cDNA cloning and expression of intracellular platelet-activating factor (PAF) acetylhydrolase II. Its homology with plasma PAF acetylhydrolase.

Platelet-activating factor (PAF) acetylhydrolase, which inactivates PAF by removing the acetyl group at the sn-2 position, is widely distributed in plasma and tissues. We previously demonstrated that tissue cytosol contains at least two types of PAF acetylhydrolase, isoforms Ib and II, and that isoform Ib is a heterotrimer comprising 45-, 30-, and 29-kDa subunits, whereas isoform II is a 40-kDa monomer. In this study, we isolated cDNA clones of bovine and human PAF acetylhydrolase isoform II. From the longest open reading frame of the cloned cDNAs, both bovine and human PAF acetylhydrolases II are predicted to contain 392 amino acid residues and to exhibit 88% identity with each other at the amino acid level. Both enzymes contain a Gly-X-Ser-X-Gly motif that is characteristic of lipases and serine esterases. Expression of isoform II cDNA in COS7 cells resulted in a marked increase in PAF acetylhydrolase activity. An immunoblot study using an established monoclonal antibody against the bovine enzyme revealed that the recombinant protein exists in the membranous fraction as well as the soluble fraction. Isoform II is expressed most abundantly in the liver and kidney in cattle, but low levels were also observed in other tissues. The amino acid sequence deduced from the cDNA of isoform II had no homology with any subunit of isoform Ib. Interestingly, however, the amino acid sequence of isoform II showed 41% identity with that of plasma PAF acetylhydrolase. Combined with previous data demonstrating that isoform II shows similar substrate specificity to plasma PAF acetylhydrolase, these results indicate that tissue type isoform II and the plasma enzyme may share a common physiologic function.

Platelet-activating factor (PAF) acetylhydrolase, which inactivates PAF by removing the acetyl group at the sn-2 position, is widely distributed in plasma and tissues. We previously demonstrated that tissue cytosol contains at least two types of PAF acetylhydrolase, isoforms Ib and II, and that isoform Ib is a heterotrimer comprising 45-, 30-, and 29-kDa subunits, whereas isoform II is a 40-kDa monomer.
In this study, we isolated cDNA clones of bovine and human PAF acetylhydrolase isoform II. From the longest open reading frame of the cloned cDNAs, both bovine and human PAF acetylhydrolases II are predicted to contain 392 amino acid residues and to exhibit 88% identity with each other at the amino acid level. Both enzymes contain a Gly-X-Ser-X-Gly motif that is characteristic of lipases and serine esterases. Expression of isoform II cDNA in COS7 cells resulted in a marked increase in PAF acetylhydrolase activity. An immunoblot study using an established monoclonal antibody against the bovine enzyme revealed that the recombinant protein exists in the membranous fraction as well as the soluble fraction. Isoform II is expressed most abundantly in the liver and kidney in cattle, but low levels were also observed in other tissues. The amino acid sequence deduced from the cDNA of isoform II had no homology with any subunit of isoform Ib. Interestingly, however, the amino acid sequence of isoform II showed 41% identity with that of plasma PAF acetylhydrolase. Combined with previous data demonstrating that isoform II shows similar substrate specificity to plasma PAF acetylhydrolase, these results indicate that tissue type isoform II and the plasma enzyme may share a common physiologic function.
PAF is degraded by hydrolysis of the acetyl group at the sn-2 position of the glycerol backbone to produce the biologically inactive lyso-PAF and acetate. This reaction is catalyzed by a specific enzyme, PAF acetylhydrolase (15)(16)(17)(18). PAF acetylhydrolase is detectable in the cytosol of tissues and cells as well as plasma. The physiologic function of PAF acetylhydrolase has not yet been established, but several possible hypotheses have been proposed. Finally, PAF production may be regulated at the levels of both synthesis and degradation. This notion is based on the observation that PAF production was greatly enhanced in cells pretreated with phenylmethanesulfonyl fluoride, an inhibitor of intracellular PAF acetylhydrolase, on stimulation of platelets with thrombin (19). It was also demonstrated that plasma PAF acetylhydrolase effectively abolishes the inflammatory effects of PAF on leukocytes and the vasculature, indicating involvement of the enzyme in the maintenance of plasma PAF at certain levels (20). In addition to the degradation of PAF, PAF acetylhydrolase has the ability to hydrolyze short chain phospholipids and oxidized fragments of polyunsaturated fatty acids at the sn-2 position in a calciumindependent manner (21)(22)(23). From these biochemical properties, it was speculated that PAF acetylhydrolase may scavenge oxidized phospholipids produced inside or outside of the cells during oxidative stress. It is well known that during oxidation of low-density lipoprotein in vitro, significant amounts of lysophosphatidylcholine are produced in the low-density lipoprotein particles (23)(24)(25). Lysophosphatidylcholine may be formed by sequential oxidation of phosphatidylcholine in the surface coat of the low-density lipoprotein particle and subsequent hydrolysis of the oxidized phospholipids by plasma acetylhydrolase, which is usually associated with plasma lipoproteins (26,27).
PAF acetylhydrolases from several sources have recently been purified and their cDNAs have been cloned. Plasma PAF acetylhydrolase is a 45-kDa monomeric enzyme (20) that is usually associated with plasma lipoproteins, such as low-and high-density lipoproteins (20,26,27). The predicted amino acid sequence deduced from its isolated cDNA is unique and unrelated to that of any known lipase or phospholipase (20,28). We previously demonstrated that the PAF acetylhydrolase present in the soluble fraction of bovine brain cortex can be separated into three isoforms, designated isoforms Ia, Ib, and II (29). In contrast to plasma PAF acetylhydrolase, isoform Ib, which is the most abundant form in bovine brain, is a heterotrimeric enzyme composed of 29-(␥), 30-(␤) and 45-kDa (␣) subunits (29). A heterodimer of the ␤ and ␥ subunits forms a catalytic unit in the native complex. These two catalytic subunits are homologous with each other (63% identity), but share no homology with any other protein, including plasma PAF acetylhydrolase, except that a sequence of about 30 amino acids located 6 residues downstream from the active serine exhibits significant homology to the first transmembrane region of the PAF receptor (30 -32). The ␣ subunit has a 7 tandem WD-40 repeat (33,34), which is often found in proteins that function through interaction with other protein components and is identical to the product of the causative gene for Miller-Dieker lissencephaly, a malformation of the brain cortex (35). Thus, this enzyme appears to play an important role in the signal transduction system for developement of the central nervous system. Very recently, we also succeeded in purifying isoform II and revealed that it differs from isoform Ib with respect to its polypeptide composition, substrate specificity, and tissue distribution, suggesting that it serves a different physiologic function from isoform Ib in tissues (36).
In this paper, we report the cloning of isoform II of tissue type PAF acetylhydrolase and demonstrate that it is distinct from isoform Ib but shows significant homology to the plasma PAF acetylhydrolase at the amino acid sequence level.
Amino Acid Sequence-PAF acetylhydrolase II was purified from bovine liver as described previously (36). Approximately 0.4 mg of purified enzyme was reduced with 1 mg of dithiothreitol for 2 h at room temperature and then S-alkylated with 0.6% (w/v) 4-vinylpyridine for 2 h at room temperature. This reaction mixture was applied to a Sephadex G-25 gel filtration column equilibrated with 70% (v/v) formic acid. The protein solution was then dried using a Speedvac concentrator, and 0.2 ml of 70% formic acid solution containing 1% CNBr was added. The reaction mixture was incubated for 16 h at room temperature then dried again and dissolved in 0.1 ml of 70% formic acid. The fragments were applied to a reverse-phase high-performance liquid chromatography system with a Vydac 304-1251 C4 column pre-equilibrated with 5% acetonitrile containing 0.1% trifluoroacetic acid, and the peptides were eluted with a linear gradient of acetonitrile (5-70%, v/v) containing 0.1% trifluoroacetic acid. The amino acid sequence was determined using a Shimadzu automated sequencer. Lysyl endopeptidase fragments of purified isoform II were also prepared as described previously (36) and their amino acid sequence was determined as well as that of the cyanogen bromide fragment.
Preparation of Antibodies-A monoclonal antibody against bovine isoform II was raised as follows. Immunization was performed by intrasplenic injection as described previously (37). Briefly, about 40 g of antigen was resolved by SDS-PAGE using a 12% acrylamide gel and transferred to a 0.45 m meshed nitrocellulose membrane. The membrane was then stained with Coomassie Brilliant Blue, and the band containing the antigen was cut out and homogenized in 800 l of saline. Then 10 g of antigen was injected into the spleens of BALB/c mice through a 26 gauge needle. After the intrasplenic injection, three further immunizations were performed by peritoneal injection every two weeks. Three days after the final immunization, 1.6 ϫ 10 8 spleen cells were fused with 4.0 ϫ 10 7 PAI mouse myeloma cells by treatment with 0.5 ml of 50% polyethylene glycol 1500 (Boeringer Mannheim) for 1 min at 37°C. The cells were then washed with 10 ml of Dulbecco's modified Eagle's medium and resuspended in Dulbecco's modified Eagle's medium with 10% fetal bovine serum and IL-6 as described previously (38). The cells were grown in 96-well culture plates. Ten days after fusion, the supernatants of the hybridomas were tested for production of anti-PAF acetylhydrolase II antibodies using an enzyme-linked immunosorbent assay. Positive cells were cloned by limiting dilution. One of the established monoclonal antibodies was named 7F7.
A polyclonal antibody against the ␥ subunit of bovine isoform Ib was prepared as follows. The cDNA for the ␥ subunit of isoform Ib was inserted into pET21a (Invitrogen) and transfected into BL21-competent cells. The recombinant ␥ subunit was expressed in histidine-tagged fusion protein form. The fusion protein was then purified by manganese chelate affinity chromatography and hydroxylapatite column chromatography. One hundred micrograms of the purified recombinant protein was homogenized with an equal volume of Freund's adjuvant solution and injected intradermally into the back of an adult female New Zealand white rabbit. Two weeks after the first injection, a second injection was carried out using Freund's incomplete adjuvant. The third and fourth immunizations were carried out at 2-week intervals. One week after the final immunization, all blood was collected from the rabbit and the serum was used for polyclonal antibody analysis. The polyclonal antibody thus obtained was named 453. cDNA Library-A bovine kidney cDNA library was synthesized from poly(A) RNA using a cDNA synthesis kit from Life Technologies, Inc. The cDNA was ligated into pSPORT 1 and then transfected into Electro Max DH 10B-competent cells (Life Technologies, Inc.). The human brain library was obtained from Life Technologies, Inc.
Isolation and DNA Sequencing of PAF Acetylhydrolase II-The reverse transcription polymerase chain reaction (PCR) was performed to clone the cDNA. Two degenerated oligodeoxyribonucleotides, CKRT-GNGGDATCCAYTC and CARGARGCNGARGARAC, were synthesized based on peptide sequences EWIPHR and EEAEAEET, respectively. These oligodeoxyribonucleotides were used as PCR primers. Initial screening of total bovine kidney cDNA revealed the presence of a 380-base pair amplified product. This PCR product was sequenced and used as a basis for the synthesis of two oligodeoxyribonucleotides, GAATGGATCCCCCACCG and CAAGAGGCAGAGGAGAC. These specific primers were used for further screening, using PCR as described previously (39,40) with a slight modification. Briefly, the cDNA plasmid library was distributed into 96-well plates (2,000 clones/well), and the supernatants were pooled in every column and row. PCR was carried out with the specific primers to identify the wells containing cDNA for isoform II. The positive pools were then plated and screened by colony hybridization method (41). The colonies were transferred to a Hybond N ϩ nylon membrane (Amersham Life Science) and then treated as described previously (41). The PCR product was used as a probe. A fragment of the PCR product was subcloned into pUC18 (Pharmacia Biotech Inc.) and labeled with ECL probe-amp reagents (Amersham Life Science). The hybridized probe was detected using the Fluorescein Gene Images labeling and detection system (Amersham Life Science).
The DNA sequence was determined by the method of Sanger et al. Human Isoform II cDNA Cloning-The coding region of bovine isoform II cDNA was used as a probe to isolate human isoform II cDNA clones by colony hybridization. After recloning into M13 mp19 vector, the cDNAs of positive clones were deleted using a double-stranded nested deletion kit (Pharmacia). DNA sequencing was carried out using a Taq Dyedeoxy Terminal Cycle Sequence kit and an Applied Biosystems 373A fluorescence DNA sequencer.
Expression of PAF Acetylhydrolase II-Full-length isoform II cDNA was subcloned into pcDNA3 (Invitrogen) using a restriction enzyme (EcoRV/NotI). COS7 cells (7.5 ϫ 10 6 cells), grown in Dulbecco's modified Eagle's medium with 10% fetal calf serum, were collected in K-PBS (30.8 mM NaCl, 120.8 mM KCl, 8.1 mM Na 2 HPO 4 , 1.46 mM KH 2 PO 4 , and 5 mM MgCl 2 ) and then transfected with 30 g of pcDNA3 containing the cDNA for isoform II by electroporation using a Genepulser system (Bio-Rad). After 48 h of culture in Dulbecco's modified Eagle's medium with 10% fetal bovine serum, the cells were collected in phosphate buffered saline. The cells were washed twice with buffer A (10 mM Tris-HCl, pH 7.4, 1 mM EDTA, 250 mM sucrose), suspended in 500 l of buffer A, and disrupted by sonication in a vessel surrounded by ice for five periods of 5 s at 15-s intervals using a Branson sonifier. The same procedure was carried out for control cells transfected only with pcDNA3. The cell lysates were assayed for PAF acetylhydrolase activity as described previously (36). The cell lysate was then centrifuged at 100,000 ϫ g for 1 h to separate the soluble and pellet fractions. The pellet fraction was suspended in buffer A using sonication to disrupt the cells. Western blot analysis was performed on the soluble (16 g of protein) and pellet (10 g of protein) fractions of the transfected cells using the monoclonal antibody 7F7.
Immunoblotting-The test samples were resolved by SDS-PAGE on a 12% acrylamide gel by the method of Laemmli (43) and blotted onto a nitrocellulose membrane. The membrane was blocked with PBS containing 5% skim milk for 2 h at room temperature and then incubated in the supernatant of the 7F7 hybridoma or rabbit anti-␥ subunit serum diluted to 1/1,000 in PBS containing 5% skim milk for 12 h at 4°C. The filter was washed four times with PBS containing 0.05% Tween 20 (T-PBS) and incubated for 1 h with horseradish peroxidase-conjugated goat anti-mouse or anti-rabbit Ig polyclonal antibodies, diluted to 1/2,000 in PBS containing 5% skim milk. After washing the filter six times with T-PBS, the blots were detected by an enhanced chemiluminescence method using an ECL Western blotting detection set from Amersham Life Science. The results were visualized by fluorography using Hyperfilm -ECL (Amersham Life Science).
Preparation of Tissue Homogenate of Isoform II-Fresh bovine tissues were obtained from a local slaughterhouse and were processed within 3 h of slaughter. All procedures were carried out at 0 -4°C. The tissues were homogenized with a Waring blender in 2 volumes of buffer A, and then the homogenates were centrifuged at 10,000 ϫ g for 30 min to remove the bulk of the solid material. The protein concentrations of the homogenates were determined using BCA protein assay reagent with BSA as a standard. Western blot analysis was performed on the tissue homogenate (100 g of protein) with the monoclonal antibody 7F7 or the polyclonal antibody 453.
Analysis of DNA and Protein Sequences-Nucleotide and predicted amino acid sequences were analyzed using GENETYX programs (Software Development, Tokyo, Japan). Computer searches for protein sequences were performed using the Genome Net World-Wide Web server (address, http://www.genome.ad.jp/).

RESULTS
Sequence of the Peptide Fragments of PAF Acetylhydrolase II-As previously reported, the N-terminal amino acid sequence of the purified 40-kDa polypeptide could not be determined. One fragment obtained by digestion of the polypeptide with lysyl endopeptidase showed no perfect match with any sequence ever reported, indicating that this is a new enzyme (36). In this study, purified PAF acetylhydrolase II was again digested with lysyl endopeptidase and cyanogen bromide, and the peptide fragments were separated by reverse-phase highperformance liquid chromatography. Five new peptide sequences were determined.
Cloning of the PAF Acetylhydrolase II Gene-Using the pep- Nucleotide residues are numbered from 5Ј to 3Ј; the first residue of the ATG codon encodes the initiating methionine. The predicted amino acid sequence is displayed below the nucleotide sequence as a one-letter code starting from the methionine. Underlined sequences represent the sequences of peptides obtained from purified isoform II digested with lysyl endopeptidase or cyanogen bromide. The serine residue of the active site is circled.
tide fragment sequences as a base, degenerated oligonucleotides were synthesized and a series of PCRs was carried out, using bovine liver and kidney cDNA libraries as a template. One set of primers (CKRTGNGGDATCCAYTC and CAR-GARGCNGARGARAC) from a kidney cDNA library yielded an amplified product containing sequences corresponding to the peptides. This amplified product was used to screen a bovine kidney cDNA library to identify the clone containing the full sequence. A total of four independent clones were isolated. The clone possessing the longest insert was subcloned and sequenced. The nucleotide and predicted amino acid sequences of bovine PAF acetylhydrolase II cDNA are shown in Fig. 1. The full-length clone contained a 2.4-kilobase insert. The amino acid sequence deduced from the nucleotide sequence contained all the peptide fragment sequences. The ATG codon (Fig. 1,  1ϳ3) was designated as the translated initiation codon, since no other ATG codon existed between the methionine codon and the first in-frame stop codon TGA (Fig. 1, Ϫ89ϳϪ87). The cDNA contained an open reading frame encoding 392 amino acids, with a calculated molecular mass of 43,864 Da, which compares favorably with the molecular mass of 40 kDa estimated by SDS-PAGE. The predicted protein sequence has a Gly-X-Ser-X-Gly motif at Ser 236 , which is characteristic of lipases and serine esterases (44). The presence of this sequence explains the inhibition of the activity of this enzyme by the serine esterase inhibitor diisopropyl fluorophosphate (36). Replacement of Ser 236 in the recombinant isoform II by Cys using site-directed mutagenesis yielded a protein with no enzyme activity (data not shown), supporting the hypothesis that this serine residue is a catalytic center.
The cDNA for human PAF acetylhydrolase II was subsequently cloned from a human brain cDNA library by colony hybridization using the coding region of bovine isoform II cDNA as a probe. The predicted amino acid sequence of human isoform II is shown in Fig. 2. The cloned human PAF acetylhydrolase II cDNA encoded the same number of amino acids and exhibited 83.5% nucleotide and 88% predicted amino acid identity to bovine isoform II.
Expression of PAF Acetylhydrolase II cDNA in COS7 Cells-pcDNA3 was used to express PAF acetylhydrolase II in COS7 cells. The cDNA generated a protein that migrated to the same position as purified PAF acetylhydrolase II when subjected to SDS-PAGE followed by Western blotting (Fig. 3B). The recombinant protein showed significant PAF acetylhydro-lase activity (Fig. 3A). In contrast to the catalytic subunits of PAF acetylhydrolase Ib, both of which existed almost exclusively in the cytosolic fraction when expressed in COS7 cells (data not shown), recombinant isoform II existed in both the soluble and pellet fractions as determined by Western blot analysis (Fig. 3B).
Search for Homologous Proteins-Interestingly and unexpectedly, computer searches of protein sequences revealed that the predicted amino acid sequence for intracellular PAF acetylhydrolase II showed notable homology with plasma PAF acetylhydrolase (20) (Fig. 4). Human isoform II exhibited 43% predicted amino acid identity to human plasma PAF acetylhydrolase. A hydrophobic stretch comprising the first 20 residues of plasma PAF acetylhydrolase appears to form the signal sequence for secreted proteins. Mature protein purified from human plasma, however, starts with Ile 42 or Lys 55 , indicating cleavage by posttranslational modification or fragmentation during purification, although the mature protein showed full enzyme activity. These results indicate that the 54 N-terminal residues are not essential for expressing enzyme activity. Consistent with this, a mutation study also revealed that deletion of the first 53 amino acids rather enhances enzyme activity (28). When the amino acid sequences of both the plasma and intracellular enzymes were aligned using a computer program to maximize matching, intracellular isoform II was found to lack the first 50 residues of the plasma enzyme. Moreover, isoform II possessed no putative signal sequence motif in its N terminus, confirming the intracellular location of the enzyme. In human plasma acetylhydrolase, removal of 21 C-terminal amino acids caused only a slight loss of activity, but a 30residue deletion reduced catalysis to below the detectable limit. Interestingly, as shown in Fig. 4, the homologous regions ended with 22 C-terminal amino acids (plasma) and 8 amino acids (intracellular isoform II).
The computer search also revealed that isoform II showed homology in three distinct regions (81ϳ140, 200ϳ260, and 321ϳ380) with other proteins of mammalian and bacterial origin. A region surrounding Ser 236 in the Gly-X-Ser-X-Gly motif of isoform II exhibited high homology with the active sites of other lipases and esterase, suggesting that it is a member of the serine esterase family. A highly homologous region was observed in the N-terminal half (101ϳ142) of plasma and intracellular isoform II (Fig. 4), which is located upstream of the putative catalytic center of both enzymes. This region is in -FIG. 2. Comparison of the amino acid sequences of bovine and human PAF acetylhydrolase II. The upper and lower sequences represent PAF acetylhydrolase II from cattle and humans, respectively. Amino acid residues are numbered from the left. The asterisks and dots denote identical and similar residues, respectively. The large dot indicates the active serine residue of bovine isoform II. cluded in the 81ϳ140 region, which has some homology with lipases of bacterial origin (Fig. 5). In particular, the sequence LASXGFVV (124ϳ131) of isoform II almost completely matched a sequence from plasma PAF acetylhydrolase and other bacterial lipases. Finally, the C-terminal region of isoform II showed some homology with other, apparently unrelated proteins, such as NK-4 protein (45) and dihydrolipoamide acetyltransferase (46) (Fig. 5). The latter enzyme is a subunit of the pyruvate dehydrogenase complex. The pyruvate dehydrogenase complex in all organisms studied to date is a large multi-enzyme complex that catalyzes the oxidative decarboxylation of pyruvate, the transfer of the acetyl unit to coenzyme A, and ultimately the reduction of NAD ϩ . Dihydrolipoamide acetyltransferase catalyzes the second reaction. This enzyme also has some homology with isoform II around the catalytic serine residue. NK-4 protein was identified as the product of a gene the expression of which is increased after activation of T cells by mitogens or activation of NK cells by IL-2. The transcript, which encodes a protein with a mass of 27 kDa, appears to have a function common to the activation pathways of both NK cells and T cells, although its precise role in the cell remains unknown. The N-terminal region of this protein exhibits less, but significant, homology with the N-terminal region of isoform II.
Distribution of PAF Acetylhydrolase II in Bovine Tissues-Next, we examined the tissue distribution of isoform II by Western blot analysis using an established monoclonal antibody against purified bovine isoform II. As shown in Fig. 6, among the bovine tissues tested, isoform II was most abundant in the liver and kidney, although other tissues expressed detectable levels of the protein. We also performed Western blot analysis with a polyclonal antibody against the ␥ subunit of isoform Ib. Significant levels of the ␥ subunit were expressed in the brain, kidney, adrenal, ovary, and intestine. Interestingly, the catalytic subunit of isoform Ib was almost undetectable in bovine liver, in which isoform II is most abundant. These data thus indicate that these two intracellular PAF acetylhydrolases show a marked contrast in tissue distribution. DISCUSSION We previously demonstrated that intracellular PAF acetylhydrolase isoforms Ib and II are different from each other with respect to substrate specificity, tissue distribution, and polypeptide composition (36). In the current study, it was shown by cDNA cloning that these two intracellular PAF acetylhydrolases are also distinct from each other at the amino acid sequence level, indicating a different origin for the two enzymes. Interestingly, the amino acid sequence of isoform II exhibits 43% identity to that of plasma PAF acetylhydrolase. The plasma form of acetylhydrolase has been cloned from humans (20) and several other species (28). The human enzyme encodes a 441-amino acid protein that includes a predicted signal peptide for secretion. The deduced amino acid sequence is unique except for the Gly-X-Ser-X-Gly motif found around the active serine residue of most serine estrases and lipases. Ser 273 of the human plasma PAF acetylhydrolase was identified as an active site. The active serine residue of intracellular isoform II was inferred to be Ser 236 , as this apparently corresponds to Ser 273 of the plasma enzyme and the Gly-X-Ser-X-Gly motif occurs around this amino acid (Fig. 4). Complete loss of enzyme activity after mutation of this Ser residue to Cys also supports this hypothesis. The region surrounding the active serine residue (231ϳ243 of isoform II) exhibited an almost complete match with plasma acetylhydrolase (Fig. 4). The active site of plasma PAF acetylhydrolase, as well as those of other lipases, forms a catalytic triad comprising the nucleophilic residues Ser, Asp, and His. Using site-directed mutagenesis, it was proposed that Ser 273 , Asp 296 , and His 351 form a catalytic triad in human plasma PAF acetylhydrolase. Although we have not yet confirmed this, Asp 259 and His 314 of isoform II correspond to the amino acids comprising the catalytic triad of plasma PAF acetylhydrolase. Interestingly, the sequences surrounding these Asp and His residues exibit high homology between the plasma and intracellular enzymes, suggesting that the structures of the catalytic domains of these two enzymes are highly homologous. Crystallization and x-ray structural analysis of both enzymes should provide a definite conclusion.
The other region showing extensive homology between plasma and intracellular isoform II was observed in the Nterminal half (101ϳ142; Fig. 4). A computer search showed that this region also exhibited some homology with lipases of bacterial origin (Fig. 5). Although the function of this region is unknown at present, it may be assumed to be involved in substrate binding or recognition of lipid-water interfaces, since it is relatively hydrophobic on hydropathy plot analysis (data not shown).
As described above, isoform II and plasma PAF acetylhydrolase appear to be typical members of the serine esterase family and are possibly derived from a common ancestral gene. In contrast, several features of isoform Ib, the other form of intra-

FIG. 3. PAF acetylhydrolase activity and Western blot analysis of recombinant isoform II expressed in COS7 cells.
Cells were transfected with a control vector or a vector containing the full-length bovine isoform II cDNA as described under "Experimental Procedures." A, the PAF acetylhydrolase activity of transfected cell lysates was determined as described previously (36). 1, control; 2, transfectant. B, Western blot analysis was performed on the cytosolic and membrane fractions of the transfected cells as described under "Experimental Procedure". Briefly, the cytosolic and membrane fractions (16 and 10 g of protein, respectively) were subjected to SDS-PAGE and blotted onto a nitrocellulose membrane. Immunoblotting was performed with the monoclonal antibody 7F7. Lanes 1 and 3, cytosolic fraction transfected with control vector and the vector containing isoform II cDNA, respectively, lanes 2 and 4, membrane fraction transfected with control vector and the vector containing isoform II cDNA, respectively. cellular PAF acetylhydrolase, suggest that it does not belong to this family. First, it showed no significant homology with other lipases and esterases on computer search analysis. Second, it does not contain the typical Gly-X-Ser-X-Gly pentapeptide motif around its catalytic serine residue. We have recently succeeded in crystallizing the ␥ subunit, one of the catalytic subunits of isoform Ib, and have revealed its structure on x-ray. 2 According to this analysis, Ser 47 , Asp 192 , and His 195 form a catalytic triad. In most serine esterases, the Asp and His are separated by more than 20 amino acid residues (28,42), whereas the distance in the ␥ subunit is unusually short (only 4 residues). Finally, the tertiary folding of the catalytic subunit is strikingly similar to that of p21 ras and the GTP-binding domain of the ␣ subunit of trimeric G proteins. All of these experimental data support the notion that isoform Ib is a novel serine esterase.
Western blot analysis of recombinant bovine isoform II expressed in COS7 cells revealed that the enzyme was distributed in the membrane fraction as well as in the soluble fraction in the transfected cells. In our preliminary experiments, endogenous isoform II was also found to be distributed in both cytosolic and membranous fractions in cultured MDBK cells. Although PAF acetylhydrolase activity in rat liver is highest in the cytosolic fraction, a significant level of activity was also FIG. 4. Comparison of amino acid sequences between human PAF acetylhydrolase plasma form and isoform II. The upper and lower sequences represent human isoform II and human plasma form, respectively. Amino acid residues are numbered from the left. The asterisks and dots denote identical and similar residues, respectively. The dashes indicate gaps inserted to optimize alignment. The signal sequence is underlined. The arrows denote the N terminus of the purified PAF acetylhydrolase from human plasma. The large dots under the serine, aspartic acid, and histidine residues indicate the amino acid residues that form a putative catalytic triad.
FIG. 5. Alignment of the amino acid sequences of bovine isoform II and homologous proteins. The sequences of homologous proteins were derived from the GenBank and SwissProt protein sequence data banks. One-letter amino acid notation is used. Amino acid residue numbers are shown at both sides. Residues identical with those of isoform II are shaded. A, bovine PAF acetylhydrolase II; B, 28k lipase from Streptomyces sp.; C, lipase from Streptomyces albus; D, dihydrolipoamide acetyltransferase of Pseudomonas putida; E, lipase from Rhigomucor miehei; F, natural killer cell protein 4. detected in the light and heavy membrane fractions (16). Although the PAF acetylhydrolase activity present in membrane fraction has not yet been characterized, isoform II associated with the membrane fraction may at least in part contribute this activity. In contrast, the catalytic subunits of isoform Ib, which is expressed in MDBK cells, are exclusively located in the cytosolic fraction (data not shown). It is notable that the majority of PAF acetylhydrolase in human plasma is associated with plasma lipoproteins such as low-and high-density lipoproteins. Thus, both intracellular isoform II and plasma PAF acetylhydrolase appear to be relatively hydrophobic in nature and to have the ability to associate with membranous components inside or outside cells.
It has been proposed that one of the physiologic functions of plasma PAF acetylhydrolase is to degrade oxidized phospholipid produced in circulating lipoproteins (21). The substrate specificity of isoform II resembles that of plasma PAF acetylhydrolase, which can hydrolyze oxidatively fragmented fatty acyl chains attached to phospholipid molecules as well as to PAF. The liver, in which isoform II is most abundant, is known to lack the ability to produce PAF (4,5), indicating that isoform II in liver cells is not involved in the regulation of PAF production. Several investigators have reported that oxidized phospholipids are preferentially hydrolyzed by cellular phospholipase A2 (47)(48)(49)(50)(51). Housekeeping protective enzymes are thought to be necessary both inside and outside cells, since reactive oxygen species can be formed both intracellularly and extracellularly. The molecular nature of the enzyme responsible for this activity, however, remains to be determined. Intracellular PAF acetylhydrolase II and plasma PAF acetylhydrolase may be possible candidates for this reaction.