Molecular Cloning and Expression of Palmitoyl-protein Thioesterase 2 (PPT2), a Homolog of Lysosomal Palmitoyl-protein Thioesterase with a Distinct Substrate Specificity*

Palmitoyl-protein thioesterase is a lysosomal hydrolase that removes long chain fatty acyl groups from modified cysteine residues in proteins. Mutations in this enzyme were recently shown to underlie the hereditary neurodegenerative disorder, infantile neuronal ceroid lipofuscinosis, and lipid thioesters derived from acylated proteins were found to accumulate in lymphoblasts from individuals with the disorder. In the current study, we describe the cloning and expression of a second lysosomal thioesterase, palmitoyl-protein thioesterase 2 (PPT2), that shares an 18% identity with palmitoyl-protein thioesterase. Transient expression of a PPT2 cDNA led to the production of a glycosylated lysosomal protein with palmitoyl-CoA hydrolase activity comparable with palmitoyl-protein thioesterase. However, PPT2 did not remove palmitate groups from palmitoylated proteins that are substrates for palmitoyl-protein thioesterase. In cross-correction experiments, PPT2 did not abolish the accumulation of protein-derived lipid thioesters in palmitoyl-protein thioesterase-deficient cell lines. These results indicate that PPT2 is a lysosomal thioesterase that possesses a substrate specificity that is distinct from that of palmitoyl-protein thioesterase.

A growing number of proteins have been shown to undergo post-translational modification by fatty acids that are covalently linked to cysteine residues through a thioester bond (reviewed in Ref. 1). Usually, the modified cysteine residue is located in close proximity to the inner surface of the plasma membrane, and palmitate is the most commonly occurring fatty acid, although stearate, oleate, arachidonate, and small amounts of other fatty acids have also been described. Fatty acid modifications may contribute to intracellular protein localization by facilitating membrane binding and also by strengthening protein-protein interactions (reviewed in Ref. 2). It is also possible that the palmitate plays a structural role that is unique to each modified protein. Cycles of palmitoylation and depalmitoylation have been described for a number of intracel-lular proteins (3), but the relevant enzyme(s) that catalyze these processes have not been isolated or fully characterized, and the full significance of these cycles remains to be elucidated.
In an initial attempt to study the mechanism of depalmitoylation of Ha-Ras, we purified and cloned a thioesterase (palmitoyl-protein thioesterase, which will be referred to as PPT1) 1 that preferentially removes fatty acyl chains of 14 -18 carbons (i.e. myristate, palmitate, stearate, and oleate) from proteins in vitro (4,5). Although PPT1 was originally purified from the soluble fraction of bovine brain, it was later found to be targeted to lysosomes (6 -8). Mutations in PPT1 were recently found to cause the infantile form of neuronal ceroid lipofuscinosis (INCL) (9), a neurovisceral storage disorder. This disease is characterized by the accumulation of amorphous granular deposits in cortical neurons leading to early visual loss and progressive mental deterioration by age 3, with death by 8 -14 years of age (10,11). In vitro, small lipid thioesters that are substrates for PPT1 were found to accumulate in lymphoblasts from these patients (12). The lipid thioesters were labeled with [ 35 S]cysteine and their accumulation was blocked by cycloheximide, indicating that they are derived from acylated proteins. The accumulation of these compounds could also be prevented by the addition of recombinant PPT1 to the culture medium, via uptake of PPT1 into lysosomes through the mannose 6phosphate receptor pathway.
In this report, we present the cloning and expression of human recombinant PPT2, a homolog of PPT1 that shares an 18% identity at the amino acid level. We find that PPT1 and PPT2 have comparable palmitoyl-CoA thioesterase activities and neutral pH optima and that PPT2 (like PPT1) is a lysosomal enzyme. In contrast, PPT2 does not remove palmitate from palmitoylated protein substrates routinely used as substrates for PPT1. This result suggests that PPT2 has a distinct substrate specificity and may play a unique role in the hydrolysis of lipid thioesters in lysosomes.

EXPERIMENTAL PROCEDURES
Materials-The Genetrapper TM cDNA positive selection system, Superscript human lung cDNA library, restriction endonucleases, DNAmodifying enzymes, Lipofectin, and cell culture reagents were purchased from Life Technologies, Inc. Cloned Pfu DNA polymerase was from Stratagene. The Sequenase version 2.0 DNA sequencing kit, horseradish peroxidase-conjugated donkey anti-rabbit IgG, and enhanced chemiluminescence (ECL) reagents were purchased from Amersham Life Science. Prokaryotic expression vector pQE32 and nickelnitrolotriacetic acid-agarose were from Qiagen. Protein A-Sepharose * This work was supported by National Institutes of Health Grant NS35323 and a grant from the Robert A. Welch Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
General Methods-Standard molecular cloning techniques were employed (13). DNA or protein sequence homology searches were performed using the BLAST sequence comparison program of the National Center for Biotechnology Information (NCBI). Alignment of nucleotide residues and sequence analysis were performed using the Lasergene software package (DNASTAR, Madison, WI).
cDNA Cloning of Human PPT2-A contiguous alignment of homologous sequences was established using the published sequence of human PPT1 (14) and a number of expressed sequence-tagged human clones using the BLAST algorithm of the NCBI. Two clones (N24410 and H86385, designated A and B, respectively, in Fig. 1) were obtained from Genome Systems (St. Louis, MO) (I.M.A.G.E. Consortium ID numbers 262750 and 222677). These clones were derived from human melanocyte and retina cDNA libraries, respectively. Two additional clones (one of which is shown as clone C in Fig. 1) were obtained from a human lung cDNA library (1 ϫ 10 7 independent clones) in the double-stranded phagemid (pCMV-SPORT) using the Genetrapper TM cDNA positive selection system. Two expressed sequence tag-derived oligonucleotides (5Ј-AGATGGGACAGTATGGAGAC-3Ј and 5Ј-TGATCCCCACCACGAT-GACT-3Ј, corresponding to nucleotides 416 -435 and 540 -559 of the PPT2 coding region, Fig. 2) were used for cDNA capture and cDNA repair, respectively.
DNA Sequencing-Alkali-denatured double-stranded plasmids were sequenced using appropriate vector or synthetic internal primers and the Sequenase version 2.0 DNA sequencing kit. All clones were fully sequenced on both strands.
Expression of Recombinant PPT2 in Escherichia coli-A 0.9-kb SmaI/ PstI fragment of PPT2 (clone C, Fig. 1), corresponding to the coding region exclusive of the amino-terminal 64 residues, was excised from pCMV-SPORT and subcloned into the prokaryotic expression vector, pQE32. The recombinant protein was produced in E. coli strain M15 (pREP4) and purified under denaturing conditions on a nickel-nitrolotriacetic acid resin according to the manufacturer's instructions. Expression of this construct yielded a fusion protein with a His 6 affinity tag at the amino terminus of the protein.
Affinity Purification of Polyclonal Antiserum-The E. coli-produced recombinant protein was separated by preparative SDS-polyacrylamide gel electrophoresis and electroeluted from the gel (15). Polyclonal antiserum raised against purified recombinant human PPT2 was produced by Pocono Rabbit Farm (Canadensis, PA) using their standard fusion protein protocol. Antiserum to human PPT2 was diluted 1:1 with 0.1 M sodium phosphate buffer, pH 7.0, and loaded onto a protein A-Sepharose column. After washing with the same buffer, bound IgG was eluted from the column with 0.2 M glycine, pH 2.2. Fractions (0.5 ml) were collected into tubes containing 0.1 ml of 1 M K 2 HPO 4 . The purified IgG was dialyzed overnight against phosphate-buffered saline at 4°C and loaded onto a PPT2 affinity column. The PPT2 affinity column was prepared by coupling 1.4 mg of E. coli-produced protein to 200 mg of CNBr-activated Sepharose 4B according to the manufacturer's instructions, except that 0.5% Triton X-100 was included in the coupling buffer. The column was washed with buffer containing 25 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 1 mM EDTA, eluted with glycine buffer to recover the affinity-purified antibody as described above.
COS Cell Expression Constructs-To amplify the entire coding region of human PPT2, oligonucleotide primers corresponding to nucleotides Ϫ17 to ϩ3 and ϩ935 to ϩ954 relative to the translation start site of the PPT2 clone A (Fig. 1) (5Ј-ATCGGATCCCTTTCTCAGGCGGGAG-CATG-3Ј and 5Ј-TAAGGATCCTTGGTCTCTGGACCGAGGAG-3Ј) were employed in a polymerase chain reaction using clone A as the template in the presence of Pfu DNA polymerase (2.5 units/reaction). The 970base pair amplified product was subcloned into the BamHI site of the expression vector, pCMV5, to yield pCMV5-hPPT2. Insert orientation and polymerase fidelity was verified by restriction mapping and dideoxynucleotide sequencing. The human PPT1 expression plasmid, pCMV5-hPPT, has been previously described (12).
COS Cell Transfection-Simian COS-1 cells were maintained in monolayer culture in Dulbecco's modified Eagle's medium (high glucose) with 10% heat-inactivated fetal bovine serum, 100 units/ml penicillin, 100 g/ml streptomycin, and 0.25 g/ml amphotericin at 37°C in a humidified 5% CO 2 atmosphere. Cells growing in 60-mm plates were transiently transfected with 5 g of DNA and 10 l of Lipofectin reagent according to the manufacturer's instructions. Cells were harvested at 60 -72 h post-transfection into sonication buffer containing 50 mM Tris-HCl, pH 7.0, 150 mM NaCl, 2 mM EDTA, and 2 mM EGTA and homogenized by brief sonication for use in enzyme assays. Alternatively, cells were harvested directly for Percoll density gradient fractionation. Conditioned medium from transfected cells was pooled and centrifuged at 220 ϫ g for 5 min in a Beckman GS6R benchtop centrifuge, frozen under liquid nitrogen, and stored at Ϫ85°C. Percoll Density Gradient Fractionation-Three 60-mm dishes of COS-1 cells transiently transfected with either pCMV5 or pCMV5-hPPT2 were washed twice with phosphate-buffered saline and scraped into homogenization buffer containing 0.25 M sucrose, 20 mM Tris-HCl, pH 7.4, and 1 mM EDTA. Cells were gently lysed by 20 passages through a ball bearing homogenizer (H and Y Enterprises, Redwood city, CA) with a clearance of 51 m. Nuclei and unbroken cells were removed by centrifugation at 400 ϫ g for 10 min, and 0.8 ml of the resulting supernatant was layered over a discontinuous gradient consisting of a 1.2-ml cushion of 2.5 M sucrose and 8.5 ml of an 18% Percoll solution in homogenization buffer. The gradient was centrifuged for 60 min at 27,000 rpm (67,000 ϫ g max ) in a Sorvall T1270 rotor. The cushion and nine 1.0-ml fractions were collected from the bottom of the tube. Percoll was removed from the samples by centrifugation for 40 min at 70,000 rpm in a Beckman TL 100.3 rotor, and the resulting supernatants were homogenized by sonication. ␤-Hexosaminidase activity (lysosomal marker) was determined as described (16). Assays for other organellar marker enzymes (NADPH-cytochrome c reductase, alkaline phosphodiesterase-1, and ␣-mannosidase II for endoplasmic reticulum, plasma, and Golgi membranes, respectively) were performed as described previously (17).
Mannose 6-Phosphate-dependent Uptake of PPT2 by COS Cells-Serum-free conditioned medium containing recombinant human PPT was produced from plates of transfected COS-1 cells as follows. On day 0, five 60-mm dishes of COS-1 cells were transfected with pCMV5-PPT2 or control plasmid using Lipofectin reagent in the presence of serumfree medium according to the directions supplied by the manufacturer. On day 1, cells were fed with serum-containing medium, and on day 2, cells were changed to serum-free medium. The medium was harvested on day 3, concentrated 15-fold using a Centricon device, and diluted 1:3 (v/v) with fresh medium just prior to use. For PPT2 uptake experiments, untransfected COS-1 cells were plated at a density of 0.5 ϫ 10 6 cells/60-mm dish on day 0, and on day 1, PPT2-containing conditioned medium was added. Incubations were carried out at 37°C for 20 h in the presence or absence of 5 mM mannose 6-phosphate or 5 mM mannose. Cells were washed three times in ice-cold phosphate-buffered saline and scraped into sonication buffer. Cell extracts were homogenized by sonication and either assayed for palmitoyl-CoA hydrolase activity or analyzed by immunoblotting as described below.
Deglycosylation and Immunoblotting of Recombinant Proteins-Crude transfected COS-1 cell homogenates were digested with PNGase F or Endo H according to directions supplied by the manufacturer, electrophoresed in 12% SDS-polyacrylamide slab gels under reducing conditions, and transferred to nitrocellulose filters as described previously (6). Filters were incubated for 1 h in PBS-T (0.1% Tween-20 in phosphate-buffered saline) containing 5% (w/v) nonfat dry milk, washed in PBS-T, and incubated for 1 h with affinity-purified antibody (0.5 g/ml) in PBS-T. The filters were washed and incubated for 1 h with horseradish peroxidase-conjugated donkey anti-rabbit IgG (1:2000 in PBS-T) and developed using ECL reagents.
Palmitoyl-CoA Hydrolase and Palmitoyl-protein Thioesterase Assays-The palmitoyl-CoA hydrolase activity of transfected COS cells was determined by measuring the release of [ 3 H]palmitic acid from [ 3 H]palmitoyl-CoA, which was monitored by liquid scintillation after extraction with Dole's reagent (18). The standard reaction mixture contained 100 M palmitoyl-CoA (5,000 -6,000 cpm/nmol), 100 mM sodium acetate, pH 5.0, and enzyme in a final volume of 100 l. In some experiments, as indicated in the figure legends, the COS cell lysates were pretreated with 5 mM N-ethylmaleimide for 30 min at 37°C to reduce an endogenous palmitoyl-CoA hydrolase activity with significant interfering activity above pH 6 Cross-correction Experiments-Conditioned medium containing human recombinant PPT1 or PPT2 was prepared by transient transfection of COS-1 cells with pCMV5-hPPT1 or pCMV5-hPPT2 (or pCMV5, which served as a control). Medium was harvested at 72 h after transfection, concentrated approximately 10-fold using a Centricon concentrator, and mixed with 3 volumes of fresh medium just prior to use. The final concentrations of PPT1 and PPT2 (based on palmitoyl-CoA hydrolase activity) were 3.5 units/ml (1 unit ϭ 1 nmol of palmitate released/min) and 4.2 units/ml, respectively. (pCMV5-transfected COS-1 cell medium contained Ͻ0.1 unit/ml of palmitoyl-CoA hydrolase activity). Control and INCL human skin fibroblasts were maintained in Dulbecco's modified Eagle's medium (high glucose) with 20% heat-inactivated fetal bovine serum, 100 units/ml penicillin, 100 g/ml streptomycin, and 0.25 g/ml amphotericin at 37°C in a humidified 5% CO 2 atmosphere. Cells were plated at a density of 0.3 ϫ 10 5 cells/100 mm on day 0. On day 2, conditioned medium (10 ml) containing PPT1 or PPT2 was added. On day 4, the cells were washed twice with prewarmed cysteine and serumfree medium, resuspended in 5 ml of the same medium, and incubated at 37°C for 1 h. Fresh medium (5 ml) containing [ 35 S]cysteine (20 Ci/ml) was added, and the incubation was continued for 15 h. Labeled cells were washed twice with ice-cold phosphate-buffered saline and extracted into 6 volumes of chloroform/methanol (1:1, v/v), and the organic phase was analyzed by thin layer chromatography as described previously (12). To demonstrate uptake of recombinant PPT1 and PPT2 into fibroblasts, plates were set up in parallel, and cell lysates were analyzed for total cellular palmitoyl-CoA hydrolase activity and subjected to immunoblotting using anti-PPT1 or anti-PPT2 antibodies as described above.

RESULTS
cDNA Clones Encoding Human PPT2-A BLAST search of the Genome Database was used to identify a set of overlapping expressed sequence-tagged clones with homology to palmitoylprotein thioesterase. Fig. 1 shows three independent PPT2 cDNA clones that were fully characterized. Clones A and B (Fig. 1) were obtained from human melanocyte and human retinal libraries, respectively (through the I.M.A.G.E. Consortium). Clone C was obtained by screening a human lung cDNA library using the Genetrapper TM cDNA positive selection system, which utilizes oligonucleotide hybridization and magnetic bead separation to isolate specific cDNA-bearing plasmids. A second human lung PPT2 cDNA, identical to clone C but start-ing at nucleotide Ϫ9 relative to the initiation codon, was also fully characterized (data not shown).
Clones A and B (Fig. 1) each contained the entire coding region and 3Ј-untranslated region of human PPT2. Clones A and B were identical from nucleotide Ϫ9 to 1634 (with A of the initiator methionine as ϩ1). Clone A contained 113 nucleotides of genomic (intron) sequences at the 5Ј-end, as indicated by the dotted line in Fig. 1. This presumably arose by reverse transcription of an mRNA containing an unprocessed intron, since these sequences are also present in a genomic clone containing the PPT2 gene (see below) and contain sequences consistent with splice donor and acceptor sites. Clone C contained an additional 209 nucleotides of 5Ј-untranslated sequence as compared with clone B. In addition, the sequence of clone C diverged within the coding region starting at nucleotide 766 relative to the start codon and contained unique 3Ј-coding sequences and a unique 3Ј-untranslated region. Expression of the clone C isoform in COS cells resulted in a protein that was glycosylated and immunoreactive to anti-PPT2 antibodies but catalytically inactive in the palmitoyl-CoA assay (data not shown).
The nucleotide and predicted amino acid sequences for clone B (Fig. 1) are shown in Fig. 2. The amino acid sequence deduced from the cDNA sequence contains 302 amino acid residues. The initial 27 amino acids constitute a leader peptide (Fig. 2, panel A, underlined) as predicted from the algorithm of von Heijne (21) using the PSORT protein analysis program (22). A sequence motif characteristic of many thioesterases and lipases (⌿-X-⌿-X-Gly-X-Ser-X-Gly-Gly-X-X-⌿, where ⌿ represents a hydrophobic residue (23)) is present at residues 105-117, with two exceptions; a cysteine replaces the first glycine at residue 109, and a cysteine replaces the final hydrophobic amino acid at position 117. No discernible second thioesterase motif (Gly-Asp-His (24)) is present. Five potential N-linked glycosylation sites are found at amino acid residues 60, 190, 206, 245, and 289.
Aside from the homology to PPT1, no other significant homologies to known proteins were uncovered using the BLAST algorithm. However, exact homology to sequences in a 300-kb region of the human class III major histocompatibility complex locus (Genome Sequence Database accession number U89336) reveal that the PPT2 gene can be tentatively localized to human chromosome 6p21.3. Upon examination of this genomic sequence, we observe that clone C in Fig. 1 is an alternatively spliced form of PPT2 in which the final exon of PPT2 is skipped and the penultimate PPT2 exon is spliced into a gene that encodes a previously uncharacterized homolog of the latent transforming growth factor-␤-binding protein (LTBP). The splicing event occurs out of frame with respect to the LTBP reading frame, resulting in a novel 46 amino acids that replace 47 amino acids at the carboxyl terminus of the predicted PPT2 protein (Fig. 1, clone C, shaded box). A further six exons of the LTBP-homologous gene comprise the novel 3Ј-untranslated region of clone C. The entire clone C sequence (not shown) will be deposited with the Genome Sequence Data base. Fig. 3 shows a comparison of PPT2 with the published sequence of human PPT1 (14). The overall length of the protein, exclusive of leader peptide (275 amino acid residues), is similar to that of PPT1 (281 residues). Short contiguous regions of identical amino acid residues (shown in black boxes) are present throughout, and the overall sequence identity at the amino acid level is 18%.
RNA blot hybridization (Fig. 4) of human tissues using the clone C PPT2 cDNA (excluding the 5Ј-untranslated region) as a probe (Fig. 1)  examined, with the highest expression in skeletal muscle. Interestingly, PPT1 expression is strikingly low in skeletal muscle (14), suggesting a possible reciprocal relationship between PPT1 and PPT2 expression in some tissues. When the RNA blot shown in Fig. 4 was reprobed with sequences from various regions in clones A, B, and C (Fig. 1), the pattern of hybridization suggested that the major 2.0-kb mRNA encodes the active form of the enzyme (represented by clones A and B in Fig. 1) and that the 2.8-kb transcript encodes the inactive, alternatively spliced form (represented by clone C in Fig. 1) (data not shown). The origin of the 7.0-kb transcript remains to be determined.
To characterize the activity and properties of the human PPT2 enzyme, we amplified the entire coding region of clone A in Fig. 1 using synthetic oligonucleotides in a polymerase chain reaction. The amplified fragment was subcloned into a eukaryotic expression vector pCMV5 to yield the plasmid pCMV5-hPPT2, and expression in transiently transfected COS-1 cells was assessed by immunoblotting and enzyme assays. Immunoblotting of crude cell lysates of COS cells expressing recombi-nant PPT2 showed six immunoreactive bands, ranging in size from 31 to 42 kDa (Fig. 5, lane 1). No specific signals were observed using cells transfected with vector alone or with preimmune serum (data not shown). Incubation of the cell lysates with PNGase F, a treatment that removes asparagine-linked oligosaccharides, resulted in the collapse of all the bands to yield a single major band of 31 kDa and a minor band of 33 kDa (Fig. 5, lane 2). The 31-kDa band is in excellent agreement with the size predicted for the expressed protein after removal of the leader sequence (31 kDa). Treatment with Endo H to release  Fig. 1. The autoradiogram was exposed to film for 7 days at Ϫ85°C.

FIG. 5.
Immunoblotting of recombinant human PPT2 and sensitivity to endoglycosidases. Simian COS cells were transfected with pCMV5-hPPT2, and a crude homogenate was subjected to SDS-polyacrylamide gel electrophoresis and immunoblotting using affinity-purified polyclonal antiserum to human PPT2. Lane 1, no treatment; lane 2, after treatment with PNGase F; lane 3, after treatment with endoglycosidase H. The predicted molecular mass of the mature PPT2 polypeptide is 31 kDa.
high-mannose and hybrid-type oligosaccharides from the polypeptide, also gave similar results to PNGase F treatment (Fig. 5, lane 3). These results are consistent with the presence of high mannose or hybrid-type asparagine-linked oligosaccharide moieties attached to recombinant PPT2. Fig. 6 shows that recombinant PPT2 possesses thioesterase activity using palmitoyl-CoA as a substrate. A pH curve obtained using lysates from vector-or pCMV5-hPPT2-transfected COS-1 cells shows that the pH optimum of PPT2 is broad, with peak activity at 7.0 and half-maximal activity between 4.5 and 5.0. This neutral pH optimum was also observed previously for PPT1. Of note, the assays presented in Fig. 6 were performed after a 30-min preincubation of cell lysates (at 37°C) in the presence of 5 mM N-ethylmaleimide, to suppress an endogenous N-ethylmaleimide-sensitive palmitoyl-CoA hydrolase activity with significant interfering activity above pH 6.0. As in the case of PPT1, PPT2 is also insensitive to N-ethylmaleimide (data not shown). In addition, PPT2 (like PPT1), is sensitive to inactivation by diethylpyrocarbonate (5 mM), with 100% inactivation occurring in 30 min at pH 6.5 at 37°C (data not shown).
Lysosomal Localization of PPT2-A lysosomal localization of PPT2 seemed likely given the homology to PPT1 (a lysosomal enzyme), the presence of a signal peptide that is not present in the mature protein, the lack of a putative transmembrane domain, and the finding that the expressed enzyme is modified by Endo H-sensitive oligosaccharides, a characteristic of lysosomal (and endoplasmic reticulum) enzymes. In addition, a modest amount of the protein is secreted into the culture medium from transfected cells (data not shown), a finding also consistent with lysosomal enzyme trafficking. To determine whether PPT is targeted to lysosomes through the mannose 6-phosphate receptor pathway, we added concentrated culture medium from PPT2-overexpressing COS-1 cells to untransfected COS-1 cells and found that PPT2 was taken up and internalized by the untransfected cells (Fig. 7). There was a significant increase in palmitoyl-CoA hydrolase activity from cells treated with conditioned medium from pCMV5-hPPT2 transfected cells as compared with medium from vector-transfected cells (Fig. 7, top panel, lanes 1 and 4). This activity correlated with PPT2 immunoreactivity in lysates from those cells (Fig. 7, bottom panel, lanes 1 and 4). Furthermore, cellular uptake of PPT2 and palmitoyl-CoA hydrolase activity was inhibited by mannose 6-phosphate (Fig. 7, lane 5) but not by mannose (Fig. 7, lane 6). This result is diagnostic of uptake through the mannose 6-phosphate receptor pathway, a distinguishing feature of lysosomal enzyme transport (25).
To confirm the lysosomal localization of PPT2, we separated subcellular fractions from PPT2-transfected COS cells on a Percoll density gradient and analyzed the fractions for marker enzymes, palmitoyl-CoA hydrolase activity, and PPT2 immunoreactivity by immunoblotting. Using this technique, the pattern of palmitoyl-CoA hydrolase activity was found to colocalize with the marker enzyme for lysosomes (Fig. 8, top panel). PPT2 immunoreactivity is shown in Fig. 8, bottom panel, (Fig. 9). These substrates have all been shown to be cleaved by PPT1 in vitro (4) 2 COS-1 cells were transfected with 5 g of plasmid DNA encoding human PPT1 or PPT2, and crude lysates were analyzed in the assays. The expression of PPT1 and PPT2 protein as assessed by immunoblotting was found to be similar as determined by comparison with standards (purified recombinant Sf9-produced PPT1 and E. coli-produced PPT2 proteins) (data not shown). The palmitoyl-CoA hydrolase activity of the crude transfected COS cell lysates was found to be 3.2 nmol of palmitoyl-CoA hydrolyzed/min/mg of protein for PPT1 and 3.3 nmol/ min/mg for PPT2 (Fig. 9, upper panel). However, PPT2 did not remove palmitate bound to Ha-Ras (Fig. 9, middle panel) or human albumin (Fig. 9, lower panel) under these conditions.
We have previously shown (12)  we performed experiments designed to test whether PPT2 could substitute for PPT1 in correcting the metabolic defect in INCL cells. For technical reasons outlined below, fibroblasts were chosen for this study. (Previously, we had reported that we did not see the accumulation of [ 35 S]cysteine-labeled thioesters in INCL fibroblasts (12), but subsequently we have demonstrated this accumulation using an optimized labeling protocol, as described under "Experimental Procedures.") In the experiment shown in Fig. 10, normal human fibroblasts and INCL fibroblasts were analyzed for the accumulation of acylated [ 35 S]cysteine-containing lipids, and the effect of adding recombinant PPT1 or PPT2 was addressed. Fig. 10, lane 1 shows the background labeling that occurs in normal cells. In lane 2 (INCL fibroblasts), at least three abnormal bands are seen that are specific to the INCL cells. Conditioned medium containing PPT1 reduces the appearance of these abnormal bands back to the background level (lane 3). However, conditioned medium containing PPT2 did not change the appearance of the INCL-specific bands. To ensure that the recombinant PPT2 was internalized and active in this experiment, in parallel dishes, the fibroblasts were analyzed for PPT2 by immunoblotting (lower panels) and by palmitoyl-CoA hydrolase assays. We calculate that approximately 1 g of PPT1 and more than 1 g of PPT2 was internalized per 10 6 cells and that this amount is approximately 100-fold greater than the minimum amount of PPT1 needed to correct the defect, as determined in previous experiments using INCL lymphoblasts (12). In addition to fibroblasts, PPT2 failed to correct the defect in INCL lymphoblasts (data not shown), but in these experiments, internalization of PPT2 was more difficult to demonstrate by immunoblotting and enzyme assay, due to somewhat less efficient uptake of the PPT enzymes through the mannose 6phosphate receptor pathway, and due to a higher endogenous background thioesterase activity. DISCUSSION In the current paper, we describe the cloning, structure, and expression of PPT2, a homolog of human lysosomal palmitoylprotein thioesterase that shares an 18% identity with PPT1. Lysosomal targeting of PPT2 was demonstrated by mannose 6-phosphate-dependent uptake into COS cells and by subcellular fractionation and Percoll density gradient centrifugation. The presence of a signal sequence and high mannose oligosaccharides on the polypeptide are also consistent with a lysosomal localization. Interestingly, the PSORT sequence analysis program, which uses a discriminant score based on amino acid composition of lysosomal proteins (22), predicted a lysosomal localization for PPT2 with a discriminant score that was higher for PPT2 (0.550) than PPT1 (0.437).
PPT2 demonstrated thioesterase activity comparable with that of PPT1 when a model substrate (palmitoyl-CoA) was used in enzyme assays. Of note, the lipase consensus sequence in PPT2 is not perfectly conserved, with a cysteine for glycine substitution two residues away from the putative active site serine (Cys-X-Ser-X-Gly versus Gly-X-Ser-X-Gly). However, the relevance of this lipase consensus motif in thioesterases is in question, because mutation of the homologous serine to alanine in PPT1 does not completely abolish activity. 2 Furthermore, the first crystal structure of a thioesterase that was determined recently (myristoyl acyl carrier protein-specific thioesterase from the bacterium, Vibrio harveyi (26)) showed that a serine other than the serine in the lipase consensus sequence is present in the active site. Therefore, the substitution of cysteine for glycine at this position in PPT2 may be irrelevant to the geometry of the active site.
PPT2 does not contain a carboxyl-terminal Gly-X-His sequence, a sequence that is present in many thioesterases (24), and in which, in some cases, the histidine has been shown to participate directly in catalysis. However, like PPT1, PPT2 is sensitive to diethylpyrocarbonate, a reagent that selectively modifies histidine residues. Perhaps one of the two histidine residues (at position 283 or 287), which are located in a similar position to the active site histidine in PPT1 (at position 289), participates in catalysis. Further mutagenesis studies and future structural work will be needed to more fully characterize the catalytic mechanisms used by these two homologous thioesterases.
The neutral pH optimum observed for PPT2, which is also a characteristic of PPT1 (6), is unusual for lysosomal enzymes. Two other lysosomal enzymes (aspartylglucosaminidase (27) and sialic acid-specific O-acetylesterase (28)) have been previously reported to have neutral pH optima. Interestingly, all of these enzymes are believed to participate in the removal of post-translational protein modifications in the lysosome. One potential explanation for the existence of lysosomal enzymes that work at neutral pH is that the lysosome itself may undergo cycles of acidification, as suggested by Butor,et al. (29). Changes in pH may facilitate the sequential degradation of macromolecules. Alternatively, the neutral pH optimum may merely reflect ionizations in the enzyme-substrate complex that may differ between an acyl-CoA and any particular acylpeptide substrate. The pH optimum for the endogenous substrate(s) may well be in the acidic pH range.
The physiological function of PPT2 remains to be elucidated. PPT2 is presumably a very ancient thioesterase, since the Caenorhabditis elegans PPT1 shares over 50% amino acid identity with the human PPT1 (data not shown), yet PPT2 is only 18% identical to human and C. elegans PPT1. This finding implies that the function of PPT2 has been conserved throughout a long period of evolution that predates the divergence of humans and worms. PPT2 did not hydrolyze the acyl-cysteine bond of two of the protein substrates routinely used to assay PPT1, and it does not correct the metabolic defect in INCL cells deficient in PPT1. This finding suggests that although both enzymes possess intrinsic palmitoyl thioesterase activity, the "leaving group" recognized by the enzymes may differ. One possibility is that PPT2 recognizes palmitoylated protein substrates but that these substrates differ from those recognized by PPT1. Perhaps two enzymes are needed to recognize the broad range of protein substrates that may be present in the lysosome of a cell. A second possibility is that PPT2 recognizes a novel lipid thioester substrate that is not derived from acylated proteins. Clearly, more work is needed to completely define the substrate specificities of both of these lysosomal thioesterases.
During the cloning of PPT2, a 2.8-kb transcript that encodes an inactive form of PPT was identified in some tissues. Alternatively spliced, catalytically inactive versions of lysosomal enzymes have been observed previously (30,31), but in these instances, fusion transcripts involving downstream genes were not involved. The significance of these inactive lysosomal enzyme forms is unknown.
The chromosomal localization of the PPT2 gene on 6p21.3 argues against a major role for PPT2 in the etiology of the other major forms of neuronal ceroid lipofuscinosis where the molecular defect is still undefined. These include the classical late infantile form on 11p15 (32), the Finnish variant late infantile form on 13q.22 (33), and an atypical late infantile form on 15q21-23 (32). The relatively low expression of PPT2 in brain tissue as compared with PPT1 suggests that PPT2 deficiency may lead to a different phenotypic expression in affected humans or animals. A relevant animal model of PPT2 deficiency, coupled with studies to further define the substrate specificity of PPT2, would be useful in further defining the role of PPT2 in lysosomal function.