A decrease in remodeling accounts for the accumulation of arachidonic acid in murine mast cells undergoing apoptosis.

The goal of this study was to examine arachidonic acid (AA) metabolism by murine bone marrow-derived mast cells (BMMC) during apoptosis induced by cytokine depletion. BMMC deprived of cytokines for 12-48 h displayed apoptotic characteristics. During apoptosis, levels of AA, but not other unsaturated fatty acids, correlated with the percentage of apoptotic cells. A decrease in both cytosolic phospholipase A(2) expression and activity indicated that cytosolic phospholipase A(2) did not account for AA mobilization during apoptosis. Free AA accumulation is also unlikely to be due to decreases in 5-lipoxygenase and/or cyclooxygenase activities, since BMMC undergoing apoptosis produced similar amounts of leukotriene B(4) and significantly greater amounts of PGD(2) than control cells. Arachidonoyl-CoA synthetase and CoA-dependent transferase activities responsible for incorporating AA into phospholipids were not altered during apoptosis. However, there was an increase in arachidonate in phosphatidylcholine (PC) and neutral lipids concomitant with a 40.7 +/- 8.1% decrease in arachidonate content in phosphatidylethanolamine (PE), suggesting a diminished capacity of mast cells to remodel arachidonate from PC to PE pools. Further evidence of a decrease in AA remodeling was shown by a significant decrease in microsomal CoA-independent transacylase activity. Levels of lyso-PC and lyso-PE were not altered in cells undergoing apoptosis, suggesting that the accumulation of lysophospholipids did not account for the decrease in CoA-independent transacylase activity or the induction of apoptosis. Together, these data suggest that the mole quantities of free AA closely correlated with apoptosis and that the accumulation of AA in BMMC during apoptosis was mediated by a decreased capacity of these cells to remodel AA from PC to PE.

pholipids is the precursor of one important class of inflammatory mediators, collectively known as eicosanoids (e.g. PGD 2 , TXB 2 , LTC 4 , and LTB 4 ) (6). During mast cell activation, AA is rapidly released and utilized for the formation of eicosanoids (7). This can be accomplished by one or more of several PLA 2 enzymes with different molecular characteristics that have been described in mast cells (8 -10). These enzymes show different selectivity toward the hydrolysis of fatty acids at the sn-2-position or phosphobase moieties at the sn-3-position of phospholipids and include a relatively high molecular mass cytosolic PLA 2 (cPLA 2 ; 70 -110 kDa) and several low molecular mass or secretory PLA 2 s (ϳ14 kDa) (11)(12)(13)(14)(15).
In mast cells and other inflammatory cells, low levels of free AA are maintained during resting conditions by both CoA-dependent and CoA-independent mechanisms (16 -19). Free AA is initially converted to arachidonoyl CoA by arachidonoyl-CoA synthetase and is then rapidly incorporated into 1-acyl-2-lysophospholipids by CoA-dependent acyltransferase (20,21). AA in 1-acyl-linked phospholipids is then gradually remodeled into 1-ether-linked phospholipids by CoA-independent transacylase (CoA-IT), leading to large amounts of AA in ether lipid pools (22)(23)(24)(25).
Recent studies suggest that mast cell numbers are primarily controlled by IL-3 and SCF as mast cells have been shown to undergo apoptosis in the absence of these cytokines (26 -29). Until now, neither the inflammatory capacity nor the ability to regulate AA levels of mast cells that are undergoing apoptosis has been determined. To begin to understand how mast cells metabolize AA during apoptosis, we have utilized an in vitro model of mast cell apoptosis induced by cytokine depletion. Our data suggest that there is an increase in free AA levels outside mast cells undergoing apoptosis. Importantly, the mole quantities of AA, but not other unsaturated fatty acids, correlated with the number of apoptotic cells and were inversely proportional to the number of live cells. This accumulation of AA can be explained by a decrease in CoA-IT activity, which results in a diminished capacity of mast cells to remodel AA from phosphatidylcholine (PC) to phosphatidylethanolamine (PE).

Assessment of Apoptosis in BMMC
Caspase Activity-BMMC were cultured without or with cytokines for 24 h. Protease activity in BMMC homogenate was determined using ApoAlert TM CPP32 Colorimetric Assay Kit (CLONTECH) as directed by the manufacturers. Protease activity, defined as the amount of CPP32 required to produce 1 pmol of chromophore (p-nitroanilide, pNA) per min at 25°C, was determined using a standard curve established using different concentrations of pNA at 405 nm. In order to establish that the signal detected was attributable to protease activity, a control assay was performed in which homogenates from cells that had been placed in cytokine-depleted media were pretreated with a CPP32 inhibitor, Asp-Glu-Val-Asp-aldehyde, prior to the addition of CPP32 substrate (Asp-Glu-Val-Asp-p-nitroanilide).
Annexin V Binding to Mast Cells-The reorientation of plasma membrane phospholipids was monitored using the TACS TM annexin V-FITC binding kit from Trevigen (Gaithersburg, MD). Briefly, 1 ϫ 10 6 BMMC were cultured without cytokines or with IL-3 or SCF for 24 h. BMMC were collected by centrifugation (400 ϫ g, 10 min) and then washed once in ice-cold (4°C) phosphate-buffered saline without Ca 2ϩ and Mg 2ϩ . The cells were resuspended in 100 l of binding buffer (10 mM HEPES, pH 7.4, containing 0.15 M NaCl, 5 mM KCl, 1 mM MgCl 2 , and 1.8 mM CaCl 2 ) containing annexin V-FITC and 0.5 g of propidium iodide for 15 min at room temperature in the dark. Binding buffer (400 l) was then added to the mixture, and flow cytometry was performed using a Coulter Epics XL-MCL flow cytometer. The percentages of total cells that did not bind propidium iodide or annexin-FITC (live cells), cells that bound annexin-FITC alone (early apoptosis), cells that bound both propidium iodide and annexin-FITC (late apoptosis), and cells that bound mainly propidium iodide (necrotic cells) were determined, and the results are presented in the form of dot plots.
DNA Ladder Formation-BMMC were lysed in 10 mM Tris/HCl, pH 7.6, containing 0.2% Triton X-100 and 15 mM EDTA. The lysate was treated with 50 g of proteinase K (Promega) overnight at 50°C. Subsequently, DNA was precipitated by centrifugation (15,000 rpm, 10 min on a microcentrifuge) after the addition of 1 ml of cold 2-propanol and 0.1 ml of NaCl (5 M). The DNA was resolved on 2.0% agarose gels by electrophoresis, and DNA fragments were detected after ethidium bromide staining by UV visualization.
[ 3 H]Thymidine Incorporation into BMMC-BMMC were placed in growth medium with or without cytokines. Thymidine incorporation was determined by incubating 1 ϫ 10 6 BMMC with 1 Ci of [ 3 H]thymidine (Amersham Pharmacia Biotech) for 1 h at 37°C. Unincorporated label was removed by washing (twice) the cells with HBSS containing 0.25 mg/ml HSA. The cell pellet was lysed using 0.2 N NaOH (0.25 ml for 1 h). DNA was precipitated using 15% trichloroacetic acid (1 ml) overnight at 4°C. Cellular DNA was then trapped on glass microfiber filters (Whatman GF/C). Free cellular [ 3 H]thymidine was removed from the filters by washes (4 ml, three times), and the amount of radioactivity in DNA was determined by liquid scintillation counting.

Determination of Lipid Peroxidation
BMMC were maintained in culture media without cytokines or with cytokines for 24 h. BMMC were washed (three times) using ice-cold HBSS containing 0.25 mg/ml HSA. Subsequently, cells were lysed in 200 l of HPLC grade H 2 O by repeated (four times) freezing using liquid N 2 and thawing using a 37°C water bath. Lipid peroxidation was determined by measuring the mole quantities of MDA, an end product of the peroxidation of unsaturated fatty acids (30). Briefly, lysates (200 l, 5 ϫ 10 6 BMMC) were added to a 10 mM solution of N-methyl-2phenylindole (650 l) freshly made in acetonitrile/methanol (3:1, v/v). After the addition of 150 l of 12 N HCl, the mixture was incubated at 45°C for 45 min. The A 586 was determined for samples and 0 -10 M MDA standards. The molar amounts of MDA in samples were calculated using an extinction coefficient (gradient of concentration versus A 586 ) obtained using 0 -10 M MDA standards.

Determination of cPLA 2 Activity
BMMC were washed (twice) with HBSS containing 0.25 HSA and 5 mM dithiothreitol. Cells were then suspended at 10 7 /ml in sonication buffer (10 mM HEPES, pH 7.4, containing 80 mM KCl, 1 mM EDTA, 1 mM EGTA, 40 g/ml leupeptin, 25 g/ml pepstatin, 1 mM phenylethylsulfonyl fluoride, 10 mM NaF, 0.2 mM Na 2 VO 3 , and 5 mM dithiothreitol). Sonication (10 s, three times) was performed using a probe sonicator (Heat System Inc.) at a power setting of 2 and 10% output. Protein content of fractions was determined using the Coomassie Plus protein assay reagent (Pierce). cPLA 2 activity was determined using 400 pmol sonicated-vesicles of 1-palmitoyl-2-[1-14 C]arachidonoyl-sn-glycero-3-phosphocholine as substrate and 50 g of protein from BMMC sonicates. cPLA 2 activity was initiated by the addition of substrate to fractions that had been preincubated for 15 min at 37°C in an assay mixture that contained 5 mM dithiothreitol. This preincubation was necessary for eliminating residual secretory PLA 2 activity. The PLA 2 reaction was stopped after 15 min at 37°C by extracting lipids by the method of Bligh and Dyer (31). Free fatty acids were isolated from phospholipids by TLC on silica gel G developed in hexane/diethyl ether/ formic acid (90:60:6, v/v/v). The radioactivity in lipids was located using a radiochromatogram imaging system (Bioscan Inc., Washington, D. C.). Free AA and phospholipids were isolated using TLC zonal scraping, and the radioactivity was determined utilizing liquid scintillation counting. cPLA 2 activity was calculated and expressed as pmol of AA released/mg of protein/min.

RNA Isolation Northern Analysis of cPLA 2 from BMMC
Total RNA was extracted from BMMC cultured with different concentrations of SCF (0 -100 ng/ml) using RNazol (Tel Test Inc., Friendswood, TX). RNA (10 g) resolved by electrophoresis on formaldehyde-containing agarose was transferred onto GeneScreen Plus (PerkinElmer Life Sciences) membranes by capillary blotting. The cPLA 2 cDNA probe was labeled using a nick translation labeling kit (PerkinElmer Life Sciences) following the recommendations of the manufacturer. The membranes were then prehybridized in Quikhyb (Strategene, La Jolla, CA) for 15 min at 68°C before the direct addition of the labeled cDNA probe (1.25-1.5 ϫ 10 6 cpm/ml) to the hybridization solution. After hybridization for 1 h at 68°C, excess probe was washed (twice) with 1ϫ sodium citrate (0.15 M NaCl and 15 mM Na 3 C 6 H 5 O 7 ⅐2H 2 O) containing 0.1% SDS (SSC) at 25°C for 1 min followed by a 15-min wash using 0.2ϫ SSC, 0.1% SDS at 65°C. Signals were detected by exposing the blot to x-ray film with intensifying screens at Ϫ80°C.

Determination of AA Incorporation into BMMC
BMMC were maintained in culture without or with cytokines. After 24 or 48 h, the cells were removed from culture media and placed in HBSS containing 1 Ci of [ 3 H]AA/10 7 BMMC for 30 min at 37°C. Unincorporated label was removed by washing (three times) the cells in HBSS containing 0.25 mg/ml HSA. Glycerolipids were then extracted from the cells, and individual glycerolipid classes were isolated by normal phase HPLC (33,34). Briefly, lipid extracts were reconstituted in loading buffer (hexane/2-propanol/water, 4:5.4:0.3, v/v/v) and then loaded onto an Ultrasphere Silica column (Rainin Instrument Co., Woburn, MA) that had been conditioned with hexane/2-propanol/ethanol/50 mM phosphate buffer (pH 7.4)/acetic acid (490:367:100:30:0.6, v/v/v/v/v). After 5 min, the amount of phosphate was increased from 3 to 5% over a 5-min period. This solvent composition was then maintained until all major phospholipid classes had been eluted from the column. In radiolabeled assays, 1-min fractions were collected, and the amount of radioactivity in each lipid class was determined by liquid scintillation counting.

Determination of Mole Quantities of AA in Phospholipid Classes and Subclasses
BMMC glycerolipids extracted by the method of Bligh and Dyer were separated into classes by normal phase HPLC as described above.
[ 2 H 8 ]AA (100 ng) was added to a fraction of each glycerolipid that was subsequently subjected to base hydrolysis (34). Fatty acids were then converted to pentafluorobenzyl esters, and the mole quantities of free fatty acids were determined by NICI-GC/MS. Carboxylate anions (m/z) were monitored at 303 and 311 for AA and [ 2 H 8 ]AA, respectively, in the single ion monitoring mode, and mole quantities of AA were determined from standard curves obtained using [ 2 H 8 ]AA as an internal standard.
To determine the distribution of arachidonate into phospholipid subclasses, PC and PE fractions from normal phase HPLC were hydrolyzed using 10 or 25 units of grade 1 Bacillus cereus phospholipase C (Roche Molecular Biochemicals) for 2.5 and 6 h, respectively. Diradylglycerides obtained from the phospholipase C hydrolysis were converted to diradylglyceride acetates using acetic anhydride (500 l) and pyridine (35). 1-Acyl, 1-alkyl, and 1-alk-1Ј-enyl subclasses were separated by TLC on silica gel G developed in benzene/hexane/diethyl ether (50:25:4, v/v/v). Each subclass was extracted from the silica gel, and the mole quantities of arachidonate were determined by NICI-GC/MS as described above.

Determination of CoA-IT Activity
For cell free assays, BMMC were cultured without or with cytokines for 24 h and then washed (three times) using HBSS containing 0.25 mg/ml HSA. The cells were then suspended in CoA-IT sonication buffer (50 mM HEPES buffer, pH 7.4, containing 1 mM EDTA and 20% sucrose (w/v)). Subsequently, the cells were broken by sonication using a probe sonicator (Heat System, Inc.) as described above for cPLA 2 assays. Cytosolic and membrane fractions were obtained after ultracentrifugation (100,000 ϫ g, 1 h, 4°C). The membrane fraction was diluted in phosphate-buffered saline containing 1 mM EGTA, and 10 g of total protein was utilized for CoA-IT activity determination. The reaction was initiated by the addition of 1-[ 3 H]alkyl-2-lyso-GPC (0.1 Ci) containing 1 nmol of 1-O-hexadecyl-2-lyso-GPC in a final volume of 100 l. After 10 min at 37°C, the reaction was stopped, and lipids were extracted. Phospholipids were separated by TLC on silica gel G developed in chloroform/methanol/acetic acid/water (50:25:8: was visualized by radioscaning (Bioscan), scrapped, and then quantified by liquid scintillation spectroscopy.
For whole cell assays, BMMC were pulse-labeled with [ 3 H]AA for 30 min (1 Ci/10 7 cells). Unincorporated label was removed by washing BMMC (three times) using HBSS containing 0.25 mg/ml fatty acid-free HSA. The cells were then placed in fatty acid-enriched cell culture media (10% fetal calf serum) in the absence or presence of IL-3 or SCF (100 ng/ml). After different periods of time, BMMC were removed from cell culture media, and glycerolipid classes were obtained by normal phase HPLC as described above. The amount of radiolabel in each lipid class was then determined by liquid scintillation counting.

Determination of Lyso-PC and Lyso-PE Levels
BMMC (10 6 /ml) were incubated with 0.5 Ci of [Me 3 H]choline or 0.5 Ci of [1-3 H]ethanolamine for the determination of lyso-PC and lyso-PE, respectively. After 24 h in culture media, unincorporated label was removed by washing the cells (twice) using sterile HBSS containing 0.25 mg/ml HSA. The cells were then cultured in the absence or presence of IL-3 or SCF (100 ng/ml) for a further 24 h. Phospholipids were extracted from cells as described above. The phospholipid extract was resuspended in 50 l of chloroform containing 10 g each of PC and lyso-PC or PE and lyso-PE. Lyso-PC and lyso-PE were isolated by TLC on silica gel G using chloroform/methanol/acetic acid/water (50:25: The radioactivity in phospholipids (PC or PE) and lysophospholipids (lyso-PC or lyso-PE) were determined by liquid scintillation counting.

Effects of Exogenous AA on Cellular AA Metabolism
BMMC (10 6 /ml) were labeled with [ 3 H]AA for 24 h to obtain isotopic labeling of the major lipid classes (24). After removal of unincorporated label, cells were maintained in culture in the presence of increasing concentrations of exogenous AA (0 -50 M). After 24 h, the distribution of radiolabel into lipid classes was determined by TLC as described above.

Statistical Analysis
Data are expressed as the means Ϯ S.E. of separate experiments. Statistics (p values) were obtained using Student's t test for paired samples. Notations used on figures and legends are an asterisk for p Ͻ 0.05.

Apoptosis of BMMC in Cytokine-depleted Media-Our previ-
ous studies showed that BMMC cultured in IL-3/SCF maintained low intracellular levels of AA (36). However, the regulation of AA has not been described when these cells are undergoing apoptosis. Initial experiments were designed to obtain a reproducible model of mast cell apoptosis. In these experiments, caspase activity was examined in BMMC cultured without or with IL-3/SCF for 24 h. BMMC placed in cytokine-depleted media had higher protease activity (14.5 Ϯ 0.9 CPP32 units, n ϭ 3) compared with cells maintained in cytokines (4.8 Ϯ 1.1 CPP32 units). This protease activity was reduced to control levels (4.4 Ϯ 0.1 CPP32 units) by a caspase inhibitor. These initial data suggested that a marker of apoptosis (caspase) was increased when BMMC were placed in cytokine-depleted media for 24 h.
To determine the percentage of apoptotic cells, the orientation of PS on the surface of mast cells was examined by annexin V binding. When BMMC were placed in cytokine-depleted media for 24 h, there was an increase in the percentage of annexin-FITC binding cells (Fig. 1A). Consistent with this observation, maintaining BMMC in cytokine-depleted media caused a significant increase in the percentage of apoptotic cells concomitant with a decrease in live cells (Table I). Further evidence of apoptosis was obtained by examining DNA fragmentation. As shown in Fig. 1B, BMMC cultured for 24 h with IL-3 or SCF maintained intact high molecular weight DNA. By contrast, DNA extracted from cells placed in cytokine-depleted media displayed distinct DNA fragmentation patterns. Finally, we determined the capacity of BMMC to synthesize new DNA. As shown in Fig. 1C, the incorporation of radiolabeled thymidine into cellular DNA decreased (Ͼ80%) when BMMC were placed in cytokine-depleted media for as little as 12 h. Taken together, these data suggest that BMMC placed in cytokinedepleted media rapidly stop proliferating and then undergo apoptosis.
Mole Quantities of Arachidonic Acid Correlate with the Percentage of Apoptotic Cells-As described above, mast cells placed in cytokine-depleted media undergo apoptosis. It is known that the location of PS on the surface of cells is a marker of apoptosis (37). However, it has not been established whether changes in the phospholipid microenvironment are accompanied by changes in fatty acids within or outside mast cells. Therefore, subsequent studies examined the mole quantities of various unsaturated fatty acids within mast cells or released into the culture media. Removal of cytokines resulted in a significant increase in AA levels within mast cells, while levels of other unsaturated fatty acids (LA, OA) did not change (Table  II). Subsequent studies examined free fatty acids outside mast cells and compared these levels with the percentage of live or dead (annexin V binding) cells. As shown in Fig. 2A, resting levels of AA outside mast cells were directly proportional to the percentage of apoptotic cells ( Fig. 2A) and inversely proportional to the percentage of live cells (Fig. 2B). In both cases, there was correlation between the percentages of dead or live cells and the mole quantities of AA in cell culture media (R 2 ϭ 0.76 and 0.92, respectively). In contrast, mole quantities of other unsaturated fatty acids (LA, OA, EPA, and DHA) were not related to the percentage of dead (Fig. 2, C, D, E, and F, respectively) or living cells (data not shown). These data suggest that the selective accumulation of AA within or outside mast cells is closely associated with apoptosis.
To further examine the effects of unsaturated fatty acids on apoptosis, BMMC were incubated with different concentrations Annexin V binding was determined by flow cytometry as described under "Experimental Procedures." Quadrants 1, 2, 3, and 4 represent dead cells, cells in late apoptosis, live cells, and cells in early apoptosis, respectively. These data are representative of eight separate experiments. B, DNA was extracted from BMMC that had been placed in cell culture media for 24 h with IL-3, SCF, or without cytokines (NONE). DNA was extracted from 2 ϫ 10 6 BMMC and resolved by agarose gel electrophoresis followed by ethidium bromide staining as described under "Experimental Procedures." These data are representative of five separate experiments. C, BMMC were placed in growth media without (q) or with 100 ng/ml IL-3 (E). After different periods of time in culture, the incorporation of [ 3 H]AA thymidine into 1 ϫ 10 6 BMMC was determined. These data are the mean Ϯ S.E. of triplicate determinations and are representative of five separate experiments.

TABLE I
Cytokine depletion induces apoptosis of BMMC BMMC were cultured without cytokines (None) or with IL-3 or SCF (100 ng/ml) for 24 h. Annexin FITC binding to mast cells was determined by flow cytometry. The percentage of necrotic cells (quadrants 1 of Fig. 1A), late and early apoptotic cells (quadrants 2 and 4, respectively, of Fig. 1A), and live cells (quadrants 2 of Fig. 1A)  releasing activities (cPLA 2 ), a decrease in AA incorporation and remodeling activities or a decrease in the conversion of free AA to metabolites. Thus, we examined the stimulus-coupled AA release from BMMC that were undergoing apoptosis. BMMC placed in cytokine-depleted media released 757.2 Ϯ 112.3 (n ϭ 5) pmol of AA/5 ϫ 10 6 BMMC upon stimulation with antigen for 5 min. This level of AA release was significantly higher than AA released from cells placed in IL-3 (259.3 Ϯ 31.7 pmol/5 ϫ 10 6 BMMC) or SCF (156.8 Ϯ 50.6 pmol/5 ϫ 10 6 BMMC). Ionophore A23187 stimulation also resulted in more AA release from cytokine-depleted cells compared with cytokine-treated cells (data not shown). Taken together, these data suggest that cells cultured without cytokines have the capacity to release significantly more AA than cells cultured with cytokines.
Increase in Lipid Peroxidation when Mast Cells Are Undergoing Apoptosis-Several studies suggest that peroxidation of unsaturated fatty acids is linked to apoptosis (38 -40). To determine whether AA accumulation is accompanied by lipid peroxidation in BMMC, we examined the accumulation of MDA, the major product of lipid peroxidation reactions, within BMMC induced to undergo apoptosis by cytokine depletion. When BMMC were cultured in IL-3 or SCF-supplemented culture media, levels of MDA (6.70 Ϯ 0.71 nmol/5 ϫ 10 6 BMMC, n ϭ 6 and 6.78 Ϯ 1 nmol/5 ϫ 10 6 BMMC, n ϭ 6, respectively) were not altered within a 24-h incubation period. In contrast, there was a significant increase in MDA levels (15.43 Ϯ 1.89 nmol/5 ϫ 10 6 BMMC, n ϭ 6, p Ͻ 0.05) within BMMC that were induced to undergo apoptosis by cytokine withdrawal. Taken together, these data suggest that AA accumulation within BMMC undergoing apoptosis induced by cytokine withdrawal is accompanied by an increase in lipid peroxidation. cPLA 2 Activity and Expression Does Not Correlate with Mole Quantities of AA-As described above, BMMC placed in cytokine-depleted media for 24 h released more AA upon activation. Since stimulus-coupled release of AA is associated with cPLA 2 activation, we next examined cPLA 2 expression and activity as a possible explanation for the accumulation of AA outside mast cells. As shown in Fig. 3A (upper panel), cPLA 2 mRNA decreased when BMMC were placed in cytokine-depleted media for 24 h. Western analysis also revealed that there was a decrease in cPLA 2 protein when mast cells were placed in cytokine-depleted media (Fig. 3B). Likewise, cPLA 2 activity was lower in cells cultured without cytokines (Fig. 3C). Taken together, these data suggest that changes in cPLA 2 levels are not related to the accumulation of AA in cell culture media or within BMMC when these cells are undergoing apoptosis.
Assessment of Eicosanoid Formation by BMMC Undergoing Apoptosis-To determine whether the accumulation of AA outside mast cells was due to a decrease in their capacity to form products, levels of eicosanoids (PGD 2 and LTB 4 ) were examined by NICI-GC/MS. Stimulation by either antigen increased PGD 2 , while PGD 2 levels remained unchanged after ionophore A23187 stimulation of BMMC undergoing apoptosis. There was no significant difference in LTB 4 biosynthesis by BMMC grown with or without cytokines (Table III). These data suggest that impairment of eicosanoid biosynthesis is not responsible for the accumulation of free AA outside BMMC that are undergoing apoptosis.
Determination of Arachidonic Acid Incorporation into Lipids-Our previous data indicated that BMMC rapidly incorporated trace amounts of AA predominantly into PC and PI pools (24). Under resting conditions, AA is slowly remodeled from these early pools to PE (24). To determine whether this incorporation process was blocked in cells undergoing apoptosis, cells were pulse-labeled with [ 3 H]AA, and the incorporation of AA into lipids was monitored. BMMC cultured without cyto-  kines, with IL-3 or SCF for 24 h, incorporated the same amount of radiolabel (0.414 Ϯ 0.097, 0.519 Ϯ 0.148, and 0.512 Ϯ 0.159 Ci/10 7 BMMC, respectively, n ϭ 5). When the incorporation of [ 3 H]AA into lipid classes was examined, cells cultured without cytokines accumulated more AA in neutral lipids (Fig. 4A) than cells cultured with IL-3 or SCF (Fig. 4, B and C, respectively). In contrast, [ 3 H]AA incorporation into PE was lower in cytokine-depleted BMMC, while there was no difference in the incorporation of [ 3 H]AA into PI/PS or PC fractions when cells were maintained without or with cytokines (Fig. 4) (Fig. 5A). Changes in the distribution of AA within PE subclasses (1-acyl-2AA-GPE and 1-alk-1-enyl-2-AA-GPE) and PC subclasses (1-acyl-2-AA-GPC and 1-alkyl-2-AA-GPC) mirrored the changes in their FIG. 3. Reduction in cPLA 2 expression, protein level, and activity during BMMC apoptosis. A, total RNA was extracted from BMMC cultured with increasing concentrations of SCF. RNA (10 g) was resolved by electrophoresis on formaldehyde-containing agarose and then transferred onto GeneScreen Plus membranes by capillary blotting. GeneScreen Plus membranes were probed using a labeled cPLA 2 cDNA probe (upper panel) or glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (lower panel) as described under "Experimental Procedures." Signals were detected by exposing the blot to x-ray film with intensifying screens at Ϫ80°C. These blots are representative of five separate experiments. B, BMMC were placed in culture media with increasing concentrations of SCF for 24 h. SDS-PAGE was performed using a 4 -20% gel, and proteins were blotted onto polyvinylidene difluoride membranes. Immunodetection of cPLA 2 was accomplished using anti-cPLA 2 monoclonal antibody and peroxidase-conjugated anti-rabbit IgG, followed by enhanced chemiluminescence. These data are representative of five separate experiments. C, BMMC were cultured with increasing concentrations of SCF for 24 h. cPLA 2 activity was determined using 50 g of homogenate as described under "Experimental Procedures." GAPDH, glyceraldehyde-3-phosphate dehydrogenase. These data are the mean Ϯ S.E. of four separate experiments performed in duplicate. *, p Ͻ 0.05. respective subclasses (Fig. 5B). Compared with BMMC grown in cytokine-supplemented culture media (IL-3 or SCF), there was a significant decrease in arachidonate content of PE (40.7 Ϯ 8.1%; n ϭ 3) in BMMC maintained in cytokine-depleted media (Fig. 5C). Concomitantly, there was an increase in the arachidonate content in NL (84.2 Ϯ 12.8%, n ϭ 3) and PC (50. 2 Ϯ 18.1%, n ϭ 3). These data reveal that the cellular distribution of AA is altered when BMMC are undergoing apoptosis, such that PC, and not PE, becomes the predominant arachidonate-containing lipid pool and there is ϳ2-fold increase in the arachidonate mass in NL.
Determination of CoA-independent Transacylase Activity-As described above, the major difference between cells cultured with cytokines and cells undergoing apoptosis is the decrease in arachidonate content of PE concomitant with an increase in arachidonate content of PC and NL. Additionally, free AA levels are higher in BMMC undergoing apoptosis. AA that is initially incorporated into PC is remodeled to PE by CoA-IT activity, and the inhibition of CoA-IT has been shown to result in AA accumulation in several cell types (41). Thus, a decrease in CoA-IT activity could account for the decrease in AA content of PE and the accumulation of AA within mast cells undergoing apoptosis. We tested this hypothesis by determining AA remodeling in whole cells and also measuring microsomal CoA-IT activity. BMMC pulse-labeled with [ 3 H]AA incorporated arachidonate in rank order PC Ͼ PI/PS Ͼ PE Ͼ NL. In the absence of cytokines, there was an increase of radiolabel in NL when cells are placed in culture media for 24 and 48 h. By contrast, [ 3 H]AA was maintained at the same level in NL cultured with IL-3 or SCF (Fig. 6A). Compared with pulselabeled cells, there was a slight increase in [ 3 H]AA in PE in cytokine-depleted cells at 24 and 48 h. However, the percentage of [ 3 H]AA in PE in cytokine-depleted BMMC was significantly lower than the levels found in BMMC cultured with IL-3 or SCF (Fig. 6B).  (Fig. 6, C and D). Interestingly, the increase in the mole quantities of arachidonate in PC (Fig. 5C) is not observed in the pulse experiments (Fig. 4) or in the pulse/chase experiments in Fig. 6D, probably because the mass amounts of [ 3 H]AA are very small compared with the bulk of unlabeled arachidonate in PC. Moreover, most of [ 3 H]AA that would have resided in PC is shunted to the neutral lipid fraction in apoptotic BMMC when these labeling strategies are utilized. However, the decrease of [ 3 H]AA in PE suggests that the activity (CoA-IT) responsible for transferring/remodeling AA to PE is reduced in BMMC undergoing apoptosis. To further show that the decrease in arachidonate content in PE was due to a decrease in CoA-IT activity, microsomal fractions were prepared from cells cultured without or with cytokines. CoA-IT activity was lower in cells placed in cytokine-depleted media (0.13 Ϯ 0.05 pmol/mg/ min) compared with BMMC maintained in IL-3 (1.10 Ϯ 0.09 pmol/mg/min, n ϭ 8, p Ͻ 0.05) or SCF (0.71 Ϯ 0.16 pmol/mg/ min) for 24 h. A similar CoA-IT activity profile was determined after 48 h (data not shown). These data suggest that a decrease in CoA-IT accounts for the decrease in AA remodeling and the accumulation of free AA when BMMC undergo apoptosis following cytokine depletion.
To determine whether the decrease in CoA-IT activity, or the induction of apoptosis during cytokine depletion could be accounted for by an increase in lysophospholipid levels within BMMC, phospholipid head groups were radiolabeled, and the accumulation of radioactivity in lyso-PC or lyso-PE was determined. When cells were placed in cytokine-depleted media, less [ 3 H]choline was incorporated into lyso-PC compared with cytokine-supplemented cells (5757 Ϯ 1781, 6763 Ϯ 1909, and 9245 Ϯ 2085 dpm/5 ϫ 10 6 BMMC (n ϭ 3) for no cytokine, IL-3, and SCF, respectively). Similarly, BMMC labeled with [1-3 H]ethanolamine incorporated less radioactivity into lyso-PE (1160 Ϯ 181, 1233 Ϯ 164, and 1524 Ϯ 215 dpm/5 ϫ 10 6 BMMC (n ϭ 3) for no cytokine, IL-3, and SCF, respectively). However, expressed as a percentage of total radioactivity, there was no difference in the amount of lyso-PC (16.7 Ϯ 0.4, 13.8 Ϯ 2.1, and 17.0 Ϯ 2.2% of total radioactivity (n ϭ 3) for no cytokine, IL-3, and SCF, respectively) or lyso-PE (7.6 Ϯ 0.7, 5.8 Ϯ 0.2, and 5.8 Ϯ 0.2% of total radioactivity (n ϭ 3) for no cytokine, IL-3, and SCF, respectively) in cells cultured in cytokine-depleted media when compared with cells grown with IL-3 or SCF. These data suggest that the decrease in CoA-IT observed during apoptosis is not due to an increase in the levels of endogenous substrate (lyso-PC or lyso-PE). In addition, these data suggest that lysophospholipids do not contribute to apoptosis of BMMC induced by cytokine depletion.
AA Accumulation in Neutral Lipid Classes-Our previous studies suggested that inflammatory cells exposed to high concentrations of AA shuttled excess AA into neutral lipids, with triglycerides being the predominant class (42,43). To determine whether a similar pathway existed in BMMC, the neutral lipid fractions (Fig. 6A) were further resolved into individual classes. As shown in Table IV, there was a significant increase in [ 3 H]AA in all neutral lipid fractions obtained from cells cultured without cytokines. However, the bulk of [ 3 H]AA resided in the TG pool. These data suggest that BMMC undergoing apoptosis accumulate more AA into neutral lipid classes than do live cells.
Roles of Exogenous AA on Cellular AA Metabolism-To determine whether changes in AA metabolism described above were due to the accumulation of AA in the cell culture media, BMMC were radiolabeled to isotopic equilibrium using [ 3 H]AA. The cells were maintained in culture with different concentrations of exogenous AA. As shown in Fig. 7, the bulk of the radiolabel in BMMC resided in phospholipid classes (94.6 Ϯ 0.4%, n ϭ 3). The addition of exogenous AA resulted in a decrease in [ 3 H]AA in phospholipid classes. Concomitantly, there was a dose-dependent accumulation of radiolabel in TGs and in free fatty acids. The bulk of the TG was cell-associated while most of the free AA was released into the cell culture media. These data suggest that adding exogenous AA to BMMC can induce changes in cellular AA metabolism similar to those observed during cytokine depletion. DISCUSSION The present study demonstrates an accumulation of AA within cells and in the cell culture media of mast cells undergoing apoptosis induced by cytokine depletion. Of a series of polyunsaturated fatty acids measured, only the levels of AA directly correlated with the percentage of dead cells. Conversely, the percentage of live cells was inversely proportional to the mole quantities of AA found within apoptotic mast cells or in cell culture media. These initial data suggested that AA metabolism played a role in apoptosis. There are several mechanisms that could potentially account for the accumulation of AA by mast cells undergoing apoptosis (Fig. 8). These include a decrease in AA uptake and incorporation into phospholipids (Fig. 8a), a decrease in AA remodeling (Fig. 8b), an increase in AA release from phospholipid pools (Fig. 8c), or a decrease in the capacity of mast cells to convert AA to bioactive lipid products (Fig. 8d).
Previous studies from our laboratory and others have indi-  cated that different isotypes of PLA 2 are involved in the release of AA from mast cells (8, 44 -46). While secretory PLA 2 s are known to be nonspecific in hydrolyzing fatty acids in cell-free systems, it appears that secretory PLA 2 also recruits cPLA 2 that accounts for the selective mobilization of AA during cell activation (32). The importance of cPLA 2 in AA mobilization from mast cells has recently been confirmed using targeted disruption of the cPLA 2 gene in mice (46). In these studies, neither immediate nor delayed prostanoid formation was observed in mast cells obtained from cPLA 2 knockout mice. In another set of studies, cells that overexpressed a mutated cPLA 2 (Ser 505 3 Ala 505 ) exhibited greatly diminished capacity to release AA compared with cells overexpressing wild type cPLA 2 (47). Since the activation of cPLA 2 is important in AA release, its activation represents a possible mechanism by which high levels of AA would be found within or in the supernatant fluid around mast cells. However, our data suggest that, while there is an increase in AA levels within mast cells that are undergoing apoptosis, there is a decrease in cPLA 2 mRNA, protein and activity. Therefore, cPLA 2 is not responsible for the accumulation of AA when mast cells are undergoing apoptosis. Other studies have shown that cPLA 2 is decreased in cells undergoing apoptosis induced by Fas activation or by contact inhibition (48 -51). The decrease in cPLA 2 was shown to be due to a caspase-mediated breakdown of the enzyme (48,52). It has been implied from these studies that the decrease in cPLA 2 represents a novel mechanism by which inflammatory processes are regulated when cells are undergoing apoptosis. The present study suggests that even when cPLA 2 levels are decreased in cells, it may not be strictly correct to assume that inflammatory capacity is attenuated, since other enzymes regulate AA levels. Our study also reveals that cPLA 2 levels within mast cells are regulated by cytokines. In the presence of optimum levels of SCF or IL-3, BMMC proliferate in culture. Under these conditions, cPLA 2 expression is maintained at high levels. During cytokine depletion, there is a decrease in cPLA 2 expression. These changes support the observation of Anderson et al. (51) suggesting that cPLA 2 is linked to cell proliferation.
Recent studies also suggest that inhibitors of cyclooxygenase or 5-lipoxygenase increase free AA levels within cells (41,53). These inhibitors also induce apoptosis of cancer cells (53)(54)(55)(56)(57)(58)(59)(60)(61). Thus, changes in cyclooxygenase activity and AA levels are closely linked to apoptosis. In the present study, BMMC undergoing apoptosis produced more prostanoids than control cells. LTB 4 levels remained the same, regardless of the apoptotic status of the cells, demonstrating that decreased capacity to form eicosanoids does not account for the accumulation of AA in mast cells undergoing apoptosis induced by cytokine withdrawal.
When AA is provided to mammalian cells, it is rapidly incorporated into various glycerolipid pools through both CoA-dependent and CoA-independent acylation reactions (17,62,63). Initially, AA is converted to AA-CoA by arachidonoyl-CoA synthetase. AA-CoA is then transferred to 1-acyl-2-lyso-sn-glycero-3-phospholipids by CoA-dependent acyltransferase. Various studies have shown that when this incorporation process is inhibited with Triacsin C, there is accumulation of AA products within cells and in the culture media (64,65). Importantly, inhibition of this process has been associated with apoptosis of some cells (66). The present study suggests that this initial incorporation of AA into mast cells is not altered when these cells are undergoing apoptosis. Thus, a decrease in AA incorporation does not account for the increase in AA levels within mast cells or in culture media.
In mast cells, AA that is initially incorporated into 1-acyllinked glycerophospholipids is slowly remodeled from these initial pools into 1-ether-linked glycerophospholipid pools. Our studies suggest that the remodeling of arachidonate from 1-acyl-linked to 1-ether-linked phospholipids is orchestrated by CoA-IT (7,24). We recently demonstrated that treating cancer cell lines with CoA-IT inhibitors led to accumulation of intracellular AA (41,(67)(68)(69). Importantly, this inhibition of CoA-IT also resulted in the inhibition of cell cycle progression within breast cancer cells. The present study underscores the role of CoA-IT in AA metabolism. When mast cells were placed in cytokine-depleted media, the mole quantity of AA in PE was significantly reduced, concomitant with a decrease in CoA-IT activity in microsomal pellets. Therefore, a decrease in CoA-IT is responsible for the accumulation of AA in mast cells undergoing apoptosis. Since a decrease in CoA-IT activity results in arachidonate accumulation in PC, hydrolysis of PC by PLA 2 could also potentially account for the increase in free AA levels (Fig. 8c). Because no changes were observed in lysophospholipids levels during apoptosis, and since specific activity measurements suggest that most free AA is generated from PE (7), it is likely that hydrolysis of PC plays only a minor role in free AA accumulation. However, during BMMC activation, when there is an increase in lysophospholipid levels and release of AA from all major phospholipid pools, hydrolysis of PC by PLA 2 may contribute to free AA release (7,63,70). Various pathways have been described for the incorporation of AA into glycerophospholipids, depending upon the concentration of AA to which cells are exposed (71). Exposure of cells to low AA concentrations results in AA incorporation into primarily 1-acyl-linked phospholipid molecular species through a high affinity, low capacity enzymatic pathway. In contrast, exposure of cells to high levels of AA will result in the incorporation of AA into neutral lipid fractions (mainly TG) and the formation of 1,2-diarachidonoyl-sn-glycero-3-phosphocholine by a low affinity, high capacity pathway (42,71). Our data show FIG. 8. Proposed mechanism accounting for AA accumulation in mast cells undergoing apoptosis. Processes that increase cellular AA levels within mast cells include a decrease in the uptake and incorporation of cellular AA into phospholipids (a), a decrease in CoA-IT activity resulting in a decrease in AA remodeling (b), an increase in PLA 2 activity that results in the rapid hydrolysis of AA from cellular phospholipids (c), or the inhibition of cyclooxygenase (COX) and/or 5-lipoxygenase (5-LO) (d). Hydrolysis of PC during BMMC activation may also lead to AA accumulation (c, dashed line). When AA levels are increased, the cells respond by incorporating free AA into neutral lipids via de novo TG synthesis (e). Cellular AA may also be autoxidized to form products, which may induce apoptosis of mast cells (f). that there is an increase in AA mass in the neutral lipid fractions of mast cells undergoing apoptosis. In addition, the AA mass in 1-acyl-linked PC species also increased in mast cells undergoing apoptosis, such that this becomes the single largest phospholipid subclass instead of 1-alk-1-enyl-2-AA-GPE. These data suggest that there is induction of the low affinity, high capacity pathway of AA metabolism in mast cells undergoing apoptosis. The addition of exogenous AA increased cellular [ 3 H]AA incorporation into TG in a dose-dependent manner. Exogenous AA also induced free [ 3 H]AA accumulation in the cell culture media. However, much higher levels of exogenous AA were needed to induce these changes, because exogenous AA is bound to serum albumin and is distributed over a larger volume than cell-associated AA. Overall, the increase of AA in neutral lipid fractions via de novo synthesis (Fig. 8e) and in PC subclasses (Fig. 8a) suggests that BMMC undergoing apoptosis have been exposed to high concentrations of free AA.
In addition to the accumulation of AA in some pools of lipids, another consequence of the build up of high levels of AA within and outside mast cells is the formation of oxidized products. High levels of AA may induce the formation of proapoptotic molecules such as ceramides, while fatty acids may affect various apoptotic proteins (53,69,72,73). An important role for lipid peroxidation in apoptosis is further shown by our data showing a significant increase in lipid peroxidation only in mast cells that are undergoing apoptosis (Fig. 8f). Since lipid peroxidation has been linked to apoptosis (38), our present study suggests that mast cells may be undergoing apoptosis as a result of the formation of reactive lipid peroxides.
Overall, this study implies that perturbation of cellular arachidonate metabolism is a critical process in BMMC that are undergoing apoptosis. Importantly, AA accumulation is a predictive indicator of apoptosis in BMMC after cytokine removal. The major enzyme activities that change when mast cells are undergoing apoptosis are cPLA 2 and CoA-IT. It is likely that there is a link between these two activities, since they exhibit similar biological properties. First, our data suggest that a decrease in cPLA 2 is accompanied by a decrease in CoA-IT activity during apoptosis of mast cells. Second, both cPLA 2 and CoA-IT are selective for arachidonate (19,74). Third, inhibitors of cPLA 2 induce apoptosis of cells, as do inhibitors of CoA-IT (51,68). Fourth, inhibitors of CoA-IT prevent AA release and eicosanoid biosynthesis, as do inhibitors of cPLA 2 (75). Fifth, cytokine treatment of BMMC (present study) and neutrophils results in an increase in both CoA-IT and cPLA 2 activity (76). Sixth, both activities are heat labile and are not influenced by sulfhydryl reducing agents (77). Finally, Reynolds et al. (78) reported that cPLA 2 has weak transacylase activity. However, even with these similarities, there are some striking differences between CoA-IT and cPLA 2 . Whereas CoA-IT does not require calcium for activation and is mainly membrane-bound, the major cPLA 2 isotypes require calcium for activation and reside mainly in the cytosol. However, cPLA 2 translocates to perinuclear membranes during cell activation, and a recently cloned cPLA 2 isotype (cPLA 2 ␥) is reported to be calcium-independent and to be mainly membrane-bound (79). Since CoA-IT has not been cloned, further studies will be required to show whether these two activities cooperate within cells in regulating AA metabolism. In addition to the possible link between cPLA 2 and CoA-IT, the present studies also raise important implications concerning the role of AA in apoptosis. Further investigations will be required to determine the signal transduction processes that link AA accumulation, lipid peroxidation, and changes in arachidonate pools to apoptosis of BMMC.