Calcium-independent Phospholipases from Guinea Pig Digestive Tract as Probes to Study the Mechanism of Lipocortin”

Two calcium-independent phospholipases isolated from guinea pig pancreas (lipase Ia, 37 kDa) and from guinea pig intestine (phospholipase B, 97 kDa) have been used to probe the mechanism of phospholipase inhibition by lipocortin. In the presence of calcium, both enzymes were inhibited by lipocortin I in a manner very similar to the inhibition of pig pancreas phospholipase A2. By using phospholipases that lack a requirement for calcium, we have for the first time been able to dissociate enzymatic activity from the role of calcium in the inhibitory process. It was found that lipocortin was without effect against phospholipase A1 and phospholipase B in the absence of calcium, under which conditions the inhibitory protein is unable to interact with anionic phospholipid surfaces. The same behavior toward phospholipase A1 was observed with two other related proteins, endonexin II or lipocortin V, and p68/67-kDa calelectrin or lipocortin VI. Together with the observation that lipocortins are active only in the presence of limited amounts of substrate, these data give further support to the "surface depletion model" of lipocortin inhibition, rather than to a mechanism involving a direct interaction between enzyme and inhibitor.

Two calcium-independent phospholipases isolated from guinea pig pancreas (lipase Ia, 37 kDa) and from guinea pig intestine (phospholipase B, 97 kDa) have been used to probe the mechanism of phospholipase inhibition by lipocortin.
In the presence of calcium, both enzymes were inhibited by lipocortin I in a manner very similar to the inhibition of pig pancreas phospholipase Az. By using phospholipases that lack a requirement for calcium, we have for the first time been able to dissociate enzymatic activity from the role of calcium in the inhibitory process. It was found that lipocortin was without effect against phospholipase A1 and phospholipase B in the absence of calcium, under which conditions the inhibitory protein is unable to interact with anionic phospholipid surfaces. The same behavior toward phospholipase A1 was observed with two other related proteins, endonexin II or lipocortin V, and p68/67-kDa calelectrin or lipocortin VI. Together with the observation that lipocortins are active only in the presence of limited amounts of substrate, these data give further support to the "surface depletion model" of lipocortin inhibition, rather than to a mechanism involving a direct interaction between enzyme and inhibitor.
Lipocortins have been defined as proteins inducible by glucocorticoids and able to inhibit intracellular phospholipase AZ, resulting in a decreased synthesis of eicosanoids such as prostaglandins or leukotrienes, and of platelet-activating factor or platelet-activating factor-acether (1). A great advance about the knowledge of these proteins has been recently achieved by elucidation of the sequence of either purified proteins (2) or of their corresponding cDNA (3). As recently reviewed by Klee (4), these studies revealed that lipocortins actually belong to a protein family which includes calpactins I and II, identical to lipocortins II and I, respectively (5-7), endonexin I, protein II or chromobindin 4 (8-g), 67-kDa calelectrin or ~68 (lo-12), anchorin CII (13,14), lipocortin III (15),and uteroglobin (16). Also included in this family is endonexin II or lipocortin V, which is recognized as an inhibitor of blood coagulation and has been isolated from human placenta and from human endothelial cells (17)(18)(19)(20)(21). All of these proteins display a high degree of homology (30-60%) first described by Geisow et al. (2), who identified a 17-amino acid consensus sequence present in the four repeats (eight in the case of ~68) forming the core of these various proteins, which also have become known by the general term of annexins (22). However, annexins differ by their N-terminal end, which is the site of specific interactions with other proteins, such as protein I, the heterotetrameric form of calpactin I (23). The N-terminal end also offers specific phosphorylation sites to various protein kinases including protein kinase C (24, 25), CAMP-dependent protein kinase (25), calmodulin-dependent kinase (25), and oncogene or growth factor-related tyrosine kinases (4). In some instances, it has been suggested that these phosphorylations modulate the antiphospholipase A, activity of these proteins (26,27).
Besides being responsible for immunological cross-reactivity (28, 29), the homologous core appears to support various common properties of annexins, especially their calciumdependent binding to negatively charged phospholipids (4). This binding to phospholipids rather than a specific interaction with the enzyme has been recently proposed as the mechanism of phospholipase inhibition (30,31). In favor of this, phospholipase AP inhibition by lipocortins can be reversed by increasing the concentration of phospholipid substrate, i.e. under conditions where the total membrane surface is no longer covered by adsorbed proteins. On this basis, Davidson et al. (30) proposed a so called "surface depletion model" to explain phospholipase inhibition by lipocortins. This is consistent with the fact that all of them have been shown to achieve the same in vitro inhibition of phospholipase AZ under similar conditions (32). Furthermore, in addition to inhibiting the low molecular mass, calcium-dependent phospholipases AZ isolated from pancreas, snake venom (see Ref. 4 for review) or rat liver mitochondria (31), lipocortin was previously shown to also inhibit C~ostridiumperfringens phospholipase C (26), phosphoinositide-specific phospholipase C (33), as well as phospholipase D from cabbage (26) or from mammalian cells (34). Finally, in addition to the placental protein isolated for its anticoagulant properties (17)(18)(19)21), four of these annexins were recently found to inhibit the conversion of prothrombin into thrombin by a complex of factors X, and V,, under conditions involving a limited amount of anionic phospholipids (32).
We previously purified from guinea pig pancreas and intestine, respectively, two cationic lipases with high phospholipase A1 activity (35) and a phospholipase AZ with lysophospholipase activity or phospholipase B (36). These enzymes are structurally unrelated to classical phospholipases AZ (Mr are 37,000 and 42,000 for the phospholipases A1 from guinea pig pancreas, and 97,000 for the intestinal phospholipase B), and they are characterized by a lack of calcium requirement. The present investigation took advantage of the latter property to study in more details the mechanism of phospholipase inhibition by lipocortins. Indeed, in contrast to previous investigations dealing with calcium-dependent processes such as blood coagulation and hydrolysis by classical, low molecular mass phospholipases AZ, guinea pig phospholipases should allow the study of the interaction of these enzymes with their phospholipid substrates under conditions where their catalytic activity is unaltered by calcium concentration. deoxycholate (1.2 mM), CaClz (5 mM) or EGTA (1 mivt), bovine serum albumin (0.125 mg/ml), and lipocortin at the indicated concentrations. After incubation for 15 min at 0 "C, the reaction was started by addition of 8 milliunits of phospholipase A, or 3.5 milliunits of phospholipase B (previously determined under optimal conditions),a followed by incubation for 1 h at 37 "C with continuous shaking.
Fatty acids were then selectively extracted according to Gatt and Barenholz (43) and radioactivity determined by liquid scintillation counting as described (36). This assay reports release of the acyl group from the en-2 position of the substrate, which is the ratelimiting step in the guinea pig intestine phosp~olipase B activity. For convenience, this will be referred to as phospholip~e B activity.
Guinea I"ig Phosphol@ses-The molecular form designated lipase Ia (Mr = 37,000), which was shown to display a high phospholipase A1 activity (37) was purified from guinea pig pancreas until apparent homogeneity, as determined by polyacrylamide gel electrophoresis under nondenaturing conditions (35). Specific activity (as measured using rat liver phosphatidylcholine under optimal conditions) was 54 pm01 X min-' X mg-'. Two different preparations of guinea pig intestine phospholipase B were alternatively used. One displayed a specific activit,y of 22.9 Fmol x min-' x mg-' and was purified to about 90% purity (as detected by sodium dodecyl sulfate-polyacrylamide gel electrophoresis), as previously described (36). Another preparation, although displaying a relatively high specific activity (20.9 pmol X min-' X mg-'), still showed several contaminating bands at 110, 64, and 55 kDa, in addition to the 97-kDa protein previously shown to support the phospholipase B activity (36).
Lipocortins-Lipocortin I was purified from pig lung as described (29, 38). Briefly, this included selective extraction with EGTA, chromatography on DEAE TSK 545 (21.5 x 150 mm), gel filtration on AcA 44, affinity chromatography on polyacrylamide gel immobilized phosphatidylserine and chromatography on hydroxyapatite. Whereas lipocortin I was detected in the flow-through fractions of the DEAE TSK 545 at pH 7.0, a 33-kDa calcium-and phospholipid-binding protein was retained on the column and eluted with 0.2 M NaCl. Further purification involved affinity chromatography on polyacrylamide-immobilized phosphatidylserine as well as anion exchange chromatography on Mono Q at pH 6.9, under which conditions the protein was eluted with 0.23 u NaCl. In contrast to its bovine homolog previously mentioned (39), the pig protein displayed a single band at 33 kDa upon sodium dodecyl sulfate-polyacrylamide gel electrophoresis and was tentatively identified as lipocortin V or endonexin II, on the basis of p~ysicochemical properties, immunoreactivity, and by the fact that it was not a substrate of protein kinase C? The 67-kDa caleleetrin, also referred to as lipocortin VI (15), was purified from pig liver using the same procedure as previously described (38).

co~i-Phospholipase
AZ activity was assayed with 0.35 milliunits of guinea pig intestine phospholipase B, using [3H]oleic acid-labeled phospholipids from E. coli as a substrate. The assay mixture was contained in a final volume of 0.2 ml and consisted of Tris-HCl (50 mM, pH 8.5), phospholipids (1 &M), CHAPS (4 mMf, CaCl, (5 mM), or EGTA (1 mM), bovine serum albumin (0.125 mg/ml), and lipocortin as indicated (0.5 to 8 pg). After incubation for 15 min at 0 "C, phospholipase B was added and reaction proceeded for 15 min at 37 "C. Lipids were extracted according to Bligh and Dyer (41), separated by thin layer chromatography (42), and determined for radioactivity as previously described (36) Determination of Phosphollpase A1 Activity-Phospholipase A1 activity was assayed with 100 milliunits of enzyme, using l-[aH]palmitoyl-2-acyl-phospholipids from Is". coli as a substrate. The incubation was carried out as described above for phospholipase B with some modifications: sodium deoxycholate was used as a detergent (2.4 mM), CaClz, and EGTA were used at 1 mM, and final incubation at 37 "C was performed for 60 min.

Effects of Lipocortin I on PhospholipuseApActivity of Guinea
Pig Intestine Phospholipase B and Comparison with Pig Pancreas Phospholipase A2---In the first experiments, the phospholipid substrate consisted of sonicated phospholipid vesicles containing l-acyl-2-["4C]oleoyl-phosphatidylcholine and brain phosphatidylserine in the molar ratio 7:3, previously used to study inhibition of both phospholipase AZ and prothrombinase (32). However, owing to the poor interaction of phospholipase B with phospholipid vesicles, no hydrolytic activity could be detected under these conditions, in agreement with our previous observation that the enzyme requires mixed micelles of phospholipids with sodium deoxycholate (3644).
As shown in Fig. 1, lipocortin I promoted a dose-dependent inhibition of both pig pancreas phospholipase A, and guinea pig intestine phospholipase B under identical conditions of assay, i.e. 1.2 mM sodium deoxycholate and 5 mM CaC12. Although the specific activities of the two phospholipases measured under these conditions were different, the two inhibition curves were remarkably parallel, suggesting a similar mechanism for the inhibition of the two enzymes. Maximal inhibition (7580%) was attained for both enzymes at 10 cLg/ ml of lipocortin I, while 50% inhibition occurred in the presence of 3.75 pg/ml of lipocortin I. However, under these conditions, phospholipase B no longer displayed the calcium independence previously described under optimal conditions of assay (36). As illustrated in Fig. 2, phospholipase B activity remained &fold lower in the absence of CaCb as compared with incubations performed in the presence of 5 mM CaCl,. But lipocortin remained without effect on the low phospholipase B activity detected in the presence of EGTA (it was even slightly stimulatory), whereas a significant inhibition of phospholipase B activity by lipocortin I was observed in the presence of calcium (Fig.  2). s 1 milliunit corresponds to 1 nmol of fatty acid liberated X mix-r-'. In order to find conditions which would allow us to detect identical phospholipase I3 activities in the presence or in the absence of calcium, the effect of sodium deoxycholate concentration on phospholipase B activity was examined. From data of Fig. 3, it is clear that sodium deoxycholate concentrations producing the highest phospholipase B activities were shifted to lower values in the presence of calcium. As a result of this, calcium appeared to increase phospholipase B activity at sodium deoxycholate concentrations up to 2.4 mM, but equally high activities were measured in the presence of calcium or EGTA at detergent concentrations above 4 mM, in agreement with our previous data (36). However, at these higher concentrations of detergent, phospholipase B was no longer inhibited by lipocortin at concentrations up to 20 pg/ml (not shown). This was probably due to a large excess of anionic lipid/water interface, as previously described by others (30,31 shown in Fig. 4, the phospholipase B inhibitory effect of lipocortin I absolutely required calcium under these conditions of assay. Half-maximal inhibition occurred at 2 pg/ml lipocortin I and a total inhibition at 40 pg/ml. In contrast, lipocortin remained without effect in the presence of EGTA. Effect of Lipocortins on Guinea Pig Pancreas Phospholipase A1 Activity-At variance with the previous experiments, assay of phospholipase A1 activity showed no effect of calcium, even when using sodium deoxycholate as a detergent. As shown in Fig. 5, phospholipase A1 activity using [3H]palmitate-labeled phospholipids from E. coli was 12.6 and 10.8 nmol X min-' X mg-' in the presence or in the absence of calcium, respectively. Here again, addition of lipocortin I resulted in a dramatic inhibition of phospholipase A, activity when CaCl* was present in the assay medium, whereas EGTA abolished the inhibitory effect of lipocortin I against phospholipase A,, whose activity was slightly enhanced (Fig. 5).
The same conditions of assay were selected to compare the antiphospholipase Ai activity of three different lipocortins: I, V, and VI, corresponding, respectively, to calpactin II, endonexin II and ~68, according to Pepinsky et al. (15). As shown in Table I, the three proteins displayed very similar if not  identical inhibition of phospholipase A, in the presence of calcium, whereas EGTA completely suppressed this effect.  Phospholipase Al activity was determined as described under "Experimental Procedures" using 1-[3H]palmitoyl-2-acyl-phospholipids from E. coli as a substrate and 2.4 mM sodium deoxycholate as a detergent, in the absence (controls) or in the presence of 8 pg of lipocortin. Relative activity is expressed as percentage of specific activity in the corresponding control. Data are means of two determinations. Here again, some stimulation of phospholipase A1 was even observed.

DISCUSSION
A first purpose of this study was to see whether two phospholipases which are different from the calcium-dependent phospholipases AZ can be inhibited by lipocortin through the same mechanism. Indeed, the two phospholipases used in the present study display great differences compared with known pancreas or venom phospholipases AZ. In addition to a higher molecular mass (37 and 97 kDa for phospholipase A1 and phospholipase B, respectively, uersus 14-15 kDa for phospholipases A*), the two guinea pig phospholipases are not resistant to heat and acid treatment, differ in substrate and positional specificity, do not require calcium for catalytic activity, and are not inhibited by bromophenacyl bromide. The latter reagent is currently used to inhibit calcium-dependent low molecular mass phospholipases A, by alkylating histidine residue 48 in the active site (45), whereas both guinea pig pancreas phospholipase A1 (35) and guinea pig intestine phospholipase B are unaffected by the same treatment.4 Thus, the observation that the two phospholipases examined here displayed the same sensitivity to lipocortin as pig pancreas phospholipase A, strongly argues in favor of an inhibitory action involving an effect on the substrate rather than a direct interaction with the enzymes, whose active site is certainly different. This further extends previous reports concerning inhibition by lipocortins of phospholipases C and D (26, 33, 34). It is also in full agreement with the fact that lipocortin I behaves as a potential anticoagulant agent by suppressing the stimulatory effect of negatively charged phospholipids on prothrombin conversion by a complex of factor X, and factor V, (32). It thus appears that lipocortin probably inhibits phospholipases upon binding to anionic phospholipid surfaces in the presence of calcium, thus competing for the phospholipid substrate, which is no longer available to the enzyme. However, the role of calcium in this mechanism cannot be explored directly with calcium-dependent target proteins such as pancreatic phospholipase A, or enzymes from the blood coagulation system.
In this respect, the use of calcium-independent phospholipases provides a unique opportunity to study the effect of calcium on the inhibitory effect of lipocortin. In other words, if lipocortin depresses the activity of phospholipases according to the substrate depletion model (30), this effect should display an absolute requirement for calcium.
Such a goal required that we find conditions where the activity of the enzymes is not influenced by the presence of calcium. As illustrated in this study, this requirement is not always evident, since calcium is able to modify the activity of phospholipases by interacting with the lipid interface, particularly when the substrate or detergent carry negative charges. Such an effect of calcium involving a modification of the lipid/water interface has been previously observed with pancreatic lipases, which do not require calcium in their active site (46-48). Our data fully support the view that this is also the case for guinea pig phospholipase A, and phospholipase B, in agreement with our previous reports (35)(36)(37)44).
Taking advantage of the observation that a similar phospholipase B identified in rat intestine can be activated by sodium deoxycholate as well as by CHAPS (49), the zwitterionic detergent allowed us to find conditions where calcium effects on the substrate dispersions was minimized, resulting in identical phospholipase B activities in the presence of CaC12 and of EGTA.
Thus, with both phospholipase A1 and phospholipase B, our results clearly demonstrate that lipocortin inhibition of phospholipases absolutely requires the presence of calcium. Altogether, our data support the view that phospholipase inhibition by lipocortins involves adsorption of these proteins at the lipid/water interface. It could be argued that such an adsorption of the proteins on the phospholipid surface is just a prerequisite, which allows direct interaction of the enzyme with its inhibitor. However, such a view is not supported by the observation that various lipocortins display very similar efficiencies in the inhibition of different phospholipases. Such a conclusion is in line with previous studies (30, 31), but it differs from that of Miele et al. (16), who suggested a more specific interaction between phospholipase AZ and lipocortin, uteroglobin (a closely related protein), or some peptides common to both proteins. The reasons for such a discrepancy are not entirely clear, but it should be recalled that the in uiuo anti-inflammatory activity reported for these peptides (16) as well as for recombinant lipocortin (50) could not be observed in a similar model with purified lipocortin (51). Also, the antiphospholipase AZ activity of peptides present in lipocortins has been recently challenged by van Binsbergen et al. (52).
In the assays with phospholipase A], not only lipocortin inhibition disappeared in the presence of EGTA, but some phospholipase activation could even be detected. At the present time, this increased phospholipase activity in the absence of calcium cannot be correlated to any of the known properties of lipocortin or other calcium-and phospholipid-binding proteins. This is certainly not due to fatty acid removal by the protein, as previously observed with serum albumin (40), since the latter protein was already present in the incubation medium. We cannot exclude the possibility that some lipids bound to the lipocortins might have some activating effect.
In conclusion, the present study gives further support to the view that lipocortins inhibit phospholipase AQ by a nonspecific effect involving interaction with anionic phospholipids present in cell membranes. This is expected to occur preferentially in the inner leaflet of the plasma membrane, which was shown to contain the majority of negatively charged phospholipids such as phosphatidylserine (53-55). Lipocortins have been found to account for up to l-2% of total proteins from some cells, such as human endothelial cells (29). However, only rough estimates can be drawn on this basis to conclude whether this is sufficient to cover all of the membrane phospholipids facing the cytoplasm of a cell. Furthermore, some recent studies failed to detect any increase of lipocortin or its mRNA in cells treated by anti-inflammatory steroids (29,51,56,57). Together with other observations that inhibition of prostaglandin synthesis by glucocorticoids does not necessarily involve a reduction of arachidonic acid mobilization from phospholipids (29, 58), these facts give some doubt about the mechanism of the anti-inflammatory action of corticosteroids, which might also modulate the expression of cyclooxygenase (59). They also leave open the question about the real physiological significance of calciumand phospholipid-binding proteins, which still remains rather obscure (4).