Inhibition of cytosolic phospholipase A2 by annexin I. Specific interaction model and mapping of the interaction site.

Annexins (ANXs) display regulatory functions in diverse cellular processes, including inflammation, immune suppression, and membrane fusion. However, the exact biological functions of ANXs still remain obscure. Inhibition of phospholipase A(2) (PLA(2)) by ANX-I, a 346-amino acid protein, has been observed in studies with various forms of PLA(2). "Substrate depletion" and "specific interaction" have been proposed for the mechanism of PLA(2) inhibition by ANX-I. Previously, we proposed a specific interaction model for inhibition of a 100-kDa porcine spleen cytosolic form of PLA(2) (cPLA(2)) by ANX-I (Kim, K. M., Kim, D. K., Park, Y. M., and Na, D. S. (1994) FEBS Lett. 343, 251-255). Herein, we present an analysis of the inhibition mechanism of cPLA(2) by ANX-I in detail using ANX-I and its deletion mutants. Deletion mutants were produced in Escherichia coli, and inhibition of cPLA(2) activity was determined. The deletion mutant ANX-I-(1-274), containing the N terminus to amino acid 274, exhibited no cPLA(2) inhibitory activity, whereas the deletion mutant ANX-I-(275-346), containing amino acid 275 to the C terminus, retained full activity. The protein-protein interaction between cPLA(2) and ANX-I was examined using the deletion mutants by immunoprecipitation and mammalian two-hybrid methods. Full-length ANX-I and ANX-I-(275-346) interacted with the calcium-dependent lipid-binding domain of cPLA(2). ANX-I-(1-274) did not interact with cPLA(2). Immunoprecipitation of A549 cell lysate with anti-ANX-I antibody resulted in coprecipitation of cPLA(2). These results are consistent with the specific interaction mechanism rather than the substrate depletion model. ANX-I may function as a negative regulator of cPLA(2) in cellular signal transduction.

Annexins (ANXs) 1 are structurally related, calcium-dependent, phospholipid-binding proteins that have been implicated in diverse cellular roles, including anti-inflammation, membrane fusion, differentiation, exocytosis, calcium channels, and interaction with cytoskeletal proteins (reviewed in Refs. 1 and 2). These proteins are defined structurally by a conserved core domain that contains either four or eight repeating units of ϳ70 amino acids each (3,4). The conserved repeats account for the shared abilities of ANXs to bind phospholipids in a calciumdependent manner, whereas the specific functions of each ANX are probably related to their type-specific N-terminal regions. Despite definitive structural characterization, the relationship between structure and function or precise biological function has not been well defined for any of the ANXs.
ANX-I, a 37-kDa member of the family, has been proposed as a mediator of the anti-inflammatory actions of glucocorticoids (5). These anti-inflammatory properties have been related to the ability of ANX-I to inhibit phospholipase A 2 (PLA 2 ) activity. PLA 2 represents a growing family of enzymes with the common function of catalyzing the release of fatty acids from the sn-2position of membrane phospholipids, thereby providing production of bioactive lipid metabolites and cytoprotective functions (6,7). PLA 2 enzymes can be subdivided into several groups based on their structure and enzymatic characteristics. Secretory PLA 2 (sPLA 2 ) enzymes are low molecular mass (14 -18 kDa) enzymes with little fatty acid specificity that require a millimolar calcium concentration for catalysis. Types IIA and V sPLA 2 isozymes are known to play a role in arachidonic acid release by certain stimuli (8). On the other hand, type VI Ca 2ϩ -independent PLA 2 has been proposed to participate in fatty acid release associated with phospholipid remodeling (9,10). In contrast, type IV cytosolic PLA 2 (cPLA 2 ) is a ubiquitously distributed 85-100-kDa enzyme, the activation of which has been shown to be tightly regulated by growth factors and pro-inflammatory cytokines. cPLA 2 requires a submicromolar Ca 2ϩ concentration for effective hydrolysis of its substrate, arachidonic acid-containing glycerophospholipids (11,12). This requirement is associated with the C2 domain in the N terminus of cPLA 2 that mediates calcium-dependent phospholipid binding and translocation of cPLA 2 from the cytosol to membranes (13). In addition to the calcium-dependent translocation, cPLA 2 is phosphorylated by kinases of the mitogen-activated protein kinase family (14), which is one of the important regulatory mechanisms for in vivo activation of cPLA 2 (15,16).
The mechanism by which ANX-I inhibits PLA 2 is not fully understood. Most studies, which have been performed using a 14-kDa sPLA 2 , supported the "substrate depletion" model rather than the "specific interaction" model (17). In the presence of calcium, ANX-I tightly binds to negatively charged phospholipid substrates, which results in substrate depletion and apparent cPLA 2 inhibition (18). To the contrary, our recent study using cPLA 2 isolated from porcine spleen showed that ANX-I inhibited cPLA 2 by specific interaction (19).
An increasing number of reports have suggested that cPLA 2 is a key enzyme responsible for signal transduction in inflammation, cytotoxicity, and mitogenesis (6,7). ANX-I suppresses cPLA 2 activity not only in vitro (21,22), but also in cultured cells (23,24). Thus, ANX-I may function as an endogenous negative regulator of cPLA 2 . Herein, we have studied the inhibition of cPLA 2 by ANX-I in detail. Deletion mutants of ANX-I were constructed, and enzymatic studies were performed. Also, the protein-protein interaction between cPLA 2 and ANX-I, a prerequisite for the specific interaction mechanism, was investigated.
Preparation of ANX-I Deletion Mutants-Cloning of ANX-I cDNA and expression in E. coli have been described (3,25). Briefly, ANX-I cDNA was selected by colony hybridization from a human placenta cDNA library. ANX-I cDNA was then cloned into plasmid pET-28a(ϩ) (Novagen, Madison, WI) and expressed in E. coli. Full-length ANX-I and N-terminally deleted ANX-I were cloned into the NcoI and SalI sites of pET-28a by cloning procedures that do not involve PCR. The deletion mutants ANX-I-(1-274) and ANX-I-  were cloned by PCR amplification of the DNA fragment, followed by insertion into the NcoI and SalI sites of pET-28a. The DNA fragment encoding amino acids 1-274 was amplified using primers TTAccatggCAATGGTATCAG and TTAgtcgacTCATTTGCTTGTGGCGCA (lowercase letters represent the restriction enzyme sites). The DNA fragment encoding amino acids 1-196 was amplified using primers TTAccatggCAATGGTATCAG and TTAgtcgacTCATTCATTCACACCAAA. PCR-amplified clones were verified by nucleotide sequencing. ANX-I and the mutants were purified according to methods previously described (25).
All PCR-amplified clones were verified by nucleotide sequencing. GST fusion proteins were expressed and purified according to the manufacturer's instructions. The E. coli lysate was bound to a glutathione-Sepharose 4B column (Amersham Pharmacia Biotech) and washed three times with phosphate-buffered saline. GST fusion proteins were eluted with 10 mM reduced glutathione in 50 mM Tris-HCl (pH 8.0) and dialyzed against Tris-buffered saline containing 10% glycerol. The protein concentration was determined by the Bradford method (26).
Preparation of PLA 2 Enzymes-Bee venom sPLA 2 was purchased from Sigma. A 100-kDa cPLA 2 was partially purified from porcine spleen according to previously described methods (27). Since purification of cPLA 2 from porcine spleen is laborious and time-consuming, we cloned cPLA 2 cDNA into the baculovirus vector pFastBacHTa (Life Technologies, Inc.) and produced cPLA 2 in Sf9 cells. cPLA 2 cDNA was cloned into plasmid pGEMT (Promega, Madison, WI) by reverse transcription-PCR from U937 cell mRNA using primers ATTgtcgacATGT-CATTTATAGATCC and TAAaagcttCTATGCTTTGGGTTTACTTAG.
The nucleotide sequence was verified by DNA sequencing. Plasmid pGEMT-cPLA 2 was digested with SalI and HindIII, and the insert was cloned into the SalI and HindIII sites of pFastBacHTa to produce pBac-cPLA 2 . To induce transposition between pBac-cPLA 2 and Autographa californica nuclear polyhedrin virus DNA, pBac-cPLA 2 was transformed into E. coli DH10Bac (maximum efficiency, Life Technologies, Inc.), which harbors A. californica nuclear polyhedrin virus bacmid. Transformed E. coli was incubated on LB plates containing 50 g/ml kanamycin, 7 g/ml gentamycin, 10 g/ml tetracycline, 100 g/ml 5-bromo-4-chloro-3-indolyl ␤-D-galactopyranoside (X-gal), and 40 g/ml isopropyl-␤-D-thiogalactopyranoside for 24 h. Recombinant bacmid PLA 2 -containing white colonies were then isolated. Bacmid cPLA 2 was transfected into Sf9 cells using CellFectin (Life Technologies, Inc.) and cultured for 72 h, and recombinant virus cPLA 2 production was confirmed by PCR. The titer of the recombinant virus cPLA 2 in the culture medium was determined by a plaque assay. Fresh Sf9 cells were infected with this culture medium and cultured for 72 h. Cells from a 50-ml culture were lysed in 1 ml of buffer containing 20 mM Tris-HCl (pH 7.4), 10% glycerol, 1% Nonidet P-40, 0.1% BSA, 0.1 mM phenylmethylsulfonyl fluoride, Complete TM EDTA-free protease inhibitor mixture (Roche Molecular Biochemicals), and 0.1 M Ca 2ϩ and used as the source of Sf9 cPLA 2 .
Assay of PLA 2 Activity-PLA 2 activity was assayed using sonicated liposomes prepared as described previously (19,28). A stock solution of the substrate was prepared as follows. The substrate (10 -20 nmol) was dried under nitrogen and then suspended in 0.5-1.0 ml of distilled water by sonication (3 ϫ 10 s) in a bath-type sonicator (Ultrasonik 300, The J. M. Ney Company, Broomfield, CT). The standard reaction mixture (200 l) for the PLA 2 assay contained 0.33 nmol (1.65 M) of radioactive substrate (ϳ39,000 cpm), 200 g of fatty acid-free BSA, and 10 ng (or an equivalent amount when partially purified enzyme was used) of PLA 2 in 75 mM Tris-HCl (pH 7.5). Ten ng of purified cPLA 2 yielded ϳ3000 cpm of the product under the standard conditions with 1 M Ca 2ϩ . When partially purified porcine cPLA 2 or the total cell lysate of Sf9 cPLA 2 cells was used, the amount of cPLA 2 was estimated from the activity. The reaction was started by addition of the enzyme to the reaction mixture. Assays were incubated at 37°C for 1 h and then stopped by adding 1.25 ml of 2% NH 2 SO 4 , 20% n-heptane, and 78% isopropyl alcohol. Non-esterified fatty acid was extracted as follows. First, 0.55 ml of water was added, and the sample was Vortex-mixed and centrifuged at 5600 ϫ g for 5 min. Then, 0.75 ml of the upper phase was transferred to a new tube, to which 100 mg of silica gel and 0.75 ml of n-heptane were added. The samples were Vortex-mixed and centrifuged again for 5 min. The supernatant was dried using a SpeedVac freeze drier, and the lipid was resuspended in chloroform/methanol (1:1, v/v) that contained unlabeled arachidonic acid (1 g/l) in methanol. Phospholipid and neutral lipid were separated by migration on layers of Silica Gel 60 F 254 plates (Merck, Darmstadt, Germany) in petroleum ether/ethyl ether/acetic acid (80:20:1, v/v/v). After drying, the plates were subjected to iodine vapor, and lipids were identified by their comigration with unlabeled arachidonic acids. Products were quantified by scraping their corresponding spots into counting vials containing 2 ml of Aquasol-2. Radioactivity was determined using a Packard Tri-Carb scintillation spectrophotometer. For analysis in which the substrate concentration dependence was determined, unlabeled phospholipid was added to the labeled phospholipid to produce a designated final concentration. For accurate control of the Ca 2ϩ concentration, a CaCl 2 /EGTA buffering system was used (29). In all analyses, samples were tested in triplicate and adjusted for nonspecific release by subtracting a control value in which preparation of the enzyme was omitted. For inhibition assays, 5-100 nM ANX-I was added to the reaction mixture.
Effect of ANX-I on PLA 2 Activity-All analyses were performed in triplicate and repeated at least three times. The effect of ANX-I was represented by the percentage of cPLA 2 activity compared with the control value. All data shown are means Ϯ S.E. The effect of ANX-I and its deletion mutants on cPLA 2 activity was determined by the percentage of PLA 2 activity using the following equation: % of PLA 2 activity ϭ (cpm test/cpm control) ϫ 100.
Mammalian Two-hybrid Analyses-Portions of cPLA 2 and ANX-I cDNAs were cloned into the mammalian version of the bait and prey vectors, pM for GAL4 fusion and pVP16 for VP16 fusion (CLONTECH, Palo Alto, CA). To generate GAL4 fusion, ANX-I cDNA was subcloned into the BamHI and XbaI sites (full-length and C-terminal deletion mutant) or the EcoRI and XbaI sites (N-terminal deletion mutant) of pM. To generate VP16 fusion, cPLA 2 cDNA was subcloned into the SalI and HindIII sites of pVP16. Portions of cPLA 2 cDNA containing amino acids 1-80 or 81-793 were amplified with primers ATTgtcgacATGTC-ATTTATAGATCC and TAAaagcttATCCAAAATAAATTCAAA or, in the case of amino acids 81-793, primers ATTgtcgacCTTAATCAGGAAAA-TGTT and TAAaagcttTGCTTTGGGTTTACTTAG. The pG5CAT reporter was purchased from CLONTECH. Chloramphenicol acetyltransferase assays (Promega) were performed according to the manufacturer's instructions.
In Vitro Protein Binding of cPLA 2 -⌬C2 and ANX-I-The 43 amino acids spanning amino acids 38 -80 of the C2 domain of cPLA 2 (⌬C2) were amplified by PCR using cPLA 2 cDNA as a template and two oligonucleotide primers: ATAggatccATGCTTGATACTCCA and TA-AaagcttATCCAAAATAAATTCAAA. After digestion with BamHI and HindIII, the amplified fragment was inserted into the BamHI and HindIII sites of pMAL-P2X, a maltose-biding protein (MBP) fusion vector (New England Biolabs Inc., Beverly, MA), to produce pMAL-⌬C2. The MBP fusion protein MBP-⌬C2 was expressed in E. coli and purified using amylose resin (New England Biolabs Inc.) (30) according to the manufacturer's instructions. The binding mixture contained 2 g of either ANX-I or deletion mutant, 2 g of MBP-⌬C2, and 20 l of amylose resin in 300 l of 20 mM Tris-HCl (pH 8.0), 30 mM NaCl, 1 mg/ml BSA, 0.1% Triton X-100, 0.1 mM phenylmethylsulfonyl fluoride, and 0.1 M Ca 2ϩ . The mixture was incubated for 12 h at 4°C and centrifuged at 14,00 ϫ g for 15 s at 4°C. The beads were washed three times with 1 ml of the binding buffer and subjected to 12% SDSpolyacrylamide gel electrophoresis for Western blotting as follows. Proteins were transferred to a nitrocellulose membrane (Schleicher & Schü ll), probed with monoclonal antibody against ANX-I, and visualized using the ECL system (Amersham Pharmacia Biotech).
Coprecipitation of cPLA 2 in the Sf9 cPLA 2 cell lysate and purified GST-ANX-I was examined. Total cell lysate of Sf9 cPLA 2 cells was prepared in the same way as described above for the preparation of A549 cell lysate. The binding mixture contained 100 g of Sf9 cPLA 2 cell lysate, 2 g of ANX-I, 2 g of anti-GST monoclonal antibody, and 100 l of binding buffer. The immune complexes were analyzed by Western blotting using anti-cPLA 2 antibody (31).
Immunoprecipitation of the cPLA 2 ⅐ANX-I Complex from A549 Cell Lysate-Existence of the cPLA 2 ⅐ANX-I complex in cells was examined using A549 cell lysates. A549 cell lysate was prepared as described above. One-hundred l of A549 cell lysate was precipitated with anti-ANX-I antibody and protein A-agarose. cPLA 2 activity in the precipitate or supernatant was determined in a standard buffer containing 5 mM CaCl 2 . cPLA 2 in the pellet was also analyzed by Western blotting.

RESULTS
In the previous experiments of cPLA 2 inhibition by ANX-I, we used cPLA 2 purified from porcine spleen (19). Since purification of cPLA 2 from porcine spleen is laborious and results are often inconsistent, human cPLA 2 cDNA was cloned into a baculovirus vector and expressed in Sf9 cells. cPLA 2 produced in Sf9 cPLA 2 cells was characterized by the following methods using porcine spleen cPLA 2 as a reference: 1) size determination by Western blot analysis using anti-cPLA 2 antibody, 2) activity in the presence of dithiothreitol, and 3) Ca 2ϩ concentration dependence of PLA 2 activity. Although sPLA 2 activity was sensitive to dithiothreitol, both cPLA 2 enzymes from porcine spleen and Sf9 cells were essentially insensitive to the dithiothreitol concentration (data not shown). Sf9 cPLA 2 was active at Ca 2ϩ concentrations as low as 0.1 M and showed nearly identical activity compared with porcine spleen cPLA 2 at all Ca 2ϩ concentrations. On the other hand, at least 0.1 mM Ca 2ϩ was necessary for sPLA 2 activity (data not shown). Therefore, Sf9 cPLA 2 exhibited nearly identical activity compared with porcine spleen cPLA 2 .
Inhibition of Porcine Spleen cPLA 2 by ANX-I and Its Deletion Mutants-The effects of ANX-I and its deletion mutants on cPLA 2 activity were determined. Full-length ANX-I (ANX-I-(1-346)) and the deletion mutants ANX-I-(33-346), ANX-I-(1-274), and ANX-I-  were cloned, expressed in E. coli, and purified to near homogeneity (Fig. 1A). The effects of ANX-I and its mutants on cPLA 2 from porcine spleen were determined at various concentrations of ANX-I and Ca 2ϩ using 2-AA-PC as a substrate. The reaction mixtures were incubated at 37°C for 1 h, and radiolabeled arachidonic acid, produced by the hydrolyzing reaction of cPLA 2 , was measured. An incubation time of 1 h was chosen for the following reasons. First, measuring initial rates was prone to more errors due to the small counts/ min of the product; and second, the ANX-I inhibition pattern was nearly identical at all time points until 2 h. The substrate concentration was 1.65 M, which was significantly greater than the enzyme (0.5 nM) and ANX-I (5-100 nM) concentrations.  (Fig. 1C). Both ANX-I-(1-274) and ANX-I-(1-196) had no effect on cPLA 2 activity at any calcium concentration. To rule out the possibility that the cPLA 2 inhibition by ANX-I-(1-346) or ANX-I-(33-346) was due to substrate depletion, the substrate concentration was varied from 1.65 to 33 M while holding the other components constant. As shown in Fig.  1D, the inhibition of cPLA 2 by both ANX-I-(1-346) and ANX-I-(33-346) was essentially independent of the substrate concentration.
Inhibition of Sf9 cPLA 2 by GST-ANX-I and Its Deletion Mutants-The results shown in Fig. 1 demonstrate that the Nterminal 32 amino acids are not important for the cPLA 2 inhibitory activity of ANX-I, whereas the C-terminal 72 amino acids are crucial for the inhibitory activity. To further investigate the important region of ANX-I, various deletion mutants were constructed. Attempts to produce ANX-I-(275-346) or ANX-I-(197-346) in E. coli failed due to very low expression levels. Therefore, use of the GST-ANX-I fusion protein was evaluated. The validity of Sf9 cPLA 2 instead of porcine spleen cPLA 2 was also evaluated. GST-ANX-I-(1-346) exhibited effects identical to those of ANX-I-(1-346) on the activity of cPLA 2 from porcine spleen and Sf9 cPLA 2 cells (data not shown). Therefore, various ANX-I mutants were constructed as GST fusion proteins and used for inhibition studies. Fig. 2A shows a schematic representation of various GST-ANX-I deletion mutants. All mutants were produced in E. coli, purified on a glutathione-Sepharose 4B column to near homogeneity, and used without cutting off the GST tail (Fig. 2B). Fig. 3 shows the effects of GST-ANX-I mutants on cPLA 2 activity. No C-terminal deletion mutants exhibited any inhibitory activity (Fig. 3A), whereas all N-terminal deletion mutants exhibited full inhibitory activity (Fig. 3B). Therefore, the inhibitory activity is located in amino acids 275-346. Inhibition of cPLA 2 by GST-ANX-I-(1-346) or GST-ANX-I-(275-346) was further characterized at various GST-ANX-I, substrate, and calcium concentrations. GST-ANX-I showed similar patterns with ANX-I ( Fig. 1 versus Fig. 4). GST-ANX-I-(275-346) also showed similar patterns, except that the activity was independent of the calcium concentration (Fig. 4C).
Interaction of cPLA 2 -(1-80) and ANX-I-(275-346) in a Mammalian Two-hybrid System-The results presented in Figs. 1, 3, and 4 demonstrate that ANX-I inhibits cPLA 2 by specific interaction and not by substrate depletion, which requires a protein-protein interaction. A mammalian two-hybrid assay was utilized to investigate the interaction between ANX-I and cPLA 2 . Deletion mutants of ANX-I and cPLA 2 were cloned into a mammalian two-hybrid vector, and the interaction was examined as described under "Experimental Procedures." Since ANX-I-(275-346) inhibited cPLA 2 as effectively as full-length ANX-I, the effects of ANX-I-(1-274), ANX-I-(275-346), and fulllength ANX-I were examined. The C2 domain of cPLA 2 (comprising amino acids 16 -138), of which amino acids 38 -80 (⌬C2) are highly conserved, is responsible for calcium-dependent phospholipid binding (11,32). To evaluate the importance of this region, cPLA 2 -(1-80) and cPLA 2 -(81-793) were cloned. The mutants were transfected into A549 cells as well as into Rat2 cells, and protein-protein interaction was analyzed. ANX-I-(1-346) interacted with full-length cPLA 2 as well as cPLA 2 -(1-80) in both A549 cells (Fig. 5A) and Rat2 cells (Fig. 5B). ANX-I-(275-346) was almost as effective as ANX-I-(1-346), whereas ANX-I-(1-274) was much less effective for the interaction with cPLA 2 . Therefore, it can be deduced that the Cterminal 72-amino acid region of ANX-I interacts with the N-terminal 80-amino acid region of cPLA 2 . This result is con-   were cloned into pET-28a(ϩ), expressed in E. coli, and purified according to the methods described (25). Partially purified porcine spleen cPLA 2 was used. A, ANX-I and its deletion mutants analyzed by 12% SDS-polyacrylamide gel electrophoresis. B, inhibition of cPLA 2 by ANX-I and its deletion mutants. cPLA 2 inhibition was determined using 2-AA-PC as a substrate. The reaction was carried out at 37°C for 1 h in 75 mM Tris-HCl (pH 7.5) containing 0.5 nM cPLA 2 (10 ng/200 l), 1.65 M 2-AA-PC, 1 M Ca 2ϩ , and 1 mg/ml BSA. The concentration of ANX-I was varied from 5 to 100 nM. cPLA 2 activity with or without ANX-I was determined, and the percentage of the remaining activity in the presence of ANX-I was calculated. C, calcium concentration dependence of cPLA 2 inhibition. The reaction was performed as described for B, except that the substrate and ANX-I concentrations were maintained at 1.65 M and 20 nM, respectively. The Ca 2ϩ concentration was varied from 0.1 M to 10 mM. D, substrate concentration dependence of cPLA 2 inhibition by ANX-I. Assays were performed at 20 nM ANX and 1 M calcium with various amounts of substrate. Data represent means Ϯ S.E. of three independent experiments.

FIG. 2. Production of GST-ANX-I and its deletion mutants in E.
coli. Various ANX-I deletion mutants were constructed as GST fusion proteins using a GST fusion vector (pGEX-5X-1) as described under "Experimental Procedures." Each protein was produced in E. coli and purified on a glutathione-Sepharose 4B column. Coprecipitation of MBP-⌬C2 and GST-ANX-I-To further investigate the direct interaction between cPLA 2 and ANX-I or the deletion mutants, co-immunoprecipitation of ⌬C2 with GST-ANX-I was examined. The ⌬C2 domain of cPLA 2 was produced as a fusion protein with MBP. GST-ANX-I mutants (Fig. 2) and MBP-⌬C2 were produced in E. coli and purified by affinity column chromatography. GST-ANX-I and MBP-⌬C2 were mixed; the mixture was immunoprecipitated with anti-MBP antibody; and the precipitate was analyzed by Western blotting using anti-ANX-I antibody. As shown in Fig. 6A, the C-terminal deletion mutants did not interact with ⌬C2, whereas the N-terminal deletion mutants did. These results are in agreement with the cPLA 2 inhibition pattern shown in Figs. 3 and 4 and the results presented in Fig. 5. The interaction of cPLA 2 and ANX-I was further investigated using purified proteins and cell lysates: MBP-⌬C2 and A549 cell lysate or GST-ANX-I and Sf9 cPLA 2 cell lysate. In one experiment, MBP-⌬C2 was mixed with A549 cell lysate, and the mixture was immunoprecipitated with anti-MBP antibody, followed by Western blot analysis using anti-ANX-I antibody. In another experiment, GST-ANX-I-(1-346) or GST-ANX-I-(275-346) was mixed with the Sf9 cPLA 2 cell lysate, and the mixture was immunoprecipitated with anti-GST antibody, followed by Western blotting using anti-cPLA 2 antibody. As shown in Fig.  6B, ANX-I in the A549 cell lysate coprecipitated with MBP-⌬C2. Both GST-ANX-I-(1-346) and GST-ANX-I-(275-346) coprecipitated with cPLA 2 in the Sf9 cPLA 2 cell lysate (Fig. 6C).
Coprecipitation of cPLA 2 ⅐ANX-I from A549 Cell Lysate-The results shown in Fig. 6 demonstrate that ANX-I interacts with cPLA 2 . To determine whether the cPLA 2 ⅐ANX-I complex exists in vivo, A549 cell lysate was precipitated with anti-ANX-I antibody, and the precipitate was analyzed by Western blotting using anti-cPLA 2 antibody. To minimize the effect of Ca 2ϩ during immunoprecipitation, cells were lysed in buffer containing 0.1 M Ca 2ϩ . As shown in Fig. 7A, the cPLA 2 ⅐ANX-I complex was coprecipitated by anti-ANX-I antibody. To further demonstrate precipitation of the cPLA 2 ⅐ANX-I complex by anti-ANX-I antibody, cPLA 2 activity in the pellet and supernatant was determined. This approach is based upon the observation that ANX-I inhibited cPLA 2 at Ca 2ϩ concentrations below 1 M, but not at Ca 2ϩ concentrations above 1 mM, under the assay conditions of this study (Fig. 1). Thus, cPLA 2 precipitated with ANX-I at 0.1 M Ca 2ϩ , but was able to display activity under the assay condition of 5 mM Ca 2ϩ , presumably due to its spontaneous dissociation from ANX-I. To verify the validity of this method, the interaction of GST-ANX-I with MBP-⌬C2 was examined at 0, 0.1, and 1 M and 1 mM calcium. In the absence of phosphatidylcholine, the interaction was slightly influenced by calcium, whereas in its presence, this interaction depended on the calcium concentration. GST-ANX-I bound to MBP-⌬C2 at 0.1 and 1 M calcium, but not at 5 mM calcium (data not shown). Fig. 7B shows the results of the coprecipitation experiments. cPLA 2 activity was confined primarily to the pellet. DISCUSSION The mechanism by which ANX-I inhibits cPLA 2 activity is still a controversial issue. In this study, we tried to address this issue by determining the effects of ANX-I and its deletion mutants I on cPLA 2 activity. Enzymatic studies have revealed that 1) cPLA 2 activity is specifically inhibited by ANX-I; 2) deletion from the N terminus to amino acid 274 has little effect on the inhibitory activity; 3) deletion of the C-terminal 72 amino acids abolishes the activity; and 4) inhibition is independent of the substrate concentration. Studies by immunoprecipitation and mammalian two-hybrid experiments have revealed that cPLA 2 forms complexes with the C-terminal 72 amino acids of ANX-I, but not with its N-terminal 274 amino acids. These results are in agreement with the enzymatic studies. The previously proposed substrate depletion mechanism is based upon the observation that inhibition of sPLA 2 by ANX-I is abolished with an increasing substrate concentration (17,18). As shown in Figs. 1D and 4B, inhibition of cPLA 2 by ANX-I was independent of the substrate concentration, which is consistent with the specific interaction mechanism.
The results shown in Figs. 5-7 demonstrate the specific interaction between cPLA 2 and ANX-I, which further supports the specific interaction model. As shown in Figs. 1C and 4C, inhibition depended upon the calcium concentration and was greatest at 0.1-1 M calcium and negligible above 1 mM calcium, which is consistent with the previous observation (19). These results are consistent with the following interpretation. At calcium concentrations less than 1 M, binding of ANX-I to the substrate is little, and inhibition is primarily by specific interaction. The substrate binding of ANX-I increased with an increasing calcium concentration (data not shown); and at 1 mM calcium, inhibition by specific interaction was negligible (Fig.  1C), and if any inhibition is to occur, it would be by substrate depletion. Under the conditions for the substrate depletion model, apparent cPLA 2 inhibition by ANX-I is observed only with an excess amount of ANX-I and a limiting amount of the substrate (17,18). Since the inhibition assays in Figs. 1 and 4 were performed in the presence of a large excess of the substrate (5-100 nM ANX-I and 1.65 M substrate), at 1 mM Ca 2ϩ , ANX-I binds mostly to the substrate and is unavailable for cPLA 2 inhibition. This interpretation is supported by the ob-servation that at 1 mM Ca 2ϩ , 300 nM ANX-I was required to inhibit cPLA 2 in the presence of 1.65 M substrate (data not shown). Studies with various phospholipids have indicated that ANX-I binds to anionic vesicles such as phosphatidylserine, but not to neutral vesicles such as phosphatidylcholine, at calcium concentrations up to 0.37 mM (33). In studies of the substrate depletion model, vesicles including anionic phospholipids were used as substrate (17,18). The substrate depletion is due to binding of anionic substrate to a large excess of ANX-I. In contrast, since phosphatidylcholine was used in all assays in this study, the calcium dependence of cPLA 2 inhibition by ANX-I (Figs. 3 and 4) is difficult to explain based on the previous result (33). We have performed binding assays using ANX-I and phosphatidylcholine under the assay conditions of the inhibition experiments and found that ANX-I bound to phosphatidylcholine at 5 mM calcium, but not at 0.1 and 1 M calcium (data not shown). This result is in agreement with the inhibition data (Figs. 3 and 4).
As shown in Fig. 1, inhibition by ANX-I never exceeded 50%. Furthermore, an inhibitor/enzyme ratio of 40 (20:0.5 nM) was required to obtain the maximum inhibition. Reasons for both phenomena are not clear. In the in vivo experiments using cultured cells, deletion of the C2 domain abolishes translocation of cPLA 2 to the membrane and agonist-induced arachidonic acid release (11,32). The hydrophobic residues of the C2 domain in loops known as calcium-binding regions 1 and 3 are Purified GST-ANX-I mutants and MBP-⌬C2 were mixed and incubated. The mixture was immunoprecipitated (IP) with anti-MBP antibody, and the pellet was analyzed by Western blotting using anti-ANX antibody (Ab). ANX was omitted in lane C (control). B, immunoprecipitation of MBP-⌬C2 and ANX-I in A549 cell lysate. One-hundred g of A549 cell lysate was incubated with 2 g of either MBP-⌬C2 or MBP as a control. The mixture was immunoprecipitated with anti-MBP antibody, and the pellet was analyzed by Western blotting using anti-ANX antibody. Lane C, untreated A549 cell lysate. C, immunoprecipitation of GST-ANX-I and cPLA 2 in Sf9 cPLA 2 cell lysate. One-hundred g of Sf9 PLA 2 cell lysate was incubated with 2 g of GST-ANX-I-(1-346), GST-ANX-I-(275-346), or GST. The mixture was immunoprecipitated with anti-GST antibody, and the pellet was analyzed by Western blotting using anti-cPLA 2 antibody. Lane C, untreated Sf9 PLA 2 cell lysate.
important for phospholipid binding, and ANX-I bound to the ⌬C2 domain (Figs. 5 and 6), which lacks calcium-binding region 3. Thus, even if ANX-I binds to the ⌬C2 domain, calciumbinding region 3 may still be available for weak calcium and phospholipid binding, resulting in partial inhibition of the phospholipid-binding property of cPLA 2 . Another point to be mentioned is that the inhibitor/enzyme ratio is 20 (10:0.5 nM) at 85% of the maximum inhibition and 40 (20:0.5 nM) at the maximum inhibition. It is unlikely that this number directly reflects the stoichiometry of ANX-I to cPLA 2 in the inhibitorenzyme complex. At present, it is not easy to explain the reason for the high inhibitor/enzyme ratio. As mentioned above, ANX-I has higher affinity for vesicles in which anionic phospholipids are present, such as natural membranes. In mammalian cells, both ANX-I and cPLA 2 are ubiquitous; however, the expression level of ANX-I is far greater than that of cPLA 2 . The interplay among ANX-I, cPLA 2 , and phospholipids in cells seems to follow a complex rule.
It is notable that although cPLA 2 inhibition by GST-ANX-I-(1-346) depended upon the calcium concentration, inhibition by GST-ANX-I-(275-346) was independent of the calcium concentration (Fig. 4C). This may be due to the number of the calcium-binding sites of ANX-I. ANX-I has six calcium-binding sites, and each is located in the loop region of the helix-loophelix motif (3), whereas ANX-I-(275-346) has only one calciumbinding site and may bind to the substrate in a less calciumdependent manner.
ANX-I consists of four domains, and each domain has five ␣-helix motifs (3). The four domains are highly conserved in all members of the ANX family, whereas the N terminus of each ANX varies. As shown in Fig. 3, the N-terminal region is not important for the cPLA 2 inhibitory activity of ANX-I, and domain IV is the active domain. It is located at amino acids 275-346 (Fig. 3). This result is unexpected because region 275-346 lies within the core domain that is conserved in all ANXs. Even though the core domain sequences are highly conserved, there are apparently enough differences to differentiate the cPLA 2 -binding properties. This phenomenon is supported by the fact that the anti-ANX-I antibody derived from human ANX-I specifically recognizes all type I ANXs across the species from mold to human, whereas it does not recognize any other types of ANXs from any source, including human cells. 2 Inhibition of cPLA 2 by various members of the ANX family of proteins has been reported. However, most reported data are from experiments performed under conditions in which the substrate depletion mechanism dominates (34,35). Even though there have been no detailed enzymatic studies on the mechanism of cPLA 2 inhibition by ANXs other than ANX-I, most articles in the literature describe the mechanism as substrate depletion. There is only one other report (our previous research) that concludes that the inhibition mechanism is by specific interaction (19). If the mechanism is substrate depletion, all ANXs should display similar inhibition patterns because this mechanism is due to the ability of ANXs to bind to phospholipids. Inhibition studies using several ANXs have revealed that ANX-I, but neither ANX-II nor ANX-III, inhibits cPLA 2 and that ANX-V causes much less effective inhibition (36). The differential effects of ANXs on cPLA 2 activity provide evidence for the specific interaction model.
ANX-I is a major substrate for the epidermal growth factor receptor kinase, which has been implicated in membrane-related events along the endocytic pathway, in particular the sorting of internalized epidermal growth factor receptors occurring in the multivesicular body (MVB) (37). Truncation of the N-terminal 26 residues of ANX-I altered its intracellular distribution, shifting it from early to late and multivesicular endosomes, indicating the regulatory importance of the N-terminal domain and the involvement of ANX-I in early endocytic processes (38). ANX-I is associated with both the plasma membrane and MVB in a calcium-dependent manner, but can be phosphorylated only in MVBs. ANX-I and ANX-II are localized differently in human epidermal keratinocytes (39). ANX-I translocates from the cytosol to the nucleus by epidermal growth factor or stress signals through the N-terminal region (40). These evidences indicate that the variable N-terminal residue of ANXs is important for localization of ANXs and is involved in the interaction with MVBs, rather than with the plasma membrane. This interaction may be important for signal-induced phosphorylation and endocytic processing. On the other hand, the C-terminal domain of ANX-I may be involved in plasma membrane binding and in regulation of cPLA 2 . On the contrary, involvement of ANX-I domain I in the regulation of the cPLA 2 signal has been observed (24). Since the study was designed to observe phorbol myristate acetate-induced c-fos activation, it is likely that several factors other than cPLA 2 also participate in this process. The importance of the N-terminal region in mediation of the anti-inflammatory function of glucocorticoids and inflammatory signal transduction has been reported by Flower and co-workers (41)(42)(43). In the study of neutrophil migration, a mechanism has been proposed that the N-terminus of ANX-I binds to a receptor-like molecule on the surface of the neutrophils (44). Therefore, it is reasonable to assume that the N terminus is involved in the interaction with the MVB and the receptor-like molecule and that the C terminus is involved in the interaction with cPLA 2 . cPLA 2 binds to phospholipids by hydrophobic interaction. The hydrophobic residues of cPLA 2 have a significant function in membrane binding of this domain. The C2 domain of cPLA 2 preferentially binds to phospholipids with hydrophobic head groups, such as phosphatidylcholine. However, the C2 domain of protein kinase C binds to the membrane by electrostatic force (45). Whether specific interaction between cPLA 2 and ANX-I is driven by a hydrophobic or electrostatic force is not known. The C-terminal domain of ANX-I may bind to the phospholipidbinding site of cPLA 2 by hydrophobic interaction, thereby interfering with the binding of cPLA 2 to membranes. The Cterminal domain of ANX-I has five ␣-helices. Each helix may form a hydrophobic patch that binds to the ⌬C2 domain. However, ANX-I-(275-346) binding to the regulatory site of cPLA 2 by electrostatic interaction cannot be ruled out. We are conducting mutation studies of the hydrophobic region of ANX-I-(275-346) to determine the interaction mechanism between ANX-I and cPLA 2 .
The existence of the cPLA 2 ⅐ANX-I complex in the A549 cell lysate strongly suggests that this complex exists in vivo and that ANX-I regulates cPLA 2 activity in cellular signal transduction. Inhibition of cPLA 2 by ANX-I in cellular models supports this hypothesis (23,24). Upon activation of cells, intracellular calcium concentration is elevated from 0.1 to 1 M, and cPLA 2 is phosphorylated and translocated from the cytosol to the membrane (11). ANX-I is also translocated upon activation of cells (46,47). Since the degree of inhibition due to ANX-1 is similar at calcium concentrations of 0.1 and 1 M (Figs. 1 and  4), it is unlikely that elevation of calcium concentration is the sole mechanism of cPLA 2 regulation by ANX-I. It is probable that phosphorylation of ANX-I affects its protein-protein binding property. Considering that both cPLA 2 and ANX-I bind to phospholipids, it is reasonable to assume that their binding properties in the cytosol or near the membrane are different. Whatever the mechanism, its specific interaction with cPLA 2 and phosphorylation following various extracellular signals strongly suggest that ANX-I has an essential role in cell regulation, probably through the inhibition of cPLA 2 activity.
In conclusion, the mechanism of cPLA 2 inhibition by ANX-I is consistent with the specific interaction model, and the interaction between cPLA 2 inhibition and ANX-I supports this interpretation. cPLA 2 inhibition is a specific function of ANX-I and is not a general function of all ANXs. The C-terminal region is important for cPLA 2 inhibition. The results presented here are important since ANX-I specifically inhibits cPLA 2 at intracellular Ca 2ϩ concentrations. ANX-I may regulate several biological processes through regulation of cPLA 2 activity.