell type or to a specific hormonal treatment (T46, T65, Y119, S123, T281, S299, T384, S385, T434, S440, T442, T445, and T527).

Seven residues were phosphorylated in endogenous ACSL1 from mouse liver and brown adipose tissue (Table 4). Two of these phosphorylated sites were identified in overexpressed ACSL1. Three of the endogenous phosphorylated sites were unique to treatment and tissue. Interestingly, Thr219 was phosphorylated in the fed, but not in the fasted condition in both liver and brown adipose tissue (Table 4). Thr219 was also phosphorylated when ACSL1 was overexpressed in hepatocytes treated with insulin, AICAR, or cAMP.

The 26 ACS family members have considerable homology [1]. Analysis of amino acid motifs in mammalian ACS isoforms and crystallization studies of an ACS from Thermus thermophilus HB8 have led to the designation of functional motifs (Figures 1, 2) [27]. These include the ATP/AMP binding site (Motif I), Gate motif (Motif IV), Adenine binding site (Motif III), Fatty acid binding (Motif II), Linker region (continued Motif II), and Motif V. Lysine acetylation or phosphorylation at these sites is likely to regulate ACSL1 activity and, thus, long-chain fatty acid metabolism.

Figure 1: Summary of acetylation and phosphorylation sites identified by nLC-MS/MS from overexpressed rat ACSL1 in hepatocytes and endogenous ACSL1 in mouse liver for all treatment groups. The sequence shown is rat ACSL1 (P18163). However, the amino acid residues shown in parenthesis indicate sequence variations between rat and mouse that vary at the post-translational modification site. The numbers above the amino acids indicate the amino acid position of the modification; forward slashes are present to separate the numbers. The AMP binding site, Gate motif, Adenine binding motif, Fatty acid binding site, Linker region, and the consensus fragment of Motif V are underlined; Roman numerals designate the motifs, as described in reference [27].

Figure 2: Summary of acetylation and phosphorylation sites identified by nLC-MS/MS in overexpressed rat ACSL1 in brown adipocytes in cell culture and in endogenous ACSL1 in mouse brown adipose tissue for all treatment groups. The sequence shown is rat ACSL1 (P18163). However, the amino acid residues in parenthesis indicate the sequence variations between rat and mouse that vary at the post-translational modification site. The numbers above the amino acids indicate the amino acid position of the modification; forward slashes are present to separate the numbers. The AMP binding site, Gate motif, Adenine binding motif, Fatty acid binding site, Linker region, and the consensus fragment of Motif V are underlined; Roman numerals designate the motifs, as described in reference [27].

Modifications in the fatty acid binding region (Motif II)

Lys533 and Lys544 were acetylated in the fatty acid binding region of ACSL1 (Figure 1). Both Lys533 and Lys544 were acetylated in one replicate of overexpressed ACSL1 in hepatocytes incubated with insulin. A previous study found that Lys544 of ACSL1 is acetylated in mouse liver mitochondria before and after a 12 h fast [26], and that after a 2-wk calorie-restricted diet, the amount of acetylation on Lys544 decreased 2.7-fold while protein expression of SIRT3 and lysine acetylation of acyl-CoA synthetase medium chain-1 (ACSM1) and ACSS2 increased [25]. SIRT3 also deacetylates and activates ACCS2 [28]; although the modified ACCS2 lysine is not structurally comparable, the data suggest a role for lysine deacetylation and up-regulation of fatty acid activation to acyl-CoA during caloric restriction. Thus, the relative amount of acetylation may be important at this residue. Changes in the relative amounts of lysine acetylation of ACS family members have also been associated with ethanol exposure. In rats, ethanol exposure increases total lysine acetylation of ACSS2 more than 4-fold, and it was suggested that hyperacetylation and inactivation of ACSS2 might play a role in hepatotoxicity [29]. Because ethanol-induced hyperacetylation is independent of SIRT3 expression levels, the sirtuin deacetylases may respond to different physiological stimuli.

Because ACSL1 channels fatty acids towards oxidation in heart and adipose tissue [8,30], activation of ACSL1 is likely to occur during fasting or caloric restriction when cellular energy is low. Because these conditions increase the expression and activity of the sirtuin deacetylases [31] and activate AMPK, it seems likely that deacetylation would increase ß-oxidation.

Modifications in the linker region (Motif II)

Lys562, which located just after the Linker region, a six amino acid sequence that connects the N- and C-terminal regions (Figures 1, 2) [32], was acetylated in all treatment groups of ACSL1 overexpressed in hepatocytes, in untreated and di8-treated 3T3-L1 adipocytes, and in ACSL1 overexpressed in untreated and CL-316243-treated brown adipocytes. Lys562 was also acetylated under most conditions in endogenous ACSL1 from crude liver mitochondria. The Linker region is structurally important for catalysis; after the adenylate is formed, the C-terminal domain rotates ~140° to expose the active site to the opposite side of the C-terminal domain [27]. Lys562 is equivalent to Lys609 in Salmonella enterica ACSS. Acetylation at S. enterica Lys609 prevents adenylate formation without affecting the formation of the thioester [33]. Lys609 is part of the PX₄GKθX(R/K) consensus sequence within the Fatty acid binding site/ Linker motif (where θ is the amino acid alanine, valine, leucine, isoleucine, or methionine) within the active site [32]. Further, mutating the homologous lysine in Arabidopsis thaliana 4-coumarate-CoA ligase abolishes its catalytic function [34].

Other modified sites

Similar to the decreased amount of Lys544 found after caloric restriction (see above), the amount of acetylation on Lys387 decreased 2.4-fold [25]. As predicted for possible AMPK-mediated modifications, a decrease in Lys387 acetylation with caloric restriction may enhance ß-oxidation. Lys387 was uniquely acetylated in overexpressed ACSL1 in untreated brown adipocytes.

Unique modified sites

Several additional acetylated and phosphorylated residues were unique to a particular treatment or cell type. Lys641 was uniquely acetylated in brown adipose tissue from fed mice (Figure 2). This acetylation site lies near a fragment of Motif V, which is conserved in ACS family members and contains a lysine that is required for activity [1].

Several of the acetylation and phosphorylation sites identified in the overexpressed ACSL1 samples were not observed in the endogenous samples. These types of discrepancies could result if the amount of modified peptide were below the detection limit, if sample loss occurred during preparation, if the modification was lost during sample handling or detection, or if actual biological differences were present. Because few acetylated lysines and phosphorylated residues were unique to a cell type or hormone treatment, we cannot exclude the possibility that ACSL1 is regulated quantitatively by some of these posttranslational modifications. Additionally, more biological replicates will be necessary to discern the reproducibility of the modifications under specific physiological conditions.

Mutations of Lys285 Ser278, and Lys676

To determine whether specific post-translational modifications alter ACSL1 activity, CHO cells were transfected with wild-type or mutated rat ACSL1. We elected to examine Lys285, Ser278, and Lys676. Lys285 and Ser278 are highly conserved residues that are located within the ATP/AMP binding site and were uniquely modified in hepatocytes by AICAR although the modifications had medium or low confidence peptide identifications (falling below the q value = 0.1). We also modified Lys676 because it is a highly conserved residue in the motif KXXK/R that is present in all ACSs [1]. Ser278 was mutated to mimic the phosphorylated (S278D) and non-phosphorylated (S278A) states, whereas Lys285 and Lys676 were mutated to mimic the acetylated (K285A, K285Q, K676A, K676Q) or non-acetylated (K285R, K676R) states [35]. Transfected wild-type rat ACSL1 doubled ACSL total activity in CHO cells, but mutations of Ser278 diminished the total ACSL activity to the level observed with the empty vector, indicating total inhibition of ACSL1 activity (Figure 3A) and showing that Ser278 is critical. Lys172 in FadD13 from Mycobacterium tuberculosis is equivalent to Lys285, and has been proposed to bind to the phosphate group of ATP via electrostatic interactions [36]. Both K285A and K285Q, which should mimic acetylated Lys285, decreased the overexpressed ACSL1 activity by 60-70%, whereas K285R, which mimics deacetylated Lys285, maintained the activity increase caused by overexpressed ACSL1. These results suggest that acetylation of Lys285 would decrease ACSL1 activity. The fact that Ser278 and Lys285 were modified in concert in hepatocytes that had been treated with AICAR, strongly suggests that ACSL1 might be activated in liver under conditions when the cellular energy supply is deficient. Further investigation of these sites is necessary. Because these peptides were low confidence, a multi-enzyme approach or high mass accuracy instrument may help improve the identification of these modifications.

Figure 3: ACSL activity change by ACSL1 mutants in CHO cells. CHO cells were transfected with plasmids carrying empty vector (EV), or a full-length rat ACSL1 cDNA either without mutation (NM), or with mutations to replace single amino acids A) at Ser278 with alanine (S278A) or aspartic acid (S278D), at Lys285 with alanine (K285A), glutamine (K285Q), or arginine (K285R), or B) at Lys676 with alanine (K676A), glutamine (K676Q), or arginine (K676R). Cell homogenates were collected 24h posttransfection. Total ACSL activity was measured using [1-14C]palmitate, data are means ± S.E. (n=3-6). * p<0.05, ** p<0.01. Western blotting with anti-ACSL1 antibody was performed with GAPDH as loading control, which has been repeated three times.

According to modeling of M. tuberculosis FadD13 [35], Lys676 would lie very near the phosphate group of ATP. Acetylation of the comparable Lys residue on the mitochondrial ACSS2 completely inactivates this enzyme, and deacetylation of the site by SIRT3, reactivates it [28]. Although our MS coverage of ACSL1 included this site, Lys676 was not acetylated under any condition in any of the tissues examined. However, all three mutations, Lys676Ala, Lys676Gln, and Lys676Arg, totally inhibited overexpressed ACSL activity (Figure 3B). Thus, similar to the comparable Lys487 residue in FadD13 [35], Lys676 appears to be essential for ACSL1 activity.

Recent studies of global lysine acetylation modifications found in the HeLa and mouse liver cell proteomes reveal the complexity and ubiquity of this modification (reviewed in [37]). Lysine acetylation is an important regulator of metabolism [38,39] and is prevalent in mitochondria where more than 20% of proteins are estimated to contain acetyllysine modifications [26]. Several acetylated lysine residues from members of the acyl-CoA synthetase superfamily have been identified in these global acetylome studies (Supplemental Table 4). Similarly, although phosphorylation is a well-known regulator of cellular function and several phosphorylation sites in ACSL1 and other ACS family members were previously identified (Supplemental Table 5), little is known about the effect of phosphorylation on ACSL1 activity.

Functional effects of posttranslational modifications have been investigated in only a handful of ACS family members. Acetylation within the Linker region on Lys661 of mouse ACSS1 or on Lys635 in ACSS2, a mitochondrial isoform, deactivates these enzymes [40] and ACSS1 and ACSS2 can be reactivated by the sirtuin deacetylases SIRT1 and SIRT3, respectively [40]. Human ACSS2 is regulated similarly by SIRT3 on Lys642 [41]. In mice, when SIRT1 deacetylates and activates ACSS1, fatty acid synthesis increases [40]. During fasting, deacetylation and activation of ACSS would produce acetyl-CoA, a substrate used for both ketone production and the citric acid cycle. Other increases in fatty acid oxidation are also linked to SIRT-mediated deacetylation [42]. In our study, the comparable highly conserved lysine residue, Lys676, was not acetylated under any condition. However, mutating this lysine totally abrogated ACSL1 activity. Thus, if Lys676 were acetylated under some physiological condition, it is highly likely that ACSL1 would be inactivated, as this region seems to be critical for ACSL1 activity.

Although we identified a large number of lysine acetylation and phosphorylation sites of ACSL1, further studies will be required to understand the functional consequences of the majority of these modifications. Because few studies have found physiologically induced changes in ACSL1 specific activity, we suspect that many of the identified phosphorylations and acetylations do not simply alter enzyme activity but, instead, may act in concert under specific conditions. Further, because ACSL1 directs its acyl-CoA product towards both triacylglycerol synthesis and ß-oxidation in liver [6], but only towards ß-oxidation in adipocytes [7] and heart [8], it is likely that some of the modifications, rather than altering enzyme activity, might affect ACSL1 protein stability or protein-protein interactions.

This work was supported by NIH grants DK59935 and DK59935-S1 (ARRA) (RAC), DK40936, U24-DK59635, a proteomics minigrant from the UNC Center for Gastrointestinal Biology & Disease (P30 DK34987), a postdoctoral fellowship DK082099 (JLF), the UNC Nutrition Obesity Research Center (P30 DK56350), and a postdoctoral fellowship (LOL) from the American Heart Association-Mid-Atlantic Region.