Inhibition of phospholipase A2 by "lipocortins" and calpactins. An effect of binding to substrate phospholipids.

The “lipocortins” are a group of proteins that have been reported to inhibit phospholipase A2 by direct interaction with enzyme. Two proteins which have been identified as lipocortin on the basis of inhibition of phospholipase A2 activity have recently been cloned and sequenced. These have been shown to be identical to the calpactins, which are membrane cytoskeletal proteins serving as major substrates of the tyrosine protein kinases. We have now found that two forms of calpactin (I and 11) inhibit porcine pancreatic phospholipase A2 in an assay using Escherichia coli cells or extracted phospholipid vesicles as substrate, but only when the substrate concentration is very low. Both calpactins, as well as another, 73-kDa inhibitory protein, were found to bind purified phospholipids and E. coli cell membranes directly. Kinetic studies show that the inhibition of phospholipase Az by calpactin can be overcome by high phospholipid substrate concentrations, whether E. coli cells or isolated phospholipid vesicles are used. For example, in the presence of 5 X 10”’ M phospholipase A2 and 1 X 10” M calpactin, the inhibition decreases from 100 to 0% as phospholipid in vesicles is raised from 2 to €4 PM. The evidence reported

The "lipocortins" are a group of proteins that have been reported to inhibit phospholipase A2 by direct interaction with enzyme. Two proteins which have been identified as lipocortin on the basis of inhibition of phospholipase A2 activity have recently been cloned and sequenced. These have been shown to be identical to the calpactins, which are membrane cytoskeletal proteins serving as major substrates of the tyrosine protein kinases. We have now found that two forms of calpactin (I and 11) inhibit porcine pancreatic phospholipase A2 in an assay using Escherichia coli cells or extracted phospholipid vesicles as substrate, but only when the substrate concentration is very low. Both calpactins, as well as another, 73-kDa inhibitory protein, were found to bind purified phospholipids and E. coli cell membranes directly. Kinetic studies show that the inhibition of phospholipase Az by calpactin can be overcome by high phospholipid substrate concentrations, whether E. coli cells or isolated phospholipid vesicles are used. For example, in the presence of 5 X 10"' M phospholipase A2 and 1 X 10" M calpactin, the inhibition decreases from 100 to 0% as phospholipid in vesicles is raised from 2 to €4 PM. The evidence reported here strongly suggests that in vitro inhibition of phospholipase A2 by lipocortin is due to sequestering of the phospholipid substrate by the inhibitor protein, rather than a direct interaction with the phospholipase. These results raise questions about the physiological significance of the inhibition of phospholipases by calpactins.
The lipocortins are defined as anti-inflammatory proteins which are inducible by steroids, secreted by cells, and thought to act prior to the cyclooxygenase pathway by inhibition of a phospholipase A,. In the search for the purified form of such an inhibitor protein, several 36-kDa proteins have been isolated, using as an assay their ability to inhibit phospholipase Az in vitro. While there was initial evidence for smaller molecular weight proteins possessing lipocortin-like activity, these proteins are now thought to be proteolytic fragments of a 36-kDa parent form. The first 36-kDa purified protein to be assigned the name lipocortin is now known as lipocortin I and * This work was supported in part by Grants GM 20,501 (to E. A. D.) and GM 32,866 (to J. R. G.) from the National Institutes of Health and by Grant DMB 18684 from the National Science Foundation (to E. A. D.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. (1 Postdoctoral fellow of the American Cancer Society. ** To whom correspondence should be addressed. is thought to be a potent phospholipase Az inhibitor regulated in vivo by phosphorylation (for reviews, see Refs. 1 and 2).
Recently, sequence analysis has united the study of lipocortin with another line of research, that involving membrane-associated cytoskeletal proteins that serve as substrates of viral and growth factor receptor tyrosine kinases. These membrane cytoskeletal proteins have been termed calpactins I and I1 to denote their calcium-dependent phospholipid and actin binding properties and are also known as p36 and p35. Cloning and sequencing of the cDNA for calpactin I has revealed the surprising result that it is greater than 50% homologous to lipocortin I (3,4). Further sequence results have shown that calpactin I1 is equivalent to lipocortin I (5), and a second form of "lipocortin" is equivalent to calpactin I (6). In other words, "calpactin I" equals "lipocortin 11" and "calpactin 11" equals "lipocortin I." For simplicity, we will refer to these proteins as calpactins.
Calpactin I as isolated from intestine, lung, or lymphocytes can occur as a 38-kDa monomer or as a heterologous tetramer of two 38-kDa subunits and two ll-kDa subunits, the latter bearing homology to the SI00 proteins (7)(8)(9). Calpactin I1 occurs only as a 35-37-kDa monomer and was originally identified in A431 cells on the basis of shared functional and antigenic properties with calpactin I. Further investigation revealed it to be the same protein as the 35-kDa epidermal growth factor receptor substrate described by Fava and Cohen (10,11). Some of the early controversy as to the size of lipocortin I in different cell types may have its roots in the coexistence in many tissues (3) of the multisubunit antigenically related calpactin I, which in addition to its native subunits has also a protease-sensitive region separating a 33-kDa core from a 5-kDa tail (7). Sequence homology between calpactins I and I1 is greater than 50% in this central core region (3,4), where Ca2+ and phospholipid are known to bind (12,13) and less than 20% in the amino-terminal tail which in calpactin I contains the site of tyrosine phosphorylation (7) and the binding site for the Il-kDa light chain (14).
In an effort to elucidate the functions of these related proteins, we have initiated studies on the nature of the inhibition of phospholipase A, by calpactins. We and others have developed kinetic approaches for analyzing phospholipase A, action (15-20), but due to the insoluble nature of the substrate, these analyses can be complex (21). Different extracellular phospholipase Azs show variations in specificity for different polar head groups (22)(23)(24), so that caution must be applied when analyzing even simple substrate analogues (25), and the ability of an inhibitor to partition between aqueous and lipid phases or otherwise affect the aggregated state of the substrate must always be considered. Although it has been suggested (26) that the mechanism of lipocortin action is by direct binding to phospholipase A, to inactivate the enzyme, no experimental evidence has been given to support this claim.

Phospholipase Inhibition by Lipocortins and Calpactins
Furthermore, calpactin I is known to interact with the cytoskeletal proteins spectrin and actin in the presence of high Ca2+ and with phospholipids at micromolar concentrations of free Ca2+ (12). Thus, we questioned whether direct binding of calpactin to the phospholipid membrane might not be responsible for phospholipase A, inhibition in vitro. We now report that calpactin I1 and the 38-kDa subunit of calpactin I both bind phospholipid in a Ca2+-dependent manner and that both calpactins are equally potent in inhibiting phospholipase Az under certain assay conditions. A related 73-kDa protein is even more potent at inhibiting phospholipase A, and also binds to phospholipids. We show that calpactin inhibition of phospholipase A2 is dependent on substrate concentration and is most probably due to substrate depletion by direct substrate/inhibitor complexation not involving a specific interaction with the enzyme.

EXPERIMENTAL PROCEDURES
Isolation of Proteins-Calpactins I and I1 and the related 73-kDa protein were isolated from bovine lung by a modification of the previously described method (12), the details of which are provided elsewhere (5). The proteins were characterized by two-dimensional gel electrophoresis, where they are known to be resolved (lo), and by Western blotting using Cohen's anti-p35 (27) and our anti-intestinal calpactin I (12) antibodies. By these criteria there was no detectable cross-contamination of the two calpactins. Peptide mapping of the calpactins was performed as described previously (10). Phospholipid Binding by Calpactins and the 73-kDa Protein-Phospholipid binding by calpactin was monitored by sucrose flotation gradients at a defined free Ca2+ concentration. Calpactin I, calpactin 11, the 73-kDa protein, or cytochrome c (25 pg each) were mixed with phospholipid vesicles (50 pg) prepared by sonication in 10 mM imidazole, 40 mM KC1,2 mM MgCl,, 10 mM EGTA,' and either 5 mM CaC12 (1 p~ free CaZ+) or 8.8 mM CaC12 (10 p M free Ca2+) at pH 6.8. Vesicles were prepared by mixing 2 mg of phosphatidylserine (PS) with 2 mg of phosphatidylethanolamine (PE) (both from Sigma) and 1 pg (0.5 mCi) of 1,2-dioleoyl-~-3-phosphatidyl-~-[U-'~CC]serine (["CIPS) (Amersham Corp.), drying under vacuum, and sonicating into 1 ml of 10 mM Tris-HC1, 50 mM NaC1, pH 8.0. The calpactin plus lipid solution was then adjusted to 50% sucrose in the same buffer using an 80% sucrose stock solution. This was overlaid with a step gradient of sucrose consisting of 150 pl each of 40% sucrose, 30% sucrose, 20% sucrose, and buffer only in a 5 X 41-mm ultracentrifuge tube and centrifuged at 50,000 rpm (Beckman SW 50.1 rotor) for 2 h. After centrifugation, the gradients were immersed in liquid Nz and 8 fractions prepared by cutting into sections with a razor blade. The fractions were added to SDS-sample buffer, and equal amounts of each fraction were analyzed by SDS-PAGE, Coomassie Blue Staining, and quantitation by densitometry. The distribution of phospholipid in the gradients was determined by liquid scintillation counting.
Preparation of 3H-LabeZed E. coli-Radiolabeled bacteria were obtained by a modified version of the procedure of Patriarca et al. (28). E. coli (K-12C600) were grown to log phase in L-broth (400 ml), pelleted, and resuspended to one-tenth the original volume with fresh broth. [9,10-3H]Oleic acid (2.5 mCi of 2.5 Ci/mmol, Amersham Corp.) which had been dried from toluene and sonicated into 2.5 ml of L broth was added to the culture, the cells allowed to grow an additional 3 h at 37 "C, and aliquots (0.1 ml) removed for determination of cell density immediately prior to autoclaving the cell preparation (15 min at 120 "C, 2.7 kg/cmz). The cell density, determined by serial dilution and plating out on agar, was 1.5 X lo9 cells/ml for the 40-ml culture.
After autoclaving, the cells were washed by centrifugation (13,000 X g for 5 min) up to 10 times with sterile 100 mM Tris-HC1, pH 8.5, 20 mg/ml bovine serum albumin (BSA), and then up to 10 times with the same buffer without BSA, resuspended in 40 ml of the same buffer, and stored at 4 "C. Total protein was measured by the BCA protein assay (Pierce Chemical Co.) and found to be 0.74 mg/ml. Quantitation of phospholipid in the cells (see below) yielded a protein/ phospholipid ratio of 8 pg of protein/nmol of phospholipid.
Phospholipase A, Activity Toward 'H-lubekd E. coli Celk"Ph6spholipase A, was assayed by the method of Rothhut et al. (29). For the dose-response experiments with inhibitor proteins and competition with PS, 3H-labeled bacteria were used as substrate in an assay containing 50 ng of porcine pancreatic phospholipase A, (Sigma, 690 units/mg), 3H-labeled E. coli (1 pg of protein, 0.16 nmol of phospholipid), and the specified concentration of inhibitor protein in a final volume of 350 pl with 100 mM Tris-HC1, pH 8.0, 10 mM CaC1,. Substrate was added to start the reaction and after 5 min at 0 "C, 100 pl of 2 N HCl was added to stop the reaction. BSA was then added (100 pl of 100 mg/ml, and the tubes were capped, vortexed, and centrifuged at 13,000 X g for 5 min to remove the bacteria. 3H release was determined by scintillation counting of 400-pl aliquots of the supernatant. Each assay result presented represents the average of three separate determinations performed on different days. Controls always included tubes in which the phospholipase or inhibitor were omitted. In all other experiments involving whole E.. coli cells, the standard conditions were slightly modified in order to compare with vesicle assays. The reaction volume was 150 pllassay with the cells and phospholipase and/or calpactin incubated in 100 mM Tris-HC1, pH 8.5,l mM CaCl, at 0 "C for the required times. Reactions were stopped with the addition of 50 pl of 2 N HC1 and 50 pl of BSA (100 mg/ml), and 2 0 0 4 aliquots of the supernatant were used for liquid scintillation counting. The amount of E. colilassay was varied, as were reaction times in some instances, and the reaction was started with either the addition of 174 ng of phospholipase A, or with the addition of cells. Results are reported as the average of triplicates k S.D. of the mean percent inhibition. Percent inhibition was calculated from the 3H cpm released in the calpactin and phospholipase A,-containing sample, minus the blank, divided by 3H cpm released in the sample containing phospholipase A, only minus 3H cpm in the blank, where the blank had no phospholipase A2.
Two-dimensional Thin Layer Chromatography of Phospholipase A,reacted E. coli-E. coli lipids were extracted by the method of Bligh and Dyer (30) following phospholipase A, treatment. Briefly, porcine pancreatic phospholipase A, (174 ng) was added to 150 pl of 100 mM Tris-HC1, pH 8.5, 10 mM CaCl, with or without calpactin I (7.5 pg), and the reaction started with 3H-labeled E. coli cells (2X = 2 pg of protein, 0.32 nmol of phospholipid). One set of reactions was stopped after 10 min at 0 "C with 50 pl of 2 N HC1 and analyzed as above. An identical set of reactions was stopped after 10 min at 0 "C with 9 pl of 0.2 M EDTA plus 605 pl of chloroform/methanol, 1:2, giving a final ratio of 1:2:0.8 CHC13/CH30H/H20. The one-phase extractions were allowed to sit for 10 min at 0 "C, and then CHC1, and H20 were added (100 p1 each) to give a final ratio of 2:2:1.8 CHC13/CH30H/ H20, and the samples were centrifuged (13,000 X g for 1 min) to separate the two phases. The chloroform layers were taken, dried under a stream of nitrogen and then in vacuo, and the lipid residues resuspended in 20 p1 of CHC13/CH30H, 1:2. 3H label left in the aqueous phase was always less than 3% of total radioactivity. Each sample was subjected to two-dimensional TLC on Silica Gel G plates (20 X 20 cm, Analtech), and phospholipids were identified by reference to known standards (28, 31). Development in the first dimension was with CHC13/CH30H/CH3COOH (65:25:6) and in the second dimension with petroleum ether/ether/acetic acid (80201).
Co-chromatographed standards were visualized with I , , and the phospholipid zones were scraped into vials for liquid scintillation counting. In this TLC system, cardiolipin comigrates with phosphatidic acid, and PE and PG are incompletely separated. The results are expressed as 3H cpm for a given phospholipid zone divided by the total cpm recovered from the plate times 100. Values reported are the average of duplicate assays run on separate plates. The average error was ? 15% of the value given, and in no case did the deviation represent more than 4% of total radioactivity.

Extraction of E. coli Phospholipid and Determination of Specific
Radioactiuity-Cells (8 ml of the 40-ml suspension described above) were pelleted, resuspended in 1.6 ml of buffer, and extracted according to the method of Bligh and Dyer (30). The phospholipid so obtained (0.7 pmol) was resuspended in CHC13, quantitated on the basis of inorganic phosphate (32,331, and stored under N2 at -20 "C. Extraction was 96% complete based on radioactive yield. The specific radioactivity of this lipid was 2.0 X lo6 cpm/nmol of Pi. Extrapolation back to the original cell culture indicated that 1.6 X 10" cells contained approximately 10 nmol of phospholipid, in reasonable agreement with previously reported results (34).

Phospholipase Inhibition by Lipocortins and Calpactins
Phospholipase A2 Activity Toward pH]Phosphlipid Vesicles-To prepare stock solutions of E. coli-derived vesicles, phospholipid was extracted as described above, dried under Nz and in vacuo, resuspended in 100 mM Tris-HC1, pH 8.5, 1 mM CaC12, and sonicated for 5 min, on ice, just before use. Assays were carried out with appropriate amounts of [3H]phospholipid with or without calpactin I in 100 mM Tris-HCI, pH 8.5, 1 mM CaClZ in a total volume of 150 pl at 0 "C. Reactions were started with the addition of 1 ng of phospholipase A2 and stopped by the addition of 563 pl of CHC13/MeOH, 1:2, containing 4 mM EDTA followed by standard Bligh and Dyer (30) extraction. Time points were chosen to achieve six to eight points per assay within a (linear) hydrolysis range of 1-5% (background hydrolysis was 4 % ) .
Chloroform layers containing phospholipase reaction products were subjected to one-dimensional thin layer chromatography (34) on Silica Gel G plates in petroleum ether/ether/acetic acid, 80:201. The fatty acid and diacylglycerol plus phospholipid spots were retrieved and subjected to liquid scintillation counting. Triacylglycerol radioactivity was found to be negligible. Percent hydrolysis was calculated as cpm of fatty acid divided by (cpm of fatty acid + cpm of phospholipid). For each assay condition, percent hydrolysis was plotted as a function of time and subjected to linear least squares analysis ( N = 6-8). The slope thus obtained was used to calculate the hydrolysis rate (nmol min" mg"). Nonzero slopes all gave PN( I r I > ro) < 0.05, and the average PN (1 rl > TO) was less than 0.02. Error bars represent standard deviations from the calculated least squares fit. In experiments where there was no apparent phospholipase A2 activity, least squares analysis also showed no statistical correlation for hydrolysis uersus time (PN > 0.15), and slopes of zero were assigned. The standard deviation from the mean percent hydrolysis was always less than 0.10 in these cases.
Binding of Inhibitory Proteins to E. coli-To monitor the binding of phospholipase inhibitory proteins to bacteria, E. coli were grown in the absence of [3H]oleic acid followed by autoclaving and washing as above. Bacteria (20 pg of protein) and inhibitors (20 pg) were adjusted to 100 pl with 100 mM Tris-HC1,lO mM CaC12 and incubated for 10 min on ice. The solutions were centrifuged at 13,000 X g for 5 min, and supernatant and pellet fractions were analyzed by SDS-PAGE and Coomassie Blue staining. Controls included a protein which does not bind lipid (cytochrome c ) and solutions in which EGTA was substituted for Ca2+ in the buffer.
Affinity Chromatography of Calpactin on Phospholipase A2-Sephnrose-Affinity chromatography was attempted by coupling 5.6 mg of phospholipase A2 to 1 ml of CNBr-activated Sepharose 4B (Pharmacia P-L Biochemicals) according to the manufacturer's directions. Calpactin I1 (120 pg) was then passed through the column in 10 mM CaC12, 100 mM Tris-HC1, pH 8.0. After washing with buffer alone, the column was eluted with buffer containing EGTA instead of Ca2+ and then with the same buffer with 1% SDS.

RESULTS
Characterization of Calpactins-Calpactins I and I1 were purified from bovine lung and were homogeneous as assayed by high-resolution one-and two-dimensional SDS-polyacrylamide gel electrophoresis. Peptide mapping (Fig. 1) demonstrates that calpactins I and I1 are distinct from each other, whereas the map of bovine calpactin I1 is strikingly similar to the peptide map of human calpactin I1 (p35) from A431 adenocarcinoma cells. Furthermore, amino acid sequence analysis of a truncated form of this molecule has revealed a 90% sequence identity to human Iipocortin ( 5 ) . Thus, the calpactin I1 used in this study is most probably the bovine equivalent of p35 or lipocortin. Although calpactin I was isolated as a complex of light and heavy chains, we removed the light chain subunit by gel filtration in 6 M guanidine followed by renaturation. This was done in order to compare calpactin I to calpactin 11, which exists only as a monomer. All experiments in this report in which calpactin I was used were performed with this renatured preparation, although phospholipase A, inhibition was also observed with the complex of heavy and light chains (not shown).
Inhibition of Phospholipase A2 by Calpactins and 73-kDa Protein-When calpactins I and I1 were used in any of the standard phospholipase Az kinetic assay systems employed in our laboratory on radioactive (35) or thiol-labeled substrate (161, containing 0.5-10 mM PE, PC, Ps, or PI in vesicles or micelles with 1-10 mM Ca2+ and concentrations of calpactin up to 10 times greater than the enzyme (10-8-10-6 M), no inhibition was observed (data not shown). In striking contrast, when the substrate used was E. coli cells in amounts consistent with previously published work (29, 34), all three proteins displayed an apparent dose-dependent inhibition of phospholipase A2 (Fig. 2). As shown in Fig. 2, increasing amounts of calpactin lead to a progressive decrease in the amount of 3H label released into the assay supernatant by phospholipase A2, and the two calpactins display an identical inhibitory activity over a broad concentration range. Surprisingly, the 73-kDa protein is a more potent inhibitor of phospholipase A2 in this system than the calpactins. Since this protein is approximately twice the molecular weight of calpactin, the half-maximal inhibition observed at an approximately 3-fold lower concentration by weight represents a 6-fold lower molar concentration when compared to calpactin.
Because Interaction of Calpactins and 73-kDa Protein with Phospholipid Vesicles-Calpactin I is known to bind to acidic phospholipids, and this property has been used to identify calpactins I and I1 in A431 cells. Since the ability to bind to phospholipids may be important for calpactin's phospholipase inhibitory activity, we tested calpactins I and I1 and the 73-kDa protein for this property. The proteins were mixed with phospholipid vesicles (using [14C]PS as tracer), adjusted to 60% sucrose, and centrifuged through a sucrose flotation gradient. As shown in Fig. 3, the extent of association of calpactins with vesicles is dependent on the free Ca2+ in the solution. At 1 p~ free Ca2+, <25% of the calpactin is associated with the lipid, whereas a t 10 pM free Ca2+, >95% of the calpactin is vesicle bound. Both calpactin I and I1 behave identically in this assay. In addition, the 73-kDa protein is also tightly associated with vesicles at 10 pt M free Ca2+, whereas the control protein, cytochrome c, is not.
Binding of Calpactins to Substrate Rather than Enzyme in E. coli Assay-An obvious question was whether calpactins I and I1 and the 73-kDa protein bind directly to the E. coli which are used as substrate in the phospholipase assay. The interaction of calpactin I and I1 and the 73-kDa protein with the E. coli membrane was monitored using a simple centrifugation method. Bacteria with bound protein were pelleted  by low speed centrifugation, and aliquots of the supernatant and pellet fractions were analyzed by SDS-PAGE and Coomassie Blue staining. As shown in Fig. 4, all 3 of the proteins which inhibit phospholipase A2 also bind to the bacteria.
Consistent with the 73-kDa protein being a more potent inhibitor of phospholipase, a higher percentage of 73-kDa protein binds to the bacteria than is the case with calpactin. When the bacteria or Ca2+ are omitted, negligible amounts of inhibitory protein are found in the pellet fraction. Cytochrome c, which did not inhibit the phospholipase A2, also does not bind to the bacteria (under conditions of high Ca2+).
To assess whether calpactin I1 binds to phospholipase A2, a solution of calpactin in the same buffer used for phospholipase A, assays was applied to a column of phospholipase A, immobilized on Sepharose. Although phospholipase A, on the column was present in greater than 100-fold molar excess to calpactin, more than 95% of the calpactin passed through unretarded, and no further protein was eluted with EGTA or 1% SDS (data not shown).
Effects of Preincubation with Enzyme or Substrate-Typically, when E. coli are used to assay lipocortin inhibition, there is a preincubation period of 10 min for inhibitor protein plus phospholipase A, before the reaction is started. However, we found that preincubation of calpactin I with phospholipase A2 up to 60 min does not affect the inhibition when concentrations of enzyme, inhibitor, and substrate are constant (data not shown). On the other hand, since calpactin binds tightly to phosphatidylserine (PS), we asked whether preincubation of calpactin with unlabeled PS has any effect on phospholipase A2 inhibition by calpactin. As shown in Fig. 5, when PS vesicles alone are added to the phospholipase assay mix, the amount of 'H label released from the E. coli is reduced by 40-70%, probably due to competition with and dilution of E. coli lipids. When the same levels of PS are added to the assay in 1 a Proteins were incubated in 150 pl of 100 mM Tris-HC1 buffer, pH 8.5, containing 10 mM CaCI2 for 10 min a t 62. Then 'H-labeled E. coli cells were added (2 pg of protein, 0.32 nmol of phospholipid), and the mixtures were incubated 10 min further a t 0°C before addition of CHC13/MeOH, 1:2, and extraction as described under "Experimental Procedures." Calpactin I, when used, was 7.5 pglassay, and phospholipase Az, when used, was 174 ng/assay. The average of duplicates is reported. Data are expressed as percent of total radioactivity.  ca2+ (b, d, e, f ) in 60% sucrose. The solution was overlaid with a 0-50s sucrose step gradient and centrifuged a t 120,000 X g for 2 h. Fractions were analyzed for radioactivity ( . ) by scintillation counting and for protein (0) by SDS-PAGE, Coomassie staining, and quantitation with a densitometer. The phospholipid codistributes with the calpactins and 73-kDa protein only a t 10 p~ Ca2+. combination with calpactin I (at a concentration which previously resulted in 65% phospholipase inhibition) the inhibitory effects of both agents are abrogated. This suggests that calpactin and PS form a complex which is not an inhibitor of phospholipase A,.
Inhibition by Calpactin Is Dependent on Phospholipid Substrate Concentration-The entire E. coli membrane is not needed in order to observe inhibition. E. coli were extracted with chloroform/methanol to obtain their phospholipid components free of proteins and polysaccharides. When the phospholipid mixture was reconstituted into vesicles and used to assay phospholipase A,, calpactin I still showed an ability to inhibit the enzyme in a sigmoidal dose-dependent manner if the substrate concentration was very low, as is shown in Fig.  6. The ratio of calpactin I to phospholipase A, a t half-maximal inhibition is 10001 when the phospholipid is 5 PM. Time courses of phospholipase A, action on this substrate are linear up to about 10% hydrolysis with the exception of a small rapid burst immediately after addition of the enzyme, lasting less than 15 s and accounting for less than 1% hydrolysis (data not shown). Thus, by taking 6-8 time points/assay in the linear region, it is possible to determine rates of hydrolysis with which to carry out a kinetic analysis of the inhibition. When this was done, it was found that inhibition by calpactin I is totally abolished at sufficiently high concentrations of substrate (Fig. 7A). Furthermore, the shape of the inhibition curve is sigmoidal and, significantly, not hyperbolic. In this experiment, calpactin and phospholipase A, concentrations ["HjOleic acid-labeled E. coli was employed as substrate as described in the legend to Fig. 2. Increasing amounts of PS lead to decreased inhibition. As a control, the same amount of PS alone (A) was added during the assay where it reduces the apparent activity of the phospholipase A,.
are held constant, and only the substrate concentration is varied. If the inhibition were due to a simple noncompetitive mechanism, as has been suggested (26), then under these conditions the inhibited sample should show a hyperbolic substrate dependence, and the rates of hydrolysis in inhibited uersus control samples should be a constant ratio regardless of substrate concentration. But, as can be seen in Fig. 7B, the percent inhibition by calpactin I goes from 100 to 0% as the substrate concentration is raised. The result so obtained for inhibition as a function of E.
coli-derived phospholipid concentration is consistent with the preliminary observation that micromolar amounts of calpactin I did not inhibit phospholipase A2 at millimolar amounts of pure phospholipid substrate. To confirm that this substrate concentration effect is also operative in the E. coli whole cell assay, the corollary experiment was done (Fig. 8). As can be seen, a t high concentrations of E. coli, the inhibition is reduced to zero, although a much broader substrate concentration range must be covered in order to achieve the effect.

DISCUSSION
Lipocortins and Calpactins-The "lipocortins" comprise a class of proteins which are thought to be induced by steroids, potent inhibitors of phospholipase A2, and regulated by phos- phorylation (1,2). These observations have been incorporated into a general model for cellular responses such as the activation of thymocytes and platelets (26, 36). According to this '""I scheme, cellular activation leads to the phosphorylation of lipocortin which then loses its inhibitory activity toward phospholipase A,. This would cause an increase in phospholipase A2 activity, thereby increasing formation of one of its products, arachidonic acid, which is a precursor of prostaglandins and leukotrienes. Although this is an attractive model, key features have not been rigorously established. Most importantly, it has not yet been demonstrated that lipocortin is a specific inhibitor of phospholipase A2 in vitro or in uiuo. In fact, earlier work (26) suggested that it may inhibit phospholipases C and D equally as well as phospholipase A,.
In our recent investigations on Ca2+ binding by calpactins, proteins which have been shown to be equivalent to lipocortins, it was noticed that phospholipids increase the affinity of calpactin for Ca2+ (12). This was due to the direct interaction of calpactin with lipid and suggested that this interaction could also be responsible for the phospholipase A2 inhibitory effect others have found. Three inhibitory proteins were tested herein: (i) calpactin I which is equivalent to the 36-kDa tyrosine kinase substrate first identified in chick embryo fibroblasts transformed by Rous sarcoma virus; (ii) calpactin 11, a substrate of the epidermal growth factor receptor tyrosine protein kinase; and (ii) a related 73-kDa protein which has been shown to co-purify with calpactin and partially colocalize with spectrin in fibroblasts (37). This protein is probably one of the calcimedins (38) or calelectrins (39), has also been isolated from lymphocyte membranes (40), and has the property of binding to hydrophobic resins in a Ca2+dependent manner. It may also be the 70-kDa inhibitor of phospholipase A, that has been noted in cell extracts in studies of lipocortin. When assayed for inhibition of phospholipase A2 using E. coli as substrate, all three proteins were found to be inhibitory, although at ratios far from stoichiometric with the enzyme.
In order to analyze the mechanism of this inhibition, we tested calpactins for (i) their ability to interact with enzyme and/or substrate, (ii) their ability to bind to pure phospholipid, (iii) inhibition of phospholipase A2 after preincubation of inhibitor with phosphatidylserine, (iv) dose-dependent inhibition of phospholipase A2 using lipids extracted from E. coli, and (v) inhibition of phospholipase A, as a function of substrate concentration, using both E. coli cells and extracted lipid. All three proteins were found to bind E. coli substrate and synthetic PS/PE vesicles directly. PS was capable of abolishing the inhibitory effect of calpactin I, perhaps by removing the protein from solution, and no evidence was seen for a phospholipase A,-calpactin complex.
Phospholipase A , Inhibition toward E. coli Membranes-In initial experiments using standard assays for phospholipase A2, calpactins were found to have no significant effect on synthetic phospholipid hydrolysis. However, a commonly used method to test for lipocortin-type activity involves the use of [3H]oleic acid-labeled E. coli, and when we tested the calpactins against this substrate, we found a marked inhibition of phospholipase A, in the assay. I n uztro assays for phospholipase A2 inhibition by lipocortin, which have relied primarily on the activity of an extracellular phospholipase A2 toward phospholipid in these radiolabeled autoclaved E. coli or else toward synthetic PC, have one striking feature in common.
They employ very low concentrations of a substrate on which the enzyme has unusually low activity. Phosphatidylcholine is generally 0.2 mM or less (26,41,42), while published reports of lipocortin inhibition on E. coli have not always provided exact growth conditions or specific radioactivities of the labeled phospholipid with which to reproduce substrate concentrations exactly (for example, see Ref. 29). Following the well established methods of Elsbach and co-workers (28,34), however, we obtained [3H]oleic acid-labeled cells that are probably labeled to an extent similar to that for cells used in previous lipocortin studies. The amounts used in our assay were chosen on the basis of total cpm to correspond with them, and the resulting concentration of lipid was always less than 10 pM. Whole E. coli cells did not, however, make a satisfactory substrate with which to explore the kinetics of inhibition further. For this reason, the bacterial phospholipids were quantitatively extracted and reconstituted into sonicated vesicles. On this substrate, inhibition of phospholipase A, was still observed by calpactin I so long as the substrate concentration was very low. It is noteworthy that although the calpactin inhibition was dose-dependent, the midpoint of inhibition occurred at a molar ratio of 1OOO:l calpactin to phospholipase A2 when the substrate was 5 p~, and we have observed that this ratio varies with the concentration of substrate used.
Kinetics of Phospholipase AP Inhibition-Previous reports suggested that the inhibition of phospholipase A2 by lipocortin is noncompetitive (26, 41). At least in the case of one form, calpactin I, our results are in substantial disagreement. This is seen in the sigmoidal shape of the velocity versus substrate concentration curve for the calpactin-containing sample in Fig. 7A. This sigmoid behavior could not be obtained for simple noncompetitive inhibition in which the inhibitor alters the V,,, but not the K , of the enzyme for substrate. In addition, competitive inhibition and inhibition requiring substrate plus calpactin to be the true inhibitor should both produce hyperbolically shaped curves in an experiment of this type, in contrast to the observed results. On the other hand, the sigmoidal shape observed here is that which would be expected for inhibition due to substrate depletion. Such a kinetic scheme i s shown in Equation 1.

E + S E t E S + E + P
where E is phospholipase A,, S is substrate, P is product, and I is calpactin, ES is the Michaelis complex, and SI the substrate-calpactin complex which is inactive. As in standard Michaelis-Menten kinetics, the velocity (u) is still given by Equation 2, but [SI, the concentration of free substrate is now given by where [SI, is the total substrate concentration and [IIt is the total calpactin concentration (43). The degree of sigmoidicity of the resulting velocity curve is dependent on the relative values of K, and KO as well as the concentrations of enzyme and inhibitor used in the assay. No attempt was made to determine K, and KO in this complex lipid system since these constants would in fact be complex functions of the lipid composition. However, it should now be possible to investigate this distinctive kinetic behavior on pure lipid substrates, for which these constants can be determined. Although the sigmoidal velocity curve obtained argues against simple competitive or noncompetitive inhibition mechanisms, it will be fully diagnostic for substrate depletion only if the exact velocity curve can be predicted on the basis of independently determined constants, K,, K, and V,,,.
The substrate depletion model appears the most attractive at this time, especially in light of the binding and competition studies with PS, but it is important to know why the inhibition was previously attributed to noncompetitive inhibition. This might be explained by the very low affinities of phospholipase A2 and possible differences in affinities of the calpactins for the different substrates used. For example, in the substrate dependence curve for E. coli whole cells, the same type of concentration dependence is seen as with vesicles, but an unusually Iarge concentration range has to be covered, on the order of 200-fold. If only the lower substrate concentrations (1-100 p~) had been investigated, the percent inhibition could have been construed to be the same, within error, regardless of substrate concentration. This wouId have been the expected result for noncompetitive inhibition. The reason that such a large concentration range must be covered, relative to the vesicles, in order to see depletion of inhibition could be due to different K, and/or KO values for the two substrate mixtures. A different composition and smaller quantity of phospholipids may be exposed to the surface in cells than is the case with vesicles at a given total lipid concentration (44). In fact, it has been our observation that the enzyme appears to have a 10-fold lower apparent K, for the sonicated vesicles than for the cells, although the highest overall rate achieved is approximately the same on both substrates. In a similar vein, long-chain phosphatidylcholine, the other substrate used in lipocortin studies, usually displays a I(, for porcine pancreatic phospholipase A2 that is in the millimolar range, in contrast to the micromolar apparent K , observed for phospholipase A2 on these E. coli lipid vesicles, although differences in lipid preparation can drastically affect these constants (21, 45).
It is well known that extracellular phospholipase A2 activity is highly dependent on the physical state of th substrate (21, 45). The hydrolysis products, lysophospholipid and fatty acid, when present together, cause concentration-dependent changes in the phase transition properties of aggregated phos-pholipid (46) which have been linked to inhibition or activation of phospholipase A,s, depending on the enzyme source (15, 23). In addition, some phospholipase A2s appear to possess specific binding sites for activator phospholipids. Thus, any protein which selectively binds either substrate or products is potentially capable of inhibiting or activating phospholipase A, in vitro. For instance, serum albumin has been reported as an inhibitor of porcine pancreatic phospholipase A2 (47), but this is not surprising since the pancreatic enzyme requires negatively charged groups at the lipid-water interface in order to achieve optimal activity (23, 24) and shows nonlinear time courses in the absence of albumin due to activation by fatty acid (23). On the other hand, BSA is an activator of the cobra venom phospholipase A,, which is inhibited by high concentrations of fatty acid (23). Overall, both cells and extracted E. coli phospholipids are poor substrates for pancreatic phospholipase A,. In addition to the PE and PG which appear to be the major lipids hydrolyzed, phosphatidic acid and cardiolipin are also present, both of which bind phospholipase A, but are hydrolyzed at extremely low rates (48,49). It must be noted that the activity of pancreatic phospholipase A, toward sonicated vesicles of purified PE is much higher than any activities observed here and that its natural substrates are phospholipids in bile-salt micelles, not bilayers (23).
Substrate Depletion Model for Lipocortin Inhibition-In summary, the most likely explanation for calpactin I inhibition of pancreatic phospholipase A2 in vitro is substrate depletion. The physiological significance of this mechanism of inhibition is not yet clear. It does appear that inhibition by calpactin I, and perhaps also calpactin 11, is not specific for phospholipase A, in terms of enzyme-inhibitor interactions. Rather, the inhibitor appears to be specific for the phospholipid substrate. If in fact calpactin does inhibit phospholipase A, in vivo, it may be due to interactions of calpactin with phospholipids (and possibly the cytoskeleton) that affect local membrane organization or else block access to the enzyme. But if this were the case, other lipolytic enzymes and other membrane components besides phospholipase A, would be expected also to be affected. This type of activity could explain the early observation by Hirata (26) of phospholipase C and phospholipase D inhibition by lipomodulin in vitro (26). It is still possible that crude cellular extracts contain a mixture of the calpactins and one or more other similarly sized proteins that specifically bind phospholipase A,. As the assays used until now to monitor lipocortin activity do not distinguish between lipid-binding and phospholipase A,-binding activities, investigators may have been misled, particularly in view of the relative abundance of the calpactins (which account for up to 1% of cellular protein). It is thus evident that the role of lipocortin in phospholipase A, inhibition requires further definition. Of primary importance will be the evaluation of lipocortins from different sources to see whether they too show the behavior demonstrated here for bovine lung calpactin I.