Regulation of mitochondrial carbamoyl-phosphate synthetase 1 activity by active site fatty acylation.

In addition to its role in reversible membrane localization of signal-transducing proteins, protein fatty acylation could play a role in the regulation of mitochondrial metabolism. Previous studies have shown that several acylated proteins exist in mitochondria isolated from COS-7 cells and rat liver. Here, a prominent fatty-acylated 165-kDa protein from rat liver mitochondria was identified as carbamoyl-phosphate synthetase 1 (CPS 1). Covalently attached palmitate was linked to CPS 1 via a thioester bond resulting in an inhibition of CPS 1 activity at physiological concentrations of palmitoyl-CoA. This inhibition corresponds to irreversible inactivation of CPS 1 and occurred in a time- and concentration-dependent manner. Fatty acylation of CPS 1 was prevented by preincubation with N-ethylmaleimide and 5'-p-fluorosulfonylbenzoyladenosine, an ATP analog that reacts with CPS 1 active site cysteine residues. Our results suggest that fatty acylation of CPS 1 is specific for long-chain fatty acyl-CoA and very likely occurs on at least one of the essential cysteine residues inhibiting the catalytic activity of CPS 1. Inhibition of CPS 1 by long-chain fatty acyl-CoAs could reduce amino acid degradation and urea secretion, thereby contributing to nitrogen sparing during starvation.

In addition to its role in reversible membrane localization of signal-transducing proteins, protein fatty acylation could play a role in the regulation of mitochondrial metabolism. Previous studies have shown that several acylated proteins exist in mitochondria isolated from COS-7 cells and rat liver. Here, a prominent fatty-acylated 165-kDa protein from rat liver mitochondria was identified as carbamoyl-phosphate synthetase 1 (CPS 1). Covalently attached palmitate was linked to CPS 1 via a thioester bond resulting in an inhibition of CPS 1 activity at physiological concentrations of palmitoyl-CoA. This inhibition corresponds to irreversible inactivation of CPS 1 and occurred in a time-and concentration-dependent manner. Fatty acylation of CPS 1 was prevented by preincubation with N-ethylmaleimide and 5-p-fluorosulfonylbenzoyladenosine, an ATP analog that reacts with CPS 1 active site cysteine residues. Our results suggest that fatty acylation of CPS 1 is specific for long-chain fatty acyl-CoA and very likely occurs on at least one of the essential cysteine residues inhibiting the catalytic activity of CPS 1. Inhibition of CPS 1 by long-chain fatty acyl-CoAs could reduce amino acid degradation and urea secretion, thereby contributing to nitrogen sparing during starvation.
The covalent modification of proteins by lipids alters their physical and functional properties. Several types of lipids are covalently bound to proteins as follows: isoprenoids, glycosylphosphatidylinositols, cholesterol, and fatty acids (1)(2)(3)(4). Protein fatty acylation is the modification of proteins by fatty acids. It is divided into two categories, myristoylation and palmitoylation. In myristoylation, the 14-carbon myristate is co-translationally attached to the N-terminal glycine residue of a protein via a stable amide bond. Palmitoylation is characterized by the post-translational attachment of the 16-carbon fatty acid palmitate to cysteine residues of a protein via a thioester bond. Due to its reversible nature, palmitoylation has been shown recently to regulate the subcellular localization of several proteins involved in signal transduction processes (1,5,6). For protein palmitoylation to occur, palmitate needs to be activated in the form of its coenzyme A derivative, palmitoyl-CoA.
In addition to playing a role in signal transduction, dynamic protein fatty acylation has been postulated to play a potential role in the regulation of amino acid catabolism (10,11). In the well fed state, amino acids originating from digestion of dietary proteins in the gastrointestinal tract are absorbed into the bloodstream and may be used for protein synthesis. Alternatively, amino acid surplus can be metabolized to glucose, be used for fatty acid synthesis, or be catabolized to generate energy. The amino group of amino acids is removed by transamination and deamination prior to urea synthesis in periportal hepatocytes, although the residual carbon skeleton is metabolized to gluconeogenic precursors. In starvation, proteolysis of muscle protein is the main source of circulating amino acids. Transamination of the amino acids, particularly branched-chain amino acids (BCAA), occurs in muscle, whereas liver is postulated to lack BCAA aminotransferase activity. However, liver has branched-chain ketoacid dehydrogenase activity and can oxidize branched-chain ketoacids in the mitochondria. During the early stages of fasting, the body draws selectively on its supply of energy in the form of triacylglycerol in adipose tissue, sparing at first the breakdown of vitally needed proteins (12). In this case, ␤-oxidation from fatty acids is very active in the liver, and large quantities of acetyl-CoA are formed that promote gluconeogenesis. The bulk of energy production from fatty acid and amino acid carbon skeletons occurs in the mitochondria.
In mitochondria, enzymes that are part of different catabolic pathways must compete for a common pool of reduced coenzyme A, NAD ϩ , and FAD ϩ cofactors and for a common electron transport chain (13). As such, the variable availability of such cofactors, appropriate catabolites, and the activity of the competing enzymes can affect the rate of a given catabolic pathway.
However, little is known about the coordination mechanisms that are responsible for the regulation of the degradation of the substrates to be oxidized. The mitochondrial protein methylmalonyl semialdehyde dehydrogenase (MMSDH), an enzyme of the valine and pyrimidine catabolic pathways, was shown to be fatty-acylated on its active site cysteine residue thereby inhibiting its enzymatic activity (10,11). Because the extent of fatty acylation of MMSDH varied with mitochondrial energy level, was reversible, and was specific for long-chain fatty acyl-CoAs, it was proposed that fatty acylation of MMSDH could act as a novel mode of regulation of enzymatic activity. In addition, as a way to prioritize substrate degradation, this novel mode of regulation has been proposed to mediate a metabolic cross-talk between amino acid and fatty acid catabolism (10,11). The fatty acylation of MMSDH required only fatty acyl-CoA and seemed to be spontaneous in vitro in apparently pure preparations of MMSDH. Thus, protein fatty acylation of a few select mitochondrial proteins could perhaps be regulated by intramitochondrial levels of long-chain fatty acyl-CoAs.
Upon incubation of mitochondria with radiolabeled fatty acids, several proteins have been shown to incorporate fatty acids in a covalent manner in mitochondria isolated from rat liver and COS-7 cells (10,14), but to date, only one fatty-acylated mitochondrial protein has been identified, MMSDH. We report the apparent similarity in protein fatty acylation electrophoretic patterns in mitochondria isolated from different rat tissues. In addition, we found that the majority of fatty-acylated proteins are found in the mitochondrial matrix and the inner mitochondrial membrane. A new fatty-acylated 165-kDa protein was identified as rat liver carbamoyl-phosphate synthetase 1 (CPS 1). CPS 1 is the first and rate-limiting step of the urea synthesis (15)(16)(17) and catalyzes the removal of ammonia, a by-product of amino acid catabolism. Based on our kinetic results, palmitoyl-CoA inhibition of CPS 1 was time-and concentration-dependent and corresponded to an irreversible inactivation of the enzyme. Further fatty acylation studies indicated that CPS 1 is likely fatty-acylated on at least one of its essential cysteine residues. The fact that both MMSDH and CPS 1 are involved in amino acid catabolism, are fatty-acylated on cysteine residues, and inhibited by physiological concentrations of palmitoyl-CoA further reinforces the possibility of a metabolic cross-talk between amino acid and fatty acid catabolic pathways.
Preparation of Mitochondria-Mitochondria were purified from rat liver, heart, brain, kidney, and gastrocnemius leg muscle from five Sprague-Dawley rats (ϳ200 g), using a combination of differential and Percoll gradient centrifugation (18). All buffers contained freshly added 1 mM dithiothreitol, 20 g/ml leupeptin, and 1 mM phenylmethylsulfonyl fluoride. The only modification to the procedure was at the homogenization step, in which the rat heart and leg muscle were also minced prior to homogenization with a Polytron at level 2.5 for 30 s. All isolation procedures were carried out at 4°C.
Liver mitochondrial subfractions were prepared from two rat livers by the established swell shrink-sonication procedure of Ohlendieck et al. (19) followed by discontinuous density gradient centrifugation with the following modifications: mitochondria were swollen in 6 ml of swell medium (10 mM potassium phosphate buffer, pH 7.4) with subsequent addition of 6 ml of shrink medium (10 mM potassium phosphate, 10 mM MgCl 2 , 30% glycerol, 32% sucrose, pH 7.4) after 15 min.
Preparation of Radiolabeled Fatty Acids and in Vitro Fatty Acylation Reactions-Radioiodination of the iodopalmitate with [ 125 I]NaI (2.14 Ci/mmol) (Amersham Pharmacia Biotech) was performed as described in Berthiaume et al. (20) without the high pressure liquid chromatography purification step. Typical specific activity of [ 125 I]iodopalmitate was 2 Ci/mmol. Synthesis of the [ 125 I]iodopalmitoyl-CoA derivative was carried out using acyl-CoA ligase as reported previously (20).
Highly purified mitochondria were incubated in the mitochondrial acylation buffer (MAB) described by Stucki et al. (14), as modified in Berthiaume et al. (10). The MAB buffer contained 60 mM KCl, 7.5 mM potassium phosphate, pH 7.4, 40 mM triethanolamine HCl neutralized with KOH to pH 7.4, 15 mM potassium succinate, 2 mM potassium glutamate, 2 mM potassium malate, 1 mM K-ATP, 1 mM MgCl 2 , 1 mM CoA, 1 mM carnitine, and 0.65 g/ml rotenone. All in vitro fatty acylation assays with radiolabeled palmitate analog were carried out in a final volume of 50 l. In mitochondrial protein fatty acylation assays, 40 g of purified mitochondria were incubated with 10 Ci of [ 125 I]iodopalmitate (ϳ100 M) in 1ϫ MAB from a 5ϫ MAB stock in 1.5-ml Eppendorf tubes. The incubation was allowed to proceed for 30 min at room temperature (typically 25°C) and then stopped by the addition of 5ϫ SDS-PAGE loading buffer. Samples were incubated for 2 min at 95°C, loaded onto a 12% SDS-polyacrylamide gel, and electrophoresed. The gel was fixed, stained with Coomassie G-250, destained, dried, and exposed to autoradiographic film.
Fatty acylation assays of chromatographic fractions were carried out in a final volume of 50 l with typically 10 g of purified proteins (or the amount of proteins necessary to contain 1 g of CPS 1) or with 0.1 Ci of [ 125 I]iodopalmitoyl-CoA in 20 mM Tris-HCl, pH 7.4, 1 mM DTT buffer (final concentrations were achieved by using a 5ϫ buffer stock solution) for 30 min at 25°C. The reaction was then stopped by the addition of 5ϫ SDS-PAGE loading buffer (10) and processed as described above.
Fatty acylation assays of CPS 1 in the presence of various substrates, cofactors, and reagents were carried out in a final volume of 50 l using Reactions were stopped and processed as described above. For reactions with FSBA or NEM, CPS 1 was previously dialyzed for 16 h at 4°C to remove DTT. FSBA was dissolved in dimethylformamide (21).
Small Scale Purification of p165-All manipulations and purification steps were carried out at 4°C. Mitochondria were prepared from 30 g of rat liver as described above using the Percoll gradient method. Purified rat liver mitochondria were solubilized in 20 mM Tris-HCl, pH 7.4, 0.1% Nonidet P-40, 1 mM PMSF, and 1 mM DTT for 30 min. The homogenate was centrifuged at 12,000 ϫ g for 30 min in a Beckman JA20 rotor. The resulting supernatant was adjusted to 20 mM K-MES (MES neutralized with KOH), pH 6.0, by addition of a 10ϫ concentrated 200 mM K-MES, pH 6.0, buffer. The supernatant was dialyzed against S buffer (20 mM K-MES, 1 mM DTT, and 1 mM PMSF).
The dialyzed protein solution containing 80 mg of protein was loaded at a flow rate of 0.5 ml/min onto a prepacked 1-ml SP-Sepharose HP column, equilibrated with S buffer. By using FPLC, a linear NaCl gradient (0 -500 mM) was applied to the column over 17 ml, and 1-ml fractions were collected. Aliquots of chromatographic fractions were labeled with the [ 125 I]iodopalmitoyl-CoA as described above and separated on a 12% SDS-PAGE. Fractions containing the radiolabeled p165-kDa protein were pooled.
The SP-Sepharose protein pool was then equilibrated with TD buffer (20 mM Tris-HCl, pH 8.0, 1 mM DTT) by dialysis, overnight at 4°C, and then applied to a Mono Q HR 5/5 column at a flow rate of 0.5 ml/min. By using FPLC, the column was washed with 5 column volumes of TD buffer and then eluted stepwise with a 2-ml gradient (0 -80 mM NaCl in TD buffer) followed by a 10-ml gradient (80 -500 mM NaCl in TD buffer). One-ml fractions were collected. Aliquots of chromatographic fractions were radiolabeled as described above, and fractions containing the radiolabeled p165-kDa protein were pooled.
The protein pool from the Mono Q column was loaded onto a 0.5ϫ 5-cm column of Macro-Prep ceramic hydroxyapatite used on FPLC and equilibrated in 10 mM K 2 HPO 4 , pH 6.8. The column was eluted with a 16-ml linear gradient (0 -300 mM K 2 HPO 4 ). Aliquots of chromatographic fractions were analyzed as described above for p165 content. 4 M ammonium sulfate solution was added to the protein pool from the hydroxyapatite column to yield a 2 M ammonium sulfate solution and applied at a flow rate of 0.5 ml/min using the FPLC to a phenyl-Superose HR 5/5 column equilibrated with 2 M ammonium sulfate in 100 mM K 2 HPO 4 buffer, pH 7.0. The column was eluted with a 15-ml linear gradient of 2.0 -0 M ammonium sulfate in 100 mM K 2 HPO 4 , pH 7.0. One-ml fractions were collected. Aliquots of chromatographic fractions were analyzed as described above for p165 content. All manipulations were carried out at 4°C.
Edman Degradation of p165-An apparently homogeneous sample of p165 was analyzed by the Alberta Peptide Institute (University of Alberta) using Edman degradation with a HP-G-1005A (Hewlett-Packard) protein sequencing system using version 3.0 chemistry.
Large Scale Purification of Carbamoyl-phosphate Synthetase 1-Rat liver carbamoyl-phosphate synthetase 1 was isolated at 4°C as described previously (21,22) with the following changes. Crude mitochondrial fractions were prepared from livers of 10 male Sprague-Dawley rats (300 -350 g) by differential centrifugation (18,23). The final pellet enriched in mitochondria was suspended in IM buffer (250 mM mannitol, 5 mM HEPES pH 7.4, 0.5 mM EDTA, 0.1% albumin (essentially fatty acid free), 1 mM DTT, 1 mM PMSF, 20 g/ml leupeptin) and stored frozen at Ϫ80°C in aliquots containing 1 g of total mitochondrial protein. For each preparation of CPS 1, 1 g of crude mitochondria aliquot was thawed and swollen in 90 ml of hypotonic buffer (1 mM PMSF, 1 mM DTT, 20 g/ml leupeptin) for 30 min on ice. The suspension was then sonicated twice for 2 min at setting 2 on a Heat Systems sonicator. Nine ml of 200 mM Tris, pH 7.4, was then added to the solution, and the lysed mitochondria were centrifuged at 100,000 ϫ g for 60 min in a Beckman Ti45 rotor. The pellet was discarded.
Solid ammonium sulfate was added to the supernatant to give a 35% saturated solution; the extract was stirred for 30 min at 4°C; the suspension was centrifuged at 20,500 ϫ g for 20 min, and the pellet was discarded. Solid ammonium sulfate was added to the supernatant to make it an 80% saturated solution; the extract was stirred for 30 min at 4°C, and the suspension was centrifuged as above. The supernatant was discarded, and the pellet was resuspended in 3 ml of 0.3 M K 2 HPO 4 , pH 7.6, buffer and dialyzed for 2 h against K 2 HPO 4 buffer (20 mM K 2 HPO 4 , pH 7.6, 1 mM DTT, 1 mM PMSF). The solution was loaded either onto a Hiprep Sephacryl S-200 26/60 or Ultrogel AcA 34 packed in a XK 16/100 column equilibrated with 0.3 M K 2 HPO 4 , pH 7.6, 0.5 mM EDTA, 0.5 mM EGTA, 1 mM DTT, 1 mM PMSF, and 20 g/ml leupeptin and eluted with the same buffer at 1 ml/min; 4-ml fractions were collected. Aliquots of chromatographic fractions were analyzed by 12% SDS-PAGE as described above.
The CPS 1 pool from the gel filtration column was then equilibrated with TD buffer (20 mM Tris-HCl, pH 8.0, 1 mM DTT) by dialysis overnight at 4°C and applied to a Mono Q HR 5/5 column. The column was washed with 5 column volumes of TD buffer (20 mM Tris-HCl, pH 8.5, 1 mM DTT) and eluted with a 2-ml gradient of 0 -80 mM, followed by a 10-ml gradient of 80 -300 mM, followed by a 2-ml gradient from 300 to 500 mM NaCl in TD buffer. One-ml fractions were collected at a flow rate of 0.5 ml/min. Aliquots of chromatographic fractions were analyzed by 12% SDS-PAGE as described above.
Neutral Hydroxylamine Treatment, Fatty Acid Analysis, and NEM Treatment-Highly purified CPS 1 samples were radiolabeled with [ 125 I]iodopalmitoyl-CoA as described above and separated in different lanes on 12% SDS-PAGE. Electrophoresed samples contained in different lanes of the gel were soaked in either 1 M hydroxylamine, pH 7.0, or 1 M Tris, pH 7.0, for 48 h prior to gel drying. Residual radiolabel incorporation into proteins was visualized by autoradiography.
Alternatively, following electrophoresis, the wet gel was frozen at Ϫ80°C and exposed to x-ray film at Ϫ80°C for 2 days. The gel slice containing the radiolabeled protein was hydrolyzed with 1 M NaOH and extracted with chloroform (24) with the exception that the residue was analyzed by TLC on a 250 M/Silica Gel 60 plate (VWR Scientific) and that the TLC plate was developed with a mobile phase containing water:glacial acetic acid:acetonitrile (1:1.75:1.75. v/v/v). Following air drying, PhosphorImager analysis was performed.
Purified CPS 1 (1-2 g) was preincubated for 30 min in the absence or presence of 1, 2, 5, or 10 mM NEM diluted in 100 mM K-MOPS, pH 7.0, and then radiolabeled with 0.8 Ci of [ 125 I]iodopalmitoyl-CoA for 30 min at 25°C in a final volume of 50 l. Reactions were stopped by the addition of 5ϫ SDS-PAGE loading buffer and processed as above.
Enzymatic Activity and Protein Determination-The enzymatic activity was measured by the pyruvate kinase/lactate dehydrogenase coupled assay (25). Briefly, 20 g of CPS 1 was incubated with the reaction mixture in a final volume of 1 ml at 25°C. This mixture contained 50 mM Tris-HCl, pH 7.6, 50 mM KHCO 3 Ϫ , 35 mM SO 4 (NH 4 ) 2 , 15 mM SO 4 Mg, 10 mM ATP, 5 mM phosphoenolpyruvate, 0.5 mM NADH, 10 mM acetylglutamate, 33 units/ml pyruvate kinase, and 67.5 units/ml lactate dehydrogenase. The activity was expressed as the amount of carbamoyl phosphate produced, assuming a stoichiometry of 1 mol of carbamoyl phosphate produced per 2 mol of ADP released. CPS 1 was also incubated at 25°C in the absence or presence of palmitoyl-CoA (5, 10, 20, 50, and 100 M) or 50 M FSBA with or without 2 mM AGA in 100 l of 50 mM Tris-HCl, pH 7.6, buffer and assayed for enzymatic activity in a 1-ml final volume at different times as described above. For kinetic measurements, CPS 1 was incubated with 0, 20, or 50 M of palmitoyl-CoA for 60 min in a 100-l final volume and then assayed in the presence of different concentrations of AGA (0, 0.01, 0.05, 0.1, 0.5, and 1 mM), NH 4 ϩ (0, 0.2, 0.5, 1, 2, and 10 mM), HCO 3 Ϫ (0, 0.66, 1, 2.5, 5, 10, and 20 mM), and ATP (0, 1, 1.75, 2.5, 5, and 10 mM). While varying the concentrations of these substrates, the concentrations of other reagents in the CPS 1 assay were kept at the saturating concentrations indicated in the standard CPS 1 enzymatic assay mixture described above. In addition, CPS 1 was incubated with 0, 20, or 50 M FSBA for 60 min in a final volume of 100 l of 50 mM Tris-HCl, pH 7.6, buffer prior to the spectrophotometric assay in the presence of different concentrations of ATP described above. Data show the average of at least 4 independent experiments.
Substrate Protection Assay and Specificity-CPS 1 was preincubated in the presence or absence of either 2 mM AGA, 1.7 mM MgATP, or both for 60 min in 50 l of 50 mM Tris-HCl, pH 7.6, and then either 50 M palmitoyl-CoA or MgFSBA were subsequently added to the mixture and incubated for another 60 min in a total volume of 100 l. The mixture was then assayed for residual enzymatic activity in a 1-ml reaction volume. In all assays, the concentration of MgCl 2 was kept constant at 10 mM. For specificity assays, CPS 1 was incubated with 50 M coenzyme A, 50 M acetyl-CoA A, 50 M palmitoyl-CoA, or 50 M palmitate for 60 min and then assayed for enzymatic activity. All incubations were carried out at room temperature (ϳ25°C). Protein concentration was measured (26) using bovine serum albumin as a standard.

Protein Fatty Acylation Patterns Are Similar in Various
Types of Rat Mitochondria-Several proteins were shown previously to incorporate radiolabeled fatty acids in mitochondria isolated from rat liver and COS-7 cells (10,14). To determine whether similar mitochondrial fatty-acylated proteins exist in mitochondria isolated from different tissue origins, isolated mitochondria were incubated with [ 125 I]iodopalmitate in a mixture containing all of the necessary cofactors for mitochondrial fatty acid import. Radiolabel incorporation into proteins after electrophoresis and autoradiography was taken as evidence these proteins have been acylated by the [ 125 I]iodopalmitate analog. Furthermore, soaking of gel slices containing radiolabeled rat liver mitochondrial proteins in 1 M neutral hydroxylamine prior to autoradiography removed more than 95% of the label from the majority of proteins. 2 This suggested that the radiolabel was incorporated into these proteins via a thioester bond. We show that there are many proteins that could be labeled in this manner in mitochondria isolated from rat heart, liver, brain, kidney, and leg gastrocnemius muscle. Radiolabel incorporation into mitochondrial proteins appeared to be similar among the tissues examined (Fig. 1, A and B), with only a few tissue-specific acylated proteins. For the most part, acylated proteins analyzed by SDS-PAGE were concentrated within two groups based on their molecular masses. The first group ranged from 30 to 34 kDa, and the second ranged from 45 to 75 kDa. Of interest was an apparently liver-specific abundant 165-kDa protein. In addition, the overall fatty acylation levels and the number of fatty-acylated proteins were highest in the liver and lowest in brain mitochondria. Similar results were also observed when mitochondria were prepared from mouse tissues (data not shown).

Localization of Radiolabeled Proteins in Rat Liver
Submitochondrial Compartments-To establish the localization of mitochondrial fatty-acylated proteins, rat liver submitochondrial compartments were prepared (19,27) and then labeled with [ 125 I]iodopalmitoyl-CoA. Radiolabel incorporation into equivalent amounts of proteins from various compartments was visualized by autoradiography after SDS-PAGE analysis. A variety of marker enzymes were used to assess the purity of the mitochondrial subfractions. These were citrate synthase (matrix) (28), succinate-cytochrome c reductase (inner membrane) (29), and NADH cytochrome c reductase (outer membrane) (30). According to the marker assays, all submitochondrial fractions prepared were of greater than 90% purity. 2 We found that radiolabeled proteins were present mostly in the mitochondrial matrix fraction but were also observed in the inner mitochondrial membrane fraction (Fig. 2). The appearance of labeled bands in the inner mitochondrial fraction may also reflect contamination by matrix proteins because the main labeled bands of the matrix are those observed with much less intensity in the inner membrane fraction. Overall, our fractionation results confirmed those reported by Stucki et al. (14). Little to no radiolabeled protein was detected in the outer mitochondrial membrane or the intermembrane mitochondrial space fractions (Fig. 2).
Purification of p165 from Rat Liver Mitochondria-To determine whether mitochondrial protein fatty acylation is a general mechanism of metabolic regulation, the role of protein fatty acylation must be investigated for different fatty-acylated mitochondrial proteins. To do so, purification and identification of new fatty-acylated mitochondrial proteins was needed. In our analysis, we found an abundant 165-kDa fatty-acylated protein present in rat liver mitochondria. By using incorporation of [ 125 I]iodopalmitate into p165 as an assay, we purified this 165-kDa protein to apparent homogeneity from rat liver mitochondria (Fig. 3). Edman degradation sequence analysis of the protein was obtained for the N-terminal 15 amino acids. A search of the translated GenBank TM data base revealed 100% identity with the N-terminal sequence of the mature (processed) rat liver CPS 1. Final preparations of CPS 1 were more than 99% pure with specific activities from 0.30 to 0.46 mol of carbamoyl phosphate produced per min per mg of CPS 1 at 25°C. These specific activities are in agreement with that of Lusty (31) of 4.0 mol/mg/10 min at 25°C but lower than those measured at 37°C by various sources (25,(32)(33)(34)(35). These varied from 1.87 to 5 mol/min/mg of CPS 1 protein at 37°C. The difference in assay temperatures may account for a significant proportion of these apparent differences in specific activities.
Hydroxylamine Treatment, NEM Treatment, and Fatty Acid Analysis of Rat Liver CPS 1-To investigate the chemical nature of the bond linking the radiolabel to CPS 1, we incubated crushed polyacrylamide gel slices containing CPS 1 radiolabeled with [ 125 I]iodopalmitoyl-CoA with 1 M hydroxylamine, pH 7.0, for 48 h. This treatment, known to cleave thioester and not ester (oxyester) bonds (36), removed 87 Ϯ 6% (n ϭ 7) of the radiolabel in comparison to the control treated with 1 M Tris, pH 7.0 (Fig. 4, A and B). This result suggests that the radiolabel is bound to CPS 1 via a hydroxylamine-sensitive thioester bond and likely identified a cysteine residue as the fatty acyl acceptor.
It was demonstrated previously that the pretreatment of bovine liver MMSDH (11) and glutamate dehydrogenase (10) with N-ethylmaleimide (NEM), which alkylates the sulfhydryl group of cysteine residues, blocked covalent acylation of these proteins. Therefore, to confirm further whether fatty acylation of CPS 1 occurred on a cysteine residue, CPS 1 was preincubated with increasing concentrations of NEM (0 (control), 1, 2, 5, and 10 mM) prior to incubation with the [ 125 I]iodopalmitoyl-CoA analog (Fig. 4, C and D). We observed an inhibition of [ 125 I]iodopalmitate incorporation into CPS 1 in samples pretreated with NEM at all concentrations (Fig. 4, C and D). The extent of inhibition of [ 125 I]iodopalmitate incorporation into CPS 1 was greater than 95% (n ϭ 4). The residual labeling shown in Fig. 4D may be due to nonspecific hydrophobic interactions between radiolabel and CPS 1 protein. Overall, our results confirm that the fatty acylation of CPS 1 occurs on a cysteine residue(s).
To confirm the nature of the radiolabel bound to CPS 1, we hydrolyzed gel slices containing radiolabeled CPS 1 with 1 M NaOH and analyzed the extract by TLC. As shown in Fig. 5, the radioactive label present in the hydrolysate co-migrated with the [ 125 I]iodopalmitic acid used as a standard, suggesting that the unmodified [ 125 I]iodopalmitate was the chemical entity covalently bound to CPS 1.
Inhibition of CPS 1 Catalytic Activity by Palmitoyl-CoA and FSBA-To investigate the effect of palmitoyl-CoA on CPS 1, CPS 1 enzymatic activity assays were performed in the presence of increasing concentrations of palmitoyl-CoA for variable periods (Fig. 6). We found time-and concentration-dependent loss of CPS 1 activity and showed that concentrations of palmitoyl-CoA well within physiological mitochondrial concentrations (up to 230 M) (37) can inhibit CPS 1 catalytic activity. In more detail, incubation of CPS 1 with 100 M palmitoyl-CoA for 60 min inhibited 89 Ϯ 2% (n ϭ 4) of CPS 1 catalytic activity, whereas incubation of CPS 1 with as little as 5 M palmitoyl-CoA resulted in a 22 Ϯ 3% (n ϭ 4) reduction in activity. In comparison, preincubation of CPS 1 with 50 M FSBA, a known ATP analog that selectively reacts with CPS 1 active site cysteine residues (38) in the presence or absence of the allosteric activator AGA, resulted in inactivation of 94 Ϯ 2 (n ϭ 4) and 67 Ϯ 2% (n ϭ 4) of CPS 1 activity, respectively. The efficiency of the coupled assay measuring CPS 1 activity was not impeded by the presence of palmitoyl-CoA in the concentration range tested (data not shown).
CPS 1 Kinetic Analysis-CPS 1 was inactivated by incubation with palmitoyl-CoA. The inactivation was reflected in kinetic assays at a variable concentration of the substrates and AGA by a decrease in V m without substantial effect on K m values (Fig. 7). The concentration of palmitoyl-CoA that, when incubated with the enzyme for 60 min, decreases the V m to one-half is defined here for brevity as K i value and was found to be 18, 15, and 24 M when the varied substrates were MgATP, NH 4 ϩ , or HCO 3 Ϫ respectively. K m for MgATP extrapolated from Fig. 7, A and E, was 1.0 mM in both cases, which is in accordance with that of Lusty (31). K m for AGA was 0.38 mM (Fig. 7B), for NH 4 ϩ was 0.21 mM (Fig. 7C), and for HCO 3 Ϫ was 1.9 mM (Fig.  7D), and values were within the range reported by other groups (31).
The rate of inactivation of CPS 1 by 50 M palmitoyl-CoA in the absence of AGA was faster than that observed with 50 M FSBA in the absence of AGA. This suggests that palmitoyl-CoA is a better inhibitor than FSBA under these conditions. This is confirmed by our calculated inhibitory constants (K i ) as the K i determined for FSBA was 131 Ϯ 20 M (in the absence of AGA) and for palmitoyl-CoA the K i was 19 ؎ 5 M.
Palmitoyl-CoA Specificity and Protection of CPS 1 Inactivation by Palmitoyl-CoA or FSBA by Various Substrates and Cofactors-Incubation of CPS 1 with either 50 M coenzyme A, 50 M acetyl-CoA, or 50 M palmitate for 60 min did not modify CPS 1 activity, whereas a similar incubation with 50 M palmitoyl-CoA showed a 63% reduction in CPS 1 activity, suggesting that the inhibitory effect of palmitoyl-CoA requires a longchain fatty acid activated in its CoA derivative and does not simply reflect the high reactivity of some cysteine residue(s) (Fig. 8A). The remaining activity after we preincubated CPS 1 with palmitoyl-CoA in the absence of other ligands or in presence of AGA was ϳ30% ((31 Ϯ 6% (ϩ AGA) versus 31 Ϯ 3% (Ϫ AGA) of control; n ϭ 6), whereas the remaining activity found in the presence of AGA and MgATP was 76 Ϯ 6% (n ϭ 6). The remaining activity after we preincubated CPS 1 with FSBA in the presence AGA and MgATP was 36 Ϯ 2% (n ϭ 4) of the controls, respectively. Whereas AGA by itself did not protect against inactivation of CPS 1 by palmitoyl-CoA, AGA stimulated the FSBA inactivation rate of CPS 1 (3.3 Ϯ 3% (ϩ AGA) versus 21 Ϯ 4% (Ϫ AGA) of control; n ϭ 4).
Fatty Acylation of CPS 1 in the Presence of Substrates, Allosteric Activator, and FSBA-CPS 1 catalyzes the three-step conversion of HCO 3 Ϫ , NH 4 ϩ , and ATP into carbamoyl phosphate in the presence of the allosteric activator N-acetylglutamate FIG. 5. Thin layer chromatography analysis of fatty acid hydrolyzed from radiolabeled p165. Gel slice containing the radiolabeled CPS 1 was hydrolyzed as described under "Experimental Procedures" and extracted with chloroform. The chloroform extract was reduced to dryness, and the residue (fatty acid) was analyzed by TLC beside an [ 125 I]iodopalmitate standard (IC16). Following air drying, PhosphorImager analysis was performed.
(AGA). To test whether formation of complexes between the enzyme and ATP or its allosteric activator AGA would interfere with the fatty acylation process of CPS 1, we preincubated CPS 1 with saturating concentrations of reactants (2 mM AGA, 1.7 mM MgATP) prior to addition of [ 125 I]iodopalmitoyl-CoA (Fig.  9). When CPS 1 was first incubated individually with reactants and then radiolabeled with the [ 125 I]iodopalmitoyl-CoA, AGA and ATP treatment reduced CPS 1 acylation by 55 Ϯ 2 and 39 Ϯ 2% (n ϭ 4), thus leaving 45 Ϯ 2 and 61 Ϯ 2% non-acylated enzyme, respectively. Addition of AGA and ATP together leads to a near complete prevention of fatty acylation of CPS 1 (86 Ϯ 4% reduction).
To test the possibility that CPS 1 may be acylated on its active site cysteine residue(s), we preincubated CPS 1 with FSBA in the presence or absence of AGA. Preincubation of CPS 1 with FSBA and AGA, which leads to a selective inactivation of two essential cysteine residues (38), almost completely prevented the acylation of CPS 1 (greater than 95% inhibition). This exciting result suggested that fatty acylation of CPS 1 was very likely occurring at the active site of the enzyme. In the absence of AGA, the extent of FSBA protection against acylation was not as strong (60% inhibition of acylation), suggesting once again that binding of AGA facilitated the binding of the ATP analog FSBA to CPS 1 active site and thus increased the protection of some active site-related cysteine residues against acylation observed when both are present. Preincubation with 50 M palmitoyl-CoA for 60 min, prior to the addition of radiolabeled [ 125 I]iodopalmitoyl-CoA, resulted in a 63 Ϯ 2% (n ϭ 4) reduction in labeling in comparison to the control. Interestingly, this result corroborated the kinetic results obtained in Fig. 6, in which we showed that a 60-min preincubation of CPS 1 with 50 M palmitoyl-CoA resulted in 37% remaining CPS 1 activity.
Reversibility Assay-To test if palmitoyl-CoA binds to CPS 1 in a reversible manner, we labeled CPS 1 to saturation with [ 125 I]iodopalmitoyl-CoA and then incubated it for 1 h in the presence of different concentrations of palmitoyl-CoA. As shown in Fig. 10, the label on CPS 1 could not be displaced even in the presence of 200 M palmitoyl-CoA. This result indicates that the attached palmitate binds in an apparent irreversible manner during this period. DISCUSSION To characterize the general occurrence of mitochondrial protein fatty acylation, we analyzed the distribution of fatty-acylated proteins in mitochondria isolated from different tissues. We found that electrophoretic patterns of fatty-acylated proteins in mitochondria isolated from different rat and mouse tissues were quite similar but interestingly quite different in mitochondria isolated from transformed versus normal hepatocytes. Although the electrophoretic distribution of fatty-acylated proteins was similar in different tissues, the acylation levels of bands of similar apparent molecular weights varied quite significantly. This variation could be explained by differences in metabolism in different tissues as well as differences in tissue-specific expression levels of given proteins or perhaps by a combination of both. The fact that electrophoretic protein palmitoylation patterns were quite different in transformed and normal hepatocytes (data not shown) further supports these possibilities because transformed cells generally have altered metabolic and protein profiles and often display increased rates of glucose uptake and glycolysis (39). Further analysis of fatty-acylated protein content in mitochondrial subcompartments indicated that the majority of acylated proteins reside in the matrix with fewer localized to the inner membrane. This finding corroborates similar results obtained by Stucki et al. (14) who also showed that the majority of acylated mitochondrial proteins were in these fractions. Mitochondrial protein fatty acylation has been suggested to play a role in regulation of mitochondrial metabolism (10,11). The localization of acylated proteins within mitochondria is consistent with this possibility because most acylated proteins are found in areas of intense mitochondrial metabolic activity (12).
In rat liver mitochondria, a major radiolabeled acylated 165-kDa protein was isolated, and based upon amino acid sequences obtained from this purified acylated protein, we identified it as rat liver mitochondrial CPS 1. CPS 1 is involved in amino acid catabolism where it catalyzes the first and ratelimiting step of urea synthesis (15)(16)(17).
We feel confident that fatty acylation of CPS 1 is likely occurring on at least one of the CPS 1 active site cysteine residues because of the following: 1) inhibition of CPS 1 by palmitoyl-CoA was specific and dependent on time and concentration and corresponded to an irreversible inactivation of the enzyme; 2) fatty acylation of CPS 1 was largely prevented by preincubations with either NEM or FSBA/AGA; 3) covalently bound fatty acid was removed from CPS 1 using neutral hydroxylamine treatment; and 4) preincubation of CPS 1 with a variety of substrates and its allosteric activator reduced CPS 1 acylation to various extents. This is also corroborated by the similar levels of protection of CPS 1 against inactivation and fatty acylation afforded by AGA or MgATP that linked inactivation and fatty acylation together. Thus our results suggest that the activity of CPS 1 is negatively regulated by active site fatty acylation. This would make CPS 1 the second enzyme to be regulated and modified in such manner, since MMSDH was also previously shown to be acylated on its active cysteine residue, Cys-319 (10). Fatty acylation of CPS 1 active site residues (Cys-1327, Cys-1337, or both) could thus potentially account for the observed inhibition of CPS 1 at concentrations of palmitoyl-CoA well within the physiological range of mitochondrial palmitoyl-CoA levels (below 230 M) (37). Alternatively, a conformational change could occur in CPS 1 in the presence of AGA and/or MgATP and MgFSBA which could prevent the fatty acylation of a specific acyl-acceptor cysteine residue(s). The palmitoyl-CoA inhibition of CPS 1 as part of an enzymatic complex with aspartate aminotransferase and/or glutamate dehydrogenase has been shown to contain specific and nonspecific components and to occur at a concentration significantly below the critical micellar concentration of palmitoyl-CoA (40). Critical micellar concentrations for palmitoyl-CoA have been established by a variety of methods to be no lower than 30 -60 M (41). These values are more than double that of our calculated K i of 19 M for palmitoyl-CoA inhibition of CPS 1. As such, inhibition of CPS 1 by palmitoyl-CoA is believed to be specific and not due to detergent-like properties of longchain acyl-CoA molecules that could occur at concentrations above the critical micellar concentration.
The fatty acylation of bovine liver MMSDH was apparently spontaneous in vitro. The fact that recombinant bovine liver MMSDH prepared from Escherichia coli lysates was readily fatty-acylated in the presence of [I 125 ]iodopalmitoyl-CoA suggests that a mammalian mitochondrial acyltransferase activity may not be necessary for fatty acylation of MMSDH. 3 Preparations of CPS 1 used in our fatty acylation assays were greater than 99% homogeneous, and unless a minute amount of a given transferase is present in our CPS 1 preparations, we believe fatty acylation of CPS 1 also appears to be spontaneous. In addition, inhibition of CPS 1 activity by palmitoyl-CoA required the CoA-activated form of long-chain fatty acids. Furthermore, fatty acylation appeared to be irreversible in the presence of an excess of palmitoyl-CoA in vitro using highly purified enzyme. Perhaps inside the mitochondria a protein fatty acylthioesterase exists that can remove the fatty acid moieties bound to CPS 1 and allows its reactivation.
In diluted solutions, inhibition of CPS 1 by palmitoyl-CoA could appear to be paradoxical as our results would predict that the enzyme should be almost completely inhibited under physiological conditions. Indeed, the liver mitochondrial long-chain acyl-CoA concentration in the fed state has been estimated at 230 M (37) and is postulated to increase as much as 3-fold (up to ϳ600 M) during starvation. This increase corresponds to an augmentation of the serum-free fatty acid concentration from 0.66 mM (fed) to 1.60 mM during starvation (42). Nonetheless, CPS 1 is a very abundant mitochondrial protein; it represents 15-26% of mitochondrial matrix protein and its concentration is estimated at 0.4 to 1.5 mM (43). As such, in the fasted state, the concentration of long-chain acyl-CoAs in the mitochondria would be sufficient to inhibit a large proportion of CPS 1 activity but potentially not all of it. This observation could thus explain the partial but not total reduction in urea secretion observed during starvation (42,44,45).
In starvation, the reduction of amino acid utilization is essential for survival because it spares vitally needed proteins. This nitrogen sparing strategy is accompanied by a 60 -80% diminution in urinary nitrogen secretion, and most of this reduction comes from a 75-95% reduction in urea secretion (42,44,45). During starvation, palmitoyl-CoA-dependent inhibition of CPS 1 could account for part of the reduction in urea cycle activity. Thus, the switch responsible for a reduction in amino acid oxidation and urea synthesis activity could be mediated by fatty acylation and inhibition of a few key metabolic enzymes (namely CPS 1, MMSDH, and perhaps even glutamate dehydrogenase). Fatty acylation of these enzymes could contribute 3 L. G. Berthiaume, unpublished observations. to the nitrogen sparing effect observed during starvation.
As such, concerted inhibition of CPS 1 and glutamate dehydrogenase by palmitoyl-CoA would predict an accumulation of glutamate and ammonia in the liver mitochondria in the fasted state. These compounds could then be exported by the liver in the form of glutamine or glutamate ϩ ammonia. These predictions are physiologically consistent with the established fact that the liver has a net production of glutamate (46) and glutamine (47) during starvation. In more detail, the excess glutamate or ammonia produced in the fasted state (e.g. in periportal hepatocytes) is known to be transformed locally into glutamine by perivenous hepatocytes that are rich in glutamine synthetase (48). Alternatively, glutamine synthesis could occur at a distal site in skeletal muscles (12,43,49).
An interesting concept emerging from this study is that protein fatty acylation may be a key player in the regulation of metabolic cross-talk between several catabolic pathways in the mitochondria including fatty acid oxidation, amino acid degradation, and the urea cycle. Some evidences supporting this possibility are as follows. 1) Ammonia has been shown to inhibit fatty acid oxidation, whereas no effect was observed with the oxidation of succinate or malate and glutamate (50,51). 2) Patients suffering from Reye's syndrome often exhibit hyperammonemia and display microvesicular fatty metamorphosis of the liver (52). 3) Patients or animals with genetic alteration of enzymes involved in the ␤-oxidation cascade (e.g. mediumchain acyl-CoA dehydrogenases or long-chain acyl-CoA dehydrogenases) often suffer from hyperammonemia (53). 4) Excess fatty acid oxidation has been documented to decrease the oxidation of deaminated branched-chain amino acids (e.g. Leu, Ile, and Val) (44,54).
Furthermore, accumulation of long-chain acyl-CoAs could play a role in some pathophysiological states such as those found in patients suffering from very long-chain acyl-CoA dehydrogenases or long-chain acyl-CoA dehydrogenase deficiencies and even in ischemic heart injury where mitochondrial concentrations of palmitoyl-CoA have been reported to be as high as 1.0 mM (55). In addition to oxygen depletion, such high concentrations of palmitoyl-CoA in the ischemic heart could thus inhibit several catabolic pathways and hinder energy production.