Protein-facilitated export of arachidonic acid from pig neutrophils.

Activated neutrophils release a variety of eicosanoids into the extracellular medium including arachidonic acid, 5-hydroxyicosatetraenoic acid, and leukotriene A4 and B4. In this study, the mechanism of arachidonic acid export has been examined using inside-out plasma membrane vesicles from pig polymorphonuclear leukocytes. Tritiated arachidonic acid associated rapidly with the membrane vesicles and crossed the membrane into the intravesicular space in a time-dependent and saturable manner. Half the maximal influx rate was measured at an arachidonate concentration of 5.7 microM, and a maximal influx velocity of 3.0 nmol/mg x min was determined at pH 6.8. Influx into vesicles was sensitive to a number of common anion transport inhibitors including pentachlorophenol, phloretin, diiodosalicylic acid, and quercetin as well as to the proteases trypsin and Pronase, suggesting a protein-dependent process. Furthermore, influx was temperature-sensitive with an energy of activation of 11.6 kcal/mol. Varying extravesicular concentration of ATP, Na+, or K+ had no impact on arachidonate influx, whereas changes in pH had a profound effect; optimum transport activity was observed at an extravesicular pH of 6, whereas raising the pH to 9.5 essentially abolished uptake. These results indicate and initially characterize a novel protein-facilitated arachidonate export mechanism in pig neutrophils.

Activated neutrophils release a variety of eicosanoids into the extracellular medium including arachidonic acid, 5-hydroxyicosatetraenoic acid, and leukotriene A 4 and B 4 . In this study, the mechanism of arachidonic acid export has been examined using inside-out plasma membrane vesicles from pig polymorphonuclear leukocytes. Tritiated arachidonic acid associated rapidly with the membrane vesicles and crossed the membrane into the intravesicular space in a time-dependent and saturable manner. Half the maximal influx rate was measured at an arachidonate concentration of 5.7 M, and a maximal influx velocity of 3.0 nmol/mg ؋ min was determined at pH 6.8. Influx into vesicles was sensitive to a number of common anion transport inhibitors including pentachlorophenol, phloretin, diiodosalicylic acid, and quercetin as well as to the proteases trypsin and Pronase, suggesting a protein-dependent process. Furthermore, influx was temperature-sensitive with an energy of activation of 11.6 kcal/mol. Varying extravesicular concentration of ATP, Na ؉ , or K ؉ had no impact on arachidonate influx, whereas changes in pH had a profound effect; optimum transport activity was observed at an extravesicular pH of 6, whereas raising the pH to 9.5 essentially abolished uptake. These results indicate and initially characterize a novel protein-facilitated arachidonate export mechanism in pig neutrophils.
Polymorphonuclear leukocytes (PMN) 1 are the predominant cell type present in areas of acute inflammation. Activated neutrophils release a number of eicosanoids such as arachidonic acid (1-3), 5-hydroxyicosatetraenoic acid, and leukotriene A 4 (4) and B 4 (5) into the plasma where they may amplify or perpetuate the acute inflammatory response. Recent progress in leukotriene research has led to a more detailed understanding of the molecular biology and enzymology of the oxidative metabolism of arachidonic acid and to the development of potent inhibitors of biosynthesis as well as receptor antagonists interfering with signal transduction (6,7). The mechanism of how eicosanoids, once biosynthesized, leave the producer cell to reach their target cell is, however, still poorly understood. Recently, an ATP-driven export pump for cysteinyl leukotrienes has been identified in hepatocytes (8), erythrocytes (9), and heart cells (9,10). In addition, in intact PMN a LTB 4 export carrier has been identified and characterized (11). To study the cellular export mechanism, whole cell systems are too complex to allow the accurate measurement of transport kinetics and of the effects of inhibitors. Furthermore, the situation for PMN is particularly complex since these cells do not only biosynthesize and release eicosanoids but also avidly take up and metabolize arachidonic acid, LTA 4 (1,4), and LTB 4 (12).
To overcome those restrictions imposed by whole cell systems to the investigation of transport mechanisms, we prepared inside-out plasma membrane vesicles from PMN and investigated the usefulness of this system to study eicosanoid transport using arachidonic acid as a ligand. Inside-out plasma membrane vesicles without contamination with right side-out vesicles were prepared by isolating phagocytic vesicles from PMN. When a PMN engulfs particles, the ingested material is encased within a vesicle derived from the plasma membrane with an inside-out orientation (13). After phagocytosis of albumin-coated mineral oil droplets, those phagocytic vesicles can readily be isolated and have been used for Ca 2ϩ export studies (14).
The mechanism of the translocation process of fatty acids through the plasma membrane in general is controversial (15). On one hand, it is argued that fatty acid movement through the phospholipid bilayer is an entirely unregulated process in which diffusion of fatty acids in either direction of the membrane is driven by concentration gradients (16). Juxtaposed to the diffusion hypotheses, there is increasing evidence for a facilitated protein-dependent transport pathway for long chain fatty acids. Five putative fatty acid transport proteins have been identified in the plasma membrane of several tissues including adipocytes, liver, endothelium, brain, intestine, kidney, lung, skeletal, and cardiac muscle. For three of these cDNA clones have been isolated (reviewed in Ref. 15). Schaffer and Lodish (17) first succeeded not only in expression cloning and purification of one putative carrier protein but could also reconstitute transport function, thus directly demonstrating its physiological role.
The results reported herein meet several criteria for a protein-facilitated export mechanism for arachidonic acid in PMN plasma membrane vesicles, such as substrate saturation, inhibition by a variety of anion transport inhibitors (18), and sensitivity toward proteases. The ionic mechanism seems to be that of an H ϩ ϩ arachidonate cotransport system operating with an energy of activation of 11.6 kcal/mol. With the transport assay presented here, it may be possible to separate the transport processes out of the complex cascade of events starting after activation of PMN by various stimuli to study the mechanism(s) of putative eicosanoid transporters and to investigate the impact of leukotriene inhibitors on transport.
Pig Polymorphonuclear Leukocytes (PMN)-Arterial blood was collected from 3 to 4-month old pigs using sterile Na 2 EDTA (2.14 g/liter) as an anticoagulant. The whole blood was centrifuged at 200 ϫ g for 15 min, and the platelet-rich supernatant was discarded. PMN were prepared by dextran sedimentation, hypotonic lysis of remaining erythrocytes, and centrifugation on Ficoll-Paque, as described (19).
Preparation of Phagocytic Vesicles-Purified PMN (50 ϫ 10 6 /ml) were resuspended in modified Hank's buffer (124 mM NaCl, 4.9 mM KCl, 0.66 mM Na 2 HPO 4 , 0.64 mM KH 2 PO 4 , 15 mM NaHCO 3 , 10 mM Tris-HCl, pH 7.4) and incubated with a mineral oil emulsion, essentially as described (13). Briefly, 1 volume of mineral oil and 4 volumes of modified Hank's buffer containing 20 mg/ml fatty acid-free albumin were mixed and sonicated in a Branson Sonicator (Cell Disrupter B15), with a micro tip for 2 ϫ 1 min, setting 5. The oil emulsion was centrifuged for 5 min at 1600 ϫ g, and the thin layer of coacervated droplets at the surface was removed. The remaining stable emulsion was mixed with 1 volume of fresh pig serum. One volume of this oil emulsion was added to 10 volumes of PMN suspension at 4°C. Phagocytosis was initiated by the addition of 0.9 mM CaCl 2 , 1.27 mM MgCl 2 , and 10 ng/ml lipopolysaccharide and warmed to 37°C. After 30 min incubation at 37°C under gentle stirring, cells were separated from residual emulsion by centrifugation at 250 ϫ g for 5 min at 4°C and washed twice with Hank's buffer. Finally, the cells were suspended in modified relaxation buffer (20) (100 mM KCl, 3 mM NaCl, 3.5 mM MgCl 2 , 10 mM PIPES-HCl, pH 6.8), treated with 5 mM MgCl 2 , 1 mM 1,4-dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and 0.25 mg/ml soybean trypsin inhibitor and subjected to nitrogen cavitation (200 p.s.i. for 20 min). After addition of 1.25 mM EGTA, the suspension was centrifuged at 48,000 ϫ g for 30 min. Liberated phagocytic vesicles floating on the top of the tubes were collected with a spatula and resuspended in 250 mM sucrose, 10 mM Tris-HCl, pH 7.4. Inside-out plasma membrane vesicles were frozen in liquid nitrogen and stored at Ϫ20°C until use. Protein was determined using the bicinchoninic acid assay (21).
Purification of Phagocytic Vesicles by ConA Chromatography-ConA affinity chromatography was performed at 4°C as described by Del Buono et al. (22). ConA-Sepharose was added to polypropylene columns to give a packed bed volume of 3 ml, and columns were washed with degassed relaxation buffer with and without addition of 0.1% BSA. Samples of membrane vesicles or intact PMN in modified relaxation buffer were added to the columns in a volume of 1-3 ml and incubated for 10 min. Nonadherent vesicles were eluted with ice-cold modified relaxation buffer, and protein in the fractions was quantified by the bicinchoninic acid assay (21). About 95-98% of the phagocytic vesicles were found in the first 6 ml of the eluate, whereas intact PMN almost completely (Ͼ95%) adhered to the columns. Based on the assumption that ConA receptors are found exclusively on the extracellular surface of the plasma membrane of PMN (22), from these experiments a low contamination of the inside-out membrane vesicles by right side-out vesicles or plasma membrane fragments was concluded.
Electron Microscopy of Isolated Vesicles-Vesicles in modified relaxation buffer were fixed in 3% glutaraldehyde at 4°C overnight. Material was rinsed with relaxation buffer, followed by a 1.5-h fixation with osmium tetroxide in the same buffer. For en bloc staining, vesicles were incubated with 1% uranyl acetate in 20% acetone for 1.5 h, and dehydration was performed with a graded acetone series. Samples were then infiltrated and embedded in Spurr's low viscosity resin. After polymerization, sections were cut, mounted on colloidin-coated copper grids, and post-stained with aqueous lead citrate (3%, pH 13.0). All micrographs were taken on an EM 109 electron microscope (Zeiss, Oberkochen, Germany) Arachidonic Acid Uptake into Inside-out Plasma Membrane Vesicles-The standard incubation medium contained plasma membrane vesicles (20 g of protein) in modified relaxation buffer (100 mM KCl, 3 mM NaCl, 3.5 mM MgCl 2 , 10 mM PIPES-HCl, pH 6.8) in a final volume of 200 l. Tritiated arachidonic acid was added as a 50% ethanolic solution. The final ethanol concentrations never exceeded 0.5% and did not influence transport activity. Since a substantial portion of the added arachidonate bound to the walls of the container, the actual arachidonate concentration in each experiment was determined by scintillation counting of an aliquot of the incubations. All concentrations of arachidonate were corrected for adsorption losses to container walls. The incubations were carried out at 37°C and terminated by the addition of 800 l of ice-cold stop solution containing modified relaxation buffer, pH 6.8, 0.5 mM pentachlorophenol, and 0.5% albumin and then further incubated for 5 min on ice. The sample was pipetted onto a Schleicher & Schuell filter GF 52 (24 mm) and filtered using a filtration manifold (DN 025/1, Schleicher & Schuell, Dassel, Germany). Subsequently, vesicles were washed with 1 ml of stop solution and 3 ml of relaxation buffer. The filters were placed in scintillation vials and left overnight in 5 ml of Rotizint Eco Plus (Roth, Karlsruhe, Germany) before radioactivity was counted and corrected for adsorption of unbound arachidonate to the filters.

RESULTS
Vesicle Preparation-Vesicles prepared for electron microscopy consisted primarily of membrane-bound vesicles (0.1-1.2 m) and some associated granules. These isolated vesicles were identical to the phagocytic vesicles seen in intact cells after phagocytosis of albumin-coated mineral oil droplets. The vesicle membrane was clearly trilaminar and identical in appearance and dimensions to the plasma membrane. The membrane encloses the lipid droplet with an amorphous electron-dense periphery representing the albumin coat of the lipid droplet. There was essentially no contamination of the vesicles by mitochondria, glycogen particles, nuclei, or other cytoplasmic matter except for some granules associated with the vesicles, Fig. 1. Application of phagosomes to a ConA affinity column revealed that only a small fraction (Ͻ2-5%) of the vesicles was bound to the column, whereas intact PMN adhered to the affinity material (Ͼ95%), demonstrating that the vesicles had indeed the inside-out orientation with little contamination of plasma membranes with exposed ConA receptors.
Vesicle Association of Arachidonic Acid-Various concentrations of [ 3 H]arachidonic acid were added as an ethanolic solution to a vesicle suspension (20 g of protein in modified 200 l of modified relaxation buffer) and incubated at 37°C. A significant and constant portion of the added [ 3 H]arachidonate was bound to the walls of the incubation container. About 50% of the added [ 3 H]arachidonate was recovered when the incubation mixture from the Eppendorf tube was transferred to a scintillation vial and counted for radioactivity. From the residual arachidonate in the buffer (50%), a constant arachidonate portion (70.1%) associated with the vesicles in the concentration range tested (0.24 -9772 pmol/sample) and was recovered with the vesicle membranes after rapid filtration (Fig. 2). During a 30-min incubation, the vesicle-associated arachidonate concentrations did not change.
Adsorption of arachidonate to container walls is a common observation (23), which makes it necessary to correct arachido- The picture on the left shows isolated vesicles with some adhering granules. On the right, a single isolated vesicle shows the trilaminar membrane (arrows) enclosing the albumin-coated mineral oil droplet from which the lipid has been extracted during sample preparation.
nate concentrations for adsorption losses in every experiment. Vesicle association of unbound arachidonate is thought to represent absorption to the plasma membrane (24) and uptake into the intravesicular space. To measure uptake accurately and to distinguish internalized ligand from that merely bound to the external surface of the vesicles, it is essential to employ a wash solution that blocks influx and efflux after certain periods of incubation and removes ligand bound to the external membrane of the vesicles (25).
In previous studies, washes with a 0.1-0.5% albumin solution with or without addition of 200 M of the transport inhibitor phloretin were generally used to remove fatty acids, which are exchangeably bound to the outer membrane leaflet, while preventing efflux (25)(26)(27)(28)(29). We employed a 0.4% albumin solution containing 0.4 mM pentachlorophenol (PCP) for this purpose since PCP is a stronger influx/efflux inhibitor in our system.
To evaluate the effectiveness of this wash solution in incubations with vesicle preparations, vesicle aliquots (20 g of membrane protein) were incubated at 37°C with various amounts of [ 3 H]arachidonic acid at different time points, and vesicle association of radioactivity was measured after treatment with ice-cold BSA/PCP (stop/chase solution) as described under "Materials and Methods." The vesicle-associated arachidonate resistance toward BSA/PCP extraction increased linearly within the first 120 s of incubation at 37°C. The linear regression line of this time course was extrapolated to zero time to give the amount of arachidonate that could not be extracted by the BSA/PCP wash (determined for every arachidonate concentration tested). As much as 98.3% of the merely bound arachidonate at zero time could be extracted from the vesicles with BSA/PCP stop solution. A similar result was obtained when vesicles were incubated at 0°C instead of 37°C, even after prolonged incubation times.
The non-extractable arachidonate increased in a linear fashion with increasing amounts of arachidonate (0.24 -9772 pmol), Fig. 2. Thus, under the conditions employed, no saturable binding to putative receptors became apparent in the concentration range (1.22 nM to 48.9 M) tested.

Arachidonic Acid Influx into Inside-out Membrane Vesicles-
Vesicles were incubated with 20 nM [ 3 H]arachidonate (4 pmol/ 200 l) in modified relaxation buffer, pH 6.8 at 37°C, and reactions were terminated after various time points by treatment with ice-cold BSA/PCP stop solution as described under "Materials and Methods." During incubation, the non-extractable portion of vesicle-associated arachidonate increased from 68 fmol (1.7%) at zero time to approximately 1 pmol (58.8%) after 60 min (total arachidonate corrected for losses to the test tube represents 1.7 pmol ϭ 100%), Fig. 3. Increase of nonextractable arachidonate associated with the membrane vesicles could either reflect a time-dependent tight binding to high affinity binding sites at or in the membrane vesicles or influx to an as yet undefined compartment within the vesicles that is not accessible to BSA/PCP extraction. To distinguish between these possibilities, incubation of vesicles at 37°C was terminated after 150 min by addition of ice-cold BSA/PCP stop solution and subjected to a 3 ϫ 10-s sonication treatment on ice and filtered. Almost complete loss of vesicle-associated arachidonate that had been accumulated during incubation at 37°C was observed after physical destruction of the vesicles by sonication, Fig. 3. To check whether the loss of radioactivity was due to the loss of sonicated membranes, aliquots of the same incubation mixture were subjected to sonication and filtration in the same way as before, except that the BSA/PCP solution was replaced by albumin-free modified relaxation buffer. These control experiments revealed that the recovery of vesicle membranes was not influenced by the sonication procedure, suggesting that the loss of radioactivity was due to liberation of arachidonic acid from the intravesicular space. Thus, the timedependent vesicle association of non-extractable arachidonate is consistent with influx but not with binding to high affinity binding sites on the vesicle membranes. to aliquots of the suspension after 0 or 75 s. Addition of 0.4 mM PCP prevented influx of arachidonate. Furthermore, internalized arachidonate concentrations at the time of PCP addition remained almost at the same level over the entire incubation period of 1000 s, demonstrating an efficient and long lasting influx inhibition effect of PCP, Fig. 4.
Since PCP did not only block influx into vesicles over prolonged incubation times at 37°C, but also prevented losses of internalized arachidonic acid during the incubation, we tested the hypothesis whether PCP could also prevent vesicle efflux. Incubation of vesicles (20 g of protein/200 l) with 4 pmol of arachidonic acid resulted in a time-dependent increase of internalized arachidonate. Addition of BSA (0.4% final concentration) stopped the accumulation of arachidonate in the vesicles, and a time-dependent loss of intravesicular arachidonate was observed, reaching close to zero time levels after 50 min. In contrast, addition of BSA/PCP (0.4 mM PCP, 0.4% BSA) only resulted in a very little loss of internalized arachidonate.
Inhibition of vesicle association as well as vesicle dissociation of the non-BSA/PCP extractable arachidonate cannot be explained on the basis of high affinity binding to putative receptors. Furthermore, inhibition of influx and efflux of arachidonate by the anion transport inhibitor PCP suggested a nondiffusional uptake and release of arachidonate.
Effect of Anion Transport Inhibitors on Arachidonate Influx-To study the effect of common anion transport inhibitors on arachidonate transport, a variety of structurally different inhibitors were preincubated with vesicles in modified relax-ation buffer for 15 min before 4 pmol/200 l of [ 3 H]arachidonic acid were added. To determine internalized arachidonate after various incubation periods, 200-l aliquots of the suspension (20 g protein) were pipetted at desired time points into 0.8 ml of cold 0.5% BSA, 0.5 mM PCP and filtered after 5 min extraction time. Incubations were performed in triplicate, and initial influx rates were computed from the slope of the linear regression line over the first 120 s after arachidonate addition. Strong inhibition of the initial influx was observed by 1 mM quercetin (11.5 Ϯ 1.5% of control, mean Ϯ S.D.), 0.5 mM phloretin (9 Ϯ 5.6%), 0.5 mM diiodosalicylic acid (10.3 Ϯ 6.8%), and 0.5 mM pentachlorophenol (1.3 Ϯ 0%), respectively (Fig. 5). N-Ethylmaleimide had little, if any, effect on arachidonate influx. However, three other compounds that inhibit monocarboxylate (18) or prostaglandin transport (30), 4-acetamido-4Ј-isothiocyanostilbene-2,2Ј-disulfonic acid, ␣-cyano-4-hydroxycinnamate, and the dye bromcresol green, showed no effect on arachidonate influx. The effectiveness of common anion transport inhibitors is suggestive of non-diffusive uptake of arachidonic acid into the vesicles.
Evidence for a Protein-facilitated Transport of Arachidonate-Proteolytic digestion of vesicles is often employed as a key experiment to decide whether or not proteins are involved in transport processes. Since proteins facilitating transmembrane movement of ligands are often partially exposed on the membrane surface, they can be digested by proteases without damaging the membrane barrier (31). Vesicles were preincubated in modified relaxation buffer with or without addition of Pronase (1 mg/ml) or trypsin (2.5 mg/ml) overnight. Preincubation with either Pronase or trypsin at 37°C strongly inhibited the initial arachidonate influx into the vesicles. Pronase and trypsin digestion reduced initial influx rates to 12.7 Ϯ 4.6 and 5.6 Ϯ 2.1% of control (mean Ϯ S.D., n ϭ 3), respectively. Preincubation at 0°C had no effect on the influx rates, suggesting that influx inhibition is indeed mediated by the enzymatic properties of the proteases and not by unspecific protein binding as seen with albumin.
Kinetics of Arachidonate Influx-Inhibition of arachidonate influx by proteases suggested that at least one step in the mechanism of arachidonate influx is protein-dependent. Thus, arachidonic acid influx should be saturable by arachidonate. Initial arachidonic acid influx rates were determined as described above after addition of various amounts of arachidonate (final concentration 1.2 nM to 86.5 M, corrected for adsorption losses to the container). At low ligand concentrations, the initial influx rates increased in a linear fashion with increasing arachidonic acid concentrations and appeared to saturate at higher arachidonate concentration (Fig. 6). The data were fitted against the Michaelis-Menten equation by nonlinear regression analysis. At a vesicle protein concentration of 100 g/ml, a half-maximal influx rate was calculated for an arachidonate concentration of 5.7 Ϯ 1.9 M, and a V max of 3.0 Ϯ 0.2 nmol/min ϫ mg was obtained. The data were replotted according to Lineweaver-Burk giving a linear regression line (Fig. 6,  inset).
Temperature Dependence of Arachidonate Influx-To determine whether arachidonate influx is a temperature-dependent process, vesicles were incubated at temperatures between 0 and 71°C, and initial influx rates were determined over the first 120 s as described above. Arachidonate influx demonstrated a marked temperature dependence, Fig. 7. At 0°C influx was almost completely abrogated (0.22 pmol/min ϫ mg), and an optimum transport rate was measured at 51°C (7.2 pmol/min ϫ mg), Fig. 7, inset. The temperature dependence of the initial influx rates (0 -47°C) was tested according to Arrhenius where ln V max is plotted against 1/T as shown in Fig. 7, inset. A linear relationship was found (r ϭ Ϫ0.9890), and from the slope of the Arrhenius plot an activation energy of 11.6 Ϯ 0.3 kcal/mol (mean Ϯ S.D., n ϭ 3) could be calculated.
Essentially no influx could be observed at 71°C (0 pmol/ min ϫ mg) and transport competence could not be restored by lowering the temperature back to 37°C. Inactivation of transport function might be due to denaturation of a putative carrier protein or physical damage of the vesicles. Electron micrographic pictures taken from the vesicles before and after a 10-min incubation at 70°C showed that the vesicles were indeed heavily damaged by heat treatment. In fact, it was not possible to find any intact vesicle in representative pictures of more than 100 vesicles (data not shown).
Effect of pH on Arachidonic Acid Influx-To study the effect of pH on initial arachidonic acid influx, the extravesicular pH was varied from pH 9.8 to 3. At an alkaline pH higher than 8 almost no influx could be measured (0.1 pmol/mg ϫ min at pH 9.8) and was indistinguishable from similar incubations where PCP had been added. Lowering the pH to 6.0 increased the initial influx (12.7 pmol/mg ϫ min) which could completely be inhibited by PCP, Fig. 8. At pH values below 6, initial influx is again suppressed, and influx at pH 3 is indistinguishable from that observed in the presence of PCP (0.45 pmol/mg ϫ min). The enhancement of arachidonate influx by acidification and suppression by alkalinization in the pH range from 6 to 9 is compatible with an H ϩ ϩ arachidonate cotransport system but may also be due to allosteric effects of pH on the hypothetical carrier protein. Electron microscopic examination of vesicle integrity at low pH values revealed that vesicles preincubated at pH 4.5 were almost indistinguishable from those incubated at neutral pH, whereas at pH 3.5 about 20% of the vesicles appeared to be damaged. Thus, decrease of transport activity at low pH values appears not to be predominantly due to vesicle destruction. Notably, PCP inhibited influx almost completely over the whole pH range tested, further supporting the hypotheses that arachidonic acid influx is non-diffusive in nature even at pH values below 6 where arachidonic acid should be in a non-ionic state. A pH-induced change in the binding of arachidonate to putative receptors appears unlikely since sonication of vesicles just prior to filtration liberates almost all of the arachidonate (Fig. 2) without affecting membrane recovery.
Effect of Extravesicular ATP, Na ϩ and K ϩ Concentration on Arachidonic Acid Influx-We next investigated the influence of extravesicular Na ϩ and K ϩ concentrations. The relaxation buffer was replaced in different experiments by buffers (10 mM HEPES-KOH, pH 6.8) containing 0 -200 mM NaCl or buffers (10 mM HEPES-NaOH, pH 6.8) containing 0 -200 mM KCl. In a series of experiments, modified relaxation buffer containing various ATP concentrations (0.5-10 mM) or an ATP-regenerating system was used as external medium. All these replacements had no impact on the initial arachidonate influx or efflux rates (data not shown).

DISCUSSION
Fatty acid trafficking through membranes is very often thought to be purely by passive diffusion driven by fatty acid gradients (16). In contrast, our findings on arachidonate export support the view (15, 28, 31) that transmembrane fatty acid movement is a protein-dependent and saturable process that cannot be explained by free diffusion. Although a protein that functions as long chain fatty acid transporter has recently been cloned, the translocation mechanism still remains to be elucidated. Three models for the protein-facilitated transport mechanism of hydrophobic fatty acids across the phospholipid bilayer that are compatible with the data collected so far have been proposed recently (32,33). First, the transport protein may form a pore through which fatty acids pass, similar to the function of classical facilitated diffusion transporters, or the transport protein may bind specifically fatty acids from the aqueous phase to its external surface. The fatty acid moves along the surface of the protein until it reaches the interface of the protein and the phospholipid bilayer. The transport protein catalyzes a "flip" of the fatty acid from the outer to the inner leaflet of the bilayer, and the fatty acid is then released into the aqueous phase; or the transport protein may bind the fatty acid already associated with one leaflet of the plasma membrane and catalyze a "flip flop" of the fatty acid in the membrane.
The studies presented herein indicate that pig neutrophils possess a transport mechanism for arachidonic acid that facilitates the efflux of arachidonic acid into the extracellular medium. In the experimental model used, arachidonic acid associated rapidly with the outer leaflet of the inverted plasma membrane vesicles from where it was translocated by a protease-sensitive, time-dependent, and saturable process to the vesicle interior. Arachidonate associated with the external membrane leaflet is extracted by extravesicular albumin, whereas internalized arachidonate is inaccessible. Thus, albumin extraction in the presence of an appropriate efflux inhibitor can be used to distinguish internalized arachidonate from that associated with the external membrane leaflet. In PMN, arachidonic acid once translocated to the outer leaflet of the plasma membrane may easily partition into the plasma and bind to plasma albumin. Because of its high binding affinity for fatty acids with a K d of about 15 nM (34), albumin has been found to extract fatty acids from membranes (24).
In activated neutrophils, activation of various phospholipases (PLA) (35) leads to increased intracellular levels of free arachidonate, some of which is used for leukotriene biosynthesis and some of which is released into the extracellular medium (3,36,37). However, the sources of arachidonate released into the medium as well as the lipase(s) involved in this process are still a matter of controversy. Studies using activated mast cells and macrophages support evidence for the possibility of different pools of arachidonate generated by different lipases (38,39). Thus in macrophages, most of the arachidonate released into the medium is derived from a secreted type II PLA 2 (sPLA 2 ) acting on the outer surface of the cell, while a group IV PLA 2 (cPLA 2 ) acting intracellularly is coupled to eicosanoid biosynthesis and is only responsible for a portion of the arachidonate released into the medium. In PMN, the situation may be similar, although direct evidence for secretion of sPLA 2 activity has been elusive (40). Alternatively, activated neutrophils may also generate arachidonate by the concerted action of phospholipase C or D (41) and a lipase located at the inner leaflet of the plasma membrane.
Regardless of the source, unesterified arachidonate dissolved in plasma membranes has been shown to influence a variety of membrane properties in different tissues including K ϩ channels (42,43), Cl Ϫ channels (44), proton pumping (45), Ca 2ϩ permeability (46), and membrane ATPase (47). Therefore, it may be important that unesterified arachidonate in the inner leaflet of the plasma membrane is removed by an efficient export mechanism to clear the membrane from arachidonic acid to prevent disturbance of normal membrane function. Thus, export might be an additional mechanism apart from the rapid reacylation in the cytosol to keep the arachidonate level well below 3 pmol/million PMN (48) and may contribute to the release of arachidonate from PMN during cell activation.
Differential sensitivity toward anion transport inhibitors (18,30,49) makes it possible to discriminate between different transport pathways. So far, three different Cl Ϫ channels and four different anion transport activities have been identified in neutrophils as follows: a Cl Ϫ /HCO 3 Ϫ exchanger (50), a sulfate carrier (51), a H ϩ ϩ lactate-carrier (52), and a H ϩ ϩ lipoxin A 4 cotransporter (49). Lactic acid fluxes are completely blocked by N-ethylmaleimide, and the sulfate carrier is very sensitive to 4-acetamido-4Ј-isothiocyanostilbene-2,2Ј-disulfonic acid, whereas arachidonate export is sensitive to neither of these inhibitors at roughly comparable concentrations. The Cl Ϫ / HCO 3 Ϫ exchanger displays a pH dependence opposite to that for arachidonate transport. In contrast, many of the kinetic characteristics and the inhibitor profile for arachidonate export are shared by the H ϩ ϩ LXA 4 cotransporter that imports the eicosanoid LXA 4 into PMN. Similar to LXA 4 uptake into PMN, arachidonate influx into phagosomes displays a striking pH dependence with almost no transport activity at pH 8 and highest activity at pH 6. ATP, Na ϩ , or K ϩ dependence was not noted in either case. With respect to the inhibitors, both transporters are sensitive toward PCP, 3,5-diiodosalicylic acid, but not to N-ethylmaleimide, iodoacetate, and 4-acetamido-4Јisothiocyanostilbene-2,2Ј-disulfonic acid. The energy of activation for both transporters was relatively low, 15.0 kcal/mol for the trihydroxylated eicosanoid LXA 4 and 11.6 kcal/mol for arachidonic acid. Substrate saturation was not seen with the LXA 4 transporter but was evident in the case of arachidonate transport at relatively high concentrations. It has not yet been clarified if arachidonate influx and efflux in PMN is mediated by the same protein, facilitating the electroneutral diffusion of arachidonate ϩ H ϩ species through the membrane similar to the lactate ϩ H ϩ cotransporter in PMN (52) and erythrocytes (18). Alternatively, two independent transport systems may exist. The question of which other eicosanoids and fatty acids are accepted by the arachidonate transporter has not been addressed in this investigation and remains to be elucidated in the future.