Identification of Peroxisomal Acyl-CoA Thioesterases in Yeast and Humans*

A computer-based screen of theSaccharomyces cerevisiae genome identifiedYJR019C as a candidate oleate-induced gene.YJR019C mRNA levels were increased significantly during growth on fatty acids, suggesting that it may play a role in fatty acid metabolism. The YJR019C product is highly similar to tesB, a bacterial acyl-CoA thioesterase, and carries a tripeptide sequence, alanine-lysine-phenylalanineCOOH, that closely resembles the consensus sequence for type-1 peroxisomal targeting signals. YJR019C directed green fluorescence protein to peroxisomes, and biochemical studies revealed that YJR019C is an abundant component of purified yeast peroxisomes. Disruption of the YJR019C gene caused a significant decrease in total cellular thioesterase activity, and recombinant YJR019C was found to exhibit intrinsic acyl-CoA thioesterase activity of 6 units/mg. YJR019C also shared significant sequence similarity with hTE, a human thioesterase that was previously identified because of its interaction with human immunodeficiency virus-Nef in the yeast two-hybrid assay. We report here that hTE is also a peroxisomal protein, demonstrating that thioesterase activity is a conserved feature of peroxisomes. We propose that YJR019Cand hTE be renamed as yeast and human PTE1 to reflect the fact that they encode peroxisomal thioesterases. The physical segregation of yeast and human PTE1 from the cytosolic fatty acid synthase suggests that these enzymes are unlikely to play a role in formation of fatty acids. Instead, the observation that PTE1 contributes to growth on fatty acids implicates this thioesterase in fatty acid oxidation.

Acyl-CoA thioesterases catalyze the hydrolysis of acyl-CoA molecules to free fatty acids and CoA. This enzymatic activity is an intrinsic component of animal fatty acid synthetase and in this context serves to terminate chain elongation (1). Additional thioesterases may associate noncovalently with fatty acid synthetase to modify the length of fatty acids produced by fatty acid synthetase in certain tissues. For instance, a mammary gland-specific thioesterase has been shown to direct the production of medium chain fatty acids during lactation (2).
Although these fatty acid synthetase-associated thioesterases perform a relatively well understood role in cellular metabolism, eukaryotic cells contain a number of additional thioesterase activities with less well defined functions. The most notable of these are the intraorganellar thioesterases of peroxisomes and mitochondria.
Both peroxisomal and mitochondrial acyl-CoA thioesterases have been reported in mammalian cells (3)(4)(5)(6)(7). Peroxisomes and mitochondria do not contain fatty acid synthetase, and their thioesterases would not be expected to participate in the synthesis of fatty acids. In fact, the mere presence of thioesterases in these organelles seems counterproductive, because the oxidation of acyl-CoAs is a main function of both peroxisomes and mitochondria. Although the molecular cloning of mitochondrial fatty acyl-CoA thioesterase has been reported recently (5), the structural genes for peroxisomal acyl-CoA thioesterases remain to be identified. Here we report that YJR019C is a novel fatty acid-induced gene of the yeast Saccharomyces cerevisiae. This gene encodes a peroxisomal protein, and both genetic and biochemical evidence demonstrate that the YJR019C product is an acyl-CoA thioesterase. Loss of YJR019C interferes with the ability of yeast to grow on fatty acids, suggesting that this thioesterase plays an ancillary role in fatty acid oxidation rather than fatty acid synthesis. Furthermore, we report that this gene is conserved in humans and that the human form of this gene also encodes a peroxisomal thioesterase. The implications of these results for the role of peroxisomal thioesterases are discussed.

EXPERIMENTAL PROCEDURES
Plasmids-The YJR019C open reading frame (ORF) 1 was amplified from S. cerevisiae genomic DNA using the primers 5Ј-GGGAGATCTAT-GAGTGCTTCCAAAATGGCG-3Ј and 5Ј-CCCGAGCTCAGGCTCCTC-CCATTGCGAG-3Ј. The PCR product was digested with BglII and SacI (sites underlined) and cloned into the BglII and SacI sites of the yeast GFP fusion vector pGFP-X 2 to make pGFP-YJR019C. The YJR019C ORF was reamplified by PCR from S. cerevisiae genomic DNA using the primers 5Ј-CCCGTCGACCATGAGTGCTTCCAAAATGGCCATG-3Ј and 5Ј-CCCGCGGCCGCTCAGAACTTGGCTCGAATGTCTCG-3Ј. This PCR product was digested with SalI and NotI (sites underlined) and cloned into the SalI and NotI sites of pMBP, a modified form of the pMALc2 expression vector (New England Biolabs) to make pJMJ19.
The entire hTE ORF was amplified by PCR from human muscle cDNA using the primers 5Ј-CCCGGATCCGGCTCGAGCATGTCGTC-CCCGCAGGCCCCAGAAG-3Ј and 5Ј-CCCTCTAGAGCGGCCGCCTCT-GGCTAAAGCTTGCTCTCTG-3Ј. A form of hTE lacking the final two codons of the ORF (hTE⌬KL) was amplified using the oligonucleotide 5Ј-CCCTCTAGAGCGGCCGCCTCTGGCTAAAGCTAGCTCTCTG-3Ј in conjunction with the first hTE primer above. These oligonucleotides append BamHI and XhoI sites to the 5Ј end and XbaI and NotI sites to the 3Ј end of the ORF. The PCR product was digested with BamHI and XbaI and cloned into the corresponding sites in pMALc2. The resulting plasmids were denoted pMBP-hTE for the complete hTE ORF and pMBP-hTE⌬KL for the truncated form. The sequences of all ORFs in pMBP and pMALc2 were confirmed by automated fluorescent sequencing. The BamHI-XbaI fragment from pMBP-hTE was excised and transferred to the mammalian expression vector pcDNA3-Nmyc, a modified version of pcDNA3 (Invitrogen). This plasmid contains the sequence 5Ј-AAGCTTACCATGGCGGAGCAGAAGCTGATCTCCGAGGA-GGACCTGCTGGGATCC-3Ј between the HindIII and BamHI sites of pcDNA3 and is designed to express proteins in fusion with an aminoterminal myc epitope.
Northern Blot Analysis-BY4733 cells were maintained in midlog phase growth for 24 -48 h in YPOLT or YPE medium. Two-liter cultures of cells were then harvested at an A 600 of 1.0 and RNA was extracted using standard procedures (12). Poly(A) ϩ RNA was purified using Dynabeads according to the manufacturer's directions (Dynal, Inc.). 0.5 g of poly(A) ϩ RNA was loaded per lane, separated by denaturing agarose gel electrophoresis, and transferred to nylon membranes. Filters were prepared for hybridization and hybridized with radiolabeled DNA fragments using standard protocols (12).
Localization of GFP-YJR019C-The plasmid pGFP-YJR019C was used to transform the strain FY86 and its pex3 derivative to uracil prototrophy. The resulting strains were grown for 24 h in minimal medium lacking uracil. Cells were washed once in minimal medium and then transferred to YPOLT and grown for an additional 24 h. Cells were harvested and attached to cover glasses coated with poly-L-lysine. The subcellular distribution of GFP-YJR019C was determined by confocal fluorescence and phase contrast microscopy.
Purification and Amino Acid Sequencing of the YJR019C Gene Product-Preparation of high salt-extracted peroxisomal proteins from oleic acid-induced SKQ2N cells and their further separation by reverse phase HPLC was performed as described by Erdmann and Blobel (11). HPLC fractions were examined by SDS-PAGE. Fractions 26 -28 were pooled, separated on a 12% polyacrylamide gel, and electrophoretically transferred to a polyvinylidene difluoride membrane. Protein bands were visualized with 0.1% Amido Black in 10% acetic acid and subjected to amino-terminal sequence analysis (13). The amino-terminal 12 amino acids of a 40-kDa protein in these fractions corresponded precisely to the amino-terminal 12 amino acids of the deduced YJR019C sequence.
Growth Curves-Wild-type BY4733 cells, pex8 cells (BY4733, pex8⌬::HIS3), and yjr019c cells (BY4733, yjr019c⌬::HIS3) were maintained in midlog phase for 2 days in YPD or selective medium. Cells were washed with water and used to inoculate 30 ml of SYOLT medium at an initial A 600 of 0.001. Growth was assessed by removing 1-ml aliquots, washing the cells twice with water, and measuring the A 600 . Aliquots were removed at 24-h intervals for 168 h.
Expression and Purification of Recombinant Proteins-50-ml cultures of DH10B cells containing either the pMBP-hTE⌬KL or the pMBP-YJR019C plasmid were grown overnight at 37°C in Luria broth supplemented with 100 g/ml ampicillin. 40 ml of this culture were diluted into 1 liter of 2YT media (12) supplemented with 0.2% glucose and 100 g/ml ampicillin. This culture was grown at 37°C until the A 600 reached 0.4, at which time isopropyl-1-thio-␤-D-galactopyranoside was added to 1 mM to induce protein expression. The induced culture was incubated at 37°C for 4 h for cells containing pMBP-hTE⌬KL or 30°C for 15 h for cells containing pMBP-YJR019C. Cells were harvested, washed with Luria broth, and resuspended in 25 ml of amylose column buffer (20 mM Tris-HCl, pH 7.5, 200 mM NaCl, 1 mM EDTA, and 10 mM 2-mercaptoethanol). The cell suspension was incubated on ice with 0.4 mg/ml lysozyme for 20 min. After this incubation the cells were frozen in liquid N 2 , thawed at 37°C, and sonicated briefly at high intensity. This cycle of freezing, thawing, and sonication was repeated twice for a total of three cycles. A cleared lysate was generated by centrifugation at 17,500 ϫ g for 20 min. Proteins were purified by one-step affinity chromatography using a 10-ml amylose-agarose (New England Biolabs) column according to the manufacturer's instructions. Fractions were analyzed by SDS-PAGE, and those containing highly purified (Ͼ90% purity) MBP-YJR019C (M r Ӎ 83,000) or MBP-hTE⌬KL (M r Ӎ 79,000) were pooled and precipitated with 0.4 g/ml (NH 4 ) 2 SO 4 . Aliquots of the purified fusion proteins were stored at Ϫ70 C until needed.
Whole Cell Lysates and Enzyme Assays-For preparations of whole cell lysates, BY4733 cells or their yjr019c derivative were grown for 24 h in YPD, washed once with water, and transferred to YPOLT for an additional 24 h. Cells were incubated in reducing buffer (50 mM KP i , pH 7.4, 1 mM EDTA, and 10 mM 2-mercaptoethanol) for 20 min at 20°C. The cells were then converted to spheroplasts in KP i -sorbitol buffer (50 mM KP i , pH 7.4, and 1.2 M sorbitol) using 0.1 unit of Zymolase (ICN Biomedical) per A 600 equivalent of cells for 30 min at 30°C. Spheroplasts were pelleted and resuspended in 40 ml of Dounce buffer (5 mM 4-morpholineethanesulfonic acid, pH 6.0, 0.6 M sorbitol, 1 mM KCl, 0.5 mM EDTA, and 0.1% ethanol) supplemented with protease inhibitors (0.2 mM phenylmethylsulfonyl fluoride and 25 g/ml each aprotinin and leupeptin). The spheroplasts were then homogenized by 10 passes in a tight-fitting Dounce homogenizer (Wheaton) followed by brief high intensity sonication. A cleared lysate was generated by centrifugation at 3000 ϫ g for 10 min at 4°C. Total protein concentration in each lysate was determined by Bio-Rad protein assay (New England Biolabs). 18.1 g of total protein from each lysate were added to the assay mixture (100 M n-decanoyl-CoA, 0.1 mM 5,5Ј-dithiobis(2-nitrobenzoic acid), 50 mM KP i , pH 8.0) and the absorbance at 412 nm (A 412 ) was read at 5-s second intervals for 5 min.
For assays of purified proteins, 0.75 g of the two recombinant MBP fusion proteins were added to 50 mM KP i , pH 8.0, containing a mixture of 0.1 mM 5,5Јdithiobis(2-nitrobenzoic acid), 20 g/ml bovine serum albumin, and five different concentrations (2,5,20,50, and 100 M) of n-decanoyl-CoA. The A 412 was monitored at 5-s intervals for 2 min. Recombinant MBP-YJR019C was assayed at 30°C, whereas recombinant MBP-hTE was assayed at 37°C. One unit of thioesterase activity is defined as the amount of enzyme required to hydrolyze 1 mol of substrate/min.
Transfections, Indirect Immunofluorescence, Antibodies, and Fluorescence Microscopy-Indirect immunofluorescence studies were done in transformed human skin fibroblasts (GM5756-T) or the human hepatoblastoma cell line HepG2. The cell line PBD100 was derived from a Zellweger syndrome patient, has inactivating mutations in PEX10, and has been described previously (14). Cell lines were cultured and transfected as described (15). After electroporation, cells were resuspended in complete medium, transferred onto cover glasses, and incubated at 37°C for 2 days. The cells were fixed, permeabilized, and processed for indirect immunofluorescence as described (15). Permeabilization was normally performed for 5 min with 1% Triton X-100, which permeabilizes both plasma and peroxisome membranes. For selective permeabilization of only the plasma membrane, the cells were incubated for 5 min in 25 g/ml digitonin instead of Triton X-100. Rabbit antibodies to hTE were generated against purified MBP-hTE⌬KL. Immune serum was used directly for immunofluorescence experiments. Rabbit anti-PMP70 antibodies were obtained from Dave Valle and Gerardo Jimenez-Sanchez. The anti-myc mouse monoclonal antibody, anti-catalase sheep monoclonal antibody, and fluorescent secondary antibodies were obtained from commercial sources.
Immunoblot and Subcellular Fractionation-For immunoblot experiments, anti-hTE antibodies were affinity purified using MBP-hTE⌬KL covalently linked to CNBr-activated Sepharose resin (Amersham Pharmacia Biotech). The protein-coupled column was prepared and used according to the manufacturer's specifications. Immune serum was diluted 1:15 in TBST (25 mM Tris-HCl, pH 7.5, 140 mM NaCl, and 0.1% Tween 20) and passed over the column. The column was washed with 30 bed volumes of TBST, and the purified anti-hTE antibodies were eluted with 100 mM glycine, pH 2.5, neutralized with 1 M Tris-HCl, and precipitated with 0.4 g/ml (NH 4 ) 2 SO 4 for storage. Anti-myc polyclonal antibodies were obtained from Santa Cruz Biotechnology. Expression of proteins in rabbit reticulocyte lysates (Promega) was according to the manufacturer's instructions, using the pcDNA3-Nmyc plasmid containing either the hTE open reading frame or that of an unrelated peroxisomal enzyme. Subcellular fractionation of rat liver was as described by Mihalik (16). Immunoblots were performed as described by Crane et al. (17).

RESULTS
YJR019C Is an Oleate-induced Gene-A screen for potential fatty acid-regulated genes in yeast identified a consensus oleate response element (CGGN 3 TNAN (6 -13) (C/G)CG) in the YJR019C promoter region (Fig. 1). 2 To test whether YJR019C is regulated by fatty acids, we determined the relative abun- dance of YJR019C mRNA in yeast grown in medium containing either oleic acid or ethanol. Both oleic acid and ethanol are converted to acetyl-CoA before use in further metabolic or biosynthetic pathways. Therefore, any increases in mRNA levels in yeast grown on oleic acid versus those grown on ethanol should reflect changes specific to fatty acid metabolism. Polyadenylated RNA was extracted from log phase yeast grown in either oleic acid medium (YPOLT) or ethanol medium (YPE) and analyzed by Northern blot with the YJR019C gene as probe (Fig. 2). Although a significant amount of YJR019C mRNA was present in ethanol-grown cells, the level of YJR019C mRNA was increased greatly in oleate-grown cells. The promoter region of YEL020C lacks ORE-like elements, and its mRNA was present at similar levels in cells grown in either oleic acid or ethanol medium.
The YJR019C Gene Encodes a Peroxisomal Protein-Growth of yeast on fatty acids requires a wide array of peroxisomal proteins. Because YJR019C was induced by fatty acids and encodes a protein with a PTS1-like sequence at its carboxyl terminus (alanine-lysine-phenylalanine COOH ), we tested whether the YJR019C gene product might be located in peroxisomes. Previous studies have established that GFP can be targeted to peroxisomes by fusion to peroxisomal proteins (18). We therefore fused the GFP ORF to the amino-terminal end of the YJR019C ORF in an oleate-inducible expression vector and introduced this vector into the wild-type yeast strain FY86. The distribution of the GFP-YJR019C fusion protein was assessed by confocal and phase contrast microscopy (Fig. 3A). GFP fluorescence was present in discrete punctate structures typical of S. cerevisiae peroxisomes. To determine whether these structures were in fact peroxisomes, we next examined the distribution of this fusion protein in a pex3⌬ derivative of FY86 (PEX3 is required for peroxisomal matrix protein import; Ref. 19). In the pex3 strain, GFP fluorescence was found throughout the cytoplasm (Fig. 3B), a distribution that is expected for a peroxisomal matrix protein.
The identification of YJR019C as a peroxisomal protein was also accomplished by a direct biochemical approach. In an independent attempt to identify the constituents of peroxisomes, highly purified peroxisomes were isolated from yeast grown in oleic acid medium. Soluble matrix proteins were removed by hypotonic lysis of the peroxisomes, and membraneassociated proteins were released by high salt wash of the peroxisomal membrane. The proteins released by high salt extraction were then separated by reverse phase HPLC as described previously (11). HPLC fractions were separated further by SDS-PAGE, transferred to a membrane, and visualized by Amido Black staining. Individual bands were then subjected to amino-terminal sequence analysis. The amino-terminal sequence of an ϳ40-kDa protein, which eluted in fractions 26 -28, corresponded to the deduced amino terminus of YJR019C (Fig.  4). These results demonstrated that YJR019C is an abundant membrane-associated component of yeast peroxisomes.
YJR019C Encodes an Acyl-CoA Thioesterase and Contributes to Growth on Fatty Acids-Because YJR019C mRNA was highly induced by oleic acid and encoded a peroxisomal protein, we next tested whether YJR019C might have a role in growth on fatty acids. A yjr019c⌬ derivative of BY4733 was created, and its growth on oleic acid medium was compared with that of the wild-type BY4733 strain and a pex8 derivative of BY4733. The yjr019c cells exhibited a partial growth defect, growing to 53% of the final wild-type density, whereas the pex8 strain showed the typical pex phenotype, growing to ϳ20% of wildtype density (Fig. 5A). Thus, YJR019C appears to play an ancillary role in growth on fatty acids.
The deduced product of YJR019C shared significant amino acid similarity with tesB, a bacterial acyl-CoA thioesterase (20). Disruption of YJR019C resulted in a loss of ϳ80% of the total n-decanoyl-CoA esterase activity in yeast cells (Fig. 5B). To test directly whether YJR019C is an acyl-CoA thioesterase, we expressed and purified YJR019C in fusion with maltosebinding protein. MBP-YJR019C had significant acyl-CoA thioesterase activity using n-decanoyl-CoA as substrate, exhibiting a specific activity of 6.1 units/mg at 30°C.
The Human Acyl-CoA Thioesterase hTE Is Homologous to YJR019C-The amino acid sequence of YJR019C was used to scan the data base of expressed sequence tags for any human homologues of this yeast thioesterase. Multiple human expressed sequence tags were identified, all of which corresponded to a single, previously characterized gene, hTE (21,22). The deduced amino acid sequence of hTE shows 26.3% identity to YJR019C and, like its yeast counterpart, contains a PTS1 motif (serine-lysine-leucine COOH ) at its carboxyl terminus (Fig. 6). Although hTE has been shown previously to have acyl-CoA thioesterase activity, we tested its activity as an MBP fusion for direct comparison with YJR019C. MBP-hTE displayed a specific activity of 12 units/mg hTE at 37°C, an activity similar to that of MBP-YJR019C.
The hTE Gene Encodes a Peroxisomal Protein-Given that the hTE gene product is similar to YJR019C and has a canonical PTS1 sequence, we tested whether hTE encoded a peroxisomal protein. To address this issue, we first modified the hTE ORF to include a 10-amino acid myc tag at its 5Ј end. A plasmid designed to express this myc-hTE fusion, pcDNA3-Nmyc-hTE, was introduced into human skin fibroblasts. Indirect immunofluorescence experiments revealed that Nmyc-hTE was localized to peroxisomes (Fig. 7A), as determined by colocalization with the peroxisomal marker protein PMP70 (Fig. 7B). To assess whether Nmyc-hTE was imported into the peroxisome lumen, we repeated these experiments under differential permeabilization conditions. Skin fibroblasts expressing pcDNA3-Nmyc-hTE were permeabilized with a limiting amount of digitonin, which permeabilizes the plasma membrane but does not permeabilize the peroxisomal membrane. Indirect immunofluorescent labeling of Nmyc-hTE in digitoninpermeabilized cells showed only background cytosolic staining, even though PMP70-containing peroxisomes were readily detected (Fig. 7, C and D). The PMP70 antibodies that were used in these experiments recognize the cytosolic domain of this peroxisomal membrane protein. These results demonstrate that Nmyc-hTE is translocated into the peroxisome lumen.
Zellweger syndrome is caused by defects in genes that are required for peroxisome biogenesis, and cells from Zellweger syndrome patients display defects in peroxisomal matrix protein import (23). As an independent test of whether hTE is a peroxisomal protein, we expressed Nmyc-hTE in the Zellweger syndrome cell line PBD100. This cell line is homozygous for an inactivating mutation in PEX10 and is unable to import peroxisomal matrix proteins, although it does contain numerous peroxisomes and imports peroxisomal membrane proteins normally (14). Nmyc-hTE accumulated in the cytosol of PBD100 cells, as would be expected for a peroxisomal matrix protein, although intact peroxisomes were detected using antibodies to PMP70 (Fig. 7, E and F).
Previous studies identified hTE as a protein that interacts with HIV-Nef (21,22). Given that there is no evidence that HIV-Nef is peroxisomal, we tested whether endogenously synthesized hTE was located in peroxisomes using antibodies raised against a bacterially synthesized version of hTE. The immune sera recognize peroxisomes, as evident from their colocalization with the peroxisomal marker protein catalase (Fig.  8, A and B). In contrast, the preimmune sera do not recognize peroxisomes (Fig. 8, C and D).
The distribution of this mammalian thioesterase was also assessed by subcellular fractionation experiments. We first The yjr019c⌬ cells grew to 53% of the final wild-type density. B, lysates were prepared from oleate-grown wild-type (WT) and yjr019c⌬ cells. Equal amounts of protein from each lysate were assayed for acyl-CoA thioesterase activity. The lysate of the wild-type strain had a specific activity of 1.8 ϫ 10 Ϫ2 unit/mg, whereas the lysate of yjr019c⌬ strain had a specific activity of 4.0 ϫ 10 Ϫ3 unit/mg. generated affinity-purified anti-hTE antibodies and tested whether they recognized the product of the hTE gene. Rabbit reticulocyte lysates were used to synthesize either Nmyc-hTE or an unrelated myc-tagged protein, and the two lysates were then examined by immunoblot. Affinity purified anti-hTE antibodies recognized a protein of the correct molecular mass only in the lysate in which Nmyc-hTE was synthesized (Fig. 9A). Anti-myc polyclonal antibodies confirmed expression of both proteins (data not shown). Next, a postnuclear supernatant was prepared from rat liver and separated by Nycodenz density gradient centrifugation. Fractions across the gradient were   FIG. 7. hTE is a peroxisomal matrix protein. Human skin fibroblasts expressing Nmyc-hTE were processed for double indirect immunofluorescence by fixing cells and permeabilizing with 1% Triton X-100. The distribution of Nmyc-hTE was examined using anti-myc (A) and anti-PMP70 (B) antibodies. Additional cells from the same set expressing Nmyc-hTE were permeabilized with 25 g/ml digitonin and examined again using anti-myc (C) and anti-PMP70 (D) antibodies. The distribution of Nmyc-hTE was also examined in the PEX10-deficient cell line PBD100, again by double indirect immunofluorescence using anti-myc (E) and anti-PMP70 (F) antibodies. Scale bar, 25 m. assayed for peroxisomal and mitochondrial markers as well as by immunoblot with the affinity purified anti-hTE antibodies (Fig. 9B). These antibodies detected a 36-kDa protein in rat liver peroxisomes, as well as in cytosolic fractions at the top of the gradient. Peroxisomes will rupture during homogenization and centrifugation, releasing peroxisomal matrix proteins to the cytosol. Peroxisome rupture in this experiment is confirmed by the presence of significant catalase activity in the fractions at the top of the gradient. Although the amount of cytoplasmic thioesterase may be partly explained by peroxisome rupture, we cannot rule out the possibility that a small pool of thioesterase exists in the cytoplasm at steady state. These results support the hypothesis that hTE is a peroxisomal protein but do not rule out the possibility that some hTE may be cytosolic and available to interact with HIV-Nef. DISCUSSION In this report we have demonstrated that YJR019C is an oleate-regulated gene that encodes a novel peroxisomal protein. Furthermore, we find that YJR019C is required for the majority of acyl-CoA thioesterase activity in yeast cells and show that recombinant YJR019C displays significant acyl-CoA thioesterase activity in vitro. The physical segregation of peroxisomal YJR019C from cytosolic fatty acid synthetase, the enzyme required for fatty acid synthesis, provides compelling evidence that the peroxisomal thioesterase is not involved in fatty acid synthesis. Our finding that loss of YJR019C results in impaired growth on fatty acids implicates this gene in fatty acid oxidation, a process that is exclusively peroxisomal in yeast. The fact that the human homologue of YJR019C, the previously identified hTE gene, also encodes a peroxisomal thioesterase suggests that there may be a conserved role for this enzyme in peroxisomal fatty acid oxidation. We propose that YJR019C and human hTE both be renamed PTE1 to reflect the peroxisomal distribution and acyl-CoA thioesterase activity of their gene products.
Fatty acids must be esterified with CoA before their ␤ oxidation. Thus, it is somewhat paradoxical that a peroxisomal acyl-CoA thioesterase is required for normal growth of yeast on fatty acids. In fact, our current understanding of peroxisomal fatty acid oxidation in yeast provides no direct role for thioesterase activity in fatty acid ␤ oxidation or growth on fatty acids. Therefore, it is important to consider possible indirect roles for PTE1 in fatty acid oxidation. Previous studies have established that the peroxisome membrane is impermeable to small molecules such as NAD and NADP (24, 25). As a result, loss of enzymes involved in NAD regeneration lead to loss of NAD-dependent activities within the peroxisome, including fatty acid oxidation. If the peroxisome membrane is also impermeable to CoASH, then the intraperoxisomal free CoASH pool may also be dependent on constant regeneration. Previous studies in yeast have established that CoASH regeneration would occur via complete oxidation of acyl-CoAs to acetyl-CoA, followed by transfer of acetate to carnitine and release of CoASH in a reaction catalyzed by CAT1 (26). However, if CoA were appended to poorly metabolized or nonmetabolizable fatty acids, CoASH levels would be expected to fall as more and more CoA was incorporated into these metabolic sinks. In this scenario, a peroxisomal thioesterase that acted primarily at high acyl-CoA and/or CoASH levels might serve a stimulatory role in fatty acid oxidation by providing an alternative mechanism for generating the free CoASH that is necessary for fatty acid ␤ oxidation. As for the fate of the fatty acids that may be released, these would be free to equilibrate with cellular and extracellular pools, reducing their concentration in the peroxisome.
This indirect model for PTE function has the potential to explain the fact that loss of PTE1 impairs growth of yeast on fatty acids, but we readily admit that there is not yet any independent corroborating evidence in its favor. As such, it should be viewed with skepticism. At the same time, it is worthwhile to consider this model as a possible explanation for the role of mammalian PTE1. If CoASH depletion does serve to inhibit peroxisomal fatty acid ␤ oxidation in yeast, we might also expect CoASH depletion to pose problems for fatty acid oxidation in mammalian peroxisomes and mitochondria. In fact, a similar hypothesis has been proposed to explain the function of rat mitochondrial thioesterase (6).
An alternative role for mammalian peroxisomal thioesterases has also been suggested (7). Peroxisomal ␤ oxidation in mammalian cells differs from the yeast system in that it does not oxidize fatty acids completely. Instead, it transfers medium chain fatty acids out of the peroxisome as acyl-carnitine (27). These are then exported to mitochondria where their oxidation is completed. This facet of mammalian peroxisomal fatty acid oxidation led to the hypothesis that peroxisomal thioesterases may serve to regulate the chain length at which fatty acids are exported (7). However, there are three flaws with this model. First, mammalian peroxisomal carnitine acyltransferase accepts fatty acids in the form of CoA esters, not as free acids (28). Thus, it is unclear how a thioesterase would contribute to acyl-carnitine export. Second, the chain length specificities of mammalian peroxisomal ␤-oxidation enzymes are thought to be sufficient to control the extent to which fatty acyl-CoAs are shortened in peroxisomes (7,27). Third, this model cannot explain the presence of a thioesterase in yeast peroxisomes, because they oxidize fatty acids to completion (29).
The identification of PTE1 in both yeast and human cells FIG. 9. Anti-hTE antibodies recognize a 36-kDa protein in peroxisomal and cytoplasmic subcellular fractions. A, rabbit reticulocyte lysate was used to synthesize Nmyc-hTE (lane 1) or an unrelated peroxisomal protein (lane 2). Equal amounts of each lysate were separated by SDS-PAGE and analyzed by immunoblot with affinity-purified anti-hTE antibodies. B, postnuclear supernatant from rat liver was fractionated by Nycodenz density centrifugation. Equal amounts of fractions were assayed for catalase (dark bars) and succinate dehydrogenase (light bars) activity. Equal amounts of fractions were also analyzed by immunoblot using affinity-purified anti-hTE antibodies. The bar graph shows the relative amounts of the peroxisomal and mitochondrial marker enzyme activities in each fraction. The lower panels show the distribution of the ϳ36 kDa rat protein recognized by the anti-hTE antibodies.
raises the question of whether they correspond to previously described peroxisomal enzymes. This is clearly not the case for yeast PTE1, because this report is the first description of this enzyme activity in yeast peroxisomes. However, there have been several reports of peroxisomal thioesterases in mammalian cells (4,7). A purification of rat liver peroxisomal thioesterases revealed that myristoyl-CoA thioesterase activity eluted in two peaks, one of ϳ35 kDa and one of larger size (4). The product of human PTE1 has a deduced M r of 36 kDa and may represent the human homologue of the smaller rat peroxisomal thioesterase. A second study found peroxisomal thioesterase activity to be present in a broad peak of ϳ40 kDa (7). However, subsequent fractionation of this sample led to the purification of a single thioesterase of ϳ46 kDa. Interestingly, the putative cytosolic thioesterase 1 contains a PTS1-like sequence at its carboxyl terminus, proline-lysine-isoleucine COOH , and has a deduced M r of 46 kDa (30). A further analysis of cytosolic thioesterase 1 distribution may resolve the question of whether it actually represents a second, 46-kDa thioesterase of peroxisomes.
The localization of human PTE1 (hTE) to the peroxisome lumen was observed for an overexpressed, epitope-tagged version of HsPTE1 as well as for the endogenously synthesized, wild-type HsPTE1 of human fibroblasts. Thus, there is little doubt that this protein is predominantly peroxisomal. However, it is also true that human PTE1 (hTE) was first identified because of its interaction with HIV-Nef in the yeast two-hybrid system (21,22). There is no indication that Nef is associated with peroxisomes, nor is there any known role for peroxisomes in the life cycle of HIV. Nevertheless, the interaction between PTE1 (hTE) and HIV-Nef was also detected in lysates of CEM cells that stably expressed HIV-Nef. Although the PTE1-Nef interaction might have occurred after the formation of the cell-free lysate in these experiments, it is also quite possible that small amounts of PTE1 are present in the cytosol. Our fractionation data cannot rule out the possibility that some PTE1 may be cytoplasmic at steady state and available for interaction with HIV-Nef. Also, it is possible that HIV-Nef may bind PTE1 before its import into peroxisomes.