The peroxisome proliferator-induced cytosolic type I acyl-CoA thioesterase (CTE-I) is a serine-histidine-aspartic acid alpha /beta hydrolase.

Long-chain acyl-CoA thioesterases hydrolyze long-chain acyl-CoAs to the corresponding free fatty acid and CoASH and may therefore play important roles in regulation of lipid metabolism. We have recently cloned four members of a highly conserved acyl-CoA thioesterase multigene family expressed in cytosol (CTE-I), mitochondria (MTE-I), and peroxisomes (PTE-Ia and -Ib), all of which are regulated via the peroxisome proliferator-activated receptor alpha (Hunt, M. C., Nousiainen, S. E. B., Huttunen, M. K., Orii, K. E., Svensson, L. T., and Alexson, S. E. H. (1999) J. Biol. Chem. 274, 34317-34326). Sequence comparison revealed the presence of putative active-site serine motifs (GXSXG) in all four acyl-CoA thioesterases. In the present study we have expressed CTE-I in Escherichia coli and characterized the recombinant protein with respect to sensitivity to various amino acid reactive compounds. The recombinant CTE-I was inhibited by phenylmethylsulfonyl fluoride and diethyl pyrocarbonate, suggesting the involvement of serine and histidine residues for the activity. Extensive sequence analysis pinpointed Ser(232), Asp(324), and His(358) as the likely components of a catalytic triad, and site-directed mutagenesis verified the importance of these residues for the catalytic activity. A S232C mutant retained about 2% of the wild type activity and incubation with (14)C-palmitoyl-CoA strongly labeled this mutant protein, in contrast to wild-type enzyme, indicating that deacylation of the acyl-enzyme intermediate becomes rate-limiting in this mutant protein. These data are discussed in relation to the structure/function of acyl-CoA thioesterases versus acyltransferases. Furthermore, kinetic characterization of recombinant CTE-I showed that this enzyme appears to be a true acyl-CoA thioesterase being highly specific for C(12)-C(20) acyl-CoAs.

Long-chain fatty acids entering cells are rapidly esterified to the corresponding CoA esters prior to degradation via ␤-oxidation in mitochondria or peroxisomes or esterification into triacylglycerol, phospholipids, cholesterol esters, or pro-teins (for review, see Ref. 1). However, long-chain acyl-CoAs have also been shown to exert a number of regulatory functions such as allosteric regulators of various enzyme activities and ion channels, membrane fusion, cell signaling, etc. (2)(3)(4). Recently free fatty acids were shown to activate and bind as ligands to the peroxisome proliferator-activated receptors (PPARs) 1 (5)(6)(7), and long-chain acyl-CoAs bind as ligands to the hepatocyte nuclear factor 4␣ (HNF-4␣) (8). Interestingly, long-chain acyl-CoA esters act as antagonists for the PPARs (9,10), suggesting that free fatty acid/acyl-CoA levels may be important regulators of gene transcription. Thus, the diversity of cellular processes involving activated fatty acyl-CoA esters underscores the central role of acyl-CoA thioesters in lipid metabolism.
Acyl-CoA thioesterases are enzymes that catalyze the hydrolysis of acyl-CoA thioesters to the corresponding free fatty acid and CoASH, and are therefore likely to regulate lipid metabolism as well as a number of cellular processes (for review, see Ref. 11). This enzyme activity is widely distributed among organisms and shows multiple subcellular localizations. The enzyme activity is found in peroxisomes, mitochondria, cytosol, and endoplasmic reticulum and can be induced in liver by feeding peroxisome proliferators (12-15). These peroxisome proliferators (e.g. clofibrate) are potent modulators of lipid metabolism, and treatment of rodents with these compounds results in hypolipidemic effects and induction of fatty acid-catabolizing enzymes. The acyl-CoA thioesterase activity induced in mouse liver cytosol by treatment with these peroxisome proliferators is mainly because of induction of an ϳ45-kDa protein called CTE-I (16,17). CTE-I is closely related to an acyl-CoA thioesterase expressed in mitochondria (MTE-I) (15,18). We and others (17)(18)(19)(20)(21) have recently isolated and cloned type I long-chain acyl-CoA thioesterases from rat and mouse liver mitochondria and cytosol. Cloning of the corresponding genes from mouse revealed a unique, highly conserved gene family encoding acyl-CoA thioesterases with putative expression in cytosol (CTE-I), mitochondria (MTE-I), and peroxisomes (PTE-Ia and -Ib), all of which can be regulated by feeding peroxisome proliferators and by fasting (22,23). Based on sequence similarity, these enzymes belong to a unique family of proteins that show sequence homology only to a bile acid-CoA:amino acid N-acyltransferase (BAAT), which catalyzes the conjugation of bile acids to glycine or taurine. Still, however, the functions of these acyl-CoA thioesterases are largely unknown.
So far, the best characterized thioesterases are thioesterases I and II, which are involved in chain termination of fatty acid synthesis by cleavage of the newly synthesized fatty acid from the fatty acid synthase complex (24,25). These thioesterases also hydrolyze acyl-CoAs and, similar to serine proteases, lipases, and cholinesterases, they have been proposed to contain a Ser-His-Asp/Glu catalytic triad, which catalyzes the ester hydrolysis. A nucleophilic serine is thought to be part of a charge-relay system along with a conserved histidine and an aspartic or glutamic acid residue (26 -28). Previous site-directed mutagenesis experiments of thioesterase II have identified two of the residues that are involved in catalysis: a serine residue found within a so-called esterase consensus sequence, GXSXG (29), and a histidine residue present within a GXH motif (27,28). The GXSXG motif seems to be conserved among esterases, and in mouse CTE-I Ser 232 is found within a GXSXG motif. To gain further insight into the catalytic mechanism of the type I acyl-CoA thioesterases, we have characterized recombinant CTE-I with respect to chemical modifiers and used site-directed mutagenesis to identify the catalytic amino acid residues. Our results demonstrate that CTE-I is a member of the ␣/␤ hydrolase superfamily of esterases and that Ser 232 , Asp 324 , and His 358 constitute the catalytic triad. Interestingly, the active-site serine is apparently replaced by a cysteine in the BAAT enzyme, which may mediate the different activities of these enzymes.

EXPERIMENTAL PROCEDURES
Identification of Putative Candidates for the Catalytic Triad of Mouse CTE-I-Fasta3 (30) (European Bioinformatics Institute server) was used for the generation of a multiple sequence alignment between mouse CTE-I and all of its homologous sequences in the GenBank TM . Each of the homologous sequences detected by Fasta3 was used independently as a query to search for homologous proteins in the threedimensional structure protein data bank at Brookhaven. Secondary structure elements in mouse CTE-I were predicted with the PHD Pre-dictProtein server (31,32). 2 Generation of Mouse CTE-I Mutants-The full-length cDNA for mouse CTE-1 was cloned into the pET16B vector (Novagen, Madison, WI) as described previously (17), producing a His-tagged fusion protein. Point mutations were introduced by PCR using the QuickChange TM site-directed mutagenesis kit (Stratagene, La Jolla, CA). The mutagenic sense oligonucleotides used are shown in Table I. PCR reactions for single-base mutations were run for 30 s at 95°C (hot start) followed by 12 cycles of 30 s at 95°C, 1 min at 55°C, and 20 min at 68°C with subsequent cooling at 4°C. For two-base substitutions, 16 cycles were used. The methylated parent plasmid strands were then degraded using the endonuclease DpnI before transformation of the mutant plasmids into XL-1 Blue cells (Stratagene). The mutant plasmids were sequenced using the ABI Prism dye terminator ready-reaction kit (PerkinElmer Life Sciences) at Cybergene AB (Novum, Sweden). Sequences were analyzed using the LaserGene software package (DNAStar).

Expression of Recombinant Wild-type and Mutant CTE-I Proteins-
The mutant and wild-type plasmids were transformed into Escherichia coli BL21(DE3)pLysS cells. Single colonies of bacteria were amplified by overnight culture, transferred to 250 -2000 ml of Luria-Bertani medium, and grown in the presence of 34 g/ml chloramphenicol and 50 g/ml ampicillin at 37°C for approximately 3 h until an A 600 nm of 0.6 -1.0 was reached; expression of recombinant proteins was induced by the addition 1 mM isopropy1-1-thio-␤-D-galactoside (IPTG, Sigma). The cultures were incubated for a further 3 h, and bacteria were harvested by centrifugation and frozen at Ϫ20°C for at least 1 h. For the initial screening of all mutations, the effects of chemical modifiers, and the acylation experiments, the bacteria were thawed and solubi-lized by sonication (2 ϫ 30 s at level 2 of 10 using a Sonicator XL2020 from Heat Systems), and the activity was measured in the supernatants (bacterial extracts) following centrifugation at 36 000 ϫ g for 1 h at ϩ4°C. For kinetic characterization of wild-type and mutant CTE-I, bacteria were solubilized using the Bugbuster TM protein extraction kit and Benzonase nuclease (Novagen) and centrifuged for 60 min at 36 000 ϫ g at ϩ4°C. The supernatants were filtered through a 0.22-m filter, and His-tagged recombinant proteins were purified using Hi-Trap™ chelating columns according to the manufacturer's instructions (Amersham Biosciences AB). The columns were first washed with phosphate buffer containing 0.5 M NaCl and 20 mM imidazole (pH 7.4) and then sequentially eluted with increasing concentrations of imidazole. The recombinant proteins were eluted at concentrations of between 300 and 500 mM imidazole.
Gel Electrophoresis and Western Blotting-Purified recombinant proteins or bacterial protein extracts were separated by SDS-PAGE on 10% polyacrylamide gels. Western blotting was performed by electrophoretic transfer of the proteins onto nitrocellulose filters (Nitropure). The blots were probed with anti-MTE-I antibodies (18), which strongly crossreact with CTE-I, and subsequently with horseradish peroxidase-conjugated secondary antibodies. The signal was visualized by enhanced chemiluminescence (ECL, Amersham Biosciences, Inc.) using x-ray film.
Acylation and Deacylation of Recombinant Protein-Protein extracts (100 g) of expressed wild-type and mutated CTE-I were incubated with 25 M 14 C-palmitoyl-CoA (Sigma) for 1.5 h at 37°C. The samples were split, and one aliquot of each incubation was treated with 1 M hydroxylamine (adjusted to pH 7.5 with NaOH). Hydroxylamine-treated and untreated samples were subjected to SDS-PAGE followed by electrophoretic transfer onto nitrocellulose membranes, which were exposed to phosphorimaging plates (Fuji film). Signals were analyzed in a BAS-1800 (Fuji Photo Film Co.) and quantitated using Image Gauge software, version 3.0. Protein loading was quantitated by Western blotting followed by analysis using an Image Master VDS (Amersham Biosciences, Inc.) and Image Master Program, version 3.0. The incorporation of labeled palmitate was normalized to protein amount.
Enzyme Activity Determinations-Acyl-CoA thioesterase activity was measured spectrophotometrically at 412 nm using 5,5Ј-dithiobis-(2nitrobenzoic acid) (DTNB). The medium contained 200 mM potassium chloride, 10 mM Hepes, and 0.05 mM DTNB. An ⑀ 412 nm ϭ 13,600 M Ϫ1 cm Ϫ1 was used to calculate the activity. Because the CTE-I thioesterase activity was inhibited at substrate concentrations higher than 5-10 M with acyl-CoAs longer than C 12 , bovine serum albumin (BSA) was added to a molar ratio of BSA/acyl-CoA of 1:4.5. To test the effects of amino acid residue modifiers (diethyl pyrocarbonate, diisopropyl fluorophosphate (DFP), bis-(4-nitrophenyl)phosphate (BNPP), phenylmethylsulfonyl fluoride, p-chloromercuribenzoic acid (pCMB), dithiothreitol (DTT), or DTNB), the enzyme was preincubated with the compound for 10 min before the addition of substrate. When acyl-CoA thioesterase activity was measured with the serine to cysteine mutant protein, and when the effects of pCMB or DTT were tested on wild-type CTE-I protein, the decrease in absorbance due to cleavage of the thioester bond was followed at 232 nm, and activity was calculated using an ⑀ 232 nm ϭ 4250 M Ϫ1 cm Ϫ1 . Enzyme kinetic data were analyzed using the Sigma Plot Enzyme Kinetics program. Protein concentrations were determined according to Bradford (33).

RESULTS
We recently cloned four highly conserved members of a peroxisome proliferator-induced acyl-CoA thioesterase gene family (22). These genes encode putative mitochondrial, peroxisomal, and cytosolic thioesterases based on the presence or absence of targeting signals. The encoded proteins show high homology, and sequence comparisons show that they are closely related only to BAAT, an enzyme involved in conjugation of bile acids (34,35). Sequence analysis identified a putative active-site serine within a consensus motif (Gly-Xaa-Ser/Cys-Xaa-Gly) frequently found in the active site of ␣/␤ hydrolases (36). This superfamily of proteins is one of the largest known and includes esterases, lipases, transferases, thioesterases, haloperoxidases, lyases, etc. All of the enzymes in the ␣/␤ hydrolase superfamily share a common fold and harbor a catalytic triad formed by Ser/Cys, His, Asp/Glu. There is only one crystal structure available for a mammalian thioesterase, the palmitoyl protein thioesterase 1, shown to belong to the ␣/␤ hydro-lase family (37). The presence of the GXSXG consensus motif suggests that also the type I thioesterases may conform to the paradigm of the ␣/␤ hydrolase fold. Furthermore, biochemical and site-directed mutagenic analyses have provided evidence for the involvement of a serine and a histidine residue in the catalytic triad of thioesterase II, an enzyme involved in termination of fatty acid synthesis (27,28). When mouse type I thioesterase sequences were aligned to the mouse BAAT sequence (Fig. 1), a reduced number of conserved carboxylic acid, serine, and histidine residues was identified that may constitute a catalytic triad in type I thioesterases and BAAT, although Ser 232 aligned with a cysteine residue in BAAT. Fasta3 searches of the GenBank TM detected a number of proteins sharing significant overall sequence homology with mouse CTE-I (Fig. 1). The alignment further suggested Ser 232 , Asp 324 , and His 358 as a tentative triad, because these are the only residues of this type conserved in all of the sequences found. When each of the sequences shown in Fig. 1 was used to search for homologous proteins in the three-dimensional structure data bank, one of the Caenorhabditis elegans proteins (T16563) showed a distant but unequivocal homology with a bacterial dienelactone hydrolase of known three-dimensional structure (Protein Data Bank code 1DIN). This hydrolytic enzyme belongs to the ␣/␤ hydrolase superfamily, and therefore its evolutionary relationship to CTE-I confirms that this thioesterase also shares the ␣/␤ hydrolase fold. The sequence homology between 1DIN and CTE-I is too low to permit homology modeling, but a secondary structure prediction of CTE-I allowed us to identify some of the central elements of the fold in CTE-I. A partial secondary structure guided alignment of 1DIN and CTE-I is shown in Fig. 2. A reduced number of conserved residues can be detected in strategic positions of the structure. Among these, Ser 232 , Asp 324 , and His 358 in CTE-I aligned to the corresponding residues in the catalytic triad of 1DIN, further supporting the theory that these residues constitute the catalytic triad of CTE-I.
Effects of Chemical Modifiers on Mouse CTE-I Thioesterase Activity-The finding of a putative active-site serine was surprising, as previous characterizations of purified type I acyl-CoA thioesterases showed that the activity is not inhibited by DFP or BNPP, common serine esterase inhibitors, but is very sensitive to pCMB, a cysteine-reactive compound (18,19). To further investigate amino acids involved in the active site, recombinant mouse CTE-I protein was produced to characterize the enzyme with respect to sensitivity to common amino acid modifying reagents. The recombinant protein was found to be insensitive to the serine reactive agents DFP and BNPP when tested at concentrations up to 1 mM (Fig. 3A). However, phenylmethylsulfonyl fluoride, another serine-reactive agent, inhibited the activity with an IC 50 Ϸ 0.5 mM (Fig. 3A). The histidine-reacting reagent diethyl pyrocarbonate was also found to abolish CTE-I thioesterase activity (Fig. 3A). These results are consistent with the presence of a Ser-His-Asp/Glu catalytic triad. In addition, the cysteine reducing reagent DTT had no effect on CTE-I thioesterase activity, whereas pCMB and DTNB were found to be potent inhibitors (Fig. 3B), suggesting that a cysteine residue may be located near the active site.
Site-directed Mutagenesis and Expression of Mouse CTE-I-As discussed above, multiple sequence alignments strongly suggested that Ser 232 , Asp 324 , and His 358 constituted the active-site amino acids. Therefore, PCR-based sitedirected mutagenesis was performed as described under "Experimental Procedures" to verify the identity of these amino acids as being part of the active site. Sequencing of all mutant plasmids verified the introduction of the correct mutations (shown in Table I), and wild-type and mutant plasmids were expressed in E. coli. Acyl-CoA thioesterase activity was measured in crude extracts using myristoyl-CoA, which is the best substrate for CTE-I. The specific activity of CTE-I wildtype enzyme in the crude extract was about 140 nmol/min/mg protein, as compared with ϳ17 nmol/min/mg protein in nonexpressing bacterial extracts (data not shown). Mutation of Ser 232 to Ala, His 358 to Gln, and Asp 324 to Ala abolished the activity, suggesting that Ser 232 , Asp 324 , and His 358 indeed constitute the active-site amino acids, whereas none of the other mutations had any major effect on activity (data not shown). Expressed His-tagged recombinant wild-type CTE-I, S232C, S232A, D324A, and H358Q were purified and characterized kinetically. Expression of the recombinant proteins were verified by SDS-PAGE (Fig. 4), and these proteins were used for further kinetic characterization. The V max of wildtype CTE-I was about 1.2 mol/min/mg protein when measured with palmitoyl-CoA, with a calculated K m of about 2.6  (Table II). However, the specific activities of the S232A, D324A, and H358Q were about 0.1 nmol/min/mg protein (when measured with 10 M palmitoyl-CoA), which is about four orders of magnitude lower than the activity of the wildtype CTE-I. The activities of the mutants were at the baseline detection level for the assay, and more detailed kinetic studies could therefore not be performed. Interestingly, the S232C mutant retained about 2% of the wild-type activity, with a calculated K m of about 0.6 M, suggesting that this mutant is still active as an acyl-CoA thioesterase, albeit with profoundly decreased activity. A possible explanation for the lower activity of the S232C mutant may be that the deacylation step becomes rate-limiting and that this mutant therefore becomes acylated as the acyl-enzyme intermediate. To test this possibility, we carried out experiments to compare acylation of the wild-type CTE-I and the various mutants.
Acylation of CTE-I Mutants-Acylation of CTE-I was analyzed as covalent incorporation of labeled palmitate after incubation with 14 C-palmitoyl-CoA. Neither wild-type CTE-I nor the S232A mutant contained any detectable incorporation of palmitate (Fig. 5). However, the S232C mutant, which showed low palmitoyl-CoA thioesterase activity (see above) was strongly labeled with radioactive palmitate, and the H358Q and D324A mutants were also labeled, albeit much more weakly. Quantitation of the acylation (normalized to protein amount) showed that the H358Q D324A mutants contained about 14 and 10% radioactivity, respectively, of the labeling seen with the S232C mutant, whereas the wild-type CTE-I and the S232A mutants contained Ͻ2.6% (indistinguishable from background) of the radioactivity found in the S232C mutant. The incorporated fatty acids were covalently bound, as the labeling could be removed completely by treatment with neutral hydroxylamine (data not shown).
Acyl-CoA Substrate Specificity of CTE-I-Previous studies on CTE-I (corresponding to ACH2 of Yamada et al. (19)) showed that the enzyme is inhibited at higher substrate concentrations. We therefore tested the addition of BSA to the incubations, which at a molar ratio of 1:4.5 showed a strong protective effect against substrate inhibition (data not shown). Therefore, albumin was added at a constant molar ratio of albumin to substrate when the acyl-CoA chain length specificity was tested with acyl-CoAs longer than lauroyl-CoA (C 12 -CoA). When measured at 10 M substrate, CTE-I is active on saturated acyl-CoAs of 12-20 carbon atoms, with only very low activity with decanoyl-CoA as substrate (Fig.  6). Introduction of one or two double bonds decreased the activity to about half or less compared with the corresponding saturated acyl-CoA. However, the activity is negligible with arachidonoyl-CoA (C 20:4 ), about 35-fold lower than the activity with arachidoyl-CoA (C 20:0 ). A summary of V max and K m values with various acyl-CoAs is shown in Table III. The highest V max values were obtained with C 12 -C 18 saturated and monounsaturated acyl-CoAs with K m values below 4.5 M. We also tested a number of other acyl-CoA substrates (all measured at 10 M) of varying structures; two methylbranched chain acyl-CoAs, 4,8-dimethylnonanoyl-CoA and 2-methylstearoyl-CoA, which were both poor substrates. Also the ␤-oxidation intermediates 2-trans-decenoyl-CoA and 3-hydroxypalmitoyl-CoA were much poorer substrates than the corresponding saturated acyl-CoAs. Clofibroyl-CoA thioesterase activity has been reported to be localized mainly in cytosol and to be induced by clofibrate treatment in rats (39). However, CTE-I showed almost no activity with clofibroyl-CoA, which excludes CTE-I as the enzyme responsible for the activity seen in rodent liver. In addition, because of the sequence similarity between CTE-I and the BAAT enzyme (which uses choloyl-CoA and chenodeoxycholoyl-CoA as sub-

FIG. 4. SDS-PAGE analysis of expressed and affinity-purified wild-type and mutant CTE-I proteins used for kinetic characterization.
Wild-type and single amino acid-specific mutants of CTE-I, as indicated, were expressed in E. coli after which each protein was purified using affinity chromatography as described under "Experimental Procedures." 2-8 g of protein was electrophoresed in 10% polyacrylamide gels and stained with Coomassie Brilliant Blue. Wt, wild type.

TABLE II
Comparison of enzyme kinetics of recombinantly expressed CTE-I and mutant proteins Wild-type and various mutants of CTE-I were expressed and purified as described under "Experimental Procedures." The purified proteins were analyzed for acyl-CoA thioesterase activity using palmitoyl-CoA as substrate. The activities of the S232A, D324A, and H358Q mutants were at the base-line detection level, and therefore proper kinetic characterization could not be carried out. NA, not attainable.

FIG. 5. Acylation of wild-type and various mutants of CTE-I.
Wild-type (Wt) and mutated CTE-I cDNAs, as indicated, were expressed in E. coli, and bacterial extracts were prepared and incubated with 14 C-palmitoyl-CoA. Aliquots were subjected to SDS-PAGE and blotted onto nitrocellulose membranes, and incorporation of 14 C-palmitate was analyzed using a phosphorimaging device (upper panel). Aliquots were also analyzed by Western blotting, quantitated by scanning, and used to calculate the relative incorporation of radioactivity (lower panel). strates) we tested the activity of CTE-I with these bile acid intermediates. Somewhat surprisingly we did not detect any activity with choloyl-CoA or chenodeoxycholoyl-CoA. Furthermore, CTE-I showed no phospholipase A 2 or diacylglycerol lipase activity. 3 Taken together, the kinetic characterization suggests that CTE-I functions as a highly specific long-chain acyl-CoA thioesterase with very low activity for "bulky" substrates.
In view of our observation that the activity of a recently characterized peroxisomal acyl-CoA thioesterase (PTE-2) is highly regulated by CoASH, 4 we tested the effect of CoASH on CTE-I activity. In contrast to PTE-2, CTE-I activity was not affected at any of the concentrations tested (up to 500 M, data not shown).

CTE-I Is a Ser-His-Asp Triad Containing ␣/␤ Hydrolase-
Thioesterases are ubiquitous enzymes that appear to have diverse functions in a number of processes such as fatty acid and polyketide synthesis, removal of acyl chains from palmitoylated proteins, and bioluminescence and turnover of acyl-CoA. A number of acyl-CoA thioesterases have been characterized with possible functions in lipid metabolism (for review, see Ref. 11). To date, only four thioesterases have been characterized structurally by means of x-ray crystallography. The myristoyl acyl carrier protein thioesterase from Vibrio harveyi, which is involved in bioluminescence (40), and the mammalian palmitoyl protein thioesterase PPT1 (37) both belong to the large ␣/␤ hydrolase superfamily, which contains a catalytic Ser-His-Asp triad in the active site. However, the serine in the V. harveyi enzyme is not found in the common esterase/lipase consensus sequence Gly-X-Ser-X-Gly. Recently the structure of E. coli thioesterase II was solved (41) revealing a tertiary structure similar to ␤-hydroxydecanoyl thiol ester dehydratase (42) and 4-hydroxybenzoyl-CoA thioesterase (43). The catalytic site of E. coli TE II involves a novel chemistry and includes Asp 204 , Gln 278 , and Thr 228 , which synergistically activate a nucleophilic water molecule (41). However, the mechanism by which the 4-hydroxybenzoyl-CoA thioesterase hydrolyzes the thioester bond is not yet fully understood. In addition, thioesterases I and II of the fatty acid synthesis system have been identified as serine esterases. Site-directed mutagenesis experiments on thioesterase II have highlighted the active-site serine as well as a histidine residue, both of which are crucial for catalytic activity (26 -28, 44, 45), suggesting that the active site of the thioesterases may consist of a catalytic triad similar to ␣/␤ hydrolase enzymes (Fig. 7). In the present study we have identified mouse CTE-I as a novel member of the ␣/␤ hydrolase superfamily with Ser 232 , Asp 324 , and His 358 constituting the catalytic triad. This conclusion is based on sequence alignments and secondary structure prediction, and the importance of these amino acids for activity was verified by site-directed

TABLE III
Kinetic characterization of wild-type CTE-I CTE-I was expressed and purified on an affinity column as described under "Experimental Procedures." Thioesterase activity was measured at various concentrations of straight-chain acyl-CoAs of different chainlengths (C 10 -C 20:4 ). Activity was not detectable with acyl-CoAs of 8 or less carbon atoms in length. K m and V max were calculated using the Sigma Plot Enzyme Kinetics program. Recombinant CTE-I was also tested for activity with some other CoA-esters at 10 M; 4,8-CH 3 -C 9 -CoA, 4,8-dimethylnonanoyl-CoA; 2-trans-C 10 -CoA, 2-transdecenoyl-CoA; 3-OH-C 16 -CoA, 3-hydroxypalmitoyl-CoA; 2-CH 3 -C 18 -CoA, 2-methylstearoyl-CoA.  7. Alignment of the catalytic triad amino acids of the type I acyl-CoA thioesterases and BAAT. Alignment of the GXSXG motif of the active-site serine and the active-site aspartic acid and histidine residues (shown in boldface type) of the type I acyl-CoA thioesterases and the BAAT enzyme from mouse and the rat thioesterase I and II involved in fatty acid synthesis. The numbering of the mouse enzymes refers to the amino acid sequence of CTE-I. Note that the BAAT enzyme contains a cysteine residue instead of the active-site serine and that the first glycine of the GXSXG motif is replaced by a serine (indicated in italics). The active-site serines of thioesterases I and II are at positions 101 and 103 respectively, and the active-site histidines are at positions 276 and 237, respectively. mutagenesis as the mutations S232A, D324A, and H358Q decreased the activity by about four orders of magnitude. The strong inhibitory effect of pCMB on wild-type CTE-I suggests the presence of a cysteine in, or close, to the active site. Cys 378 of CTE-I is conserved in all of the type I acyl-CoA thioesterases and in the BAAT enzyme, thus being a strong candidate. However, despite several attempts we were not able to mutate this cysteine, most likely because of the difficulty in producing mutagenic primers in this area without secondary structure formation that would prevent proper PCR amplification.
Acyl-CoA Thioesterase Versus Acyltransferase Activity-CTE-I belongs to a highly conserved gene family with four more members, MTE-I, PTE-Ia, and PTE-Ib, and BAAT, the latter which is an acyltransferase involved in conjugation of bile acids. All four acyl-CoA thioesterases contain the Ser-His-Asp triad amino acids, whereas the BAAT enzyme contains a cysteine instead of the active-site serine (see Fig. 7), providing a possible mechanism for the transferase activity. In addition, the first glycine of the active-site cysteine motif is replaced by a serine. Interestingly, the BAAT enzyme slowly hydrolyzes choloyl-CoA to cholic acid and CoASH, the rate being about 2% of the transferase activity in the presence of 50 mM glycine (46) with formation of an acyl-enzyme intermediate (47). Replacement of the active-site cysteine in BAAT with a serine would be expected to generate an excellent bile acid-CoA thioesterase. Interestingly, the S232C CTE-I mutant retained about 2% of the wild type activity, which was still about 200-fold higher than the activity of the other mutants. The acylation experiments showed that wild-type CTE-I is not acylated under the conditions used, whereas the S232C mutant becomes very strongly acylated. In addition, the S232A mutant was not acylated (consistent with this serine being the nucleophilic residue) whereas the H358Q and D324A mutants were both acylated, suggesting that these two residues are also involved in the deacylation reaction. The rate-limiting step in the enzymatic mechanism for mammalian thioesterases is the acylation of the protein, as only low levels of the acylated serine intermediate can be detected (26,44). In agreement with this hypothesis, our experiments described here show that incubation of wild-type CTE-I with radiolabeled palmitoyl-CoA results in no detectable labeling of the CTE-I protein.
Our finding that mutation of the active-site serine to cysteine in CTE-I almost abolished acyl-CoA thioesterase activity is similar to results obtained by others, e.g. the S203C mutant of acetyl cholinesterase (48) and the S114C mutant of the V. harveyi myristoyl-ACP thioesterase (49), which were devoid of thioesterase activity. In contrast, mutation of the active-site serine to cysteine (S101C) in the thioesterase domain of chicken fatty acid synthase and thioesterase II caused only about a 10 -50% reduction in activity compared with thioesterase activities of the wild-type enzymes, with no apparent increase in acylation (26,28,44). It therefore seems very difficult to predict the effects on enzymatic activity of replacing the active-site serine with a cysteine residue. In most cases it appears that a cysteine residue forms a more stable acyl-enzyme intermediate, which may allow an acceptor molecule (e.g. glycine as in the case of the BAAT enzyme) to act as the second nucleophile and to complete an acyl transfer reaction. Attempts to test whether S232C CTE-I enzyme could catalyze BAAT activity failed as wild-type CTE-I is not able to hydrolyze choloyl-CoA or chenodeoxycholoyl-CoA (data not shown), probably because of very different substrate specificities of these enzymes (see the discussion below). In an elegant study, Witkowski et al. (44) were able to engineer thioesterase II into a highly active acytransferase by creating a S101C,H237R double mutant that was almost devoid of thioesterase activity but acted as an excellent acyltransferase (44). Thus, additional modifications may be required to engineer CTE-I into an acyltransferase. It could also be of interest to test whether CTE-I is able to conjugate fatty acids to glycine, although amidation may be more restricted to conjugation of arylacetic acids and alkyl carboxylic acids (50). Acyl-CoA Substrate Specificity of CTE-I-Previous kinetic characterization of CTE-I (corresponding also to rat ACH2 of Yamada et al. (19)) have indicated that the enzyme activity drops quickly with acyl-CoAs longer than C 16 (19), mainly because of strong substrate inhibition. Characterization of recombinant CTE-I showed that it is also strongly inhibited at concentrations higher than ϳ5 M acyl-CoAs that have chain lengths longer than C 12 . However, inclusion of BSA into the assay medium largely prevented this inhibition, allowing a more extensive and detailed kinetic characterization to be carried out. CTE-I is most active on myristoyl-and palmitoyl-CoA, but also shows appreciable activity also on stearoyl-and Free fatty acids (FFA) entering the cell may directly bind to fatty acid-binding proteins (FABP) or are esterified to CoASH by long-chain acyl-CoA synthetases (LACS). Formed acyl-CoAs may undergo ␤-oxidation in mitochondria or peroxisomes, bind to acyl-CoA-binding protein (ACBP), which may act as a CoA pool-forming protein in the cell. Alternatively, the acyl-CoAs may be hydrolyzed by CTE-I and, bound to fatty acid-binding protein, be transported to the nucleus and delivered to PPAR␣/PPAR␥ as ligands. Free fatty acids are also subject to -oxidation by CYP4A1, which is the initial step in the formation of dicarboxylic acids. CTE-I may also interfere with triacylglycerol (TG) and very low density lipoprotein (VLDL) synthesis by removing the substrate for the SCD1. arachidoyl-CoA. However, in contrast to rat ACH2, the activity of CTE-I was lower with unsaturated acyl-CoAs (irrespective of cis or trans double bonds), and in particular arachidonoyl-CoA turned out to be a very poor substrate. We also tested a number of other CoA esters, such as methyl-branched acyl-CoAs, ␤-oxidation intermediates, clofibryl-CoA, and CoA esters of bile acids, which were all found to be very poor substrates. This is in contrast to the acyl-CoA specificity of a peroxisomal acyl-CoA thioesterase, PTE-2, which we recently characterized and showed could hydrolyze all tested CoA esters. The best substrates for PTE-2 were the bile acid intermediates choloyl-and chenodeoxycholoyl-CoA and branched chain acyl-CoA esters (51). In addition, whereas PTE-2 is inhibited by CoASH, CTE-I appears insensitive to CoASH (data not shown), indicating that CTE-I does not regulate CoASH levels but rather hydrolyzes acyl-CoAs when available. CTE-I did not show any detectable phospholipase A 2 or diglyceride lipase activities or esterase activity on nitrophenyl esters (data not shown). The narrow substrate specificity suggests that CTE-I is highly specific as a long-chain acyl-CoA thioesterase and may also explain the apparent insensitivity to common serine esterase inhibitors, as these may not bind to the CTE-I enzyme for structural reasons, similar to results obtained with the PPT1 protein (52).
Possible Functions of CTE-I-Despite an increasing number of identified thioesterases, the functions are largely unknown. Given the importance and many functions of acyl-CoAs and fatty acids in metabolism and other cellular functions, it is obvious that acyl-CoA thioesterases may play diverse and important functions in vivo (for review, see Ref. 11). Our finding in the present study that CTE-I is not negatively regulated by CoASH suggests that CTE-I is not involved in the regulation of CoASH levels but rather that it hydrolyzes acyl-CoA when available. A cytosolic acyl-CoA thioesterase can play an important role in supply of ligands for the PPAR family of nuclear receptors (as outlined in Fig. 8). Recently the liver fatty acid-binding protein was shown to translocate to the nucleus and to interact with PPARs, and thereby liver fatty acid-binding protein may function as a cytosolic gateway for transport of fatty acids to the nucleus to be delivered to PPARs (53,54). Two features that support such a function of CTE-I are the tissue expression and the strong and rapid regulation of expression caused by peroxisome proliferators and by fasting (22,23). These conditions are associated with increased activation of the PPAR␣ and increased expression of a large number of PPAR␣-regulated genes. Thus, the up-regulation of the CTE-I may act as an amplification system for PPAR␣ activation. Fasting is also associated with increased flux of unesterified fatty acids into the liver and increased formation of dicarboxylic acids, the initial -hydroxylation being catalyzed mainly by CYP4A1, another PPAR␣ target gene (55). The microsomal -hydroxylases utilizes the free fatty acid rather than the CoA ester as substrate, and therefore CTE-I may supply substrate to the -hydroxylases.
The kinetic characterization of the CTE-I enzyme in the present study may support yet another function for the CTE-I. The acyl-CoA substrate specificity suggests that CTE-I is mainly active on long-chain saturated acyl-CoAs, which are substrates for stearoyl-CoA desaturase 1 (SCD1). SCD1 catalyzes the introduction of a double bond, mainly in palmitic and stearic acid in the ⌬9 position. Expression of SCD1 is highly regulated by nutritional conditions in mice, being strongly reduced by fasting 5 and strongly increased by refeeding a fat-free diet to starved mice (56). Recent data show that disruption of the SCD1 gene largely impairs the biosynthesis of hepatic triglycerides and cholesteryl esters, which are normally rich in oleic acid (38). The dietary regulation of the CTE-I gene is largely opposite to the SCD1 gene (23), and CTE-I may therefore have a function in channeling fatty acids toward degradation instead of esterification.