Characterization of Saccharomyces cerevisiaeAcyl-protein Thioesterase 1, the Enzyme Responsible for G Protein α Subunit Deacylation in Vivo *

Thioacylation is a reversible lipid modification of proteins that plays a role in the regulation of signal transduction. Acyl-protein thioesterase 1 (APT1) was identified as an enzyme capable of deacylating some thioacylated proteins in vitro. Saccharomyces cerevisiae open reading frame YLR118c encodes an enzyme homologous to Rattus norvegicus APT1. We demonstrate that the catalytic activity of the protein encoded by the yeast open reading frame is similar to that of rat APT1, and we designate the protein S. cerevisiae Apt1p. Yeasts bearing a disruption of the APT1 gene lack significant biochemically detectable acyl-protein thioesterase activity. They also fail to deacylate Gpa1p, the yeast Gα subunit, in metabolic radiolabeling studies. We conclude that native APT1 is the enzyme responsible for Gα subunit deacylation in S. cerevisiae and presumably other eukaryotes as well.

The covalent attachment of lipid moieties to polypeptide backbones is required for the biological activity of many eukaryotic proteins. Lipid modification of proteins has been found to play roles in protein stability, catalytic activity, proteinprotein interactions, and subcellular localization. Isoprenylation, N-myristoylation, and thioacylation are the best characterized forms of lipid modifications of intracellular proteins. Of these, only thioacylation (also called palmitoylation or S-acylation) is metabolically reversible. Thioacylated proteins cycle between acylated and deacylated states, allowing thioacylation (like other, better characterized, reversible covalent modifications) to act as a regulated switch of protein activity.
Proteins involved in many signal transduction pathways are thioacylated, including Ras proteins, nonreceptor tyrosine kinases, and nitric-oxide synthase. Heterotrimeric G proteinregulated signaling systems have a relative abundance of thioacylated proteins, including many G protein-coupled receptors, G␣ subunits, regulator of G protein signaling (RGS) 1 proteins, and G protein-coupled receptor kinases (1)(2)(3)(4). Some thioacylated G␣ proteins have a higher affinity for G␤␥ complexes and are relatively resistant to the GTPase-accelerating properties of RGS proteins when compared with their nonacylated counterparts in vitro (5,6). These alterations in the biochemical properties of G␣ proteins suggest that thioacylation has an effect on G protein-coupled signaling in vivo. Additionally, mutant G␣ proteins that lack thioacylated cysteine residue(s) are localized to internal rather than plasma membranes, indicating that thioacylation probably plays a role in the subcellular localization of signaling events (7,8).
The deacylation of G␣ proteins (and other thioacylated proteins) is the regulated step of the thioacylation cycle. In the basal state, G␣ proteins are thought to be thioacylated stoichiometrically and associated with G␤␥. Upon activation by ligandbound receptor, G␣ proteins dissociate (at least partially) from G␤␥, and the rate of deacylation (determined by metabolic radiolabeling) increases dramatically (9,10). Similar increases in thioester-bound palmitate turnover have been noted for other thioacylated proteins involved in signal transduction pathways, including several G protein-coupled receptors and endothelial cell nitric-oxide synthase (11)(12)(13). The role that the regulated deacylation of G␣ plays in the signal transduction process has yet to be determined.
The regulated deacylation of G␣ and other proteins is presumed to be carried out by an intracellular thioesterase. An enzyme, designated acyl-protein thioesterase 1 or APT1, that performs regulated deacylation of Gs␣ and other thioacylated proteins in vitro was isolated from rat liver (14). Although this enzyme had previously been identified as a source of lysophospholipase activity, palmitoyl-G␣ is a better substrate for APT1 than lysophospholipids by several hundred-fold (14,15). Furthermore, overexpression of APT1 in cultured cells accelerates palmitate turnover on G␣ s and endothelial cell nitric-oxide synthase. Although overexpressed APT1 is capable of deacylating proteins in vivo, proof of the involvement of the native enzyme in this process has been lacking until now.
Putative APT1 orthologs exist in many eukaryotic organisms, including the budding yeast Saccharomyces cerevisiae. We now have demonstrated that YLR118c, a S. cerevisiae open reading frame, encodes an enzyme with both lysophospholipase and acyl-protein thioesterase activity in vitro. Targeted disruption of the gene encoding Apt1p (YLR118c) reveals that S. cerevisiae Apt1p appears to be responsible for virtually all of the acyl-G␣ thioesterase activity detectable in yeast extracts. Furthermore, the turnover of palmitate associated with the yeast G␣, Gpa1p, is dramatically reduced in yeast lacking the APT1 gene, demonstrating that Apt1p is actually involved in deacylation of G␣ subunits in vivo.

EXPERIMENTAL PROCEDURES
Materials and Miscellaneous Procedures-Ni 2ϩ -NTA-agarose was purchased from Qiagen. All other chromatographic resins were acquired from Amersham Biosciences. Isotopes were from PerkinElmer Life Sciences. All chemicals were supplied by Sigma or Calbiochem, unless otherwise noted. Media for growth of S. cerevisiae were obtained from either CLONTECH or Bio 101, when available; other media were prepared as described by Guthrie and Fink (16). Yeast strain YPH501 was obtained from Dr. Henrik Dohlman (Yale University School of Medicine, New Haven, CT). The plasmid pRS303 was obtained from Dr. Joel Goodman (University of Texas Southwestern Medical Center, Dallas, TX). The plasmid pKM1362-2 was from Carol Manahan (Washington University School of Medicine, St. Louis, MO). The FUS1-␤-galactosidase pheromone-responsive reporter plasmid, pSB234, was provided by Dr. James Konopka (State University of New York, Stoneybrook, NY). Molecular biological procedures were carried out using standard methods (17). Qiagen products were used for DNA preparation. The polymerase chain reaction was performed using VENT thermostable polymerase (New England Biolabs) under conditions recommended by the manufacturer. DNA sequencing was carried out by Lynda Doolittle (Department of Pharmacology DNA Sequencing Core Facility, University of Texas Southwestern Medical Center) using a 373A DNA Sequencer (Applied Biosystems).
Procedures for culturing and manipulating S. cerevisiae were modified from standard protocols (16). Poststationary phase survival assays were performed as described by Ashrafi et al. (18). The names and genotypes of S. cerevisiae strains utilized in this work are detailed in Table I. Acyl-protein thioesterase, lysophospholipase, and palmitoyl-CoA hydrolase assays were carried out as described (14). Recombinant (untagged) G␣ subunits were purified as described by Lee et al. (19). The recombinant G␤1␥2 complex was isolated from recombinant baculovirus-infected Sf9 cells (20). Hexahistidine-tagged RGS4 and hexahistidine-tagged, myristoylated G␣ i1 (with an internal hexahistidine tag) were produced in Escherichia coli and purified as described (21,22). Protein concentrations were determined using Bradford protein reagent (Bio-Rad) and albumin standards.
Construction of Plasmid for Bacterial Expression of S. cerevisiae APT1-The vector pQE60-6HB was created to facilitate the cloning of S. cerevisiae APT1. Two oligonucleotides (5Ј-CATGGTCGACCCGC-3Ј and 5Ј-GGGTCGAC-3Ј) were phosphorylated using T4 polynucleotide kinase, annealed together, and ligated into the vector pQE60 -6H, which had been digested with NcoI and SacII. The resulting plasmid, pQE60-6HB, has the same features as the parent, pQE60-6H, and eliminates a stop codon between the translation start site in the vector and the multiple cloning site (19).
The polymerase chain reaction was used to amplify the S. cerevisiae open reading frame YLR118c from genomic DNA isolated from S. cerevisiae strain YPH501. The sequence of the sense primer was 5Ј-GAAC-CGCGGATGAATGGACTTAGA-3Ј, corresponding to a SacII restriction endonuclease site (in frame with the SacII site within the multiple cloning site of pQE60-6HB), followed by nucleotides 1-15 of YLR118c. The sequence of the antisense primer was 5Ј-GGCCTGCAGCATCAC-CCCAACTAT-3Ј, correlating to a PstI restriction endonuclease site followed by the antisense sequence to nucleotide bases 106 -130 distal to the stop codon of YLR118c. The polymerase chain reaction product was digested with SacII and PstI and ligated into the vector pQE60-6HB, which had been digested with the same enzymes. The resulting bacterial expression plasmid, pQE60-6H-scAPT1, encodes the protein sequence MHHHHHHAMVDPR followed by amino acids 1-228 of S. cerevisiae Apt1p.
Expression and Purification of Recombinant APT1-E. coli strain XL1-Blue harboring the plasmid pQE60 -6H-scAPT1 or pQE60 -6H-APT1 (encoding the rat enzyme) was grown in 6 liters of LB medium containing 50 mg/liter ampicillin. Recombinant protein expression was induced as previously described for the expression of G␣ i1 (19). The bacterial pellet was harvested and frozen in liquid N 2 . The cells were thawed in 250 ml of buffer TP (50 mM Tris-HCl (pH 8.0) and proteinase inhibitors (leupeptin, 6 g/ml; lima bean trypsin inhibitor, 6 g/ml; L-1-tosylamido-2-phenylethyl chloromethyl ketone, 32 g/ml; 1-chloro-3-tosylamido-7-amino-2-heptanone, 32 g/ml; and aprotinin, 2 g/ml)) and lysed by treatment with lysozyme (50 mg), followed by 500 ng of DNase I. A soluble lysate was generated by centrifugation of the lysed cells at 100,000 ϫ g for 30 min. Recombinant APT1 was purified by application of the soluble lysate to a column containing 20 ml of Ni 2ϩ -NTA-agarose. The resin was washed by adding 50 ml of buffer B (50 mM Tris-HCl (pH 8.0), 10 mM imidazole, and 100 mM NaCl) and eluted with 50 ml of buffer C (50 mM Tris HCl (pH 8.0) with 100 mM imidazole). The eluted protein was concentrated to 20 ml in a Centriprep 30 (Amicon) and dialyzed against two changes of HME (20 mM Na-HEPES (pH 8.0), 2 mM MgCl 2 , and 1 mM EDTA) buffer (2 liters each) over 16 h at 4°C. Aliquots of this dialyzed protein (ϳ19 mg/ml) were frozen in liquid N 2 and stored at Ϫ80°C. Roughly 3 g of each recombinant APT1 was recovered using this protocol.
Preparation of Acyl-protein Substrates-Palmitoylated Ha-Ras was prepared essentially as described by Camp and Hofmann; however, CHAPS detergent (1% (w/v) for extraction, 10 mM elsewhere) was substituted for n-octyl glucoside (23). Palmitoyl-G␣ i1 and palmitoyl-RGS4 were both prepared using hexahistidine-tagged, bacterially expressed recombinant proteins. Each protein was diluted to ϳ50 M in HMEC (HME supplemented with 10 mM CHAPS) buffer containing 500 M [ 3 H]palmitoyl-CoA (200 -2000 cpm/pmol) and incubated at 30°C for 60 min. Incorporation of palmitate into these proteins was assessed by precipitation with trichloroacetic acid and found to be ϳ0.7 mol of palmitate/mol of G␣ i1 and ϳ1.3 mol of palmitate/mol of RGS4 (24). The reaction mixtures were diluted to 5 ml in TC buffer (50 mM Tris-HCl (pH 8.0) and 10 mM CHAPS) and applied to 300 l of Ni 2ϩ -NTA-agarose equilibrated in the same buffer. The columns were washed five times with 3 ml of TC buffer and eluted with 1.5 ml of HMEC buffer supplemented with 100 mM EDTA. The eluted palmitoylated proteins were diluted 1:20 in HMEC buffer and concentrated in a Centricon 30 (Amicon) to ϳ50 M palmitoyl-protein (assessed by liquid scintillation spectrometry). The concentrated acyl-protein substrates were frozen in liquid N 2 and stored at Ϫ80°C until use.
Construction of APT1 Targeting Plasmid-The following DNA fragments were amplified using the polymerase chain reaction with primer sequences listed in Table II and template DNA indicated in parentheses: the 5Ј-flanking region of YLR118c (5FR fragment, genomic DNA isolated from YPH501), the 3Ј-flanking region of YLR118c (3FR fragment, genomic DNA isolated from YPH501), and the entire HIS3 gene (HIS3 fragment, plasmid pRS303). The oligonucleotide primers were designed to produce restriction endonuclease sites on each end of the DNA fragments to facilitate cloning. The 5FR fragment was digested with PstI and BamHI and ligated into the plasmid pGEM-4Z (Promega), which had been digested with the same enzymes. The resulting plasmid, pGEM-5FR, was cut with EcoRI and BamHI and ligated with the similarly digested 3FR fragment. The product of this ligation, pGEM-5FR-3FR, was opened at the BamHI site lying between 5FR and 3FR, and the HIS3 fragment was inserted, producing the complete targeting plasmid, pGEM-5FR-HIS3-3FR. The linear targeting fragment, used for transformation of S. cerevisiae strains, could then be excised from pGEM-5FR-HIS3-3FR using EcoRI and PstI. Preparation of S. cerevisiae Extract-Cultures of S. cerevisiae, typically 50 -250 ml, were grown in either YPD medium (or synthetic dropout medium for maintenance of the indicated plasmids) at 30°C to stationary phase (culture A 600 of Ͼ2.5). The cells were then pelleted by centrifugation for 30 min at 2,500 ϫ g. The cell pellet was washed with H 2 O, collected by centrifugation, and frozen in liquid N 2 . Frozen yeast were suspended in five packed cell volumes of HEC buffer (50 mM Na-HEPES (pH 8.0), 10 mM EDTA, and 10 mM CHAPS) supplemented with proteinase inhibitors (see above). The cell suspension was frozen in liquid N 2 and thawed at 37°C three times. The extraction proceeded by rocking the suspension at 4°C for 30 min. The CHAPS-solubilized cell extract was finally isolated by centrifugation of the cell suspension for 45 min at 35,000 ϫ g and collection of the supernatant fraction.

Metabolic Labeling with [ 3 H]Palmitate and Analysis of Yeast
Lipids-Indicated strains of S. cerevisiae were grown in complete synthetic medium. When the cultures reached an A 600 of 0.5, [ 3 H]palmitate in ethanol (50 Ci/ml, 1% ethanol final) was added to the culture. At the indicated times, 25 l of each culture was removed and frozen in liquid N 2 . These samples were thawed by the addition of 25 l of PDL buffer (100 mM NaPO 4 (pH 7.4), 10 mM dithiothreitol, and 5 mg/ml yeast lytic enzyme) and incubated for 1 min at 37°C. The total lipids were extracted by the addition of 500 l of chloroform/methanol (1:1), vigorous vortexing, and incubation for 5 min at 45°C. The samples were centrifuged at 13,000 ϫ g for 10 min, and the lipid-containing supernatants were analyzed by thin layer chromatography (24).

Metabolic Labeling with [ 14 C]Acetate or [ 32 P]Orthophosphate and
Analysis of Yeast Lipids-S. cerevisiae were grown to an A 600 of 0.5 in complete synthetic medium. The yeast were then transferred into synthetic minimal medium and grown at 30°C for 45 min. Metabolic labeling reagents ([ 14 C]acetate (10 Ci/ml) or [ 32 P]orthophosphate (10 Ci/ml)) were added to the appropriate cultures, and the yeast were grown for an additional 2 h. The cells from 5 ml of culture were collected by centrifugation at 4,000 ϫ g and resuspended in 100 l of PDL buffer. The lipid extraction proceeded as described above, except 1 ml of chloroform/methanol was used. The lipid extracts were then dried under N 2 and suspended in 50 l of chloroform/methanol (2:1). The resuspended lipids (10 l) were subjected to thin layer chromatography using PE Sil G thin layer chromatography plates (Whatman) developed with chloroform/methanol/ammonium hydroxide/H 2 O (65:25:2:2) mobile phase. The migration of radiolabeled lipids was detected using an MP1000 phosphorimaging screen and BAS1500 scanner (Fuji Medical Systems).
Metabolic Radiolabeling of Gpa1p with [ 3 H]Palmitate-S. cerevisiae strains YAD101 (APT1) and YAD103 (apt1⌬::HIS3) were transformed with the plasmid pKM1362-2, which expresses carboxyl-terminal HAtagged Gpa1p from a copper-inducible promoter (25). The yeast were grown in synthetic medium lacking leucine (to maintain pKM1362-2) until the culture reached an A 600 of 0.5. Expression of Gpa1p was induced by the addition of 100 M CuSO 4 for 1 h. Cerulenin was added to the culture to a concentration of 25 M and grown for an additional 15 min. [ 3 H]Palmitate (50 Ci/mmol; PerkinElmer Life Sciences) was added to each culture (100 Ci/ml, 2% ethanol final concentration), and metabolic labeling was allowed to proceed for 20 min. The yeast were collected by centrifugation at 2,500 ϫ g for 10 min and resuspended in "chase" medium (synthetic medium lacking leucine and supplemented with 0.25% Brij 58, 0.1 mg/ml palmitate, 0.05 mg/ml myristate, and 100 M CuSO 4 ). At the indicated times, 25 ml of each culture was removed, and the yeasts were collected by centrifugation at 13,000 ϫ g for 15 s. After discarding the supernatant, the yeast were frozen in liquid N 2 and stored at Ϫ80°C until all time points were collected.
Metabolic Radiolabeling of Gpa1p with [ 35 S]Methionine and Cysteine-Gpa1p was overexpressed in S. cerevisiae strains YAD101 (APT1) and YAD103 (apt1⌬::HIS3) as described above. One hour after induction of Gpa1p expression with CuSO 4 , the yeast were transferred from leucine-deficient medium to M/C/L/CU medium (synthetic medium lacking methionine, cysteine, and leucine but supplemented with 100 M CuSO 4 ) and grown for 15 min. [ 35 S]methionine/cysteine (Tran 35 S label; 11,175 Ci/mmol; ICN) was added to each culture (100 Ci/ml final), and metabolic labeling was allowed to proceed for 20 min. The yeasts were collected by centrifugation at 2,500 ϫ g for 10 min and resuspended in "chase" medium (synthetic medium lacking leucine and supplemented with 50 g/ml methionine, 10 g/ml cysteine, and 100 M CuSO 4 ). At the indicated times, 1 ml of each culture was removed, and the yeast were collected by centrifugation at 13,000 ϫ g for 15 s, frozen in liquid N 2 , and stored at Ϫ80°C until all time points were collected.
Analysis of Metabolically Radiolabeled Gpa1p-Radiolabeled Gpa1p was immunoprecipitated utilizing the following protocol. The frozen yeasts were lysed by suspension and incubation on ice for 15 min in 50 l of permeabilization reagent (Y-PER; Pierce) supplemented with protease inhibitors (Complete TM protease inhibitors (Roche Molecular Biochemicals) and phenylmethylsulfonyl fluoride (70 g/ml)). Each sample was then vortexed vigorously after the addition of 25 l of glass beads (Bio 101) and SDS (final concentration of 1%). These broken cell suspensions were diluted with 500 l of radioimmune precipitation buffer (50 mM Na-HEPES (pH 7.4), 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS), and solubilized protein was obtained after centrifugation at 13,000 ϫ g for 15 min. Immunoprecipitation and analysis by SDS-PAGE of HA-tagged Gpa1p from these extracts proceeded as previously described for the immunoprecipitation of HAtagged Gs␣ (14), except that Western blots to quantify the amount of immunoprecipitated Gpa1p were carried out using goat IgG directed against Gpa1p, YN19 (Santa Cruz Biotechnology). A TR1 phosphorimaging screen (Fuji Medical Systems) with exposure times of 60 days was used for imaging of 3 H-labeled proteins. An MP1000 phosphorimaging screen (Fuji Medical Systems) with exposure times ranging from 16 to 24 h was used for imaging of 35 S-labeled proteins. lating to levels of 25-40% of the soluble cellular protein (data not shown). Each protein was purified essentially to homogeneity using a single step of Ni 2ϩ -NTA-agarose affinity chromatography (Fig. 1). Both enzymes migrated as a single peak of ϳ29 kDa upon Superdex 75 gel filtration chromatography (data not shown). These enzymes also demonstrated chromatographic properties very similar to the native rat APT1 isolated from liver (data not shown). They bound tightly to phenyl-Sepharose, eluting between 35 and 50% ethylene glycol. Even at pH 6.0, both failed to bind to cation exchange resins (SP-Sepharose and mono S). Despite only 33% amino acid sequence identity, these recombinant proteins exhibit remarkably similar chromatographic profiles.

Isolation of Recombinant
Acyl-protein Substrate Preference Is Conserved between Rat and Yeast APT1-The native rat APT1 exhibits both acylthioesterase activity (toward acyl-proteins and acyl-CoA) and acyl-esterase activity (toward lysophospholipids). Under the in vitro assay conditions utilized, APT1 has a marked preference for the acyl-protein substrates (14). The recombinant rat and yeast enzymes were assayed using varying concentrations of three substrates: lysophosphatidylcholine, palmitoyl-CoA, and palmitoyl-G␣ i1 (Fig. 2). Like APT1 isolated from rat liver, re- combinant rat APT1 preferred palmitoyl-G␣ i1 substrate over either lysophosphatidylcholine or palmitoyl-CoA. For the yeast enzyme, this preference was even more pronounced, with a Ͼ2,000-fold difference in catalytic efficiency (V max /K m ) between palmitoyl-G␣ i1 and lysophosphatidylcholine. Having established that the enzyme encoded by S. cerevisiae open reading frame YLR118c has both biochemical activity and amino acid sequence similar to rat APT1, we will refer to YLR118C as the S. cerevisiae APT1 gene in the further description of our work.
Rat and Yeast APT1 Orthologs Differ in Acyl-protein Substrate Preference-The recombinant APT1 enzymes were assayed for acyl-protein thioesterase activity using a panel of palmitoyl-protein substrates: palmitoyl-Gi␣1, in which the thioacylated cysteine is adjacent to a myristoylated, aminoterminal glycine (26); palmitoyl-RGS4, in which multiple thioacylated cysteines are found several residues away from the amino terminus of the protein (27); and palmitoyl-Ras, in which two thioacylated cysteines are located near a farnesylated carboxyl-terminal cysteine (28). These assays were performed at substrate concentrations substantially below the enzymes' K m value for each, demonstrated by a linear relationship between substrate concentration and initial reaction rate (data not shown). Under these conditions, the relative catalytic efficiency of the enzyme for each substrate can be assessed by comparing the t for the completion of each reaction. Both rat APT1 and yeast Apt1p depalmitoylated these substrates in vitro. The recombinant rat APT1 showed a 10-fold variance in catalytic efficiency between the best (palmitoyl-G␣ i1 ) and worst (palmitoyl-Ras) acyl-protein substrates (Fig. 3A), roughly comparable with the values observed previously with native rat APT1 (24). Surprisingly, the yeast Apt1p demonstrated more specificity among acyl-protein substrates. The catalytic effi-ciency of yeast Apt1p for palmitoyl-G␣ i1 was at least 35-and 70-fold higher than that for palmitoyl-RGS4 and palmitoyl-Ras, respectively (Fig. 3A). Both recombinant enzymes were also able to discriminate between heterotrimeric and free palmitoyl-G␣ i1 (Fig. 3B). Again, as observed with the native enzyme, free palmitoyl-G␣ i1 was used as a substrate more efficiently than heterotrimeric palmitoyl-Gi␣1 by recombinant APT1 in this assay (14). The difference in the activity of yeast Apt1p toward heterotrimeric and free palmitoyl-G␣ i1 was greater than observed for the rat enzyme. We note, however, that assays of the yeast enzyme were performed only with mammalian proteins as substrates.
Yeast Strains Bearing a Deletion of the APT1 Gene-Having established that the protein encoded by YLR118C did indeed have acyl-protein thioesterase activity, we proceeded to create a yeast strain that lacked the APT1 gene to assess further the in vivo function of Apt1p. An APT1 targeting vector was created by adding ϳ350 base pairs of DNA from the 5Ј and 3Ј regions flanking the S. cerevisiae APT1 open reading frame to the ends of the HIS3 gene. The integration of this DNA fragment into the S. cerevisiae genome results in the removal of the entire open reading frame encoding Apt1p (Fig. 4A). The diploid S. cerevisiae strain, YPH501, was transformed with the APT1 targeting construct and selected for growth on medium lacking histidine. The resulting yeast colonies were confirmed as APT1/apt1⌬::HIS3 (referred to as the apt1⌬ allele) heterozygotes by performing the polymerase chain reaction on their genomic DNA with primer sets specific for either the wild type or disrupted allele. (Note that these primers cannot amplify the nonintegrated targeting vector; Fig. 4, A and B.) To obtain haploid yeast bearing the apt1⌬ allele, the diploid APT1/apt1⌬ yeast were induced to sporulate, and the resulting meiotic tetrads were dissected. The germinating spores all appeared to grow at the same rates, and the disruption marker (HIS3) appeared with equal frequency in both a-type and ␣-type haploids (data not shown). The presence of either APT1 or apt1⌬ alleles was confirmed in a set of haploid yeast germinated from a single tetrad utilizing the polymerase chain reaction (Fig. 4B,  lanes 4 -7).
Growth of Haploid Yeast Carrying the APT1 or apt1⌬ Allele-Because thioacylated proteins are involved in a variety of cellular processes, the apt1⌬ S. cerevisiae strain was tested for defects in normal growth. There were no obvious abnormalities observed in yeast strains that lacked the APT1 gene, including normal budding morphology (data not shown). The growth rates in complete synthetic medium of APT1 and apt1⌬ strains were essentially identical at three temperatures (25, 30, and 37°C; Fig. 5, A and B). There was also no difference in the growth rates of the APT1 and apt1⌬ strains in YPD, an enriched medium (data not shown). Normal haploid S. cerevisiae, cultured in the laboratory, exhibit three phases of growth. Initially, they expand exponentially, rapidly exhausting the dextrose included in the medium. As the dextrose in the culture is depleted, the growth of the yeast slows to a more linear rate. This slowing of growth is accompanied by an alteration in metabolism known as the diauxic shift. During growth after the diauxic shift, the glycolytic products present from nonoxidative utilization of dextrose are metabolized through an oxidative pathway. When the carbon source is exhausted, the yeast enter a dormant but viable state known as stationary phase, from which they emerge when a nutrient-rich environment is restored. The apt1⌬ yeast strains maintained their viability for at least 1 month after entering the stationary phase, similar to the wild type strains (Fig. 5C). Thus, it appears that APT1 does not play an important role in the normal growth stages of the haploid S. cerevisiae organism, including spore germination (observed during the isolation of apt1⌬ haploid yeast), normal growth, diauxic shift, or poststationary phase survival.
Apparently Normal Lipid Metabolism in apt1⌬ Yeast Strains-In addition to acyl-protein thioesterase activity, recombinant Apt1p has esterase activity (although less efficient) toward a number of other substrates. Several assays unrelated to acyl-protein metabolism were performed to determine whether Apt1p might be involved in some form of lipid (particularly lysophospholipid or acyl-CoA) metabolism within the yeast. There was no detectable difference in the levels of lysophospholipase activity in cell extracts derived from either APT1 or apt1 ⌬ strains of yeast (data not shown). Application of exogenous lysophospholipids to the medium of a yeast culture results in inhibition of growth (data not shown), a toxicity presumably attributable to the detergent like characteristics of the lysophospholipids. APT1 and apt1⌬ yeast had identical sensitivity to the growth-inhibitory effects of exogenous lysophospholipids (data not shown). To test for gross perturbations in acyl-CoA metabolism, the incorporation of radiolabeled fatty acids into cellular lipid pools was measured. No differences were observed between wild type and apt1⌬ yeast strains in the   FIG. 5. Normal growth of apt1⌬ haploid S. cerevisiae. A and B, the growth of yeast strains YAD101 (APT1; circles) and YAD103 (apt1 ⌬; squares) in minimal medium at three separate temperatures (filled black, 37°C; filled gray, 30°C; unfilled, 25°C) was measured as described under "Experimental Procedures." Growth from 0 to 14 h is plotted in A, showing the exponential phase of growth for these strains. Growth from 0 to 72 h is plotted in B, showing the diauxic shift and final culture densities. C, S. cerevisiae strains, YAD101 (APT1; circles), YAD102 (APT1; squares), YAD103 (apt1⌬; diamonds), and YAD104 (apt1⌬; triangles) were grown to stationary phase in YPD medium. The yeast were isolated and suspended in H 2 O. After various times at 30°C, the number of surviving yeasts was assessed by plating dilutions of each sample on YPD plates and counting the yeast colonies that grew. rate or the identity of lipids into which [ 3 H]palmitate was incorporated (Fig. 6A). An aberrant accumulation of a particular lipid species in the steady state would suggest an alteration of the metabolism of a specific lipid or group of lipids. However, no difference was detected between lipids extracted from either strain following metabolic labeling for 2 h with [ 14 C]acetate or [ 32 P]orthophosphate (Fig. 6B).
Acyl-Protein Thioesterase Activity in Haploid S. cerevisiae-Although S. cerevisiae Apt1p clearly exhibits acyl-protein thioesterase activity in vitro, it is possible that other acyl-protein thioesterase activities exist within S. cerevisiae. These activities may normally (or in the absence of Apt1p) be able to carry out deacylation reactions essential to the survival of the yeast, resulting in no obvious phenotype following deletion of the APT1 gene. Alternatively, these activities could be increased in apt1⌬ yeast, allowing them to carry out reactions that would normally be the responsibility of Apt1p. Extracts from both APT1 and apt1⌬ yeast strains were prepared and assayed for acyl-protein thioesterase activity. Although the activity present in the extracts was low when compared with mammalian cell extracts, the extracts from wild type yeast could clearly deacylate palmitoyl-G␣ i1 (Fig. 7A). Extracts from apt1⌬ strains exhibited dramatically lower, nearly undetectable activity (Fig.  7A). This biochemical phenotype was observed in several extracts prepared from apt1⌬ strains derived from two separate laboratory yeast strains (YPH501 and MMY030; data not shown). The apt1⌬ extracts had between 2 and 10% of the acyl-protein thioesterase activity detected in the wild type (APT1) extracts. However, because of the small percentage of substrate utilized in the reaction, we have not been able to confirm that the organic solvent-soluble, palmitate-containing material released from palmitoyl-G␣ i1 by extracts from the apt1⌬ yeast represents free fatty acid (rather than an organic solvent-soluble proteolysis product). Because differences in S. cerevisiae Apt1p activity toward various acyl-protein substrates had been observed in vitro, these cell extracts were also tested for their ability to deacylate palmitoyl-Ras. The palmitoyl-Ras thioesterase activity in wild type extracts was very low (0.36 pmol⅐min Ϫ1 ⅐mg Ϫ1 extract protein) and could only be detected reliably when the concentration of palmitoyl-Ras substrate was roughly 6-fold higher than in similar assays utilizing palmitoyl-G␣ i1 substrate. Interestingly, the extracts from apt1⌬ yeast exhibited a low level of activity toward palmitoyl-Ras substrates that was virtually indistinguishable from APT1 strain extracts ( Fig. 7; different extract preparations ranged from 75 to 110% of that detected in wild type extracts). The specific activity of both extracts toward palmitoyl-Ras, when corrected for the different concentrations of substrates, was very similar to that for the palmitoyl-G␣ i1 thioesterase activity observed in apt1⌬ yeast strain extracts. These results indicate that the product of the APT1 gene is responsible for almost all of the palmitoyl-G␣ i1 thioesterase activity detected in the S. cerevisiae strains examined but contributes very little, if any, to the detectable palmitoyl-Ras thioesterase activity.
Thioacylation of Gpa1p in apt1⌬ S. cerevisiae-To determine whether Apt1p plays a role in thioacyl-protein metabolism in vivo, we examined metabolic labeling of the protein Gpa1p with [ 3 H]palmitate. Wild type and apt1⌬ yeast strains were transformed with plasmids that direct overexpression of the yeast heterotrimeric G protein ␣ subunit: Gpa1p (with a carboxylterminal HA tag) from a copper-inducible promoter (25). This protein, when expressed in yeast, can complement the growtharrested phenotype of gpa1⌬ yeast strains. It has also been shown to incorporate [ 3 H]palmitate (as well as a metabolic product, myristate) through a thioester bond in metabolic labeling studies (29). Overexpressed Gpa1p was metabolically labeled with [ 3 H]palmitate in wild type and apt1⌬ yeast. (This radiolabel can be removed by treatment with hydroxylamine, consistent with incorporation as a thioester.) The yeasts were subsequently washed and placed in a medium containing high concentrations of unlabeled palmitate and myristate. At various time intervals, the yeasts were removed from the medium, the Gpa1p was isolated by immunoprecipitation, and the 3 H label associated with the protein was measured. In the wild type yeast, labeled palmitate was incorporated into Gpa1p, and this label was lost in a time-dependent fashion after the addition of unlabeled fatty acids (Fig. 8, A and B). The initial rise in incorporated label between 0 and 30 min in "chase" medium is likely to represent the time required for significant uptake of exogenous, unlabeled fatty acids into the intracellular acyl-CoA pool. In the apt1⌬ yeast, the accumulation of 3 H label associated with Gpa1p was approximately one-half that seen in the wild type yeast. Furthermore, the label associated with Gpa1p was stable after the addition of unlabeled fatty acids (Fig. 8, A  and B). To ensure that the apparent difference in the metabolism of Gpa1p thioacylation was not due to alterations in the metabolism of the entire protein, we analyzed the metabolism of Gpa1p using a combination of [ 35 S]methionine/cysteine as a metabolic label. The HA-tagged Gpa1p in this experiment was lost at the same rate in both wild type and apt1⌬ yeast (Fig. 8,  C and D). Thus, S. cerevisiae bearing a deletion of the APT1 gene appear to be unable to deacylate heterotrimeric G protein ␣-subunits in vivo.
We note that the rate of loss of radiolabeled palmitate from Gpa1p in apt1⌬ yeast appeared to be less than the rate of turnover of the protein in these cells. We suspect that this is caused by the slow chase of the high concentration of [ 3 H]palmitate used in labeling, coupled with continued synthe -FIG. 7. APT1 is required for normal acyl-protein thioesterase activity in S. cerevisiae. A, extracts from YAD101 (APT1; circles) or YAD103 (apt1⌬; squares) were assayed for acyl-protein thioesterase activity using 60 nM palmitoyl-G␣ i1 as a substrate. B, palmitoyl-G␣ i1 thioesterase activity (60 nM substrate), palmitoyl-Ras thioesterase activity (340 nM substrate), and lysophospholipase activity (3 M substrate) of extracts described in A (solid, APT1; striped, apt1⌬) were assayed as described under "Experimental Procedures." Initial reaction rates were determined by linear regression analysis of palmitate release at four or more time points. sis of Gpa1p. (Note the lag in chase of [ 3 H]palmitate in the wild type cells.) It is also possible that there are two populations of Gpa1p, since the protein was overexpressed in the absence of the G protein ␤␥ subunit complex. A more rapidly turning over, nonacylated population of protein would explain this discrepancy.
Pheromone Response of apt1⌬ S. cerevisiae-At least two proteins (Gpa1p and Ste18p) involved in the yeast mating signal transduction cascade are known to be thioacylated. The pheromone response pathway in apt1⌬ strains (and their wild type counterparts) was tested using two assays: the "halo" bioassay and an assay of the ␤-galactosidase reporter gene linked to the pheromone-responsive promoter of the FUS1 gene. When mating factor from the opposite mating type is applied to a growing lawn of haploid S. cerevisiae, the yeast around the spot undergo cell cycle arrest, while the rest of the yeast continue to grow. This creates a cleared area, or halo, around the spot of pheromone. Because pheromone will diffuse from the point of application, the concentration of pheromone surrounding the spot is inversely proportional to the square of the distance from the spot. Consequently, the diameter of the halo is proportional to the sensitivity of the yeast strain to pheromone-induced growth arrest. Eventually, yeast within the halo will become desensitized to pheromone and resume growth. The filling or "turbidity" of the halo, then, is proportional to the rate at which the yeast are able to adapt to pheromone. The wild type and apt1⌬ yeast produced halos of equal size and turbidity in response to equal quantities of pheromone applied (Fig. 9A). The mating factor signaling cascade ultimately leads to the transcription of a number of genes involved in the mating process. To assay for this transcriptional activation, wild type and apt1⌬ yeast were transformed with a plasmid carrying the E. coli ␤-galactosidase cDNA transcribed from the FUS1 promoter. The level of ␤-galactosidase activity in the yeast was assayed after challenge with pheromone. In this assay, the apt1⌬ yeast consistently showed modestly elevated ␤-galactosidase activity compared with the wild type strains (Fig. 9B). However, when the ␤-galactosidase activity is plotted as a percentage of maximal activity attained in each strain, it becomes apparent that the elevation was seen at all concentrations of pheromone tested (including no pheromone; Fig. 9C). The most straightforward explanation of these results is that APT1 is not exerting an effect on the hormone responsiveness of the reporter in this assay. DISCUSSION In mammalian tissues, the bulk (if not all) of biochemically detectable acyl-protein thioesterase activity can be attributed FIG. 8. Gpa1p-associated thioacyl turnover is impaired in apt1⌬ S. cerevisiae. Wild type (APT1) and apt1⌬ yeast were transformed with the expression plasmid encoding Gpa1p (pKM1362-2), and Gpa1p expression was induced by the addition of CuSO 4 to the culture medium. A and B, the fatty acid stores of the yeast were depleted by growth in the presence of cerulenin (a fatty acid synthetase inhibitor), and Gpa1p was "pulse-labeled" with [ 3 H]palmitate for 20 min, followed by a "chase" with unlabeled palmitate and myristate. After the indicated times of chase, Gpa1p was isolated by immunoprecipitation, resolved by SDS-PAGE, transferred to nitrocellulose, and analyzed using SDS-PAGE and phosphorimaging (see "Experimental Procedures"). A, the resulting image from an experiment is shown. B, the quantitative analysis of time points performed in triplicate are plotted in arbitrary units of Gpa1p-associated [ 3 H]palmitate. C and D, the yeasts were grown in medium lacking both methionine and cysteine. Subsequently, the Gpa1p was pulse-labeled with [ 35 S]methionine/cysteine for 20 min, followed by a chase with unlabeled methionine and cysteine. After the indicated times of chase, Gpa1p-associated radiolabel was assessed as described above. C, the resulting image from an experiment is shown. D, the quantitative analysis of time points performed in quadruplicate are plotted as a percentage of the initial label incorporated into Gpa1p; label incorporated in wild type and apt1⌬ strains was equivalent in each case. The results depicted in A and B are representative of three independent experiments, and results in C and D are representative of two independent experiments. to two enzymes, PPT1 and APT1 (14). PPT1 is a lysosomal enzyme and appears to act on thioacylated lipopeptide substrates, which may be derived from the breakdown of intracellular thioacylated proteins (30). Unlike PPT1, where a mutant (essentially null) allele exists in humans, our knowledge about the substrate(s) and physiologic role of APT1 are primarily inferred from biochemical studies of the enzyme. Because the entire S. cerevisiae genome has been sequenced, it is known that the organism contains no protein homologous to PPT1 and only a single open reading frame, YLR118C, encoding a protein homologous to APT1.
We felt that S. cerevisiae was an attractive system for studies of APT1 for at least three reasons: 1) methods for genetic manipulation of the organism are well established and easy to utilize; 2) heterotrimeric G protein signaling occurs in the organism (however, unlike mammalian systems, there are only two receptor-G protein combinations (31)); and 3) thioacylation of heterotrimeric G proteins and other proteins has been reported in the organism (7). The first step in establishing S. cerevisiae as a model organism for the study of APT1 was to demonstrate that the protein encoded by the YLR118c open reading frame had biochemical properties similar to mammalian APT1, as suggested by their homologous primary amino acid sequence.
Using recombinant enzymes, we have shown that the acylprotein thioesterase activity of APT1 is conserved between R. norvegicus and S. cerevisiae and that the palmitoyl-G␣ thioesterase activity is more highly conserved than the other catalytic activities of the enzymes from these species. The activities of both recombinant and native rat APT1 toward lysophospholipids are comparable with those observed by other groups (V max between 2 and 4 mol⅐min Ϫ1 ⅐mg Ϫ1 ), suggesting that this is indeed the maximum rate at which APT can hydrolyze acyl-ester bonds (15,33). In all cases, the V max toward acyl-thioester substrates appears to be at least 10 times greater than this number. Camp and Hofmann (23) found no acylprotein thioesterase activity in several known esterases (including a B. cereus phospholipase C, Crotalus adamanteus FIG. 9. apt1⌬ yeast exhibit no gross alteration in response to pheromone. A, lawns of either YAD101 or YAD103 were grown on YPD plates. Filter discs containing either 0, 5, or 20 g of ␣-factor were placed on the plate after the yeasts were plated; no ␣-factor is on the left filter, 5 g of ␣-factor is on the upper right filter, and 20 g of ␣-factor is on the lower right filter. B and C, YAD101 (APT1; circles) and YAD103 (apt1⌬; squares) were transformed with the plasmid pSB234, which carries the ␤-galactosidase gene fused to the pheromone-responsive FUS1 promoter. After reaching an A 600 of 0.5, the indicated concentration of ␣-factor was added to each culture. After an additional 4 h at room temperature, ␤-galactosidase activity was assayed. B, total ␤-galactosidase activity is plotted. C, the ␤-galactosidase activities determined in B are expressed as a percentage of the maximal activity seen in each strain. phospholipase A 2 , Candida cylindracea lipase, and Oryctolagus cuniculus carboxylesterase); thus, thioesterase activity is not a general property of esterase enzymes. Instead, the data suggest that the active sites of at least two APT1 enzymes are better suited to catalyzing a thioesterase reaction than an esterase reaction.
Comparisons of the K m values for all of the substrates described in the literature are confounded by varying critical micellar concentrations of each substrate and differences in type and concentration of detergent used. In the work described herein, the concentration and type of detergent was constant for all substrates. Differences in the observed K m toward any substrate should then reflect a variance in the affinity of APT1 for each substrate. Again, APT1 enzymes from Rattus norvegicus and S. cerevisiae had at least a 10-fold (and as high as several hundredfold) higher apparent affinity for palmitoyl-G␣ i1 substrate than for small molecule substrates (either palmitoyl-CoA or lyso-PC).
In contrast to mammalian APT1, the yeast enzyme did exhibit a marked preference for palmitoyl-G␣ i1 over other acylprotein substrates. The physiological relevance of this substrate preference is unclear. The in vivo rate of thioacyl group turnover for most thioacylated proteins in S. cerevisiae has not been reported. It is possible that these rates may be proportional to the catalytic efficiency of S. cerevisiae Apt1p toward each particular substrate. Alternatively, other acyl-protein thioesterase activities or nonenzymatic mechanisms for protein deacylation may exist in S. cerevisiae.
To define further the function of APT1 in vivo, we generated a strain of S. cerevisiae lacking the APT1 gene. We sought evidence that Apt1p, either through protein deacylation or some other esterase activity, was important in some aspect of the growth or survival of S. cerevisiae. The phenotype of APT1/ apt1⌬ heterozygote diploids appears normal with regard to growth and sporulation when compared with the wild type strain (data not shown). Both the phenotype and levels of acyl-protein thioesterase activity of wild type and apt1⌬/apt1⌬ diploid yeast will be the subject of future investigation. Haploid apt1⌬ yeast appeared to behave identically to their wild type counterparts during all phases of growth under a variety of conditions (Fig. 5).
Apt1p is unlikely to function as a lysophospholipase in S. cerevisiae. The lysophospholipase activity of S. cerevisiae Apt1p detected in vitro can be described as poor at best. Additionally, S. cerevisiae carrying the apt1⌬ allele show no alteration in biochemically detectable lysophospholipase activity. Previous studies have shown that yeast that lack the PLB1 gene exhibit almost no detectable lysophospholipase activity in vitro and do not metabolize lysophospholipids in vivo, suggesting that Plb1p is the primary lysophospholipase activity in S. cerevisiae (34). Other roles for Apt1p in lipid metabolism also seem unlikely, since no gross alterations in palmitate metabolism or lipid accumulation were detected in the apt1⌬ strain (Fig. 6).
Biochemical analyses indicate that Apt1p is the primary enzyme in S. cerevisiae responsible for palmitoyl-G␣ i1 thioesterase activity. Extracts prepared from strains of haploid yeast carrying the apt1⌬ allele had a dramatic reduction in biochemically detectable palmitoyl-G␣ i1 thioesterase activity (Fig. 7). Interestingly, our data suggest that a low level of thioesterase activity (no more than 10% of total) that cannot discriminate between palmitoyl-Ras and palmitoyl-G␣ substrates appears to exist in S. cerevisiae. When compared with mammalian cells, even wild type haploid S. cerevisiae exhibit very low levels of acyl-protein thioesterase activity. In light of the normal growth characteristics of the apt1⌬ yeast, it is possible that the lower level of thioesterase activity in these yeast is sufficient for the protein deacylation needs of a haploid S. cerevisiae grown under laboratory conditions. We sought to determine whether Apt1p was involved in the in vivo deacylation of Gpa1p. In metabolic labeling studies utilizing [ 3 H]palmitate, apt1⌬ yeast accumulated less Gpa1pbound tritium than wild type counterparts. This finding is consistent with the incorporation of label into newly synthesized Gpa1p in the apt1⌬ strain and the incorporation of label into newly synthesized and previously existing Gpa1p in the wild type strain. In wild type yeast, the incorporated label turned over when the yeast were incubated with unlabeled fatty acids. When apt1⌬ yeasts were incubated with unlabeled fatty acids, the Gpa1p-associated tritium remained stable. These in vivo experiments demonstrate that apt1⌬ yeasts are unable to deacylate Gpa1p effectively and strongly correlate with our findings that extracts from these yeast lack significant acyl-protein thioesterase activity in vitro.
Mutations in the palmitoylation sites of three thioacylated proteins from S. cerevisiae are known to produce distinguishable phenotypes in behaviors regulated by those proteins. Mutation of Cys 318 in Ras2p produces a deleterious growth phenotype similar to but milder than that observed in the ras2 null strain (35). When the thioacylation site of S. cerevisiae G␥, STE18, is removed, yeasts carrying the mutant gene are nearly sterile, mating with 100-fold less efficiency than wild type yeast. Finally, S. cerevisiae carrying the gpa1 C3A or C3S allele exhibit supersensitivity to mating factor (7,29). These phenotypes are presumed to be the result of loss of thioacylation of the protein in question, rather than the loss of the cysteine residue per se. The APT1 gene of S. cerevisiae represents the first identified yeast protein involved in the metabolism of thioacylated proteins. Although we have shown that yeast carrying a disrupted allele of this gene have a defect in the turnover of Gpa1p-associated thioacyl groups, the yeast have no obvious phenotypic alterations, either in general growth characteristics or response to pheromone (Figs. 5 and 9). Thioacylation has been shown to affect interactions between G␣ subunits and regulatory proteins in vitro. Consequently, we hope the future use of apt1⌬ yeast in combination with the wide array of mutations in the pheromone response pathway that already exist will serve to shed further light on the in vivo role of the enigmatic process of thioacylation in heterotrimeric G protein signaling.