Cloning and expression of a cDNA encoding the beta-subunit (30-kDa subunit) of bovine brain platelet-activating factor acetylhydrolase.

Bovine brain platelet-activating factor (PAF) acetylhydrolase isoform Ib is a heterotrimeric enzyme. Its gamma-subunit (which, formerly, we called the 29-kDa subunit) acts as a catalytic subunit, whereas the alpha-subunit (45 kDa) is the bovine homolog of the product of human LIS-1, the causative gene of Miller-Dieker lissencephaly, indicating that this intracellular PAF acetylhydrolase plays a key role in brain development. In the current study, we cloned the cDNA for the beta-subunit (30 kDa) of bovine brain PAF acetylhydrolase Ib. The predicted 229-amino acid sequence was homologous (63.2% identity) to that of the gamma-subunit, especially (86% identity) in the catalytic and PAF receptor homologous domains. The recombinant beta-protein produced in Escherichia coli showed significant PAF acetylhydrolase activity. A mutant protein, in which Ser48, which corresponds to the active serine residue of the gamma-subunit, was replaced with cysteine showed no enzymatic activity, suggesting Ser48 is the active serine residue. Although the beta- and gamma-subunits form a heterocomplex in the native enzyme, both recombinant beta- and gamma-proteins exist as a homodimer. The purified recombinant beta-protein was labeled readily with [1,3-H]diisopropyl fluorophosphate, whereas the beta-subunit in the native complex was only labeled with higher concentrations of [1,3-3H]diisopropyl fluorophosphate to a lesser extent than the gamma-subunit. Combined with our previous data, the present study demonstrated that bovine brain PAF acetylhydrolase Ib is a unique enzyme possessing two catalytic subunits and another, possibly regulatory, subunit.

Moreover, PAF has been reported to serve as part of a retrograde messenger in long-term potentiation (7) and to induce differentiation of cultured neuronal cells, although high concentrations of PAF are neurotoxic (8). These data suggest that an additional role of PAF is to regulate brain development and function(s).
PAF is inactivated by a specific enzyme, PAF acetylhydrolase, which removes the acetyl moiety at the sn-2 position of the glycerol backbone (9,10). Mammalian PAF acetylhydrolase is classified into two types (11,12), plasma (extracellular) and tissue (intracellular). Recently, we demonstrated that there are at least three isoforms of PAF acetylhydrolase in bovine brain (13), kidney, and liver (14) and that one isoform (designated isoform Ib) consists of three different subunits, with molecular masses of 45, 30, and 29 kDa (in this paper, we call them the ␣-, ␤-, and ␥-subunits, respectively) (13). We have already reported cDNA cloning of the ␥- (15) and ␣-subunits (16). The ␥-subunit was found to function as a catalytic subunit, and Ser 47 of this subunit was the active serine residue, as only this residue reacted with diisopropyl fluorophosphate (DFP), a potent inhibitor of the enzyme (15). The active serine residue occurs earlier on in its sequence than it does in other serine esterases (usually 100 -200 residues from the N-terminal end), and the sequence surrounding it differs from the consensus sequence (Gly-Xaa-Active Ser-Xaa-Gly) (17) of the serine esterase family (15). Even more interesting is the demonstration that the amino acid sequence of the ␣-subunit was almost identical (410/411) to that of the human LIS-1 gene product (16). LIS-1 was identified as the causative gene of Miller-Dieker lissencephaly (18), a brain malformation manifesting as a smooth cerebral surface and abnormal neuronal migration (19,20). This finding indicates that the ␣-subunit plays an important role in brain development, probably by regulating PAF acetylhydrolase activity and thereby maintaining physiological concentrations of PAF.
In order to understand the precise role of PAF acetylhydrolase in the brain, it is very important to elucidate the particular functions of all its subunits and how they interact with each other. In this paper we report the cDNA cloning of its ␤-subunit and have now succeeded in revealing the primary structures of all the brain PAF acetylhydrolase Ib subunits. From the predicted amino acid sequence and transfection experiment results, we found that the ␤-subunit, like the ␥-subunit, can function as a catalytic subunit.

Materials-1-O-Hexadecyl-2-[ 3 H-acetyl]
-sn-glycero-3-phosphocholine and [1, H]diisopropyl fluorophosphate ([ 3 H]DFP) were purchased from DuPont NEN. Unlabeled PAF was from Bachem Feinchemikalien AG (Bubendorf, Switzerland), and endoproteinase Lys-C (sequencing grade) was from Boehringer Mannheim. All the other materials were * 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EMBL Data Bank with accession number(s) D49678.
from Wako Pure Chemical (Osaka, Japan). Bovine brain PAF acetylhydrolase Ib was purified, as described previously (13). The fractions obtained after the hydroxyapatite chromatography step were used as "partially purified PAF acetylhydrolase." General Methods-Standard molecular biological techniques were used (21). The DNA sequence was determined by the method of Sanger et al. (22) using a Taq dye primer cycle sequencing kit and an Applied Biosystems model 373A DNA sequenator. PAF acetylhydrolase activity was determined as described previously (13). Protein concentration was determined using a BCA protein assay kit (Pierce).
Amino Acid Sequences of Peptides Derived from the ␤-Subunit of Bovine Brain PAF Acetylhydrolase Ib-PAF acetylhydrolase Ib was purified from bovine brain (13) and its subunits were separated using the reverse-phase HPLC system described previously (15). About 100 g of the HPLC-purified ␤-subunit was digested with endoproteinase Lys-C and separated by reverse-phase HPLC, as described previously (15). The amino acid sequences were determined using an Applied Biosystems model 477A automated sequenator.
cDNA Cloning of the ␤-Subunit-A bovine brain cDNA library was synthesized, as described previously (14). The ␤-subunit cDNA was cloned by performing mixed oligonucleotides-primed amplification (23). Two degenerate oligonucleotides, AARGARCCNGAYGTNYT and NARNGGRTTNGGYTTKT (where R represents A or G, Y represents C or T, K represents G or T, and N represents A, G, C, or T) based on the peptide sequences KEPDVL and EKPNPL of peptides 1 and 2 (see Table I), respectively, were synthesized and used as polymerase chain reaction (PCR) primers. Initial screening of the total bovine brain cDNA revealed the presence of a 336-base pair amplified product, which was sequenced, and two oligonucleotides based on it were synthesized: CCT-GTTCGTGGGGGACTC and CCCAGACAACGATGACCT. These specific primers were used for further screening of the library utilizing PCR, as described elsewhere (24). Briefly, the cDNA plasmid library was distributed in 96-well plates, the supernatants from every column and row were pooled, and PCR was carried out using the specific primers to find wells containing ␤-subunit cDNA. The positive pooled samples were plated, and every colony was checked by PCR using specific primers.
Expression and Partial Purification of the Recombinant ␤-Protein-The coding area of the ␤-subunit cDNA was amplified by PCR. The 5Ј oligomer GTAGAATTCATGAGCCAAGGAGACTCAAACC, which contained the EcoRI site in addition to the ␤-subunit sequence, and the 3Ј oligomer CAGGTCGACTCAGGCAATGGTGG, which contained the SalI site and the complementary sequence of ␤-subunit cDNA, were synthesized and used as primers. The amplified product was subcloned into the pUC-P L -cI vector (Suntory Co. Ltd., Osaka, Japan) using (EcoRI/SalI) restriction enzymes. This vector was introduced into E. coli W3110, and the resulting transformants were cultured in LB medium containing 50 g/ml ampicillin at 32°C. At the late log phase, the temperature was increased to 42°C, and culture was continued for 2 h. Then the cells were collected by centrifugation at 5000 ϫ g for 10 min, washed twice with phosphate-buffered saline, and resuspended in SET buffer (10 mM Tris-HCl (pH 7.4), 1 mM EDTA, and 250 mM sucrose). The cells were disrupted by sonication and centrifuged at 100,000 ϫ g for 1 h, and the supernatant was collected. The recombinant ␤-protein was partially purified by subjecting the supernatant to sequential DEAE Sepharose CL-6B (Pharmacia Biotech Inc.) and hydroxyapatite column chromatographies, as described previously (14).
Mutant Construction and Expression-For site-directed mutagenesis, the coding region of the ␤-subunit cDNA, obtained as described above, was ligated into the pET-21a(ϩ) vector (Novagen) using the (EcoRI/SalI) restriction enzymes and used both as a PCR template and for positive control expression. First, two oligonucleotides, GAAGC-CCGGGCCACCATGGCTAGCATGACTGGTGGACAG and GCAT-CAACTGTACCATGCAGTCCCCCACG, were used as PCR primers to introduce a mutation (Ser 48 was replaced by Cys at the underlined site).
The amplified product and TTAGCAGCCGGATCTCAGTGGTGG were then used as PCR primers, and the resulting amplified product was ligated into the pET-21a(ϩ) vector using the (EcoRI/SalI) restriction enzymes. The nucleotide sequences of this mutant cDNA were confirmed by sequencing. The protein expressed by this system had a 16-amino acid leader peptide (T7-Tag TM ) on its N-terminal. Vectors containing cDNA for the wild type ␤-subunit (T7-␤-wt) or cDNA for the mutant (T7-␤-S48C) were introduced into Escherichia coli BL21 and cultured in LB medium containing 50 g/ml ampicillin. For expression, isopropyl-1-thio-␤-D-galactoside at a final concentration of 2 mM was added, the cells were cultured for 4 h at 37°C, and the cytosol fraction was prepared as described above.
Antibodies-Monoclonal antibodies against the ␥-subunit were prepared as follows. The partially purified recombinant ␥-protein was used to immunize BALB/c mice once every 2 weeks for a total of 5 times, after which the splenocytes obtained from the mice were fused with mouse myeloma cells (PAI) (25). An enzyme-linked immunosorbent assay and immunoblotting were utilized for screening, and two monoclonal antibodies (1F4 and 6D8) against the ␥-subunit were established. Polyclonal antisera against the ␣and ␤-subunits were obtained by immunizing New Zealand White rabbits with the purified ␣-subunit from bovine brain (16) and partially purified recombinant ␤-protein, respectively.
Immunoblotting-After SDS-PAGE was carried out, the proteins were transferred onto a nitrocellulose membrane, which was blocked with Tris-buffered saline containing 5% skim milk for 2 h. It was incubated at room temperature with the indicated antibodies for 2 h and then washed with Tris-buffered saline containing 0.05% Tween 20. It was then incubated with horseradish peroxidase-conjugated second antibodies in Tris-buffered saline for 2 h and washed with Tris-buffered saline containing 0.05% Tween 20, and the color was developed using a ECL coloring kit (Amersham Corp.).
Cross-linking of the Native Heterotrimeric Complex and the Recombinant Proteins-Purified PAF acetylhydrolase Ib, purified recombinant ␤-protein, and purified recombinant ␥-protein were dialyzed against 100 mM sodium phosphate buffer (pH 8.0) and 10% (v/v) glycerol. The samples (100 g/ml) were incubated with 2 mM BS 3 for 30 min at room temperature. To the reaction mixtures were added half of the volume of 3 ϫ Laemmli sample buffer, and SDS-PAGE was carried out, as described previously (13).
DFP Labeling-In Fig. 8, purified PAF acetylhydrolase Ib at the indicated concentration in 50 mM KH2PO4-KOH (pH 6.8) and 10% (v/v) glycerol was incubated with the indicated concentrations of [ 3 H]DFP (in propylene glycol, 222 GBq/mmol) for 20 min at room temperature. SDS-PAGE and fluorography were carried out as described previously (13). RESULTS cDNA Cloning of the ␤-Subunit and Its Homology to the ␥-Subunit-The ␣-subunit of bovine brain PAF acetylhydrolase Ib was separated by heparin-Sepharose CL-4B column chromatography as described previously (data not shown) (13,16). The complex of ␤and ␥-subunits thus obtained was reduced, Salkylated, and separated by reverse-phase HPLC (Fig. 1). Judging from the absorbance at 214 nm, the amounts of ␤and ␥-subunits were almost equal, supporting the idea that the ␤and ␥-subunits exist in a 1:1 ratio in the complex. The ␤-subunit was digested with endoproteinase Lys-C, and the peptide fragments were purified by HPLC (data not shown). The Nterminal amino acid sequence could not be obtained, possibly because of some modification. Table I shows the sequences of the three peptides obtained from the HPLC. On the basis of TABLE I Sequences of peptides derived from the ␤-subunit of bovine brain PAF acetylhydrolase Ib after digestion with endoproteinase Lys-C The sequences of HPLC-purified peptides isolated from the Lys-C-digested ␤-subunit of bovine brain PAF acetylhydrolase Ib were determined. Each peptide represents a pure species from a single HPLC peak. Asterisks denote ambiguous residues. The amino acid sequences of all three peptides were found in the cDNA clone (Fig. 2 cDNA Cloning of the ␤-Subunit of Brain PAF Acetylhydrolase these sequences, degenerate oligonucleotides were synthesized and used in a series of PCR using bovine brain cDNA as a template. One set of primers yielded an amplified product containing sequences corresponding to the peptides. Therefore, we concluded that this amplified product was generated from the ␤-subunit cDNA and utilized it to screen a bovine brain cDNA library. The nucleotide sequence of the ␤-subunit cDNA is shown in Fig. 2. The amino acid sequence deduced from the nucleotide sequence contained all the sequences of the three peptides (see Table I).
Interestingly, and unexpectedly, the deduced amino acid sequence of the ␤-subunit exhibited a high degree of similarity to that of the ␥-subunit (63.2% identity and 82.9% similarity), which previously we had concluded to be the catalytic subunit of this enzyme (15) (Fig. 3). Conservative residues were spread over almost all the sequence. The most conserved lay between the active serine residue (circled in Fig. 3) and the PAF-receptor-like domain (double underlined in Fig. 3), in which 31 of the 36 residues (86%) are identical. The total amino acid sequence of the ␤-subunit is homologous to no other proteins reported so far, including human plasma PAF acetylhydrolase (26).
Expression of the ␤-Subunit cDNA in E. coli-Two vectors, pUC-P L -cI and pET-21a(ϩ), were used for expression of the ␤-protein in E. coli. Cultured at 42°C, E. coli transfected with the ␤-subunit cDNA in the former vector generated a protein that migrated to the same position as the ␤-subunit of the native enzyme when subjected to SDS-PAGE (Fig. 4A, lanes 1  and 3). This recombinant ␤-protein showed significant PAF acetylhydrolase activity (Fig. 4B, column 3). E. coli transfected FIG. 1. Separation of the ␤-subunit from the ␥-subunit by reverse-phase HPLC. About 200 g of the ␣-subunit-depleted bovine brain PAF acetylhydrolase Ib, purified as described previously (16), was reduced and S-alkylated as described previously (15). The reaction mixture was then applied to a reverse-phase HPLC system with a 4.6 ϫ 250-mm Vydac 304-1251 C4 column previously equilibrated with 5% acetonitrile containing 0.1% trifluoroacetic acid. Proteins were eluted with a linear gradient of acetonitrile (5-50%) containing 0.1% trifluoroacetic acid. The effluent was monitored by absorbance at 214 nm. The peak fractions were collected manually and analyzed by SDS-PAGE. Inset, the peak 1 and 2 fractions were applied to SDS-PAGE using 12% acrylamide slab gel, and the gel was stained with silver.
FIG. 2. Nucleotide and predicted amino acid sequences of the cDNA encoding the ␤-subunit of bovine brain PAF acetylhydrolase Ib. Nucleotide residues are numbered on the right; amino acid residues are numbered on the left. Residue 1 is the putative initiator methionine. Underlined sequences represent the sequences of peptides obtained from the purified ␤-subunit digested by endoproteinase Lys-C. The putative active serine residue is circled.
cDNA Cloning of the ␤-Subunit of Brain PAF Acetylhydrolase with the ␤-subunit cDNA in the latter vector generated a ␤-subunit with 16 extra amino acids on its N-terminal end (T7-␤-wt), which also possessed PAF acetylhydrolase activity (Fig. 4, A and B, lane and column 6). We utilized the pET-21a(ϩ) vector for site-directed mutagenesis.
Replacement of the Putative Active Serine Residue with Cysteine-We inferred that Ser 48 was the active serine residue of the ␤-subunit, as its alignment appears to correspond to that of Ser 47 , which we determined previously to be the active serine residue of the ␥-subunit of bovine brain PAF acetylhydrolase Ib (15) (Fig. 3). In order to confirm this, we replaced Ser 48 of the recombinant ␤-protein with cysteine by site-directed mutagenesis. This mutation resulted in complete loss of enzyme activity, whereas expression levels in E. coli were not affected significantly (Fig. 4, A and B, lanes and columns 7 and 8). We confirmed that the expression level was indeed unaffected by immunoblotting with an anti-T7-Tag monoclonal antibody (data not shown).
Immunoblotting and DFP Labeling Analysis-We purified the recombinant ␤-protein (Fig. 4A, lane 4) and raised an antiserum against it in rabbits. This antiserum recognized the ␤-, but not the ␥-, subunit (Fig. 5B, left). We also established two monoclonal antibodies against the ␥-subunit, neither of which recognized the ␤-subunit (Fig. 5B, right, using monoclonal antibody 1F4. Data using monoclonal antibody 6D8 not shown).
The PAF acetylhydrolase activity of the recombinant ␤-protein was abolished by 0.1 mM DFP (data not shown). When incubated with [ 3 H]DFP at a concentration of 17 M, the purified recombinant ␤-protein was covalently labeled by this reagent (Fig. 5C, lane 2). The purified recombinant ␥-protein was also labeled under the same conditions (Fig. 5C, lane 3). However, as shown in our previous study (13), the ␤-subunit of the native heterotrimeric enzyme was hardly labeled at all under these conditions. (Fig. 5C, lane 1).
Homodimer Formation of the Recombinant ␤and ␥-Proteins-We estimated the apparent molecular masses of the recombinant ␤and ␥-proteins by gel filtration column chromatography. The apparent molecular masses of the native isoform Ib and the complex from which the ␣-subunit was dissociated (i.e. heterodimer of the ␤and ␥-subunits) were about 100 and 60 kDa, respectively (Fig. 6, a and b). On the other hand, those of both recombinant ␤and ␥-proteins were about 60 kDa, respectively (Fig. 6, c and d). These data suggest that both recombinant ␤and ␥-proteins form a homodimer. In order to confirm dimer formation of the recombinant proteins, a crosslinking experiment was also performed using the cross-linking reagent BS 3 . By mixing each recombinant protein with 2 mM BS 3 , both recombinant proteins generated ϳ60-kDa bands on SDS-PAGE, (Fig. 7A, lanes 2 and 3). These 60-kDa bands reacted with the respective specific antibodies (Fig. 7B, lanes 5  and 9), suggesting a homodimer formation of the recombinant proteins. On the other hand, when the native PAF acetylhydrolase Ib was incubated with BS 3 , two major cross-linked products were seen on SDS-PAGE; a broad band between about 80 and 100 kDa and a faint 60-kDa band (Fig. 7A, lane 1). The cross-linked product of the ϳ100-kDa band reacted with all the antibodies against the ␣-, ␤-, and ␥-subunits, suggesting that this band is a product after cross-linking of these subunits (Fig.  7B, lanes 1, 4, and 7). The cross-linked product of ϳ80 kDa reacted with the antibodies against ␣and ␥-subunits. The cross-linked product of the 60-kDa band reacted with the an- cDNA Cloning of the ␤-Subunit of Brain PAF Acetylhydrolase tibodies against ␤and ␥-subunits. These results suggest that the 80-and 60-kDa bands are composed of the ␣and ␥-subunits and the ␤and ␥-subunits, respectively. These results may indicate that the ␣-subunit interacts only with the ␥-subunit but not with the ␤-subunit in the heterotrimeric complex. When the native PAF acetylhydrolase Ib was cross-linked with BS 3 , all the ␣-subunits migrated to a larger form on SDS-PAGE (Fig. 7B, lane 1), while some ␤and ␥-subunits still remained at their original positions (Fig. 7B, lanes 4 and 7). This results may indicate that some ␣-subunit got dissociated from the complex and was cross-linked with each other during crosslinking reaction.
Preferential Incorporation of DFP into the ␥-Subunit in the Native Complex-The ␤-subunit in the native complex was not labeled by DFP under the conditions we used previously, protein and DFP concentration of 1 mg/ml (about 10 M) and 17 M (molar ratio 1:1.7), respectively (Fig. 5C, lane 1, and Ref. 13). The recombinant ␤-protein was, however, labeled readily with [ 3 H]DFP (Fig. 5C, lane 2) under the same condition. When the molar ratio of [ 3 H]DFP to the protein was changed from 1.7 to 3.3, the ␤-subunit in the native complex became weakly labeled with [ 3 H]DFP (Fig. 8, left). With the [ 3 H]DFP at a 10 molar ratio to the protein the ␤-subunit was labeled to about onethird the extent that the ␥-subunit was (Fig. 8, right). This was not artificial, nonspecific labeling resulting from high [ 3 H]DFP concentrations, because the ␣-subunit was not labeled at all under these conditions.

DISCUSSION
Of the many types of phospholipases that have been reported, bovine brain PAF acetylhydrolase Ib is the only one that consists of heterogeneous subunits. It consists of three subunits with estimated molecular masses of 45 (␣), 30 (␤), and 29 (␥) kDa, which, presumably, are present in a relative stoichiometry of one copy each (13). In the present study, cDNA cloning of the ␤-subunit of bovine brain PAF acetylhydrolase Ib showed that the ␤-subunit was highly homologous to the ␥-subunit (15) (Fig. 3). The highest homology was observed around both the active serine residue and PAF receptor-like domain (Fig. 3), and the recombinant ␤-protein exhibited significant PAF acetylhydrolase activity (Fig. 4). These data indicate that bovine brain PAF acetylhydrolase Ib has two different catalytic subunits.
The active serine residue of the ␤-subunit was inferred to be Ser 48 , as it apparently corresponded to Ser 47 , the experimentally determined active serine residue of the ␥-subunit (15) (Fig. 3). The mutant protein T7-␤-S48C, in which Ser 48 was replaced by cysteine, showed no appreciable PAF acetylhydrolase activity, which supports the idea that Ser 48 is the active serine residue of the ␤-subunit. The amino acid sequence surrounding Ser 48 differed from the well established consensus sequence of the serine esterase family (Gly-Xaa-active Ser-Xaa-Gly) (17), as did that of the ␥-subunit. This raises the possibility that intracellular PAF acetylhydrolase evolved independently of other lipases and transacylases with active serine residues. It is noteworthy that human plasma PAF acetylhydrolase (27,28), the primary structure of which was reported recently (26), also possesses the serine esterase family consensus sequence. Furthermore, there is no region of either the ␤or ␥-subunits of brain PAF acetylhydrolase that is homologous with any region of human plasma PAF acetylhydrolase. Plasma and intracellular PAF acetylhydrolases share some substrate specificity (13,29,30). In spite of this similarity, however, intracellular and extracellular PAF acetylhydrolases appear to have totally different structures and, consequently, probably serve different purposes.
Both results of gel filtration column chromatography and cross-linking experiments indicate that both recombinant ␤and ␥-proteins form a homodimer. Mixing of these recombinant proteins in vitro does not result in heterodimer formation. 2 Furthermore, we have not yet succeeded in separating the native ␤-subunit from the ␥-subunit without causing denaturation of the protein. In addition, the native ␣-subunit purified from bovine brain did not bind to either the recombinant ␤or ␥-protein. In our preliminary observation, the specific hydrolyzing activity against PAF of the recombinant ␤-protein (about 60 nmol/min/nmol protein) was higher than that of ␥-protein (about 25 nmol/min/nmol protein), and the simple sum of them (85 nmol/min/nmol protein) does not reach that of the native enzyme (about 130 nmol/min/nmol protein). When the ␤and ␥-subunits form a heterocomplex the activity may synergistically increase. Since no definite conclusion can be drawn from experiments using bacterial recombinant proteins, we are now trying to separate native ␤and ␥-subunits without denaturing and obtain stable transformants of mammalian cells that bear only one of the subunits or both of them.
Despite the high similarity between the ␤and ␥-subunits, the recombinant ␤and ␥-proteins behave very differently during their purification procedures. The NaCl concentrations with which the recombinant ␤and ␥-proteins were eluted on DEAE-Sepharose column chromatography were about 200 mM and 50 mM, respectively. The same is true of the KH2PO4 concentration on hydroxyapatite column chromatography; the recombinant ␤-protein was eluted with about 150 mM KH2PO4, cDNA Cloning of the ␤-Subunit of Brain PAF Acetylhydrolase whereas the recombinant ␥-protein eluted with about 30 mM. The concentrations of NaCl and KH2PO4 required for elution of the ␣-subunit-depleted PAF acetylhydrolase Ib (i.e. the heterodimer of ␤and ␥subunits) on DEAE-Sepharose and hydroxyapatite column chromatographies were about 100 and 50 mM, respectively. Those concentrations were just in between those of the recombinant proteins. These results suggest that PAF acetylhydrolase Ib is not a mixture of two co-purified enzymes (i.e. ␣-, ␤-, ␤-subunits and ␣-, ␥-, ␥-subunits) but a heterotrimeric enzyme.
We previously demonstrated that only the ␥-subunit in the native complex was labeled by [ 3 H]DFP (13). According to the present results, however, the recombinant ␤-protein was labeled well by this reagent. There are two possible explanations for these results. One is that the catalytic center of the ␤-subunit may be modified so that DFP cannot react with the active serine residue. The other is that although the ␤-subunit is catalytically active, the active serine residue of it is not accessible to DFP. The observation that increasing the ratio of [ 3 H]DFP to the enzyme resulted in the labeling of not only the ␥-subunit but also the ␤-subunit may support the latter hypothesis. Generally speaking, incorporation of DFP does not necessarily correlate with the extent of catalytic activity. For example, although the K m value for the plasma PAF acetylhydrolase (13.7 M) (28) and brain PAF acetylhydrolase Ib (12.5 M) are almost equal, their sensitivities to DFP are very different (plasma PAF acetylhydrolase is not completely inhibited by 1 mM DFP (31) whereas brain PAF acetylhydrolase Ib is completely inhibited by less than 0.1 mM DFP (13)).
It should be noted that the number of deduced amino acid residues and calculated molecular mass of the ␤-subunit (229 and 25,569 Da, respectively) are smaller than those of the ␥-subunit (232 and 25,865 Da, respectively), whereas the apparent molecular mass determined by SDS-PAGE of the former (about 30 kDa) is larger than that of the latter (about 29 kDa). This discrepancy does not result from errors during peptide sequencing, since the ␤-subunit was separated completely from the ␥-subunit (Fig. 1) and the amino acid sequence predicted from the cDNA contained the peptide sequences obtained from the purified ␤-subunit (Table I and Fig. 2). Furthermore, the recombinant ␤and ␥-proteins migrated to exactly the same positions as the native subunits when subjected to SDS-PAGE (Fig. 5A), and a specific antibody against the ␥-subunit did not recognize the ␤-subunit and vice versa (Fig. 5B). These results also suggest that the discrepancy does not result from posttranscriptional modification. According to our unpublished observations, 2 the native ␤-subunits and the recombinant ␤-protein were eluted by almost the same concentrations of acetonitrile on reverse-phase HPLC. Therefore, it appears that the discrepancy is due to the proteinous natures of the ␤and ␥-subunits. It is noteworthy that, despite intensive and repeated trials, we have not yet succeeded in raising an antibody or antiserum that recognizes both ␤and ␥-subunits. Immunogenicity may be poor, probably because the conserved region may be buried deep under a folded structure. An alternative possibility is that the region may also be conserved in immunizing animals, such as the rabbit and mouse. In fact, all the subunits of brain PAF acetylhydrolase are highly conserved among mammalian species (Ref. 32). 2 .
Cross-linking of the native PAF acetylhydrolase Ib resulted in generation of high molecular bands on SDS-PAGE. The ϳ100-kDa band resulted from cross-linking of the ␣-, ␤-, and ␥-subunits, supporting the idea that this enzyme is a heterotrimer of these three subunits. The ϳ80-kDa band resulted from cross-linking of the ␣and ␥-subunits, and the 60-kDa band resulted from cross-linking of the ␤and ␥-subunits. The simplest explanation for these results is that the ␤-subunit may not be directly attached to the ␣-subunit. At present, however, we cannot neglect the possibility that there are few residues available for the cross-linking reagent in the ␤-subunit of the native complex.
There is ample evidence showing that PAF and intracellular PAF acetylhydrolase play important role(s) in brain development and function. PAF receptors exist in both pre-and postsynaptic cells (6), and PAF has been suggested to be a potential retrograde messenger in CA1 hippocampal long term potentiation (7). Exposure of cultured neuronal cells to low concentrations of PAF has been reported to induce neuronal differentiation, whereas at high concentrations, PAF was toxic to the cells (8). Furthermore, we found that LIS-1, a causative gene of Miller-Dieker lissencephaly (18), is a human homolog of the bovine brain PAF acetylhydrolase ␣-subunit (16). Miller-Dieker lissencephaly manifests itself as a smooth cerebral surface and abnormal neuronal migration (19,20). Putting the data together, we conclude that it is important for neuronal cells to maintain the PAF concentration within a certain range, according to their status, and degradation by PAF acetylhydrolase is essential to this. Interestingly, 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. One candidate for the counterpart of this repeat is the PH-domain (35)(36)(37). Recently, we found that the ␣-subunit interacts strongly with ␤-spectrin (38), a cytoskeltal protein with a PH domain (39). In this context, the present finding that PAF acetylhydrolase Ib has two different catalytic units indicates that its activity is modulated in a delicate and complicated manner. Further investigation into the function of its subunits and their interactions with other cellular components is needed. cDNA Cloning of the ␤-Subunit of Brain PAF Acetylhydrolase