Binding and inhibition studies on lipocortins using phosphatidylcholine vesicles and phospholipase A2 from snake venom, pancreas, and a macrophage-like cell line.

Studies are reported on the inhibition of phospholipase A2 (PLA2) from porcine pancreas, cobra (Naja naja) venom, and the P388D1 macrophage-like cell line by human recombinant lipocortin I and bovine lung calpactin I. Membrane vesicles prepared from 1-stearoyl,2-arachidonoyl phosphatidylcholine (PC) and other PCs were utilized as substrate. Binding studies using sucrose flotation gradients showed that both lipocortin I and calpactin I bind to these vesicles although less tightly than to vesicles prepared from anionic phospholipids or fatty acids. Binding to PC was somewhat enhanced by Ca2+. Inhibition of cobra venom PLA2 was not observed when PC vesicles were used as substrate but was when dipalmitoyl phosphatidylethanolamine was used. Both the pancreatic and macrophage enzymes were inhibited when acting on PC. Interestingly, the inhibition of the macrophage enzyme toward PC depended on the fatty acid attached to the sn-2 position of PC with arachidonate greater than oleate greater than palmitate. Inhibition was also highest at low [PC]; these inhibition results can be explained by the "substrate depletion model" (Davidson, F. F., Dennis, E. A., Powell, M., and Glenney, J. (1987) J. Biol. Chem. 262, 1698-1705). Experimental and theoretical considerations suggest that the in vitro inhibition by lipocortins of this macrophage PLA2 from a cell that makes lipocortin and is active in prostaglandin production is due to effects on substrate availability rather than direct inhibition.

Experimental and theoretical considerations suggest that the in vitro inhibition by lipocortins of this macrophage PLA2 from a cell that makes lipocortin and is active in prostaglandin production is due to effects on substrate availability rather than direct inhibition.
Recent progress in the structural characterization of lipocortins I-VI (1) has shown that the species equivalents of lipocortin I and calpactin I (lipocortin II) are present in rat lymphoid organs and are expressed in murine macrophage cell lines including 5774 and P388P.l. Following our previous studies on "substrate depletion" (2), we have now investigated the ability of human recombinant lipocortin I and bovine lung calpactin I to inhibit phospholipase A2 (PLA*)' porcine pancreas, cobra venom (Naju nuju), and a partially purified PLA, from the mouse macrophage-like cell line P388D1 (4), using phosphatidylcholine (PC) vesicles as substrate. The ability of calpactin I and lipocortin I to bind to PC and to fatty acid was also investigated.
Studies of the effects of "lipocortins" on porcine pancreatic PLAZ hydrolysis of Escherichia coli phospholipids or PC in deoxycholate mixtures have indicated that PLAP inhibition in those systems may be accounted for by depletion by the inhibitor of either substrate, or the cofactor of the enzyme, and may be complicated by inhibitor-induced phase changes in the substrate (2,5). However, there have also been reports of inhibition of the pancreatic PLA, by various lipocortins when PC is used as substrate (6)(7)(8)(9)(10) or with PE using several different PLA*s (11). Although the physical form of the phospholipid was not specified in any of these studies, neither detergents nor sonication was mentioned, and it can therefore be assumed that the phospholipids were in some kind of multibilayer structure. In one case (9), the essential cofactor Ca*+ was not included in the assay, making it likely that the lipocortin inhibited solely by binding up the trace amounts of cofactor. However, the other reports were intriguing since calpactin I and lipocortin I have been reported not to bind PC (12,13).
Lipocortins are proposed to be important in in uiuo antiinflammatory reactions due to the fact that purified proteins in this family added to prelabeled cells and tissues appear to decrease the release of [3H]arachidonic acid or thromboxane in response to stimuli (14,15). The mechanism of PLA, inhibition by these proteins in the only other commonly used in uitro assay system (phospholipid vesicles) therefore warranted further study. In particular, if it could be ascertained whether in vitro inhibition always involves depletion effects or sometimes works by a different mechanism then perhaps more progress could be made toward quantitating the effect and determining the likely relevance of these proteins as inhibitors of PLA2s as discussed in a recent commentary (16). In the present studies, in all cases in which inhibition was observed, the bulk concentration of substrate was very low, confirming our earlier findings with E. coli substrate (2) and subsequent studies by Aarsman et al. (11). Furthermore, the potency of inhibition of the macrophage enzyme depended on which type of long chain PC was employed as substrate, and the inhibition could be overcome by raising the PC concentration. Inhibition of the N. naja PLA, toward PC, its preferred substrate, was not seen although the pancreatic enzyme was inhibited.
However, when other substrates were used on which the N. With the cobra venom PLA,, the reaction was allowed to proceed for 2 min at 40 "C, and hydrolysis was 2-4% above background (<I%). All reactions were stopped by the addition of 300 ~1 of chloroform/methanol/acetic acid (157:114:29, v/v), and the chloroform layers containing substrate and products were removed, dried in uucuo, and subjected to thin layer chromatography as described below. Assays

RESULTS
Binding to Phosphatidylcholine and Fatty Acid-The binding of calpactin I and lipocortin I to sonicated PC vesicles and to free fatty acid was tested using sucrose flotation gradients similar to those described previously (2) for PS/PE liposomes. However, the PC vesicles were not as buoyant as PS/PE liposomes, as might be the result of the lower propensity of PC to form very large fused phases. Therefore, it was necessary to centrifuge the PC-containing gradients for longer times (6 h instead of 2 h) in order to obtain all of the phospholipid in the top several fractions of the tubes. Calpactin I and lipocortin I stayed in the bottom two or three fractions of the gradients in the absence of lipid (data not shown). In pilot assays, calpactin I was tested for binding to PC at a phospholipid concentration equal to that used with PS/PE liposomes in previous experiments (2) (1.2 mM, 50 pg of total phospholipid/gradient).
However, under conditions in which calpactin I would have been expected to bind tightly to PS/PE liposomes, PC caused smearing of the protein throughout the gradient (data not shown). This is indicative of a looser binding phenomenon than was seen with PS. The concentration of PC was therefore raised, and at a IO-fold higher amount of PC (Fig. l), both calpactin I and lipocortin I were found in discrete zones in the gradients, with approximately 30% of the protein associated with PC and the remainder at the bottom of the tube.
The effect of [Ca"] on the binding of calpactin I to lstearoyl,2-arachidonoyl PC was also investigated. When EGTA was present, very little protein was bound to PC ( Fig.  2A). There appeared to be a weak dependence on [Ca'+] because as it was increased (B and C), so too was the binding of calpactin to a maximum of about 30% at 10 mM Ca'+. In all the binding experiments, the preparation of the gels for densitometry was done by silver staining. All the fractions of a single gradient were run together in one gel in order to assess their relative intensities because the extent of development of the silver stain could vary in different preparations, and the absolute intensities of bands containing the same amount of protein sometimes differed in gels developed at different times. Interestingly, the light chain of calpactin I always developed color much more quickly than did the heavy chain (the light chain is not stained at all by Coomassie Blue). Therefore, if the amount of heavy chain was very low in a Calpactin I (A) and lipocortin I (B) (4 pg/ gradient) were preincubated with sonicated phosphatidylcholine vesicles (535 pg/gradient) in 10 mM imidazole, pH 6.8, 40 mM KCl, 2 mM MgCl*, and 10 mM CaCl* before being adjusted to 50% sucrose, overlaid with a 40-O% sucrose gradient, and subjected to ultracentrifugation for 6 h at 4 "C. l-Stearoyl,2-[1-"'Clarachidonoyl phosphatidylcholine (x) floated to the top of the gradients, and calpactin I heavy chain (O), light chain (0), and lipocortin I (W) partitioned between the bottom of the tubes (fractions l-3) and the phosphatidylcholine-containing fractions at the top. The phospholipid and relative protein contents of each fraction were determined by liquid scintillation counting and SDS-polyacrylamide gel electrophoresis, silver staining, and densitometry, respectively. Preincubation of calpactin I with l-stearoyl,2-[ 1-Wlarachidonoyl PC, sucrose flotation gradients, and analysis of fractions was carried out as described in Fig. 1 except that the buffers contained A, 10 mM EGTA; B, 2 mM Ca'+; or C, 10 mM Ca'+. X, ["Clphosphatidylcholine; 0, calpactin I heavy chain; 0, calpactin I light chain. given lane, it may not have been possible to visualize it before overdevelopment of the rest of the gel became imminent, and the reaction was stopped. This may account for some of the apparent variations in heavy to light chain ratios as the [Ca"] was raised in the gradients, and more protein became associated with the upper phosphatidylcholine-containing fractions. In marked contrast to the extent of binding of both proteins to phosphatidylcholine, binding of calpactin I and lipocortin I to free fatty acid under the same conditions ( Fig. 3) was complete. In the experiment shown in Fig. 3, palmitic acid was used at a molar concentration equal to that of the phosphatidylcholine used in the previous experiments. Oleic acid gave similar results, but arachidonic acid was retained at the bottom of the gradients with calpactin I (data not shown). Although binding of calpactin I and lipocortin I to palmitic acid at 10 mM Ca2+ was more extensive than to PC, it was probably not as efficient as to PS/PE liposomes, judging from preliminary experiments at lower concentrations of fatty acid. Lipocortin Effects on the Hydrolysis of Sonicated Phosphatidylcholine Vesicles by the Secreted Phospholipases AZ-In Fig. 4, dose-response curves are shown for calpactin I inhibition of pancreatic and cobra venom (IV. naja) PLAP toward 10 pM 1-stearoyl,2-arachidonoyl PC vesicles. Initial activation followed by inhibition was seen with the pancreatic enzyme. However, the cobra venom PLAz was not inhibited at all under the conditions used and was, if anything, slightly activated. The apparent IC& for the pancreatic PLAZ was approximately 10 rg/ml or 0.1 FM calpactin I holoprotein, and the at Biomedical Library, UCSD on December 23, 2020 http://www.jbc.org/ Downloaded from enzymes were each present at 0.5 rig/ml or 0.94 nM. For the pancreatic PLA2, reactions were allowed to proceed for 60 min at 40 "C, resulting in 5.8 f 0.2% hydrolysis in the PLAz controls. The cobra venom PLAz is much more active than the pancreatic enzyme on PC, and after 2 min, hydrolysis in the controls was 1.8 + 0.3%. This would indicate a difference in rate of approximately lo-fold, consistent with the different rates for these enzymes typically observed with PC under other conditions (e.g. Ref. 22). It should be noted, however, that under the conditions used here, control time courses of hydrolysis by the cobra venom enzyme were not linear but curved over, reaching a maximum of about 7.5% hydrolysis in about 7.5 min. The pancreatic PLAp time courses showed a brief lag period and then were roughly linear until a little over 6% hydrolysis after which they too curved over.
Although the calpactin I did not inhibit the cobra venom PLA, toward phosphatidylcholine, it was able to inhibit this enzyme when other substrates were used (Table I). When sonicated dipalmitoyl PE was used as the substrate, under conditions in which the cobra venom and pancreatic PLA*s gave approximately equal rates of hydrolysis, a 39% inhibition of the cobra venom enzyme was seen (Table I), whereas the pancreatic PLA, was 54% inhibited (data not shown). The cobra venom PLA2 was also inhibited by calpactin I and lipocortin I when 3H-labeled E. coli cells were used as substrate. Therefore, the lack of inhibition of the cobra venom PLAY in Fig. 4 cannot be taken to imply a unique specificity of the calpactin I for the pancreatic enzyme but rather shows that the inhibition depends on which substrate is used for a given PLA*. Notably, the cobra venom enzyme was not inhibited toward the substrate on which it normally gives its best rates, whereas the pancreatic enzyme, for which PC is a poor Effects on PL& from P388D1 Cells-Both lipocortin I and calpactin I were found to inhibit a partially purified PLA, from the mouse macrophage-like cell line P388D1. The substrate used in Fig. 5 was 10 PM sonicated 1-stearoyl,2-arachidonoyl PC, and the approximate I& of both inhibitors was 5 pg/ml (approximately 0.1 PM in 38-kDa chains). Fig. 6 shows the effect on the inhibition of the macrophage PLAz by calpactin I when different PCs were used as substrates. At the highest concentration of calpactin I used, less than 20% inhibition was seen using dipalmitoyl PC as substrate. About 80% inhibition was seen with that same amount of calpactin I and at that same concentration of substrate when l-palmitoyl,2-oleoyl PC and l-stearoyl,2-arachidonoyl PC were used. However, the ICsO in the presence of oleoyl-containing PC was higher than that observed with the more unsaturated, longer chain arachidonoyl-containing phosphatidylcholine: 13 @g/ml (0.3 PM 38-kDa chain) and 3.3 pg/ml(O.O9 PM), respectively.
The dependence of the inhibition on substrate concentration was examined for PLA, hydrolysis of l-stearoyl,2-arachidonoyl PC at three different concentrations of calpactin I, and the results are shown in Fig. 7. In this experiment, the concentration of calpactin I was held constant as the concentration of substrate was raised. In a previous report (21) it was shown that the macrophage enzyme exhibits different types of kinetics depending on the range of substrate concentrations used. In the concentration range used here (Cl0 PM), the PLAz activity was shown previously to give hyperbolic velocity uersus S curves, and Michaelis-Menten kinetics were assumed. At concentrations above 10 PM substrate, the velocity uersus S curves indicated cooperativity of some sort. In the experiment shown in Fig. 7, the actual velocity curves are shown in panel A and are replotted as the double reciprocals in panel B. Within the accuracy of the assay and given the uncertainties of the macrophage PLA, mechanism, it is not possible to distinguish whether or not the inhibited samples give genuinely linear double-reciprocal plots that would fit better to Michaelis-Menten kinetics than other models (see "Discussion").
Similar substrate dependence of the inhibition by calpactin I can also be seen when 1-palmitoyl-2-oleoyl PC is used as substrate, as shown in Fig. 8  conditions in which PS was bound (13,14), we have found that the proteins do bind to PC albeit with a lower affinity than is exhibited for PS. At a lo-fold greater concentration of PC than was used with PS and using a &fold lower concentration of protein and much higher [Ca'+], 30% of the protein was bound to lipid compared with 100% when PS/PE liposomes were used (2). The results obtained are consistent with those of Schlaepfer and Haigler (12), who looked for but did not find binding of iZ51-lipocortin I to large thin walled PC vesicles using sedimentation.
At the concentration of PC and Ca2+ used by these authors, significant binding would not have been detected with sonicated PC vesicles in the sucrose flotation assay either, although it should be noted that the calpactins may have different affinities for PC in different structures. These results imply that the binding of lipocortin I and calpactin I to PC-sonicated vesicles is weaker than to anionic lipids. A dependence for binding on the concentration of Ca*+ was also seen with PC but over a millimolar rather than micromolar range. The Ca2+ requirement of calpactin I has been shown to be dependent on the concentration of PS/ PE liposomes (13), and at high enough lipid (about half of the concentration of PC used in Fig. l), no Ca2+ is required for binding to that phospholipid.
It has also been observed that the Ca2+ affinity of lipocortin I is increased by PS but not by PC (12). Under conditions in which PS increases the affinity of the calpactin for Ca2+ and vice versa, PS/Ca*+ complexes are probably formed. PC, as a zwitterion, does not bind Ca2+ to its surfaces as tightly or as extensively as does PS, nor does Ca" induce phase changes of PC. This difference may be at the root of the lower affinity of the lipocortin protein for PC as well as the broader Ca2+ concentration dependence for binding.
It is possible that the nature of the calpactin-phospholipid complex differs for PC and PS or else is the same but harder to assume with PC. Calpactin I and lipocortin I cause aggregation of PS/PE liposomes (13,18), resulting in an extended protein/phospholipid phase that sediments easily from aqueous solution upon centrifugation and floats in sucrose solutions.
The aggregation phenomenon is not surprising since negatively charged phospholipids have a natural tendency to form nonbilayer phases such as hexagonal and cubic phases under a variety of conditions. Sonicated PC vesicles, on the other hand, are generally small, far less easily aggregated, and cannot be quantitatively sedimented in the absence of protein even by ultracentrifugation.
It is presumably because of their smaller size, resulting in higher relative densities, that they float less easily than PS in sucrose gradients. Binding of calpactin to sonicated PC vesicles may not induce Lipocortin Binding and Inhibition 5607 aggregation of the vesicles, in which case binding might not be detectable by either light scattering or sedimentation from aqueous solution, and this could account for the results of previous studies in which those detection methods were used (13). Also, if no significant aggregation occurs, the protein-PC complex may tend to be more dense than allowable for flotation of both components. Thus, as the centrifugal force displaces the light lipid phase upward in the gradient, the normal on/off equilibrium of the protein bound to lipid might be shifted more to "off," as the buoyancy of the lipid is increased upon dissociation of the protein. Thus, with PC it may be necessary to have a higher total ratio of phospholipid to bound protein than is the case with PS in order to prevent dissociation of the protein as it rises through the gradient. This could explain the smearing of protein at lower PC:calpactin I ratios. For reasons such as these, another assay for PC binding is needed in order to obtain Ko values.
Fatty Acid Effects-The ability of calpactin I and lipocortin I to bind to free fatty acid may influence the kinetic results of PLA? assays. Therefore, inhibition of porcine pancreatic PLA, by lipocortin binding of fatty acid, which is the enzyme's activator on PC, was considered but seemed unlikely in the assays performed here because of the low percent hydrolysis in all assays. The macrophage PLA:! is inhibited by free fatty acid not activated by it (23), and it was also inhibited by lipocortin.
However, the potential for fatty acid binding must be considered at the higher percentages of hydrolysis which are frequently reported for control PLA, samples. Furthermore, the ability to bind fatty acids and their metabolites may impinge on the choice of techniques for the measurement of levels of these molecules in any system in which lipocortins are present.
Substrate Dependence of the Inhibition by Lipocortins-The inhibition reported herein of pancreatic and macrophage PLA, hydrolysis of PC vesicles supports previous reports of inhibition of pancreatic PLA+ by calpactin I and uteroglobin on PC substrates (6,10). However, it is evident from the results with pancreatic, cobra venom, and the macrophage PLA+ that the degree of inhibition depends on which substrate is used and its concentration.
The Therefore, if KO, K,, and V,,,., are established, then the velocity curve as a function of substrate concentration can be calculated. In Fig. 9, we have held the Vmsx and K, constant at arbitrarily chosen values and plotted the velocity curves for four different relative KO values in order to illustrate the effect of these ratios on the shapes of the inhibition curves as a function of substrate concentration.
As can be seen, as the inhibitor's affinity for substrate (KO) approaches the enzyme's K, for substrate, the potency of the inhibitor decreases, and the shape of the inhibited curve becomes less and less sigmoidal, to the point where the plots resemble competitive inhibition.
If KO is sufficiently high on a given substrate relative to the enzyme's K,, then inhibition may not be observed unless massive quantities of inhibitor are used. Such phenomena could explain the kinetic plots obtained for inhibition of the macrophage PLAz toward different PCs.
On the other hand, in the PC assay system, in which there is less affinity of the lipocortins for the substrate and in which more lipocortin or calpactin must be used in order to obtain inhibition than was the case with anionic lipids, competitive protein binding could be important.
The with such a hypothesis. However, even though the shape of the kinetic curves in Fig. 7 may be consistent with active site competitive inhibition, they are also consistent with a substrate depletion mode of inhibition in the case in which KO = K, as just described, particularly given the error of the assays for the PLA*s and limitations on the amounts of inhibitor protein available. Interestingly, the Km values of the macrophage PLA, on all the PCs used herein are on the order of lo-" M (21).
Observations that support the hypothesis of a direct phase effect on substrate by calpactin I and lipocortin I include the following.
First, association of these proteins with PC has now been demonstrated under some conditions. Furthermore, significant binding under assay conditions would require a KD of only about 10V6 M, which is conceivable by comparison with other's estimations of KD values for PS (25) and the relative affinities observed here. In addition, lipocortin by itself has been reported to have some surfactant properties (cited in Ref. 1 as a personal communication).
If this is so, then it is logical to assume that partitioning into lipid bilayers is possible. If that happened, then as the PC:lipocortin ratio was raised, the protein would become diluted in the surface, and localized phase effects could diminish. Surfactant-like effects could also explain the often observed activation of PLAz by lipocortin I and calpactin I in some concentration regimes. That is, phase changes have the potential either to activate or to inhibit PLA*s depending upon the enzyme and substrate under study (26)(27)(28)(29).
Another line of evidence supportive of lipocortin-induced substrate effects concerns the stoichiometries of protein to phospholipid when inhibition is achieved. The distinction between dissociation constants and stoichiometries should first be noted, however. In competitive inhibition, it is the concentrations of all three components S, E, and 1, and the values of K, and KI that determine the extent of inhibition.
Similarly, in substrate depletion inhibition, it is the concentrations of the three components and K, and Ko. In neither case is stoichiometry the main operative factor, but rather it is the absolute concentrations relative to the appropriate dissociation constants that are definitive. Second, it must be noted that the concentration of phospholipid added to an assay is not the concentration that the enzyme initially "sees" unless a solubilizing detergent is added. The maximum amount of surface phospholipid in vesicles will be as much as 60% of the total if small single-walled vesicles are used or as low as 10% or less of the total if multimellar vesicles (liposomes) are used. Thus, if concentrations are to be compared from one assay to the next, it is important to state the method of preparation of the phospholipid and what kind of vesicles were obtained. (The PC used in this study was sonicated and therefore is expected to be a mixture of small multilamellar and unilamellar vesicles.) Given these caveats, it appears that in all the PC systems of which we know in which a purified lipocortin-type protein was used and Ca*+ was in excess (this study and Refs. 6 and lo), the maximum probable surface PC/total inhibitor molar ratio at the I&o was as low as stoichiometric and not higher than 8:l. In making this calculation, the extreme case was assumed in which 60% of the lipid was on the surface. This recurring range of stoichiometries at the ICso values lends weight to the argument that inhibition may be due to a direct effect on substrate structure. Considering the large size of these proteins compared with lipid, molecular weight 38,000 uersus about 300-800, it is not unreasonable to assume that the binding of one lipocortin I or calpactin I molecule could effectively cover many more than one lipid headgroup in the surface at any given instant. However, the stoichiometries