Activated Mast Cells Release Extracellular Type Platelet-activating Factor Acetylhydrolase That Contributes to Autocrine Inactivation of Platelet-activating Factor*

IgE-dependent and -independent activation of mouse bone marrow-derived mast cells (BMMC) elicited rapid and transient production of platelet-activating factor (PAF), which reached a maximal level by 2–5 min and was then degraded rapidly, returning to base-line levels by 10–20 min. Inactivation of PAF was preceded by the release of PAF acetylhydrolase (PAF-AH) activity, which reached a plateau by 3–5 min and paralleled the release of β-hexosaminidase, a marker of mast cell exocytosis. Immunochemical and molecular biological studies revealed that the PAF-AH released from activated mast cells was identical to the plasma-type isoform. In support of the autocrine action of exocytosed PAF-AH, adding exogenous recombinant plasma-type PAF-AH markedly reduced PAF accumulation in activated BMMC. Furthermore, culture of BMMC with a combination of c-kit ligand, interleukin-1β and interleukin-10 for > 24 h led to an increase in plasma-type PAF-AH expression, accompanied by a reduction in stimulus-initiated PAF production. Collectively, these results suggest that plasma-type PAF-AH released from activated mast cells sequesters proinflammatory PAF produced by these cells, thereby revealing an intriguing anti-inflammatory aspect of mast cells.

Mast cells are highly specialized effector cells of the immune system which, when activated, release various biologically active molecules including histamine, proteoglycans, and proteases through exocytosis, arachidonic acid-derived mediators such as eicosanoids through activation of the cyclooxygenase and 5-lipoxygenase pathways, and preformed and newly expressed cytokines (1). IgE-dependent activation of mast cells has also been reported to elicit production of platelet-activating factor (PAF), 1 a lipid mediator with a glycerophosphocholine backbone (2). PAF has been implicated in a number of physiological and pathological processes, particularly allergy and in-flammation, affecting the respiratory, vascular, digestive, and reproductive systems (3,4). Its accumulation is usually tightly regulated at the biosynthetic and degradative levels to avoid the inappropriately high accumulation observed in many diseases. PAF production often parallels immediate eicosanoid generation (5). In the remodeling pathway for PAF biosynthesis proposed to occur in inflammatory cells, Ca 2ϩ -dependent phospholipase A 2 may play a role in the production of both lysoPAF and arachidonic acid from 1-alkyl-2-arachidonoyl-snglycero-3-phosphocholine, then acetyltransferase is thought to introduce an acetyl residue from acetyl-CoA to the sn-2 position of lysoPAF leading to the formation of biologically active PAF (3)(4)(5). In turn, PAF is degraded and inactivated by PAF acetylhydrolase (PAF-AH) enzymes, a particular group of Ca 2ϩ -independent phospholipase A 2 s that remove the acetyl moiety at the sn-2 position (6). Whereas the molecular mechanisms of arachidonic acid metabolism which lead to eicosanoid generation in mast cells have been well studied (7)(8)(9)(10)(11)(12)(13)(14)(15), those of PAF metabolism remain largely unknown.
Recent studies have revealed that mammalian PAF-AHs can be classified into intracellular and extracellular types (6). Intracellular PAF-AH type I (PAF-AH-I) is a heterotrimer complex composed of 45-, 30-, and 29-kDa subunits (16). The 45-kDa subunit, which is not essential for catalytic activity, exhibits striking homology (99%) with a protein encoded by the causal gene (LIS-1) for Miller-Dieker lissencephaly, a human brain malformation manifested by a smooth cerebral surface and abnormal neural migration (17). The 30-and 29-kDa subunits, which are highly homologous with each other, belong to a novel type of serine proteases, and a sequence of ϳ30 amino acids adjacent to their active serine residues exhibits significant similarity to the first transmembrane region of the PAF receptor (18,19). PAF-AH type II (PAF-AH-II) is a monomeric 40-kDa protein that exhibits broader substrate specificity than PAF-AH-I in that PAF-AH-II hydrolyzes oxidized phospholipids as effectively as PAF, whereas PAF-AH-I is more specific for PAF (20,21).
A secreted form of PAF-AH, which is abundantly present in plasma as a lipoprotein-associated form (22), is believed to regulate base-line circulating PAF levels and may be critical in resolving inflammation. The cDNA for this enzyme encodes a 44-kDa secretory protein that contains a typical signal sequence and a serine esterase consensus motif GXSXG (23,24). Plasma-type PAF-AH displays significant homology (ϳ40%) with intracellular PAF-AH-II, but not PAF-AH-I, over the whole sequence (21,23). In agreement with this similarity, plasma-type PAF-AH catalyzes hydrolysis of PAF and structurally related oxidized phospholipids (23,24). Pretreatment of animals with recombinant plasma-type PAF-AH has been shown to block PAF-induced inflammation (23), revealing its anti-inflammatory function. Interestingly, deficiency of plasma-type PAF-AH is an autosomal recessive syndrome that is associated with severe asthma in Japanese children (25), in which a point mutation of exon 9 of the plasma-type PAF-AH gene results in production of inactive protein (26). These observations, together with the finding that PAF receptor transgenic mice are more susceptible to methacholine-induced bronchial hypersensitivity than normal littermates (27), imply that PAF and plasma-type PAF-AH are involved in propagating and terminating allergic reactions, respectively.
Although circulating blood is rich in plasma-type PAF-AH, the source of this enzyme in the extravascular space is poorly understood. Plasma-type PAF-AH activity increased dramatically in the peritoneal cavities of guinea pigs with endotoxic shock (28), raising the possibility that PAF-AH accumulating at inflamed sites is not only exudated from plasma but is also produced by tissue cells. Macrophages are likely to be one of the main sources of extravascular plasma-type PAF-AH, since monocytes have been shown to produce plasma-type PAF-AH during differentiation into macrophages (23). Here we report that mast cells are a rich source of plasma-type PAF-AH. Plasma-type PAF-AH released from mast cells activated by various stimuli degrades PAF produced by these cells, revealing an anti-inflammatory property of mast cells during allergic and non-allergic inflammation.

EXPERIMENTAL PROCEDURES
Materials-Mouse recombinant c-kit ligand (KL), interleukin (IL)-3, and IL-10 were expressed by baculovirus-infected Sf9 cells, as described previously (7)(8)(9). Mouse IL-1␤ was purchased from Genzyme. Lysophosphatidylserine was purchased from Avanti. A cDNA probe and rabbit antiserum for guinea pig plasma-type PAF-AH were prepared as described previously (28). IgE anti-trinitrophenyl (TNP) and TNP-conjugated bovine serum albumin (TNP-BSA) were provided by Drs. J. P. Arm and H. Katz (Harvard Medical School, Boston). A cDNA probe for mouse cyclooxygenase-2 was provided by J. Trzaskos, Merck DuPont. Prostaglandin D 2 radioimmunoassay kit and ECL Western blotting kit were purchased from Amersham Corp. A23187 was purchased from Sigma.
Preparation of Mouse Bone Marrow-derived Mast Cells (BMMC)-Bone marrow cells from male BALB/cJ mice were cultured for up to 10 weeks in 50% enriched medium (RPMI 1640 supplemented with 10% fetal calf serum, 100 units/ml penicillin, 100 g/ml streptomycin, 2 mM L-glutamine, and 0.1 mM non-essential amino acids) and 50% WEHI-3 cell-conditioned medium as a source of IL-3 (7-9). After 3 weeks, Ͼ98% of the cells in the culture were BMMC, as assessed by staining with toluidine blue or Alcian blue and safranin. After washing twice with enriched medium, the cells were cultured at 1 ϫ 10 6 cells/ml in enriched medium containing 5 ng/ml IL-1␤, 100 units/ml IL-3, 100 units/ml IL-10, or 100 ng/ml KL, either alone or in combination, as required for the experiments.
Preparation of Rat Connective Tissue Mast Cells (CTMC)-CTMC were obtained from the peritoneal cavities of Wistar rats (Nippon Bio-Supply Center) as described previously (29). Briefly, rats (male, weighing Ͼ 350 g) were injected intraperitoneally with 50 ml of Hanks' balanced salt solution containing 0.1% BSA, and the peritoneal cells were harvested. After centrifuging these cells with Hanks' balanced salt solution containing 38% BSA, CTMC were collected from the bottom of the tube. The purity and viability of the cells were assessed by staining with toluidine blue and trypan blue, respectively, and CTMC with Ͼ 95% purity and viability were used for the subsequent studies.
Activation of Mast Cells-In typical experiments, mast cells at a concentration of 1 ϫ 10 6 cells/ml in Tyrode's gelatin buffer were sensitized for 30 min with 1 g/ml IgE anti-TNP and then activated for various periods at 37°C using 1-100 ng/ml TNP-BSA as an antigen (7). Alternatively, the cells were stimulated with 200 ng/ml KL or 1 M A23187. For rat CTMC activation, 1 M lysophosphatidylserine, an activation cofactor, was added together with the above stimulators (30). After activation, ␤-hexosaminidase release (8), PAF-AH release and PAF production (see below) was assessed.
PAF-AH Enzyme Assay-PAF-AH activity was measured as described previously (16). Briefly, a total volume of 250 l of sample was incubated in the standard incubation system for assaying PAF-AH, comprising 50 mM Tris-HCl (pH 7.4), 5 mM EDTA, 20 M [acetyl-3 H]PAF (1-O-hexadecyl-2-[acetyl-3 H]sn-glycero-3-phosphocholine; 5 Ci/nmol) (NEN Life Science Products) for appropriate periods at 37°C. The reaction was stopped by adding 2.5 ml of chloroform:methanol (4:1) and 0.25 ml of water. Aliquots (600 l) of the aqueous phase were subjected to radioactivity measurements to determine the amount of [ 3 H]acetic acid liberated. When 1-O-[ 3 H]octadecyl-PAF (Amersham) was used as a substrate, the reaction products were extracted by the method of Bligh and Dyer (31), spotted onto thin layer chromatography plates (Merck), and then developed with a solvent system of chloroform:methanol:acetic acid:water (50:25:8:4 v/v). Lipid spots were visualized by exposing the plates to iodine vapor. The spots corresponding to PAF and lysoPAF were identified by comparison with authentic PAF (Cayman Chemical) and lysoPAF (Avanti) standard and scraped off. The radioactivities recovered were quantified.
Detection of PAF by Bioassay-The total lipids of cells and supernatants, extracted by the method of Bligh and Dyer (31), were developed on thin layer chromatography plates with a solvent system of chloroform:methanol:acetic acid:water (50:25:8:4, v/v) and visualized by exposing the plates to iodine vapor. PAF was then extracted from the scraped silica gel powder by the method of Bligh and Dyer (31) and reconstituted in Tyrode's buffer. Rabbit platelets, prelabeled with 0.25 Ci/ml [ 14 C]serotonin (NEN Life Science Products) for 20 min at room temperature, were resuspended in Tyrode's buffer at 5 ϫ 10 8 cells/ml. A 100-l portion of the platelet suspension was mixed with 10 l of extracted sample or authentic PAF standard and incubated for 2 min at room temperature. Platelet activation was stopped by adding 150 mM formaldehyde, and the [ 14 C]serotonin released into the supernatant was quantified by ␤-scintillation counting. To correct any differences in lipid extraction efficiency between samples, a trace of [acetyl-3 H]PAF was added to each sample before extraction, and the recovery of radioactivity in the PAF fraction after the final extraction was monitored.
Detection of PAF by Radioactivity Assay-BMMC or RBL-2H3 cells (5 ϫ 10 6 cells) were preincubated for 10 min with 25 Ci/ml [ 3 H]sodium acetate (NEN Life Science Products) and then activated for various periods with IgE/antigen or A23187 in the continued presence of [ 3 H]sodium acetate. After stopping the reaction by adding 0.1% SDS, the lipids contained in the cells and/or supernatants were extracted by the method of Bligh and Dyer (31) and developed on thin layer chromatography plates. The spot corresponding to PAF was identified by comparison with an authentic PAF standard and scraped off, and the radioactivity was measured (32).
Reverse Transcriptase-Polymerase Chain Reaction (PCR)-Total RNA of BMMC, extracted using guanidinium thiocyanate with TRIzol (Life Technologies, Inc.) according to the manufacturer's instructions, was mixed with the oligo(dT) primer and avian myeloblastosis virus reverse transcriptase (TaKaRa) and incubated for 30 min at 50°C. The resulting cDNA was subjected to PCR using a 21-mer sense primer, 5Ј-AGACAAATCTGCATCGGCAAC-3Ј and an antisense primer 5Ј-TT-GGTGAGGTCGATGGCTACTC-3Ј, which correspond to nucleotides 653-673 and 1246 -1267 of mouse plasma-type PAF-AH cDNA, respectively (GenBank/EMBL Data Bank accession number U34277) (24). Thirty amplification cycles were performed at 94°C for 1 min, 55°C for 1 min, and 72°C for 2 min with exTaq polymerase (TaKaRa), after which a major product with an estimated size of approximately 600 base pairs was resolved in 1.5% (w/v) agarose gel, purified with a gel extraction kit (QIAGEN), subcloned into the pCR 3 cloning vector (Invitrogen), sequenced using a Taq cycle sequencing kit (TaKaRa) according to the manufacturer's instructions, and analyzed using an automated DNA sequencer (DSQ-1000L; Shimadzu).

SDS-Polyacrylamide
Gel Electrophoresis/Immunoblotting-The supernatants and pellets of the activated BMMC (1 ϫ 10 7 cells/ml) were applied to 10% SDS-polyacrylamide gels and electrophoresed under reducing conditions. The separated proteins were electroblotted onto nitrocellulose membranes (Schleicher & Schuell) using a semidry blotter (MilliBlot-SDE system; Millipore) according to the manufacturer's instructions. The membranes were then washed once with Tris-buffered saline (TBS) (pH 7.2) containing 0.1% Tween 20 (TBS-T) and then blocked for 1 h in TBS-T containing 3% skimmed milk. After washing the membranes with TBS-T, antibody against guinea pig plasma-type PAF-AH was added at a dilution of 1:1,000 in TBS-T and incubated for 2 h. Following three washes with TBS-T, the membranes were treated for 1 h with horseradish peroxidase-conjugated goat anti-rabbit IgG (Zymed) (1:10,000 dilution) in TBS-T. After six washes, the protein bands were visualized with the aid of an ECL Western blot analysis system.
Expression of Recombinant PAF-AH-cDNA for guinea pig PAF-AH (28) was subcloned into pCR 3 (Invitrogen) and transfected into CHO-K1 cells (RIKEN Cell Bank) using CellFectin (Life Technologies, Inc.) by a method described previously (33). Three days after transfection, the PAF-AH activity released into the supernatants was measured.
Statistical Analysis-Data were analyzed by Student's t test. Results are expressed as means Ϯ S.E., with p ϭ 0.05 as the limit of significance.

Production and Subsequent Inactivation of PAF by Activated
Mast Cells-The time course of PAF production by BMMC sensitized with IgE anti-TNP and activated with TNP-BSA as antigen is shown in Fig. 1. In this experiment, mixtures of the supernatant and the cell fraction were subjected to lipid extraction, and the PAF produced was purified by thin layer chromatography and quantified by a rabbit platelet aggregation assay as described under "Experimental Procedures." Significant production of PAF occurred immediately after antigen challenge, reaching a peak of approximately 1 nmol/10 6 cells by 2-5 min (Fig. 1A). Thereafter, PAF disappeared rapidly and returned to the basal level by 10 min. The de novo synthesis of PAF was also assessed by monitoring the incorporation of [ 3 H]acetate into PAF (32). As in the PAF bioassay (Fig. 1A), the level of [acetyl-3 H]PAF in A23187-stimulated BMMC, which was barely detectable before cell activation, reached a peak by 2 min and declined thereafter, almost returning to the basal level by 10 -20 min (Fig. 1B). PAF produced by BMMC was exclusively associated with the cells and was not released into the supernatants.
Release of PAF-AH from Activated Mast Cells-The rapid disappearance of PAF from IgE/antigen-or A23187-stimulated BMMC led us to formulate the hypothesis that BMMC might contain a PAF-inactivating enzyme. We found that when the supernatants of IgE/antigen-activated, but not unstimulated, BMMC were incubated with [acetyl-3 H]PAF (Fig. 2)  H]lysoPAF, respectively, indicating that the supernatants of activated BMMC contained PAF-AH. The release of PAF-AH activity from IgE/antigen-activated BMMC reached a maximum of ϳ40% by 3-5 min then plateaued ( Fig. 2A), preceding the inactivation of PAF by these cells (Fig. 1A). Dose-response experiments revealed that the release of PAF-AH activity peaked at 10 -100 ng/ml antigen with an EC 50 of ϳ1 ng/ml, whereas no appreciable release was observed without antigen challenge (Fig. 2B). The kinetics and dose dependence of PAF-AH release showed close correlation with those of ␤-hexosaminidase release (Fig. 2, A and B). When BMMC were stimulated with another mast cell secretagogue, KL, release of PAF-AH activity also occurred in parallel with that of ␤-hexosaminidase, reaching a maximum of ϳ20% by 3ϳ5 min (Fig.  2C). A23187 stimulation of BMMC also resulted in both PAF-AH and ␤-hexosaminidase release, which reached 20ϳ40% within 2 min (data not shown). This close correlation between PAF-AH and ␤-hexosaminidase release strongly suggests that PAF-AH is stored in the secretory granules of BMMC and is exocytosed following cell activation.
Rat serosal CTMC also released PAF-AH (Fig. 3, upper) as well as ␤-hexosaminidase (Fig. 3, lower) in response to IgE/ antigen or KL in the presence of lysophosphatidylserine added as a activation cofactor (30). This observation implies that PAF-AH release from activated mast cells is not limited to a particular mast cell phenotype but reflects a general phenomenon. However, in contrast to BMMC, in which the release percentages of PAF-AH and ␤-hexosaminidase were almost equal (Fig. 2), the amount of PAF-AH released from rat CTMC was about half of that of ␤-hexosaminidase after each stimulation (Fig. 3).
Characterization of Mast Cell Secretory PAF-AH-When the supernatant and remaining cell pellet of A23187-activated BMMC were subjected to SDS-polyacrylamide gel electrophoresis/immunoblotting using antiserum against guinea pig plasma-type PAF-AH, a single protein band with a molecular mass of ϳ58 kDa, which corresponds to the size of glycosylated plasma-type PAF-AH (28), was detected (Fig. 4A). The intensity of the band visualized in the cell pellet was three times as strong as that in the supernatant, consistent with the distri- bution of PAF-AH activities in the supernatant (ϳ24%) and pellet (ϳ76%) in this experiment. This result suggests that the PAF-AH released from activated mast cells was identical or immunochemically related to plasma-type PAF-AH.
To confirm that mast cells express plasma-type PAF-AH, reverse transcriptase-PCR analysis was carried out using a set of primers based upon the cDNA sequence of murine plasmatype PAF-AH (24). A single ϳ600-base pair fragment, consistent with the predicted size of plasma-type PAF-AH cDNA (residues 653-1267), was specifically amplified from RNA obtained from BMMC (Fig. 4B). DNA sequencing revealed that this PCR fragment indeed encoded the corresponding portion of murine plasma-type PAF-AH (data not shown).
Effect of Recombinant PAF-AH on Inactivation of PAF Produced by Activated Mast Cells-To confirm that plasma-type PAF-AH contributes to the rapid degradation of PAF produced by mast cells, the effect of exogenous recombinant plasma-type PAF-AH expressed by CHO-K1 cells transfected with PAF-AH cDNA on PAF generation by activated BMMC was examined. As shown in Fig. 1B, PAF production by activated BMMC was markedly attenuated when recombinant plasma-type PAF-AH was added to the medium, indicating that extracellular PAF-AH has the capacity to inactivate BMMC-associated PAF.
Cytokine Regulation of PAF-AH Expression in Mast Cells-When BMMC were cultured for 2 days with KL in the continued presence of IL-3, the expression level of plasma-type PAF-AH was similar to the level expressed in BMMC maintained in IL-3 alone (Fig. 5A). Culture of BMMC with KL ϩ IL-10 increased PAF-AH activity 1.6-fold, with a concomitant increase in steady-state expression of the transcript for plasma-type PAF-AH (Fig. 5A, inset). Culture of BMMC with the cytokine triad KL ϩ IL-10 ϩ IL-1␤, which has previously been shown to induce several genes related to inflammatory responses in BMMC (7,12), increased plasma-type PAF-AH expression further, to approximately 3-fold (Fig. 5A). Expression of ␤-actin transcript used as a control did not change appreciably (data not shown). The increase in PAF-AH mRNA expression after culture with KL ϩ IL-10 ϩ IL-1␤ became evident 1 day after the start of culture, delayed compared with the induction of proinflammatory proteins such as cyclooxygenase-2, which reached a peak at 2 h and disappeared by 10 h (Fig. 5B), and was accompanied by concomitant increase in PAF-AH activity (data not shown). After rapid PAF-AH release initiated by KL within a few minutes as shown in Fig. 2C, only a small (Ͻ5% of the total PAF-AH activity in the cells) amount of PAF-AH was gradually released into the culture medium during 24 -48 h of culture with KL ϩ IL-10 ϩ IL-1␤ (data not shown). These KL ϩ IL-10 ϩ IL-1␤-treated BMMC released severalfold more PAF-AH activity (6ϳ10 nmol/min/10 6 cells) than the cells maintained in IL-3 (2ϳ3 nmol/min/10 6 cells) following 10-min stimulation with 1 M A23187. In accordance with increased PAF-AH expression and secretion, A23187stimulated accumulation of PAF in BMMC, which had been cultured for 2 days with KL ϩ IL-10 ϩ IL-1␤, was significantly less than that in replicate cells maintained in IL-3 (Fig. 5C). of A23187-activated BMMC were subjected to immunoblot analysis using antiserum against guinea pig plasma-type PAF-AH. Panel B, total RNA obtained from BMMC was subjected to reverse transcriptase-PCR using mouse plasma-type PAF-AH primers. The sample was applied to 1.2% (w/v) agarose gel and stained with ethidium bromide (lane 2). Lane 1 shows molecular mass markers. bp, base pairs.

DISCUSSION
Plasma-type PAF-AH has potentially important physiological and pathological roles because of its ability to abolish the diverse effects of PAF and oxidized phospholipids, including inflammation, shock, and thrombosis (6). Accumulating evidence suggests that decreased degradation of these biologically active lipid molecules results in pathological responses. In plasma, PAF-AH is associated with lipoprotein particles and has been implicated in atherosclerosis, where it functions as a scavenger of oxidized phospholipids in modified low density lipoprotein (22). Acquired PAF-AH deficiency has been described in patients with systemic lupus erythematosus (34) and septic shock (35). Increased levels of PAF have been reported in children with acute asthmatic attacks (36), and inherited plasma-type PAF-AH deficiency, caused by a point mutation in exon 9 which leads to complete abolition of catalytic activity (26), has been observed in the Japanese population, especially in children with severe asthma (25). We have now demonstrated that mast cells are a potent source of plasma-type PAF-AH. The fact that the main effector cell of allergic inflammation, the mast cell, has the capacity to secrete this antiinflammatory enzyme into the surrounding microenvironment provides new insight into the regulation of local and systemic allergic responses and is an unexplored anti-inflammatory aspect of mast cells.
IgE-dependent and -independent activation of BMMC elicited immediate PAF generation (Fig. 1), which occurred in association with the other two well characterized immediate responses, ␤-hexosaminidase exocytosis and eicosanoid generation (8,9). Temporal and spatial differences exist in the evolution of PAF and eicosanoid production; the accumulation of PAF was transient, disappearing by 10 min after stimulation (Fig. 1), and remained cell-associated (Fig. 1B), whereas prostaglandin D 2 generation reaches a plateau lasting for several hours and is released predominantly into the supernatant (8,9). The cell-associated property of PAF has been commonly observed in a wide variety of cell types (6,32). The amount of PAF produced by activated BMMC reached nearly 1 nmol/10 6 cells, which is comparable to that produced by endothelial cells (32), and is therefore likely to be biologically significant. In the early 1980s, it was shown that PAF was released from activated BMMC into the supernatant, but the amount was about 500 times less than that observed in the present study (2). The rather low level of PAF detected in this earlier study might have been due to failure to extract most of the cell-associated PAF, since absolute ethanol, which is not an efficient extractor of phospholipids, was used to extract cell-associated PAF (2). Alternatively, the different conditions under which BMMC were cultured might have affected their capacity to produce PAF.
The rapid inactivation of PAF by activated BMMC led us to the unequivocal finding that mast cells release PAF-AH upon activation. PAF-AH release occurred in parallel with the release of ␤-hexosaminidase with identical time course, dose, and stimulus specificity criteria (Figs. 2 and 3), implying that PAF-AH is stored in mast cell secretory granules. Immunochemical and molecular biological studies revealed that this secretory PAF-AH is identical to the plasma-type enzyme (Fig.  4). The involvement of plasma-type PAF-AH in degrading mast cell-associated PAF is supported by the following three lines of evidence. First, PAF-AH release, which reached a peak by 2 min (Fig. 2), preceded the degradation of PAF, which became evident after 5 min (Fig. 1). Second, adding recombinant plasma-type PAF-AH markedly reduced the accumulation of PAF in BMMC (Fig. 1B). Third, increased expression of plasma-type PAF-AH in BMMC after culture with KL ϩ IL-10 ϩ IL-1␤ led to decreased A23187-induced PAF production (Fig. 5), even though the expression of cytosolic and secretory phospholipase A 2 s, which are implicated in PAF biosynthesis, increases in BMMC after such cytokine treatment (8,12). Nonetheless, the ability of extracellular PAF-AH to quench cell-associated PAF suggests that the PAF produced by BMMC is located predominantly in the plasma membranes of the activated cells.
Unlike liver cells (37) and macrophages (23), which sponta- neously secrete plasma-type PAF-AH into their culture supernatants, PAF-AH secretion from mast cells is principally regulated by the receptor-coupled signal transduction pathway coupled with degranulation. The observations that the release percentages of PAF-AH activity were similar to those of ␤-hexosaminidase in BMMC (Fig. 2), that the distribution of immunoreactive PAF-AH protein in the supernatants and pellets of activated BMMC correlated with that of PAF-AH activity (Fig.  4A), and that the changes in PAF-AH activity and PAF-AH mRNA levels after cytokine treatment occurred in parallel (Fig.  5A) suggest that the plasma-type isozyme is the dominant PAF-AH isoform expressed in BMMC. This conclusion is further supported by our recent preliminary study that the transcripts for the intracellular PAF-AH were barely detectable in BMMC even after culture with KL ϩ IL-10 ϩ IL-1␤ under the experimental conditions shown in Fig. 5. 2 In contrast, the fact that the release percentages of PAF-AH activity were consistently about half of those of ␤-hexosaminidase in rat CTMC (Fig. 3) may reflect either granule heterogeneity or the presence of other PAF-AH isozymes in this mast cell phenotype.
Limited information is currently available on the transcriptional regulation of PAF-AH enzymes. An anti-inflammatory glucocorticoid increased PAF-AH levels in the plasma of rats (38) and increased the secretion of PAF-AH activity from macrophage-like differentiated HL-60 cells (39). An increase in plasma-type PAF-AH transcript has been reported during differentiation or maturation of monocytes into macrophages (23). Here we showed that, among the cytokines tested, IL-10, when combined with KL, significantly increased PAF-AH at the transcript level and that the addition of IL-1␤ to KL ϩ IL-10 increased its expression further (Fig. 5A). It is notable that the cytokine triad, KL ϩ IL-10 ϩ IL-1␤, represents a potent inflammatory stimulus for BMMC, inducing proinflammatory proteins such as secretory type II phospholipase A 2 (12), cyclooxygenase-2 (7), IL-1␤ (40), and IL-6 (41). However, the induction of these proinflammatory proteins occurred within 2-10 h after the initiation of culture, whereas PAF-AH induction, which occurred after 1 day of culture (Fig. 5B), lags behind the induction of proinflammatory mediators. This relatively late induction of plasma-type PAF-AH, a presumptive antiinflammatory enzyme, might reflect a signal for the termination of inflammatory reactions that mast cells initiate.
In conclusion, a detailed analysis of PAF metabolism by activated mast cells revealed that plasma-type PAF-AH is exocytosed by these cells and contributes to autocrine degradation of PAF. Furthermore, its expression level in mast cells increases during the late phase of cell activation after inflammatory stimulus. In view of the potential anti-inflammatory properties of this enzyme, we speculate that mast cells contribute not only to the initiation of inflammation related to allergy by releasing a wide variety of inflammatory mediators, but also to its termination by sequestering the PAF produced by mast cell themselves and other effector cells at inflamed sites.