Biochemical Characterization of Various Catalytic Complexes of the Brain Platelet-activating Factor Acetylhydrolase*

Brain intracellular platelet-activating factor acetylhydrolase (PAF-AH) isoform I is a member of a family of complex enzymes composed of mutually homologous α1 and α2 subunits, both of which account for catalytic activity, and the β subunit. We previously demonstrated that the expression of one catalytic subunit, α1, is developmentally regulated, resulting in a switching of the catalytic complex from α1/α2 to α2/α2 during brain development (Manya, H., Aoki, J., Watanabe, M., Adachi, T., Asou, H., Inoue, Y., Arai, H., and Inoue, K. (1998) J. Biol. Chem. 273, 18567–18572). In this study, we explored the biochemical differences in three possible catalytic dimers, α1/α1, α1/α2, and α2/α2. The α2/α2 homodimer exhibited different substrate specificity from the α1/α1homodimer and the α1/α2 heterodimer, both of which showed similar substrate specificity. The α2/α2 homodimer hydrolyzed PAF and 1-O-alkyl-2-acetyl-sn-glycero-3-phosphorylethanolamine (AAGPE) most efficiently among 1-O-alkyl-2-acetyl-phospholipids. In contrast, both α1/α1 and α1/α2hydrolyzed 1-O-alkyl-2-acetyl-sn-glycero-3-phosphoric acid more efficiently than PAF. AAGPE was the poorest substrate for these enzymes. The β subunit bound to all three catalytic dimers but modulated the enzyme activity in a catalytic dimer composition-dependent manner. The β subunit strongly accelerated the enzyme activity of the α2/α2 homodimer but rather suppressed the activity of the α1/α1 homodimer and had little effect on that of the α1/α2heterodimer. The (His149 to Arg) mutant β, which has been recently identified in isolated lissencephaly sequence patients, lost the ability to either associate with the catalytic complexes or modulate their enzyme activity. The enzyme activity of PAF-AH isoform I may be regulated in multiple ways by switching the composition of the catalytic subunit and by manipulating the β subunit.

Brain platelet-activating factor acetylhydrolase (PAF-AH 1 isoform I) was first identified in bovine brain as an oligomeric enzyme consisting of two catalytic ␣ 1 and ␣ 2 subunits and a noncatalytic ␤ subunit (1). The oligomeric brain PAF-AH contains a dimer of two highly homologous catalytic subunits, ␣ 1 and ␣ 2 , that share 63% amino acid sequence identity but are not related to any other known proteins (2,3). The ␣ 1 and ␣ 2 subunits form a heterodimer in the PAF-AH purified from the bovine brain cortex (1). They can also form a catalytically active homodimer when expressed individually in Escherichia coli (3). Recently, the crystal structure of the ␣ 1 /␣ 1 homodimer was determined (4). Surprisingly, the ␣ 1 subunit has a tertiary fold reminiscent of small GTPases such as p21 ras and harbors a chymotrypsin-like Ser-Asp-His triad in its active site (4). Although the catalytic subunits of PAF acetylhydrolase were first identified as a ␣ 1 /␣ 2 heterodimer from young adult bovine brain (1), recent studies have revealed that the ␣ 2 /␣ 2 homodimers are present in vivo as well (5). We have demonstrated in a previous study that expression of the ␣ 1 polypeptide is restricted to actively migrating neurons at the embryonic and early postnatal stages in rats and that switching of the catalytic dimer from the ␣ 1 /␣ 2 heterodimer to the ␣ 2 /␣ 2 homodimers occurs in these cells during brain development (5,6). Although both the ␣ 1 /␣ 2 heterodimer and the ␣ 2 /␣ 2 homodimer hydrolyze PAF equally in vitro, it is essential to elucidate the differences between those catalytic dimers in order to understand the significance of the switching of catalytic subunits and the function of this enzyme. In the present study, we studied the biochemical properties of three possible catalytic dimers, ␣ 1 /␣ 1 , ␣ 1 /␣ 2 , and ␣ 2 /␣ 2 , and their complexes with the ␤ subunit.
One striking feature of brain PAF-AH isoform I is that the gene for the ␤ subunit is identical to a causative gene (LIS1) for Miller-Dieker lissencephaly (7). Miller-Dieker lissencephaly is a genetic brain malformation manifested as a smooth cerebral surface caused by abnormal neuronal migration at early developmental stages. Recent data demonstrated that deletions or mutations of the ␤/LIS1 gene accounts for approximately 40% of classic lissencephaly (8). Point mutations in the ␤ subunit were recently reported in the LIS1 gene from isolated lissencephaly sequence patients (9). In one case of the mutant ␤, histidine 149 was replaced with arginine. Although it is assumed that this amino acid substitution may produce a radical change in steric hindrance due to the loss of a imidazolic ring, direct evidence for the functional abnormality of this mutation has not been presented so far. The effect of mutation in the ␤ subunit on the interaction with ␣ subunit was also examined.

EXPERIMENTAL PROCEDURES
Baculoviruses-Recombinant ␣ 1 and ␣ 2 were expressed either by the E. coli system (3) or the baculovirus system. Recombinant ␤ was expressed by the baculovirus system, since protein was not recovered from soluble fractions of E. coli. The production of recombinant baculoviruses was designed as follows. The coding regions of bovine ␣ 1 , ␣ 2 , and ␤ * This work was supported in part by research grants from the Ministry of Education, Science, Sports, and Culture of Japan. 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.
cDNA were inserted into the SalI/NotI sites of pFASTBAC (Life Technologies, Inc.). His 149 3 Arg mutant ␤ was prepared as follows: two oligonucleotides, 5Ј-GGGGGGAATTCATGTGTCCTGGCCTCTGGGG-GACA-3Ј (nucleotide positions 124 of bovine ␤ cDNA (7)) and 5Ј-GAG-CCCCCAGAGCGACACCAATGA-3Ј (nucleotide positions 485-508 of bovine ␤ cDNA), were used as primers in a polymerase chain reaction (PCR) to introduce a mutation. The amplified PCR product and oligonucleotide, CCCCCGGATCCCTACACACAGGCCATTTTCAGGTC (nucleotide positions 1348 -1371 of bovine ␤), were then used as primers for a second PCR. The resulting PCR product was introduced into the SalI/NotI sites of the pFASTBAC plasmid. Recombinant baculoviruses were prepared according to the manufacturer's protocol (Bac to Bac System, Life Technologies, Inc.). For infection, Sf9 cells were mixed with the recombinant viruses at a multiplicity of infection of 10 in various combinations of recombinant viruses and incubated at 27°C for 72 h.
Separation of Each Catalytic Dimer (␣ 1 /␣ 1 , ␣ 2 /␣ 2 , and ␣ 1 /␣ 2 ) Expressed Using Baculovirus System-The Sf9 cells (4 ϫ 10 8 ) infected with each baculovirus were homogenized in 10 ml of SET buffer (10 mM Tris-HCl, 250 mM sucrose, pH 7.4), and the cell supernatants was recovered by ultracentrifugation at 100,000 ϫ g. Each recombinant protein was purified using DEAE ion exchange, hydroxyapatite, and anion exchange column chromatographies as follows. All column chromatographies were performed at 4°C using fast protein liquid chromatography system (Amersham Pharmacia Biotech). DEAE-ion exchange chromatography was performed essentially as described previously (5). The cell supernatants obtained from the infected Sf9 cells were loaded onto a DEAE-Sepharose Fast Flow column (5 ml, Amersham Pharmacia Biotech) (flow rate 1 ml/min), which had been equilibrated with buffer A (0.1 M NaCl, 10 mM Tris-HCl, 10% glycerol, pH 7.4) and eluted with a linear gradient of 0.1-1 M NaCl in buffer A. The active fractions as judged by PAF-AH activity assay were further applied onto hydroxyapatite column (Econo-Pac CHT II, 5 ϫ 50 mm, Bio-Rad) (flow rate 0.8 ml/min), which had been equilibrated with buffer B (1 mM KH 2 PO 4 , 10% glycerol, pH 6.8). The proteins were eluted with a linear gradient of 1-400 mM KH 2 PO 4 in buffer B. Each catalytic dimer was separated by anion exchange column (POROS HQ/M, 4.6 ϫ 100 mm) (Perspective Biosystems, Inc. Framingham, MA.). After the column was first equilibrated with buffer C (10 mM Tris-HCl, 1 mM EDTA, 10% glycerol, pH 7.4), the active fractions from hydroxyapatite column chromatography which was dialyzed against buffer C were loaded onto the column and eluted with a linear gradient of 0 -0.5 M NaCl in buffer C (flow rate 4 ml/min).
Preparation of the ␤ Polypeptide-For the production of the ␤ protein, the Sf9 cells (4 ϫ 10 8 ) infected with ␤ baculovirus were homogenized in 10 ml of SET buffer (10 mM Tris-HCl, 250 mM sucrose, pH 7.4), and the cell supernatants were recovered by ultracentrifugation at 100,000 ϫ g. The ␤ protein was purified by DEAE-Sepharose and hydroxyapatite column chromatographies, as described above.
PAF-AH Assays-PAF-AH assays were performed as described previously (5), except that the radioactivity of liberated [ 3 H]acetate was counted by a ␤-counter. A kinetic study was performed using various concentrations of PAF as a substrate of an assay as described previously (2).
Cross-linking-Cross-linking was performed as described previously (5). Briefly, each sample was dialyzed against 100 mM sodium phosphate (pH 8.0), cross-linked by adding 10 mM BS 3 (Pierce), and then subjected to SDS-PAGE and Western blotting. Polyclonal anti-␣ 1 , -␣ 2 , and ␤ were used to detect cross-linked products, since our monoclonal antibodies failed to react with the BS 3 -treated polypeptides.
Preparation of PAF Analogs-1-O-Alkyl-2-acetyl-sn-glycero-3-phosphorylethanolamine (AAGPE) and 1-O-alkyl-2-acetyl-sn-glycero-3phosphoric acid (AAGPA) were prepared according to the method described by Satouchi et al. (10) using base exchange reaction by phospholipase D (from Actinomadura, Meito Sangyo, Tokyo, Japan). Briefly, nonradioactive PAF (C16, 5 mg) mixed with 3 H-PAF (87 Ci) was subjected to phospholipase D reaction in the presence of ethanolamine (200 mM for AAGPE). For AAGPA, synthesis nonradioactive PAF (C16, 5 mg) mixed with 3 H-PAF (87 Ci) was subjected to phospholipase D reaction in the absence of alcohol. The resulting products were isolated by preparative thin layer chromatography. Activity assays using these PAF analogs were performed as described above, except that 0.05 N HCl was added when liberated acetic acid was separated from the substrate when AAGPA was used.
DFP Labeling-Chemical labeling of catalytic subunits was performed as described previously (3), except that [ 14 C]DFP (NEN Life Science Products) was used in this study. Briefly, 2 g of recombinant catalytic subunits was incubated with 0.1 Ci of [ 14 C]DFP for 30 min at room temperature. The incorporation of radiolabeled DFP into the catalytic subunits was detected by SDS-PAGE and subsequent autoradiography.

Expression of Various Catalytic Complexes in the Sf9
Cells-In order to obtain various combinations of catalytic dimers, the ␣ 1 and ␣ 2 subunits of brain PAF-AH were expressed in the Sf9 cells by infecting the Sf9 cells with recombinant baculoviruses. The expression of catalytic dimers in the cytosol of Sf9 cells infected with both the ␣ 1 and ␣ 2 viruses was examined with POROS-HQ/M anion exchange column chromatography. Both ␣ 1 and ␣ 2 subunits were detected in a single peak when expressed individually in Sf9 cells (Figs. 1, A and  B). In contrast, when these two subunits were co-expressed in the cells, three peaks of activity were detected after the column chromatography (Fig. 1C). By immunoblot analysis (Fig. 1C,  bottom), we found that the ␣ 1 polypeptide was detected in both the first and second peak fractions, and the ␣ 2 polypeptide was detected in the second and third peak fractions. These data , and ␣ 1 and ␣ 2 (C) baculoviruses were subjected sequentially to DEAE-Sepharose and hydroxyapatite column chromatography as described under "Experimental Procedures." The active fraction was finally subjected to POROS-HQ/M anion exchange column chromatography, and proteins were eluted with a linear gradient of 0 -0.5 M NaCl. Each fraction was examined for PAF acetylhydrolase activity and for protein expression by Western blotting using subunit-specific antibodies.
Dimer formation of the catalytic subunits in each fraction in Fig. 1C was examined by cross-linking studies. Incubation of each activity fraction with the cross-linking agent BS 3 yielded similar ϳ60-kDa bands. The 60-kDa bands obtained from the first and third peak fractions cross-reacted with the anti-␣ 1 and -␣ 2 antibodies, respectively, and the 60-kDa bands from the second peak fraction cross-reacted with both the anti-␣ 1 and -␣ 2 antibodies (Fig. 2). This indicates that these three peak fractions represent the ␣ 1 /␣ 1 , ␣ 1 /␣ 2 , and ␣ 2 /␣ 2 dimers.
Substrate Specificity-By using recombinant proteins, we first examined the substrate specificities of three different catalytic complexes. The substrates used in this study were 1-O-alkyl-2-acetyl-glycerophospholipids with different head groups. Each recombinant catalytic complex produced in the Sf9 cells was purified to near-homogeneity (see Fig. 4A) by chromatographies, as described under "Experimental Procedures." As shown in Fig. 3, the ␣ 2 /␣ 2 homodimer exhibited different substrate specificity from the ␣ 1 /␣ 1 homodimer and the ␣ 1 /␣ 2 heterodimer, both of which showed similar substrate specificity to each other. The ␣ 2 /␣ 2 homodimer hydrolyzed PAF and 1-O-alkyl-2-acetyl-sn-glycero-3-phosphorylethanolamine (AAGPE) more efficiently than 1-O-alkyl-2acetyl-sn-glycero-3-phosphoric acid (AAGPA). In contrast, both ␣ 1 /␣ 1 and ␣ 1 /␣ 2 hydrolyzed AAGPA more efficiently than PAF. AAGPE was the poorest substrate for these enzymes. The ␣ 2 /␣ 2 homodimer hydrolyzed PAF 3-4 times better than the ␣ 1 /␣ 2 heterodimer and the ␣ 1 /␣ 1 homodimer. When AAGPE was used as a substrate, change in composition of the catalytic dimer from ␣ 1 /␣ 2 to ␣ 2 /␣ 2 induced a dramatic increase in the catalytic efficiency. It was demonstrated from these observations that the catalytic complexes of PAF-AH isoform I can hydrolyze PAF analogs with a different head group depending on the composition of the catalytic subunit. Moreover, it was shown that the ␣ 1 /␣ 2 heterodimer and the ␣ 2 /␣ 2 homodimer, both of which were detected in vivo, exhibited different substrate specificity.
The Active Site Labeling with [ 14 C]DFP-As shown in Fig. 3, the ␣ 1 /␣ 2 heterodimer exhibited a substrate specificity similar to the ␣ 1 /␣ 1 homodimer rather than the ␣ 2 /␣ 2 homodimer, suggesting that catalysis by the ␣ 1 subunit dominates in the ␣ 1 /␣ 2 heterodimer. To test this possibility, we performed an active site labeling study using [ 14 C]DFP, which binds co-valently to an active serine residue. When the recombinant ␣ 1 /␣ 1 and ␣ 2 /␣ 2 homodimers were incubated with [ 14 C]DFP, both ␣ 1 and ␣ 2 polypeptides were labeled with [ 14 C]DFP (Fig.  4). The ␣ 1 subunit was preferentially labeled with this reagent when the ␣ 1 /␣ 2 heterodimer was used. These data are consistent with previous observations that the ␣ 1 subunit is preferentially labeled by [ 3 H]DFP in the purified PAF-AH (isoform Ib) from bovine brain (1). These data suggest that a serine residue of the ␣ 2 subunit catalytic center is shielded to be inactive in the ␣ 1 /␣ 2 heterodimer, although the same residue in the ␣ 2 /␣ 2 homodimer is catalytically active.
Effect of the ␤ Subunit on the Enzyme Activity of Each Catalytic Complex-Next, we examined whether each catalytic dimer can associate with the ␤ subunit. Each catalytic dimer was prepared by Sf9 cells as described above. The recombinant ␤ subunit also prepared by baculovirus system in Sf9 cells was purified to near-homogeneity by column chromatographies, as described under "Experimental Procedures." Each catalytic dimer was incubated with the ␤ subunit on ice for 60 min, and then the mixture was analyzed with hydroxyapatite column FIG. 3. Substrate specificities of ␣ 1 /␣ 1 , ␣ 1 /␣ 2 , and ␣ 2 /␣ 2 dimers. The catalytic activities of three possible dimers were examined using PAF, AAGPE, and AAGPA as substrates.

FIG. 2. Cross-linking of ␣ subunits.
The ␣ 1 or ␣ 2 proteins in the three peaks (peaks I, II, and III) in Fig. 1C were crosslinked using BS 3 and detected by Western blotting using anti-␣ 1 (1st to 4th lanes) or anti-␣ 2 (5th to 8th lanes) antibodies. The results before (1st, 3rd, 5th, and 7th lanes)  and after (2nd, 4th, 6th, and 8th lanes) cross-linking reaction are shown. Molecular size is shown at left. chromatography. When each purified catalytic dimer was applied to the column, the activity was eluted as a single peak (Fig. 5, A-C), respectively, whereas preincubation of the catalytic complexes with the ␤ subunit generated new activity peaks. These peaks contained the respective catalytic subunit and the ␤ subunit (Fig. 5, D-F), indicating the association of catalytic dimers with the ␤ subunit.
We then examined the effect of the ␤ subunit on the enzyme activity of each catalytic dimer. As shown in Fig. 6C, the enzyme activity of the ␣ 2 /␣ 2 homodimer increased significantly by adding the recombinant ␤. In the optimum ␤ concentration (5 g per tube), the enzyme activity of the ␣ 2 /␣ 2 homodimer increased approximately 4-fold. Increasing ␤ to more than 5 g per tube did not cause a further increase in enzyme activity (data not shown) under the present conditions. In contrast, ␤ slightly suppressed the activity of the ␣ 1 /␣ 1 homodimer (Fig.  6A) and essentially had no effect on that of the ␣ 1 /␣ 2 heterodimer (Fig. 6B). We performed the same experiments using the native ␤ purified from bovine brain instead of the recombinant ␤, obtaining essentially the same results (data not shown).
We also tested the effect of the ␤ subunit on catalytic activity in Sf9 cells. The Sf9 cells were infected with either the ␣ 1 baculovirus or the ␣ 2 baculovirus with increasing amounts of ␤ baculovirus, and the lysates of the infected cells were examined for protein expression and PAF-AH activity. The PAF-AH activity of the ␣ 1 /␣ 1 homodimer was decreased in proportion to the increase of the ␤ expression, whereas that of the ␣ 2 /␣ 2 homodimer was rather increased (data not shown). The levels of ␣ 1 and ␣ 2 subunit expression were not affected appreciably by the co-expression of the ␤ subunit. These findings also support the idea that the ␤ subunit stimulates the activity of the ␣ 2 /␣ 2 homodimer but suppresses that of the ␣ 1 /␣ 1 homodimer.
Effect of His 149 to Arg Mutation in the ␤ Subunit-Like wild-type ␤/LIS1 protein the mutant ␤ could be expressed in Sf9 cells and purified. The mutant ␤ subunit exhibited no effect on the catalytic activity of all the catalytic dimers (data not shown). Unlike the native ␤ subunit, as judged from the failure FIG. 4. DFP labeling of three catalytic dimers. Three catalytic dimers (␣ 1 /␣ 1 , ␣ 1 /␣ 2 , and ␣ 2 /␣ 2 ) expressed and purified by the baculovirus system (Fig. 1, each 2 g) were mixed with [ 14 C]DFP. The incorporation of radioactivity into each subunit was detected by SDS-PAGE followed by autoradiography (B). Proteins detected by Coomassie staining were shown in A.
of inducing a shift of the ␣ dimer peak on a hydroxyapatite column chromatography, the mutant ␤ could not associate with catalytic subunits (Fig. 7).

DISCUSSION
Brain PAF-AH (also PAF-AH isoform I) contains two mutually homologous ␣ 1 and ␣ 2 subunits, both of which can form a respective homodimer and a heterodimer, and in fact, the ␣ 1 /␣ 2 heterodimer and the ␣ 2 /␣ 2 homodimer have been detected in vivo (1,5). In rats, ␣ 1 expression is restricted in early developing neurons at embryonic stages at which the neural cell migration is most active, and switching of catalytic subunits from the ␣ 1 /␣ 2 heterodimer to the ␣ 2 /␣ 2 homodimer occurs during brain development (5). It is also well established that the other subunit of PAF-AH isoform I, ␤, is a product of the causative gene for Miller-Dieker syndrome, which has a defect in neuronal migration during brain development. It was thus essential to delineate the biochemical differences of each catalytic dimer in order to understand the biological role of PAF-AH. In the present study, we have demonstrated that each catalytic dimer exhibited distinct substrate specificity and that their enzyme activity is modulated by the ␤ subunit in catalytic subunit composition-dependent manner.
Previously, we showed that PAF-AH isoform I has strict specificity for the acetyl group attached to the sn-2 position of phosphoglyceride as a substrate (11). The essential amino acid residue involved in the recognition of acetyl moiety has also been postulated from crystallographic studies (4). However, it was unknown whether the enzyme may recognize the phospholipid substrate with a different head group. In this study, we demonstrated that the enzyme hydrolyzed the phospholipid substrate with a head group other than choline moiety. 1-O-Alkyl-2-acetyl-sn-glycero-3-phosphoric acid (AAGPA) is an intermediate for the production of PAF via the de novo pathway (12) and is synthesized by acetylation of alkylglycerophosphate. Interestingly, the enzyme catalyzing this reaction is enriched in the brain microsomal fractions and, in particular, in the brain nuclear fractions, as compared with the lyso-PAF acetyltransferase (13). The present data demonstrating that PAF-AH isoform I can hydrolyze AAGPA as well as PAF raises the possibility that the enzyme may regulate the intracellular PAF level by hydrolyzing not only PAF itself but also an inter-mediate for the de novo PAF synthesis. More interestingly, it is also possible that AAGPA itself may play a role in the cell as a new type of intracellular lipid messenger (14), the level of which is regulated by PAF-AH isoform I.
FIG. 7. H149R mutant ␤ lost activity to bind to the catalytic dimers. The mutant ␤ proteins were expressed in Sf9 cells and purified. ␣ 2 (2 g) was mixed in vitro with the mutant ␤ proteins (5 g), and the mixtures were subjected to hydroxyapatite column chromatography. The elution profiles of PAF acetylhydrolase activity (A) and each subunit detected by Western blotting are shown (B).
The substrate specificity and the specific activity of the ␣ 1 /␣ 2 heterodimer were very similar to the ␣ 1 /␣ 1 homodimer but not to the ␣ 2 /␣ 2 homodimer. We previously reported that DFP, a reagent that specifically binds to an active serine residue, reacted preferentially with the ␣ 1 subunit in the isoform Ib, suggesting that the catalytic serine residue of the ␣ 2 subunit is shielded to be inactive. A crystallographic study of the ␣ 1 /␣ 1 homodimer revealed that the two active sites are at the bottom of the catalytic gorge, only 12 Å from each other (4). This proximity suggests that these two active sites do not function independently, since this catalytic gorge is capable of accommodating only one PAF molecule. Thus we speculate that only one active serine residue is enough for catalysis. It is also possible that one active serine out of two in the catalytic gorge of all three dimers are actually shielded to be inactive. A substrate specificity study as well as DFP labeling experiments have demonstrated that catalysis by the ␣ 1 subunit may dominate in the ␣ 1 /␣ 2 heterodimer. We have detected the ␣ 1 /␣ 2 heterodimer and the ␣ 2 /␣ 2 homodimer in vivo, but the ␣ 1 /␣ 1 homodimer has not been detected yet. It is still possible that the ␣ 1 /␣ 1 homodimer exists in vivo, but even the ␣ 1 /␣ 2 heterodimer can be substituted for the ␣ 1 /␣ 1 homodimer in terms of substrate specificity. Which amino acid residue(s) is critical for determining the substrate specificity in the catalytic subunits? According to the crystal structure of the ␣ 1 /␣ 1 homodimer, the amino acid residues Tyr 191 and Tyr 193 are exposed to the catalytic gorge and possibly recognize a head group of PAF (4). Interestingly, these Tyr residues of the ␣ 1 subunit are substituted into Phe 192 and Phe 194 in the ␣ 2 subunit (4). Our preliminary mutation study demonstrated that the exchange of Tyr 191 into Phe gives the ␣ 1 subunit a substrate specificity similar to that of ␣ 2 .
We previously concluded that the ␤ subunit does not possess a regulatory role on catalytic activity, based on the observation that the ␤ subunit can be dissociated without loss of enzyme activity from the ␣ 1 ⅐␣ 2 ⅐␤ complex purified from bovine brain (7). We demonstrated in the present study that this was true only for the ␣ 1 /␣ 2 heterodimer. The rate of PAF hydrolysis by the ␣ 2 /␣ 2 homodimer was, however, accelerated about 4 times by the addition of ␤, whereas the rate of hydrolysis by the ␣ 1 /␣ 1 was slightly suppressed by it. In the rat brain, composition of the catalytic subunit of PAF-AH changes drastically from the ␣ 1 /␣ 2 heterodimer to the ␣ 2 /␣ 2 homodimer during development. These changes in the catalytic subunit from ␣ 1 /␣ 2 to ␣ 2 /␣ 2 may result in a drastic increase in the PAF-AH activity, since (i) ␣ 2 /␣ 2 hydrolyzes PAF four times better than ␣ 1 /␣ 2 does, and (ii) only the enzyme activity of ␣ 2 /␣ 2 is accelerated by ␤. Our preliminary study shows that PAF-AH activity in the rat brain is much higher in adulthood than that in embryonic stages, although the expression levels of ␣ 2 and ␤ were roughly the same during this period. Change in enzyme activity can be explained by the change in the composition of the catalytic subunit. The change in PAF-AH activity during brain development might regulate PAF content in neural cells. Consistent with this idea, Tokumura et al. (14) have reported that PAF content in the young rat brain is higher than that in the adult brain.
In conclusion, we have demonstrated that the enzyme activity of PAF-AH isoform I is regulated in multiple ways by switching the combination of the catalytic subunit and by manipulating the ␤ subunit. We have also postulated the possibility that some 1-O-alkyl-2-acetyl-phospholipids other than PAF may be genuine substrates for the enzyme. It is still unclear at present which effect is relevant in vivo and what the in vivo substrate for this enzyme is. Our next challenge will be to identify the in vivo lipid substrate for this enzyme and its cellular function.