Mechanism of human group V phospholipase A2 (PLA2)-induced leukotriene biosynthesis in human neutrophils. A potential role of heparan sulfate binding in PLA2 internalization and degradation.

Human group V phospholipase A(2) (hVPLA(2)) has been shown to have high activity to elicit leukotriene production in human neutrophils (Han, S. K., Kim, K. P., Koduri, R., Bittova, L., Munoz, N. M., Leff, A. R., Wilton, D. C., Gelb, M. H., and Cho, W. (1999) J. Biol. Chem. 274, 11881-11888). To determine the mechanism by which hVPLA(2) interacts with cell membranes to induce leukotriene formation, we mutated surface cationic residues and a catalytic residue of hVPLA(2) and measured the interactions of mutants with model membranes, immobilized heparin, and human neutrophils. These studies showed that cationic residues, Lys(7), Lys(11), and Arg(34), constitute a part of the interfacial binding surface of hVPLA(2), which accounts for its moderate preference for anionic membranes. Additionally, hVPLA(2) binds heparin with high affinity and has a well defined heparin-binding site. The site is composed of Arg(100), Lys(101), Lys(107), Arg(108), and Arg(111), and is spatially distinct from its interfacial binding surface. Importantly, the activities of the mutants to hydrolyze cell membrane phospholipids and induce leukotriene biosynthesis, when enzymes were added exogenously to neutrophils, correlated with their activities on phosphatidylcholine membranes but not with their affinities for anionic membranes and heparin. These results indicate that hVPLA(2) acts directly on the outer plasma membranes of neutrophils to release fatty acids and lysophospholipids. Further studies suggest that products of hVPLA(2) hydrolysis trigger the cellular leukotriene production by activating cellular enzymes involved in leukotriene formation. Finally, the temporal and spatial resolution of exogenously added hVPLA(2) and mutants suggests that binding to cell surface heparan sulfate proteoglycans is important for the internalization and clearance of cell surface-bound hVPLA(2).

brane phospholipids, the products of which can be transformed into potent inflammatory lipid mediators, platelet-activating factor, and eicosanoids that include prostaglandins, thromboxanes, leukotrienes, and lipoxins. Multiple forms of PLA 2 s have been found in mammalian tissues (1) , including several forms (groups Ib, IIa, IIc, IId, IIe, IIf, V, and X) of secretory PLA 2 s (sPLA 2 ) and intracellular PLA 2 s (e.g. groups IV and VI). Recent cell studies indicated that sPLA 2 works in concert with group IV cytosolic PLA 2 to induce immediate and delayed eicosanoid formation (2,3). At present, however, the identity of sPLA 2 involved in eicosanoid biosynthesis, the temporal and spatial sequences of its mobilization during inflammatory cell activation, and the mechanism by which it interacts with cell membranes to induce cellular eicosanoid formation are not fully understood. Recently, mounting evidence has pointed to the involvement of group V PLA 2 in eicosanoid formation in various mammalian cells (4 -8) but the mechanisms of its eicosanoid-inducing activities have not been elucidated. In particular, the mechanism whereby group V PLA 2 interacts with cell membranes remains unclear. There are at least three possible mechanisms for interaction of sPLA 2 with cells (see Fig. 1). In the simplest mechanism, sPLA 2 would act directly on membrane phospholipids to release fatty acids and lysophospholipids, which in turn induce eicosanoid release. However, many mammalian sPLA 2 s, most notably group Ib (9) and IIa PLA 2 s (10), strongly prefer anionic membranes to zwitterionic ones, and consequently have low affinity for and activity on intact mammalian cells, the outer plasma membranes of which are mainly composed of zwitterionic phosphatidylcholine (PC) and sphingomyelin. For this reason, indirect mechanisms, involving anionic cell surface heparan sulfate proteoglycans (HSPG) (3,(11)(12)(13)(14) and protein receptors (15), have been proposed (see Fig. 1). However, direct involvement of HSPG and receptors in the actions of sPLA 2 on cells remains controversial (16). To elucidate the origin of high activity of group V PLA 2 to release fatty acids and eicosanoids from mammalian cells, we fully characterized the enzymatic and membrane-binding properties of recombinant human group V PLA 2 (hVPLA 2 ) (17,18). These studies showed that hVPLA 2 could bind and hydrolyze PC membranes and the outer plasma membranes of mammalian cells much more efficiently than other sPLA 2 s, including human group IIa PLA 2 (hIIaPLA 2 ). Additionally, a structurefunction analysis of hVPLA 2 revealed good correlation between its affinity for PC membranes and its activity on mammalian cells, suggesting that hVPLA 2 could directly bind to the outer plasma membranes of mammalian cells to hydrolyze phospholipids and elicit eicosanoid formation (18). However, it was also reported that rat group V PLA 2 binds to human embryonic kidney (HEK) 293 cells via HSPG (3). Indeed, group V PLA 2 s, including hVPLA 2 , contain a cluster of cationic residues in the C-terminal region (see Fig. 2), which is involved in HSPG binding for group IIa PLA 2 s (16). In addition, hVPLA 2 has a few cationic residues on its putative interfacial binding surface and modestly (ϳ4-fold) prefers anionic membranes to zwitterionic membranes (18). These findings suggested that the cell binding of group V PLA 2 might involve interactions with anionic surfaces, including cell surface HSPG. To determine the exact mechanism by which hVPLA 2 interacts with mammalian cells and elucidate the potential involvement of HSPG binding in its cell activities, we mutated surface cationic residues and a catalytic residue, and measured the interactions of mutants with model membranes, immobilized heparin, and human neutrophils. Results provide new insights into the temporal and spatial sequences of events involved in hVPLA 2 -mediated cellular eicosanoid biosynthesis.
Mutagenesis, Expression, and Purification of hVPLA 2 -Mutagenesis was performed using the Sculptor in vitro mutagenesis kit from Amer-sham Pharmacia Biotech and a phagemid DNA prepared from the pSK vector in the presence of helper phage R408 as described previously (22). Wild type and mutant proteins were expressed in Escherichia coli, refolded, and purified as described previously (17). Purified proteins were lyophilized and stored at Ϫ20°C. The purity of wild type and mutant proteins, assessed by sodium dodecyl sulfate (SDS)-polyacrylamide electrophoresis, was consistently higher than 90%. Protein concentration was determined by the bicinchoninic acid method (Pierce) using BSA as standard.
Kinetic Measurements-PLA 2 -catalyzed hydrolysis of polymerized mixed liposomes was carried out at 37°C in 2 ml of 10 mM HEPES buffer, pH 7.4, containing 0.1 M pyrene-PE (1 mol %) inserted in 9.9 M BLPG, 2 M BSA, 0.16 M NaCl, and 10 mM CaCl 2 (19,20). The progress of hydrolysis was monitored as an increase in fluorescence emission at 378 nm using a Hitachi F4500 fluorescence spectrometer with the excitation wavelength set at 345 nm. Spectral bandwidth was set at 5 nm for both excitation and emission. Values of k cat */K m * were determined from reaction progress curves as described previously (10). The PLA 2 -catalyzed hydrolysis of diC 6 PE monomers was performed with 0.5 mM phospholipid, 0.16 M NaCl, and 10 mM CaCl 2 . The time course of phospholipid hydrolysis was monitored with a computer-controlled Metrom pH-stat (Brinkmann) in a thermostated vessel. Under these conditions, the hydrolysis of diC 6 PE followed the first-order kinetics because the substrate concentrations remained lower than apparent K m . Thus, the values of apparent second-order rate constant, (k cat / K m ) app , were calculated by dividing by the enzyme concentration the pseudo-first-order rate constants determined from the nonlinear leastsquares analysis of reaction progress curves. Activity of PLA 2 on zwitterionic vesicles was assayed by measuring the initial rate of [ 14 C]SAPC hydrolysis. Typically, 20 l of [ 14 C]SAPC solution in chloroform was dried in a glass vial with N 2 and hydrated in 800 l of 10 mM HEPES buffer, pH 8.0, containing 0.16 M NaCl and 10 mM CaCl 2 . After vortexing, the lipid suspension was sonicated for about 10 s, frozen in ethanoldry ice bath, and sonicated again for ϳ1 min to afford a homogenous small unilamellar vesicle solution. For assay, 50-l aliquots of [ 14 C]SAPC vesicles (ϳ10 M) solutions containing 15 M BSA were placed in Eppendorf tubes. Reactions were started by addition of enzyme to a final concentration of 10 -20 nM and quenched by adding 370 l of ice-cold chloroform/methanol/HCl (2:1:0.01, v/v/v) solution after a given period of incubation at room temperature. Subsequently, 460 l of chloroform/methanol/H 2 O (1:2:0.8, v/v/v), 240 l of chloroform, and 240 l of H 2 O were added and each reaction mixture was vortexed to extract the lipids into the organic phase. Liberated [ 14 C]AA was separated from the reaction mixtures on small silica gel columns using petroleum ether/ether/acetic acid (70:30:1, v/v/v) as an eluent. Solvents collected in scintillation vials were evaporated with N 2 and replaced by 4 ml of scintillation mixture (Sigma), and the radioactivity was measured by liquid scintillation counting.
Binding of PLA 2 to Phospholipid-coated Beads and Vesicles-The binding affinity of PLA 2 for sucrose-loaded polymerized BLPG liposomes was determined at 25°C as described previously (10). The binding assay solution contained 100 M phospholipid and varying concentrations of PLA 2 in 10 mM Tris-HCl buffer, pH 7.4, containing 0.145 M NaCl, 10 mM CaCl 2 , and 1 M BSA. For the determination of PC membrane binding affinity, PC-coated styrene-divinylbenzene beads, which can be rapidly and completely separated from the solution by low speed centrifugation, were used instead of PC vesicles that typically have low pelleting efficiency (23). Binding measurements with DHPCcoated beads were performed as described previously (24). The binding assay solution contained 100 -150 M phospholipid coated on beads and varying concentrations of PLA 2 in 10 mM Tris-HCl buffer, pH 7.4, containing 0.16 M NaCl, 10 mM CaCl 2 , and 1 M BSA. For both binding measurements, the concentration of free enzyme was determined by PLA 2 activity assays using either pyrene-PE/BLPG polymerized mixed liposomes or [ 14 C]SAPC/1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoglycerol/1,2-sn-dioleoylglycerol (10:9:1) vesicles as substrate, from which FIG. 1. Three hypothetical models of hVPLA 2 -cell surface interactions. hVPLA 2 can bind either cell surface heparan sulfate proteoglycans that function as a cell surface adapter for hVPLA 2 or a cell surface receptor specific for hVPLA 2 . Either binding can trigger intracellular signaling cascade or lead to internalization of bound proteins. Heparan sulfate binding can also enhance the apparent activity of hVPLA 2 on membrane phospholipids by increasing its local concentration. Finally, hVPLA 2 can directly bind the membrane surface and hydrolyze phospholipids. Notice that the plasma membranes of mammalian cells have asymmetric distribution of phospholipids, with the outer leaflet rich in sphingomyelin and phosphatidylcholine.
[PL] o, and [E] o are total phospholipid and enzyme concentrations, respectively. This equation assumes that each enzyme binds independently to a site on the interface composed of n phospholipids with dissociation constant of K d .
Heparin Binding-Heparin-binding affinity of W79A and other mutants was measured using a HiTrap heparin-Sepharose column (Amersham Pharmacia Biotech) connected to an Ä kta Fast Protein Liquid Chromatography system (Amersham Pharmacia Biotech). Typically, ϳ40 g of hVPLA 2 dissolved in 20 mM Tris-HCl, pH 7.4, was loaded onto an 1-ml column that was equilibrated with the same buffer solution at 4°C. The column was then eluted at 1 ml/min with a linear gradient of 0 -0.7 M KCl in the same buffer. The enzyme fractions were monitored by a PLA 2 activity assay using [ 14 C]SAPC/1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoglycerol/1,2-sn-dioleoylglycerol (10:9:1) vesicles as substrate as described above. The concentration of KCl that gives rise to a protein peak was determined for each protein as an average of duplicate measurements.
Preparation of Neutrophils-Human neutrophils were prepared from heparinized venous blood collected from medication-free donors according to the method by Hansel et al. (25) with modifications. Briefly, 30 ml of heparinized blood was diluted with the equal volume of calcium-free Hanks' balanced salt solution (HBSS), layered over 15 ml of 1.089 g/ml Percoll and centrifuged for 20 min at 900 ϫ g. The supernatant and the mononuclear cells at the interface were aspirated carefully, and the inside wall of the tube was wiped with sterile gauze to remove mononuclear cells attached to the wall. To the pellet of neutrophils and erythrocytes was added 20 ml of ice-cold water and the suspension was mixed gently for 30 s, after which 20 ml of 2ϫ HBSS was added. If the erythrocytes remained, then the procedure was repeated. After the lysis of erythrocytes, neutrophils were washed once in HBSS plus 0.2% BSA, and the total cell count was determined with a Coulter counter. The resulting cell population consisted of Ͼ95% neutrophils, as estimated by differential counts of Wright-Giemsa-stained cytospin preparations.
Fatty Acid Release and Eicosanoid Production from Neutrophils-Radiolabeling of neutrophils was achieved by incubating the cells (10 7 ) with 0.5 Ci/ml [ 3 H]AA for 90 min at 37°C. Unincorporated [ 3 H]AA was removed by washing the cells with HBSS plus 0.2% BSA. Double labeling of neutrophils with [ 3 H]AA (0.5 Ci/ml) and [ 14 C]OA (0.5 Ci/ml) was performed in the same manner. The labeled cells were resuspended in HBSS plus 0.2% BSA (10 6 cells/90 l) and incubated with 0.1 M PLA 2 at 37°C for a given period (30 min for routine assays). The reaction was quenched by centrifugation, and the supernatants were removed and their radioactivity ( 3 H and 14 C) was measured by twochannel liquid scintillation counting. For the measurements of leukotriene production from neutrophils, cells (1 ϫ 10 6 cells/ml) were incubated at 37°C in 250 l of HBSS containing CaCl 2 (1.2 mM) and 0.1 M PLA 2 . Control cells were treated with HBSS. Thereafter cells were centrifuged at 8000 g for 2 min. Leukotriene B 4 (LTB 4 ) release was determined using an enzyme immunoassay kit from Cayman Chemical Co. (Ann Arbor, MI). Typically, LTB 4 secretion reached a maximal value within 20 min of incubation under our experimental conditions. The maximal value for each incubation mixture was then corrected for a background signal from control cells.
Western Blotting Analysis of hVPLA 2 -treated Human Neutrophils-Neutrophils (2 ϫ 10 6 cells per experiment) were treated with 0.3 M hVPLA 2 for various periods, and the incubation was quenched by centrifugation at 12,000 ϫ g for 10 s. After repeatedly (more than three) washing, the pellets with HBSS containing 0.5 M NaCl, the pellet was then lysed in 70 l of lysis buffer (20 mM Tris-HCl, 30 mM Na 4 P 2 O 7 , 50 mM NaF, 40 mM NaCl, 5 mM EDTA, pH 7.4) containing 1% Nonidet P-40, 10 g/ml leupeptin, 5 g/ml aprotinin, 1 mM phenylmethylsulfonyl fluoride, 2 mM Na 3 VO 4 , and 0.5% deoxycholic acid. After 10 min on ice, the cell lysates were centrifuged at 12,000 ϫ g for 20 min to remove cell debris. The supernatants were then mixed with 14 l of gel loading buffer (0.125 M Tris-HCl, pH 6.8, 20% (v/v) glycerol, 4% SDS, 0.005% bromphenol blue), and the mixtures were boiled for 5 min. The samples were subjected to SDS-polyacrylamide gel electrophoresis under reducing condition, using 16% acrylamide gels. The electrotransfer of proteins from the gels to polyvinylidene fluoride membrane was achieved using a semidry system (200 mA, 120 min). The membrane was blocked with 2% BSA for 60 min, then incubated with 2 g/ml of anti-hVPLA 2 monoclonal antibody 3G1 (26) diluted in Tris-buffered saline plus 0.05% Tween 20 (TBS-T) overnight. The membranes were washed three times for 20 min with TBS-T. Goat anti-mouse IgG conjugated with horseradish peroxidase was diluted 1:3000 in TBS-T and incubated with polyvinylidene fluoride membrane for 60 min. The membrane was again washed three times with TBS-T and assayed with an ECL chemiluminescence system (Amersham Pharmacia Biotech).
Confocal Microscopy Imaging-Neutrophils were re-suspended in HBSS at a density of 5 ϫ 10 5 cells/ml. 300 l of neutrophils in HBSS were placed into each of eight wells on a sterile Nunc chambered cover glass and incubated at 37°C with 5% CO 2 for 45 min. After incubation, the chambers were washed twice with 37°C HBSS to remove any nonadherent neutrophils. W79A and W79A/R100/101A were diluted to 0.3 M (final concentration) in HBSS supplemented with 1.3 mM CaCl 2 , and overlaid onto the neutrophils in the appropriate wells. Neutrophils were incubated with enzyme for 10, 20, 30, and 40 min in a 37°C, 5% CO 2 , humidified incubator. At the specified time, the enzyme was removed, the cells were washed once with HBSS containing 1.3 mM CaCl 2 , and then were fixed at room temperature with 0.4% p-benzoquinone in phosphate-buffered saline (PBS) for 10 min. After fixation, the neutrophils were washed four times with PBS and were placed in blocking solution (10% normal goat serum and 100 M goat IgG/ml in PBS) at 4°C overnight. Then, the monoclonal antibodies to hVPLA 2 (26) were diluted to 2 g/ml in PBS and applied to the cells. After 1-h incubation at room temperature, the antibodies were removed, and the cells were washed six times in PBS. Secondary antibody, Alexa568 goat anti-mouse (Molecular Probes), was applied for 1/2 h at room temperature. Neutrophils were washed six times with PBS, and imaged in PBS immediately. Imaging was done with a Zeiss 510 laser scanning confocal microscope with the detector gain adjusted to eliminate background autofluorescence contributions. 2 Mutants-hVPLA 2 is a basic protein (isoelectric point Ͼ 9.0) with eight lysines and seven arginines. A model structure of hV-PLA 2 built on the basis of homology to hIIaPLA 2 is illustrated in Fig. 3. The structure suggests that some cationic residues, including Lys 6 , Lys 11 , Arg 34 , Lys 74 , and Arg 76 , might be located on its putative interfacial binding surface, whereas others, including Arg 100 , Lys 101 , Lys 107 , Arg 108 , and Arg 111 , might form a cationic patch on the opposite face. To determine the roles of these residues in the interaction of hVPLA 2 with cell membranes, we mutated these cationic residues to glutamate. We also mutated a catalytic residue, His 48 , to Ala to generate a catalytically inactive mutant. Initial attempts to prepare these mutants were hampered by extremely low refolding efficiency of solubilized inclusion bodies. To overcome this difficulty, we used W79A of hVPLA 2 , which was shown to be as active as wild type but much more stable than wild type (18), as a template for mutant preparation. All these mutants (e.g. K6E/W79A) were expressed in high yields as inclusion bodies and their refolding yields were uniformly high (i.e. Ͼ2 mg/liter of culture after purification).

Expression and Kinetic Characterization of hVPLA
To determine the effects of mutations on the enzymatic activities of hVPLA 2 , we measured the activities of mutants on three types of phospholipid substrates, a short-chain phospholipid diC 6 PE, anionic polymerized mixed liposomes, and zwitterionic [ 14 C]SAPC vesicles. Since the concentration of diC 6 PE used in this study (0.5 mM) is well below the critical micelle concentration for this short-chain phospholipid (27), diC 6 PE would exist as a soluble monomer. Although one cannot preclude the possibility of enzyme-substrate microaggregate formation (28), the relative activity of mutants determined under these conditions should thus reflect the relative catalytic efficiency of their active sites. As summarized in Table I, a majority of mutants showed activities comparable to that of W79A (and wild type hVPLA 2 ), which indicated the intactness of their active sites. However, three mutants, R34E/W79A, K74E/ W79A, and W79A/K107E/R108E, exhibited less than 35% of wild type activity, suggesting that these mutations somehow disrupt catalytic steps of hVPLA 2 . Since none of these residues are in close proximity to the active site in the model structure, their deactivating effects might derive from indirect steric or electrostatic interactions with the active site residues, as seen with the mutations of other sPLA 2 s (10, 29, 30). We then measured the activities of W79A and mutants on anionic pyrene-PE/BLPG polymerized mixed liposomes to estimate the effects of mutations on binding to anionic membranes. If any mutation has a significant effect on binding to anionic vesicles, it would reduce the activity on pyrene-PE/BLPG polymerized mixed liposomes to a larger degree than that on diC 6 PE. As summarized in the third column of Table I, the relative activity of mutants on polymerized mixed liposomes was, in general, comparable to that on the monomeric diC 6 PE. This indicated that none of mutated cationic residues is critical for the binding of hVPLA 2 to anionic membranes and the decreased activities of some mutants derive largely from their reduced catalytic efficiencies. To estimate the effects of mutations on the binding affinity of hVPLA 2 for zwitterionic membranes, we then measured the activities of mutants on SAPC vesicles. We previously showed that unlike other mammalian sPLA 2 s, hVPLA 2 has relatively high activity on zwitterionic vesicles and does not show a significant lag before the initiation of hydrolysis (17,18). W79A and mutants also showed linear progress curves up to 10 min with SAPC vesicles under our assay conditions (data not shown) and, thus, the specific activities were determined from the slopes of curves after 10 min of incubation. As summarized in the fourth column of Table I, the relative activity of most of mutants on SAPC vesicles was comparable to those on other substrates, again indicating that reduced activities of these mutants derive mainly from their low catalytic efficiencies. A notable exception was K6E/W79A that had much lower activity on SAPC vesicles than expected from its activity on diC 6 PE, suggesting that the K6E mutation might selectively impair the binding of enzyme either to zwitterionic membranes or to a PC substrate bound to the active site. Also, W79A/ R100E/K101E showed modestly higher activity than expected from its activity on other substrates. Finally, H48A/W79A showed no detectable activities on any phospholipid substrates even when excess enzyme concentrations were employed (i.e. Ͼ50-fold of W79A concentration).
Membrane Affinities of hVPLA 2 and Mutants-To quantitatively determine the effects of mutations on the interfacial binding of hVPLA 2 , we measured the binding of mutants to anionic sucrose-loaded BLPG polymerized liposomes and zwitterionic DHPC-coated beads. For PC binding, DHPC-coated beads were used instead of polymerized liposomes due to low pelleting efficiency of PC vesicles. Binding constants and relative affinities are summarized in Table II. The relative affinity for BLPG polymerized liposomes showed that the mutations of Lys 6 , Lys 11 , and Arg 34 of hVPLA 2 had modest effects on its binding to anionic membranes. This supports the notion that they are located on the interfacial binding surface of hVPLA 2 and accounts for the modest preference of hVPLA 2 for anionic membranes (18). It should be noted that the apparent lack of deactivating effects of these mutations on kinetic activity toward anionic polymerized mixed liposomes is due to our assay conditions in which a majority of enzymes are bound to vesicles (i.e. [BLPG] Ͼ Ͼ nK d for all mutants) (29). Mutations of other residues, including Lys 74 and Lys 76 , had much less effect on anionic vesicle binding, indicating that they are not directly involved in binding to anionic membranes. When the affinities of mutants for zwitterionic DHPC-coated beads were measured, none but K6E mutation had a significant effect, indicating that most of anionic residues play no role in binding to zwitterionic surfaces. The K6E mutation reduced the binding affinity by about 5-fold, which is much larger than the 2-fold   c The absolute specific activity value for W79A was 0.10 Ϯ 0.05 mol/min/mg. drop in affinity for anionic BLPG polymerized liposomes. This, in conjunction with the unusually low kinetic activity of K6E on SAPC vesicles, suggests that the K6E mutation interferes with the optimal interaction of hVPLA 2 with zwitterionic membranes. However, these results do not necessarily point to the direct involvement of Lys 6 in binding to zwitterionic membranes. Intuitively, it is more likely that the K6E mutation induces local structural changes of hVPLA 2 , which in turn interfere with the interactions of other critical residues, such as Trp 31 (18), with zwitterionic membranes.
Heparin Affinities of hVPLA 2 and Mutants-Binding of hV-PLA 2 with heparin and its derivatives is a complex process that involves multiple interaction sites, which makes it difficult to quantitatively determine the binding affinity (31,32). We previously reported the use of immobilized heparin and heparan sulfate columns to semiquantitatively assess the binding affinities of hIIaPLA 2 and mutants for heparinoids (16). These measurements showed that the relative binding affinities of hIIaPLA 2 and mutants for heparin and heparan sulfate columns were comparable (16). Thus, we measured the chromatographic behaviors of hVPLA 2 mutants using an immobilized heparin column to estimate their relative affinities for cell surface HSPG. The protein bound to the resin was eluted with a linear gradient of KCl in the elution buffer, and the concentration of KCl that gives rise to a major protein peak was determined for each mutant. Chromatograms for selected mutants were shown in Fig. 4, and the KCl concentrations corresponding to protein peaks are listed in Table II. The chromatographic data clearly showed that the C-terminal cationic residues, which form a predominant cationic patch on the rear side of the molecule (see Fig. 3), are involved in heparin binding whereas the three residues on the front side are not. The presence of a well-defined heparin-binding site in hVPLA 2 , which is spatially distinct from its interfacial binding surface, is in sharp contrast to a diffuse heparin-binding site of hIIaPLA 2 that overlaps considerably with its interfacial binding surface (16). Despite the presence of well defined heparinbinding site, hVPLA 2 showed significantly lower heparin affinity than did hIIaPLA 2 , which eluted with 0.45 Ϯ 0.02 M KCl. In that hIIaPLA 2 has a higher isoelectric point and a larger number of surface cationic residues than does hVPLA 2 , this implies that PLA 2 -heparin binding is driven by relatively nonspecific electrostatic interactions.
Activities of hVPLA 2 and Mutants to Release Fatty Acids and Leukotrienes from Neutrophils-To correlate the different properties of hVPLA 2 mutants to cellular activities, we measured their activities to release fatty acids from human neutrophils pre-incubated with radiolabeled AA and OA. With 0.1 M W79A, total fatty acid release was 4% and 2% of incorporated AA and OA, respectively, indicating the lack of pronounced sn-2 acyl selectivity. A similar result was reported for hVPLA 2 acting on the macrophage-like cell line, P338D1 (33). Relative activity of mutants in terms of AA release is listed in Table III. All mutants showed essentially the same relative activity in terms of OA release. We also measured their activities to release LTB 4 from human neutrophils, which is the main eicosanoid product of neutrophils. Results are summarized in Table III. We previously showed for hVPLA 2 and selected mutants that their activities to release fatty acids from human neutrophils correlate well with their activities to induce LTB 4 production (18). Within the range of experimental errors, the mutants listed in Table III showed essentially the same trend. Note that three mutants, including K6E/W79A, R34E/W79A, and W79A/K107E/R108E, have significantly lower cell activities than others. Importantly, all these mutants have a common property, lower activity on PC vesicles, which derives from different origins: K6E/W79A has lower PC membrane affinity whereas R34E/W79A and W79A/K107E/R108E have lower catalytic efficiencies. Among the three mutants with significantly reduced heparin affinities, only W79A/K107E/R108E with lower catalytic efficiency showed decreased cell activities. W79A/R111E with unaltered catalytic efficiencies was as active as W79A, whereas W79A/R100E/K101E showed even higher activity to release AA than wild type, which is in accordance with its activity on SAPC vesicles. Thus, the heparin binding and hIIaPLA 2 (f). Relative activity was calculated by dividing activity values by an arbitrary value (ϭ 0.03 nmol/min). Note that hIIaPLA 2 has significantly lower activity than hVPLA 2 mutants. These data were taken from a single measurement for each protein. per se does not appear to play a major role in interaction of hVPLA 2 with neutrophil cell membranes and subsequent leukotriene-inducing activities. Finally, H48A/W79A (up to 1 M) did not show detectable activity on neutrophils under our experimental conditions. Role of Heparan Sulfate Proteoglycan Binding in hVPLA 2 Internalization-The lack of correlation between the heparinbinding affinity of hVPLA 2 mutants and their activities on cells, despite relative high affinity of hVPLA 2 for heparin, suggested that the binding to cell surface HSPG results in a nonproductive process. An interesting possibility is that the HSPG binding of hVPLA 2 leads to the internalization and clearance of surface-bound proteins. The HSPG-mediated internalization of sPLA 2 has been proposed for group IIa PLA 2 (14). To explore this possibility, we first monitored the temporal and spatial sequences of W79A and W79A/R100E/K101E exogenously added to neutrophils by time-dependent immunostaining and confocal microscopic imaging. Results from the microscopic imaging of the proteins are illustrated in Fig. 5. The internalization of W79A to neutrophils was detectable after 10 min (data not shown) and completed in 20 min. As seen from the comparison between permeabilized and nonpermeabilized cells, the majority of W79A molecules were internalized after 20 min. In contrast, W79A/R100E/K101E showed the same annular distribution on the membrane whether the cells were permeabilized or not, demonstrating that the protein molecules are mainly bound to the outer plasma membranes. Dramatically reduced internalization of W79A/R100E/K101E indicates that the HSPG binding is important for the internalization of hVPLA 2 .
To determine the correlation between the internalization of hVPLA 2 and its eicosanoid-inducing activities, we treated human neutrophils with the two proteins and measured the time courses of AA and LTB 4 release and protein internalization. The latter was done by time-dependent Western blotting of cell extracts. Fig. 6 shows that the liberation of AA by W79A reached a saturated value in about 10 min. LTB 4 release was also completed in 15 min (data not shown). In contrast, the AA release by W79A/R100E/K101E continued to proceed even after 1 h, resulting in 75% more AA release than W79A at 1 h (note that data in the second column of Table III were collected at 30  min). Interestingly, however, the production of LTB 4 by W79A/ R100E/K101E ceased in about 15 min (data not shown). As a result, W79A/R100E/K101E produced higher levels of AA, but not LTB 4 , than did wild type (see Table III). Additionally, cells treated with W79A/R100E/K101E for 1 h were lysed (data not shown), whereas cells incubated with W79A remained intact under the same conditions. Fig. 7 illustrates the Western blot-ting of neutrophil extracts treated with W79A and W79A/ R100E/K101E. Since the cells were thoroughly washed with the buffer containing 0.5 M NaCl, the observed bands should mainly represent the internalized PLA 2 . This notion is also supported by the time-dependent increase in PLA 2 band up to 20 min. The internalized W79A was subsequently degraded, as indicated by the complete disappearance of PLA 2 after 45 min. In contrast, no internalized PLA 2 band was detected when cells  were treated with W79A/R100E/K101E. Since the production of AA and LTB 4 by W79A was completed before the internalization occurred, it is clear that the internalization did not result in further production of fatty acids and LTB 4 . Also, W79A/ R100E/K101E continued to release AA while acting solely on the outer plasma membranes, again demonstrating that internalization of hVPLA 2 is not required for the production of fatty acids and LTB 4 . Instead, the HSPG-mediated internalization of hVPLA 2 might serve as a mechanism to clear the cell surfacebound hVPLA 2 for cell protection.

DISCUSSION
Mammalian sPLA 2 s, including group V PLA 2 , are mobilized and secreted from various cells in response to inflammatory stimuli. So far, the mainstay of sPLA 2 research has been to study the localization of sPLA 2 s and their coupling with cytosolic PLA 2 and downstream cyclooxygenases and lipoxygenases during the activation of inflammatory cells. Although these lines of research have produced valuable information about the eicosanoid-inducing actions of sPLA 2 s, they did not deliver the detailed mechanistic information because of inherent difficulty in performing a rigorous structure-function analysis. We have taken a different approach of preparing pure recombinant wild type and mutant proteins, thoroughly characterizing their in vitro enzymatic activities, membrane affinities, and heparin affinities, and finally measuring their cellular activities when added exogenously to the cells (16,18). The physiological relevance of this approach is supported by findings that the eicosanoid-inducing actions of exogenously added sPLA 2 simulate those of endogenous sPLA 2 (12) and that sPLA 2 are released and act both in autocrine and paracrine manners under inflammatory conditions (7). It should be also stressed that physiologically relevant submicromolar enzyme concentrations (i.e. Յ0.1 M) were employed in these studies to avoid nonphysiological cell responses. Finally, although one cannot completely rule out the possibility that the trace amount of bacterial endotoxin in recombinant hVPLA 2 proteins might contribute to their cell activities, the excellent correlation between in vitro and cellular activities of all mutants, no detectable cell activity of H48A/W79A (up to 1 M) in particular, indicates that this potential endotoxin effect is insignificant under our experimental conditions.
Interfacial Binding and Heparin Binding Residues-Our model structure of hVPLA 2 (Fig. 3) suggests the presence of cationic patches on both sides of molecular surface. Membrane and heparin binding measurements indicate that cationic residues on the surface containing Trp 31 , which was shown to be critical for binding to zwitterionic membranes (18), are involved in binding to anionic membranes, whereas those on the opposite face are involved in heparin binding. The effects of the K6E, K11E, and R34E mutations on the energetics of binding to anionic vesicles, which can be estimated using an equation, ⌬⌬G 0 ϭ ϪRT ln[relative affinity] under the standard conditions with the concentration of free phospholipid set at 1 M, are 0.54 kcal/mol or less at 25°C. These values are much smaller than those calculated for cationic residues on the interfacial binding surfaces of other sPLA 2 s, which range from 1.0 to 3.6 kcal/mol (9,10,22). Assuming the additivity of their contributions, the total change of interfacial binding energy would be less than 1.5 kcal/mol at 25°C. This indicates that the membrane binding of hVPLA 2 is driven largely by nonelectrostatic forces. The fact that hVPLA 2 has about 3 times higher affinity for anionic membranes than hIIaPLA 2 , despite the much smaller contributions from its cationic residues, underscores the importance of the nonelectrostatic interactions. The interactions would derive from a number of aromatic and aliphatic residues on the putative interfacial binding surface of hVPLA 2 (see Fig. 3). For instance, the W31A mutation of hVPLA 2 alone reduced the interfacial binding energy by 1.6 kcal/mol (18).
Our heparin-binding measurements indicate that the five mutated residues in the carboxyl terminus of hVPLA 2 , which form a prominent cationic patch in the model structure, provide a binding site for heparinoids. In case of hIIaPLA 2 , both aminoterminal and C-terminal cationic patches as well as four cationic residues between residues 53 and 58 were shown to be involved in heparinoid binding (16). Although hVPLA 2 also has two cationic residues, Arg 54 and Lys 58 , on the same surface as the five C-terminal residues, these residues were not included in this study because they were shown to be involved in interactions with the head group of an active site-bound phospholipid (34). The apparent lack of heparin affinity of Lys 6 and Lys 11 of hVPLA 2 might be due to their suboptimal arrangement for heparin binding. The complete spatial separation of interfacial and heparin binding sites could, in principle, allow the protein molecules bound to cell surface HSPG to act on membrane phospholipids. However, no direct correlation was observed between the in vitro heparin affinities of mutants and their activities on cells, indicating that the binding of group V PLA 2 to cells via HSPG is not important for its actions on cells. This notion is, however, at odds with the previous report by Murakami et al. (3), in which the HSPG affinities of rat group V PLA 2 and two mutants, R94G/K95E and R101S/R102S, which correspond to our K79A/R100E/K101E and K79A/ K107E/R108E mutants (note that different numbering systems are used; see Fig. 2), are in line with their AA-releasing activities, when they were transfected into HEK 293 cells. Note, however, that the two measurements were performed using different cell types under different conditions. We performed our studies with neutrophils in the absence of agonist, whereas Murakami et al. (3) investigated the augmentation of agonistinduced AA and prostaglandin release by sPLA 2 . Although further study is needed to resolve this controversy, our studies clearly show that K79A/K107E/R108E has low AA-and LTB 4releasing activities on neutrophils because of its low catalytic efficiencies.
AA-and LTB 4 -releasing Activities of hVPLA 2 -Consistent with our previous results on W31A (18), the relative activity of and W79A/R100E/K101E (B) were incubated with neutrophils suspended in HBSS at 37°C for a given period, and cells were centrifuged. The pelleted cells were thoroughly washed with HBSS containing 0.5 M NaCl, lysed, and subjected to SDS-electrophoresis on 16% acrylamide gels. Essentially the same electropherograms were obtained from triplicate experiments. mutants on PC vesicles are well correlated with their cell activities. At present, it is not clear how exactly the K6E mutation reduces the affinity for PC membranes and how the R34E, K74E, and K107E/R108E mutations decreased the catalytic efficiencies. Comparable expression yields of these mutants and W79A suggest the absence of deleterious gross structural changes that could result in reduced protein stability. One can thus speculate that the mutations locally affect the residues involved in interfacial binding, substrate binding, or catalysis via electrostatic or steric interactions. Regardless of the origin of these effects, these studies clearly show that those mutations that reduce the activity of hVPLA 2 on zwitterionic membranes would lower the activity of hVPLA 2 to hydrolyze cell membrane phospholipids and eventually induce eicosanoid formation. This, in conjunction with the unique ability of hV-PLA 2 to induce eicosanoid production in mammalian cells in the absence of agonist (18), indicates that the role of hVPLA 2 is to directly bind to the outer plasma membranes of mammalian cells and to hydrolyze phospholipids, presumably PC. Finally, the lack of cell activities of H48A/W79A points to the absence of a receptor-mediated mechanism in human neutrophils. Although H48A/W79A has less than 1% of wild type activity, it can still bind PC membranes as well as wild type, as shown by the surface plasmon resonance analysis (data not shown). 2 It also has a circular dichroism spectrum similar to that of wild type (data not shown). One would thus expect that it should bind the receptor, if any, as effectively as the wild type protein.
A similar negative result was observed with mouse group IIa PLA 2 acting on HEK 293 cells (3).
Temporal and Spatial Resolution of hVPLA 2 Action-Our results with doubly labeled neutrophils show that exogenously added hVPLA 2 liberates OA and AA to reasonably comparable degrees. In view of lack of sn-2 acyl group specificity of sPLA 2 s, including hVPLA 2 , these findings indicate that exogenously added hVPLA 2 is primarily responsible for the release of fatty acids and lysophospholipids from neutrophils. For the reasons described above and below, the site of hVPLA 2 action should be the outer plasma membrane. The elucidation of the exact mechanisms by which these extracellular fatty acids and lysophospholipids induce cellular eicosanoid production would require further investigation. It is evident from these studies, however, that the role of hVPLA 2 is not to continuously supply AA for cellular cyclooxygenases and lipoxygenases from the outer plasma membrane. The total amount of AA released from the labeled neutrophils is typically less than 4% of total incorporated AA. The total amount of the LTB 4 release that results from this fatty acid release, however, is comparable to that induced by potent agonists, such as N-formyl-methionyl-leucylphenylalanine (18). The low AA release is not because it is mostly converted to LTB 4 , as AA and LTB4 release was measured separately under different conditions. AA release was measured with BSA in the media to ensure that high proportion of AA liberated from the outer plasma membrane was extracted into the media, whereas the LTB 4 release was measured in the absence of BSA in the media. It thus appears that the products of hVPLA 2 hydrolysis trigger the cellular eicosanoid production by activating cellular enzymes involved in eicosanoid formation. If this is the case, how could cells modulate the extent of membrane hydrolysis by the cell surfacebound hVPLA 2 ? One reasonable mechanism is to remove the cell surface-bound hVPLA 2 by internalization and degradation. Our confocal microscopic imaging and Western blotting analysis of neutrophils treated with hVPLA 2 mutants support this notion. W79A/R100R/K101E with reduced heparin binding affinity was not internalized into these cells under the conditions where W79A was internalized and degraded. As a result, the former liberated a larger amount of fatty acids from the outer plasma membrane than did the latter under the same conditions. This notion is also consistent with recent reports that group X PLA 2 , which has high affinity for the outer plasma membrane but low affinity for HSPG, is highly effective in liberating fatty acids from mammalian cells (35,36). Since the extensive hydrolysis of the plasma membrane leads to cell lysis, the overproduction of fatty acids by W79A/R100R/K101E was not conducive to LTB 4 formation, as witnessed by a comparable degree of LTB 4 release by W79A and W79A/R100R/K101E.
The physiological consequence of HSPG-mediated PLA 2 internalization has remained controversial. Our results suggest that HSPG-mediated internalization might serve as a mechanism to remove cell surface-bound sPLA 2 in neutrophils. Apparently, these results are at odds with a report by Murakami et al. (14), which indicates that a family of HSPG, glypican, facilitates the trafficking of group IIa PLA 2 into particular subcellular compartments where it produces AA and induces prostaglandin biosynthesis. In our studies, no correlation was found between the internalization of hVPLA 2 into neutrophils and the release of AA and LTB 4 . W79A completed the AA release before its internalization took place, whereas W79A/ R100R/K101E was more active in AA release than W79A despite lack of internalization. Again, the discrepancy might arise from the fact that two studies were performed using different cells under different conditions. Indeed, Enomoto et al. (37) recently found that group IIa PLA 2 was internalized into mouse bone marrow-derived mast cells and degraded in a HSPGdependent manner.
On the basis of our previous and present results, we propose a mechanism for the action of hVPLA 2 on human neutrophils. In this mechanism, exogenous hVPLA 2 directly binds the outer plasma membrane and hydrolyzes PC and other phospholipids to liberate fatty acids and lysophospholipids. This immediate release of PLA 2 products induces the activation of cellular proteins, including cytosolic PLA 2 and 5-lipoxygenase, which eventually leads to the cellular production of AA and LTB 4 . Independently, the HSPG-mediated internalization and degradation of protein serves as a cell protection mechanism. The presence of distinct interfacial binding site and HSPG binding site in hVPLA 2 is advantageous for the regulatory purpose because cell surface HSPG can bind and sequester cell membrane-bound hVPLA 2 in action, without having to compete head-to-head with membrane phospholipids. Undoubtedly, the detailed understanding of this proposed mechanism entails further investigation. Additionally, it remains to be seen whether or not the action of sPLA 2 s on other mammalian cells follows the same mechanism. As such, the mechanism provides a basis for further investigation of the cellular signaling pathways that lead to the activation of eicosanoid-producing enzymes by exogenous sPLA 2 s.