Caspase-3-dependent Activation of Calcium-independent Phospholipase A2 Enhances Cell Migration in Non-apoptotic Ovarian Cancer Cells*

Calcium-independent phospholipase A2 (iPLA2) plays a pivotal role in phospholipid remodeling and many other biological processes, including inflammation and cancer development. iPLA2 can be activated by caspase-3 via a proteolytic process in apoptotic cells. In this study we identify novel signaling and functional loops of iPLA2 activation leading to migration of non-apoptotic human ovarian cancer cells. The extracellular matrix protein, laminin-10/11, but not collagen I, induces integrin- and caspase-3-dependent cleavage and activation of overexpressed and endogenous iPLA2. The truncated iPLA2 (amino acids 514-806) generates lysophosphatidic acid and arachidonic acid. Arachidonic acid is important for enhancing cell migration toward laminin-10/11. Lysophosphatidic acid activates Akt that in turn acts in a feedback loop to block the cleavage of poly-(ADP-ribose) polymerase and DNA fragmentation factor as well as prevent apoptosis. By using pharmacological inhibitors, blocking antibodies, and genetic approaches (such as point mutations, dominant negative forms of genes, and siRNAs against specific targets), we show that β1, but not β4, integrin is involved in iPLA2 activation and cell migration to laminin-10/11. The role of caspase-3 in iPLA2 activation and cell migration are supported by several lines of evidence. 1) Point mutation of Asp513 (a cleavage site of caspase-3 in iPLA2) to Ala blocks laminin-10/11-induced cleavage and activation of overexpressed iPLA2, whereas mutation of Asp733 to Ala has no such effect, 2) treatment of inhibitors or a small interfering RNA against caspase-3 results in decreased cell migration toward laminin-10/11, and 3) selective caspase-3 inhibitor blocks cleavage of endogenous iPLA2 induced by laminin-10/11. Importantly, small interfering RNA-mediated down-regulation of endogenous iPLA2 expression in ovarian carcinoma HEY cells results in decreased migration toward laminin, suggesting that our findings are pathophysiologically important.

The phospholipase A 2 (PLA 2 ) 2 family of enzymes catalyzes the hydrolysis of the sn-2 position of phospholipids to generate free fatty acids and lysophospholipids. These enzymes are classified into three groups based on their cellular localization, substrate specificity, and calcium dependence. The secreted PLA 2 family consists of low molecular weight, calcium-requiring secretory enzymes that have been implicated in modification of eicosanoid generation, inflammation, and atherosclerosis (1). The cytosolic PLA 2 (cPLA 2 ) is believed to play an essential role in the release of arachidonic acid in response to extracellular stimuli. Intracellular activation of cPLA 2 is tightly regulated by calcium and phosphorylation (2). The third group of PLA 2 enzymes is calcium-independent PLA 2 (iPLA 2 ). One of the functions of iPLA 2 is to control phospholipid levels and maintain homeostasis through remodeling of membrane structures (3)(4)(5).
More recently, direct evidence shows that iPLA 2 can be cleaved at Asp 513 and Asp 733 by caspase-3, resulting in a truncated iPLA 2 lacking ankyrin-repeat domain (22). This truncated iPLA 2 has increased activity to produce lysophosphatidylcholine (LPC) in UV-induced apoptotic cells, where LPC serves as a "find me" signal in the phagocytic process for apoptotic cell clearance (11).
Interestingly, we have observed a potentially related but different iPLA 2 -dependent cellular process. We show that when ovarian cancer cells are in contact with laminin-10/11, iPLA 2dependent LPA production and cell migration occur (13). In addition, the requirement of iPLA 2 activity in the chemotaxis of human monocytes to monocyte chemoattractant protein 1 has also been reported (14). However, the mechanisms regarding iPLA 2 activation under these non-apoptotic conditions remain to be elucidated.
In this work we investigate the molecular mechanisms of iPLA 2 activation induced by laminin-10/11. In particular, the potential involvement of caspase-3 and ␤ 1 integrin is examined. We reveal novel caspase-dependent, but apoptosis-independent activation of iPLA 2 , which plays important roles in cell migration. These include ␤ 1 integrin-and caspase-3-dependent cleavage and production of a truncation-activated iPLA 2 that produces LPA, LPA-induced intracellular signaling resulting in arachidonic acid (AA) production and an AA-dependent cell migration, and LPA-Akt activation-mediated inhibition of cleavage of poly-(ADP-ribose) polymerase (PARP) and DNA fragmentation factor (DFF) as well as apoptosis.
Cell Culture, Transfection, and Western Blotting-Malignant ovarian HEY cells were maintained in RPMI 1640 medium containing 10% fetal bovine serum at 37°C with 5% CO 2 . For transient transfection, cells (5 ϫ 10 5 ) were cultured in 6-well plates overnight and transfected with DNA using Lipofectamine TM 2000 (Invitrogen) according to the manufacturers' instruction. iPLA 2 expression in the transfected cells was evaluated by Western blotting analyses. Collected cells were rinsed with icecold PBS, and equal number of cells from each indicated condition were lysed with SDS sample buffer (Bio-Rad). Samples were separated by 10% SDS-PAGE and transferred to polyvinylidene difluoride membrane (Bio-Rad). After blocking overnight with 5% milk in PBS containing 0.1% Tween 20, the membranes were probed with the indicated antibodies and corresponding secondary antibodies and developed using enhanced chemiluminescence reagent (Amersham Biosciences). The cells were serum-starved for 16 h before exposure to ECM proteins.
siRNA Experiments and Reverse Transcription-PCR-Sequences of siRNA are as follows: iPLA 2 , 5Ј-GGAUCUCAUGC-ACAUCUCAtt-3Ј; caspase-3 5Ј-UGAGGUAGCUUCAUA-GUGGtt-3Ј; GFP, 5Ј-GCAAGCUGACCCUGAAGUUCAT-3Ј; glyceraldehyde-3-phosphate dehydrogenase catalog #4805; targeting protein of accession number NM_ 002046. siRNA was purchased from Ambion Inc. (Austin, TX). Each indicated siRNA was delivered to cells with Lipofectamine TM 2000 (Invitrogen) according to the manufacturer's instructions. RNA was extracted 48 h after treatment using SV Total RNA Isolation System (Promega, Madison, WI). The expression levels of siRNA targets were detected by reverse transcription-PCR. Primer sequences are as follows: iPLA 2 , 5Ј-AACGTTAACCTCAGGCCTCC-3Ј and 5Ј-GAGAGTTTCTTCACCTTGTT-3Ј; glyceraldehyde-3-phosphate dehydrogenase, 5Ј-GAAGGTGAAGGTCGGAGT-3Ј and 5Ј-GAAGATGGTGATGGGATTTC-3Ј. iPLA 2 Activity Assay-Equal numbers of control and iPLA 2transfected HEY cells were washed twice in PBS. Cells were then resuspended in 200 l of buffer containing 10 mM HEPES (pH 7.5), 1 mM EDTA, 1 mM dithiothreitol, and 0.34 M sucrose followed by sonication (2 ϫ 10 s), and the resulting homogenate was centrifuged at 10,000 ϫ g for 10 min at 4°C. Collected supernatant was used for the reaction as described by Carnevale and Cathcart (14). In brief, dipalmitoyl phosphatidylcholine (100 M) and 1-palmitoyl-2-[1-14 C]palmitoyl-sn-glycero-3phopshocholine (300,000 cpm/assay) were evaporated to dryness under an N 2 stream, and a buffer (125 l) containing 800 M Triton X-100, 10 mM EDTA, and 200 mM HEPES (pH 7.5) was added. Micelles were then formed by a combination of heating above 40°C, vortexing, and water bath sonication until the solution was clarified. The reaction was initiated by mixing cell lysates with the substrate and 0.8 mM ATP in a final volume of 250 l and incubated at 40°C for 1 h. The reaction was stopped, and the lipids were extracted by adding 0.5 ml of chloroform/methanol/acetic acid (2:4:1, v/v/v) followed by 0.25 ml of chloroform and 0.25 ml of H 2 O. The lipids in the chloroform layer were dried under an N 2 stream, resuspended in chloroform/methanol (2:1, v/v), separated on Silica Gel TLC plates in a solvent mixture of chloroform/methanol/acetic acid/water (25/15/4/2, v/v/v/v), and visualized with 8-anilino-1-naphthalene sulfonic acid. Fatty acids generated by the PLA activity and phosphatidylcholine were scraped and counted in a scintillation counter. Normalized activity represents cpm of free fatty acid divided by the cpm of free fatty plus the cpm of the dipalmitoyl phosphatidylcholine times the total added counts (300,000 cpm).
Cell-based Enzymatic Activity Assay of PLA 2 -The principle of this cell-based fluorescence dequenching assay for PLA 2 has been described (15). Briefly, parental or vector-or iPLA 2 -transfected HEY cells were serum-starved overnight, collected, and washed with sterile PBS. Cells were then resuspended in PBS and allowed to attach to positively charged glass slides for 20 min. The cells were incubated with 1 g of 1-O-(6-dabcyl-aminohexanoyl)-2-O-(6-[12-BODIPY-dodecanoyl]amino-hexanoyl)-sn-3-glycerophosphatidylcholine resuspended in 30 l of PBS. Cells were then washed with PBS followed by imaging with excitation wavelength of 488 nm and filtered emission detection at 515 nm. Imaging was performed using a Leica fluorescence microscope, and fluorescence intensity was measured using Image-Pro software.
Migration Assay-Cell migration assays were performed as described previously (13). Briefly, the lower face of the top chamber of the Corning Costar Transwell migration plate (Fisher) was coated with or without laminin-10/11 at a concentration of 10 g/ml. Lipids were added to the upper or lower chamber for each indicated condition. Serum-starved cells (1 ϫ 10 5 /100 l) were added to the upper chamber. The Transwell chambers were then incubated at 37°C for 4 h. After incubation the top chambers were rinsed with PBS and swabbed with a cotton swab to remove non-migrated cells. Migrated cells were fixed by methanol for 30 min and stained with hematoxylin 7211 (Richard-Allan Scientific, Kalamazoo, MI) followed by washing with water. Migrated cells were counted in five highpower fields by eye using a light microscope after the chambers were dried. Pictures were taken with Nikon SMZ 1000 microscope. The caspase-3 activity assay was performed using specific colorimetric activity assay kits (Chemicon International) following their instructions. Briefly, samples were resuspended in chilled cell lysis buffer and incubated on ice for 10 min. After centrifugation, the cytosolic extract was collected, and the assay mixture was prepared in a 96-well plate. Inhibitors were preincubated with the caspase sample for 30 min at room temperature before adding the caspase-3 substrate solution. Assay mixtures were incubated for 2 h at 37°C, and caspase activities were measured by reading the absorbance of their products at 405 nm in a microtiter plate reader.
Lipid Extraction and Mass Spectrometric Assay-The methods for LPA extraction and mass spectroscopy assay were described previously (16). In brief, LPA was extracted from culture supernatant by acidified MeOH/chloroform and separated on a TLC plate; LPA was eluted from the plate. Electrospray ionization (ESI)-mass spectroscopy and tandem mass spectrometry (MS/MS) analyses were performed using a Micromass Quattro Ultima triple quadrupole mass spectrometer equipped with an ESI source (Micromass, Beverley, MA) in the Mass Spectrometry II Core at the Cleveland Clinic Foundation. 14:0 LPA was used as an internal standard in the negative mode of detection.
Analysis of Arachidonic Acid Using High Performance Liquid Chromatography On-line Tandem Mass Spectrometry-For extraction of arachidonic acid, a 1-ml sample was acidified with 20 l of 0.75 M HCl, mixed with 1 ml mixture of 2-isopropanol/ hexane/2 M acetic acid (80/120/4, v/v/v), then mixed with an additional 2 ml of hexane; the fatty acids were extracted to the hexane layer and further dried under nitrogen flow. Extracted fatty acids were resuspended in 85% methanol and centrifuged at 6000 ϫ g for 10 min, and 20 l of each supernatant was injected onto a Prodigy C-18 column (2.1 ϫ 150 mm, 5 m). The separation of fatty acids was performed using methanol/ water/acetic acid (85/15/0.1, v/v/v) over 4 min, a gradient from methanol/water/acetic acid (85/15/0.1, v/v/v) to methanol/acetic acid (100/1, v/v) over 2 min, and then methanol/acetic acid (100/1, v/v) over 9 min. The column effluent was introduced onto a triple quadrupole mass spectrometer. Analysis was performed using electrospray ionization in negative-ion mode with multiple reactions monitoring of parent and characteristic daughter ions specific for arachidonic acid with mass-to-charge ratio (m/z) 3033259. An external calibration curve constructed with authentic standard was used to quantify arachidonic acid.

Overexpression of iPLA 2 Increases Cellular PLA 2 and Cell
Migratory Activities-We previously reported the involvement of iPLA 2 in laminin-10/11-induced migration of HEY ovarian cancer cells (13). To further evaluate the role of iPLA 2 activity in cell migration, we overexpressed different iPLA 2 constructs (Fig. 1, A and E) in HEY cells and analyzed their effects on cell migration. These constructs include the full-length iPLA 2 , iPLA 2 (encoding aa 514 -733 that can be generated by two-site cleavage of caspase-3 (11)), iPLA 2 (encoding aa 514 -806 that can be generated by one-site cleavage of caspase-3 (11)), and iPLA 2 (encoding aa 535-733, with the catalytic site removed). Transfection of these forms of iPLA 2 into HEY cells resulted in expression of peptides with predicted molecular weights, which are recognized by the antibody against the aa 557-576 region of the iPLA 2 (Fig. 1A).
The enzymatic activities of PLA 2 in transfected HEY cells were analyzed using a well established micelle-based in vitro assay (14,17,18). HEY cells transfected with all three forms of iPLA 2 containing the consensus lipase motif GXSXG had increased PLA 2 activity as compared with parental or vectortransfected cells and the truncated forms of iPLA 2 had higher activities than that of the full-length iPLA 2 . As predicted, iPLA 2 (aa 535-733) did not shown any PLA activity (data not shown). We also conducted PLA 2 activity assays using a recently developed cell-based fluorescence method (15,19,20). The results from these assays are shown in Fig.  1B. Normalizing of expression levels and the fluorescent intensity and taking the activity of the most active iPLA 2 form (aa 514 -733) as 100%, the activity levels were 36 Ϯ 10 and 77 Ϯ 17% for the full-length and aa 514 -806 forms, respectively (Fig. 1B). These results support the notion that iPLA 2 isoforms without the ankyrin repeats are active as reported in other cell types (7,11).
To confirm the nature of the observed fluorescent signal, we pretreated cells with a selective inhibitor for iPLA 2 (HELSS). As expected, HELSS (1 M) significantly blocked the PLA 2 activity in iPLA 2 -transfected cells (Fig. 1B, 2nd column and the lower panel), indicating that the detected fluorescence is mainly due to iPLA 2 activity.
To examine the cellular effect of iPLA 2 activation on cell migration, we compared the migratory activities in HEY cells transfected with different forms of iPLA 2 . As shown in Fig. 1C, vector-transfected HEY cells had a similar basal level of migration to laminin-10/11 as that of parental HEY cells. Full-length iPLA 2 overexpressing cells showed an approximate 10-fold increase in cell migration over the parental cells. The two truncated iPLA 2 forms (aa 514 -806 and 514 -733) showed a much stronger effect on enhancing the migration of HEY cells to laminin-10/11 as compared with full-length iPLA 2 (Fig. 1C). These migratory effects correlate with the enzymatic activity of iPLA 2 (Fig. 1B), although a simple linear relationship may not exist. To directly address whether the enzymatic activity of iPLA 2 is required for its migratory effect, we constructed a truncated form of iPLA 2 (aa 535-733) in which the PLA 2 catalytic site was removed. This construct encodes an iPLA 2 of ϳ22 kDa, which was expressed in transfected HEY cells (Fig. 1A, the second lane from the right). As expected, this iPLA 2 form had neither detectable enzymatic activity (Fig. 1B) nor any effect on cell migration (Fig.  1C), suggesting that the enzymatic activity is required for migration.
It is possible that the truncated form of iPLA 2 exerts an unknown cellular effect(s). To address the relationship between the enzymatic activity and the migratory effect of iPLA 2 more specifically, we constructed a point mutation at the catalytic site of iPLA 2 by altering Ser 519 to Ala. This mutant (S519A) expressed normally in cells (Fig.  1A, last lane) but did not have any increased enzymatic activity (Fig. 1B, last panel of the left column) nor did it increase cell migration above the control level (Fig. 1C), confirming that the enzymatic activity of iPLA 2 is required for its migratory effect.
We have previously shown that HEY cells have differential haptotactic activities toward different ECM proteins (33). Thus, we compared the migratory potential toward collagen I versus laminin-10/11 in cells transfected with three iPLA 2 isoforms. As shown in Fig. 1D, parental HEY cells as well as cells transfected with full-length iPLA 2 had a higher haptotactic activity toward laminin-10/11 than toward collagen I. On the other hand, the haptotactic activities toward these two matrix proteins were similar in cells transfected with either of the truncated active iPLA 2 forms. These results suggest that whereas laminin-10/11, but not collagen I, induces activation of fulllength iPLA 2 , the product(s) derived from the active forms of iPLA 2 stimulates cell migration to both laminin-10/11 and collagen I. This is consistent with our observation that collagen I does not induce cleavage of full-length iPLA 2 (see Fig. 3A). Col-lectively, these data suggest that increased iPLA 2 activity is correlated with the enhancement of HEY cell migration to laminin-10/11 (summarized in Fig. 1E).
Active iPLA 2 Increases LPA and AA Production, and AA Is an Important Stimulator for Cell Migration-To test whether iPLA 2 activation resulted in production of secreted factors that enhance cell migration, we collected the supernatant from different HEY cells transfected with different forms of iPLA 2 . The supernatants from full-length iPLA 2 and two truncated active iPLA 2 -transfected samples but not from the catalytically dead isoform-or vectortransfected cells stimulated HEY cell migration to laminin-10/11 ( Fig. 2A). These data suggest that a secreted factor(s) produced by iPLA 2 -transfected cells contributes to the migratory effect.
Our previous study showed that LPA is produced upon iPLA 2 activation in HEY cells (13). Thus, we tested whether the supernatant of HEY cells transfected with iPLA 2 (aa 514 -733) (since it showed the highest enzymatic activity and enhancement of cell migration in our study) contained more LPA than vector-transfected HEY cells. Increased production of 18:0-alkenyl-LPA, 18:0-alkyl-LPA, and 18:2-LPA were detected in the medium from iPLA 2 (aa 514 -733)-transfected cells when compared with the control (Fig. 2B). The most abundant LPA species produced was 18:1-LPA (Fig. 2B, lower panel). On the other hand, the levels of individual and total LPC species were not significantly different in iPLA 2 -transfected and control cells (data not shown). We then tested the effect of LPA and LPC on cell migration when added to the upper chamber of our assays, since cells expressing active iPLA 2 should produce lipids in the upper chamber first. Consistent with our previous report (13), LPA had a weak chemokinetic effect on HEY cell migration, whereas the addition of LPC to the upper chamber had no detectable effect on HEY cell migration (Fig. 2C). Other than lysophospholipids, fatty acids and, in particular, unsaturated fatty acids such as AA are also the products of PLA 2 action. When added to the upper chamber, AA (10 M) caused strong HEY cell migration (Fig. 2C). This effect was mainly chemokinetic, because when AA was added only to the lower chamber, the stimulated migration was much lower compared with when AA was added to only the upper chamber or both the upper and lower chambers (Fig. 2D). We found that overexpression of iPLA 2 resulted in an increased release of AA (Fig. 2E). Importantly, the concentration range of AA detected in the supernatant collected from iPLA 2 -transfected HEY cells was 15-35 M, similar to or higher than the concentrations used to stimulate cell migration (Fig. 2C).
Laminin-10/11-mediated Caspase-dependent Cleavage of iPLA 2 -One of the critical differences between apoptosis-and laminin-induced iPLA 2 activation (13) is that cells do not appear to undergo apoptosis in the latter case (see Fig. 7). Thus, it is important to determine how iPLA 2 is activated under nonapoptotic conditions and whether caspase-3 is involved. To address this question, we first examined the effect of laminin on proteolysis of iPLA 2 in HEY cells. As shown in Fig. 3A, all three forms of iPLA 2 -transfected cells had an overexpressed corresponding band of iPLA 2 . When cells were in contact with laminin-10/11, but not collagen I, a proteolytic product (the arrowhead) of the same size as iPLA 2 (aa 514 -806) was observed (Fig.  3A). Moreover, after contacting laminin-10/11, the enzymatic activity of iPLA 2 was increased (Fig. 3B). These results suggest that iPLA 2 may be activated by laminin-10/11 through a proteolytic process similar to that in apoptotic cells (7,11) that involves caspase-3. To test this we first examined the effect of the broad-spectrum caspase inhibitor (benzyloxycarbonyl-VAD-fluoromethyl ketone) and a selective inhibitor of caspase-3 (DEVD-CHO) on laminin-10/11-induced iPLA 2 cleavage. Both of these inhibitors blocked the cleavage of iPLA 2 induced by laminin-10/11 (Fig. 3C), suggesting that caspase 3 is involved in this cleavage.
In apoptotic cells caspase-3 cleaves iPLA 2 at two sites (Asp 513 and Asp 733 ) to generate a 26-kDa protein that is highly active in terms of cellular PLA 2 and migratory activities (Ref. 11 and Figs. 1 and 2). In HEY cells contacting laminin-10/11, only a 32-kDa cleavage product was observed in full-length iPLA 2 -expressing cells (Fig. 3A). This suggests that in these non-apoptotic cells caspase-3 only cleaves at Asp 513 but not at Asp 733 . The predicted truncated protein would have moderate activity. To test this hypothesis we produced alanine substitution mutants at Asp 513 and Asp 733 . We found that laminin-10/11 failed to induce the production of the 32-kDa protein in iPLA 2 -1-806/ D513A-expressing cells (Fig. 4A). Concomitantly, these cells lost their ability to enhance cell migration and PLA 2 activity when compared with wild type iPLA 2 expressing cells (Figs. 4, B and C). In contrast, iPLA 2 -1-806/D733A showed levels of migratory and PLA 2 activities that are similar to those of wild type iPLA 2 (Figs. 4, B and C). These data support the hypothesis that cleavage by caspase 3 at Asp 513 is required for production of the activated form of iPLA 2 in HEY cells.
Endogenous iPLA 2 Is Involved in Cell Migration to Laminin-10/11 via a Caspase-3-mediated Action-To investigate the role of endogenous iPLA 2 in cell migration, we used a siRNA strategy. siRNA against iPLA 2 down-regulated endogenous iPLA 2 at both RNA and protein levels (Fig. 5A). This siRNA, but not the siRNA against GFP (used as a negative control), reduced cell migration to laminin-10/11 (Fig. 5B). The involvement of caspase-3 in endogenous iPLA 2 activation and iPLA 2 -dependent cell migration was confirmed by showing that (a) laminin-10/11-induced cleavage of endogenous iPLA 2 was reduced when cells were pretreated with a caspase-3-selective inhibitor DEVD-CHO (Fig. 5C, left panel), and (b) both DEVD-CHO and benzyloxycarbonyl-VAD-fluoromethyl ketone blocked cell migration (in HEY cells without overexpression of exogenous iPLA 2 ) to laminin-10/11 (Fig. 5C, right panel). To confirm the role of caspase-3 in cell migration, we used a siRNA strategy. Pretreatment of cells with a specific siRNA against caspase-3, but not the control siRNA (against GFP), dramatically reduced the caspase-3 protein expression (Fig. 5D, left panel) and significantly inhibited the HEY cell migration to laminin-10/11 (Fig.  5D, right panel). These data on endogenous iPLA 2 and in nontransfected cells suggest that laminin-10/11-induced, caspase-3-mediated iPLA 2 activation is likely physiologically relevant. . Laminin-10/11-induced cleavage and activation of iPLA 2 is caspase-dependent. A, cleavage of full-length iPLA 2 in HEY cells upon contact with laminin-10/11. Parental and iPLA 2 -transfected cell lines were starved from serum for 16 h before culturing in plates precoated with laminin-10/11 (10 g/ml) or collagen I (10 g/ml) for 4 h. iPLA 2 expression and cleavage were detected in collected cell lysates. Ab, antibody. B, representative pictures of cell-based enzymatic activity assay of PLA 2 in iPLA 2 (aa 1-806)transfected HEY cells contacting laminin-10/11 (Ln). The same transfected cells cultured in regular plates were used as control. Fluorescence intensity was measured as described for Fig. 1B, and three pictures were used for statistical analysis. C, caspase-3-dependent cleavage of iPLA 2 . Equal numbers of full-length iPLA 2 -transfected HEY cells were starved from serum for 16 h, pretreated with an inhibitor (DEVD-CHO (2 M) or benzyloxycarbonyl (Z)-VAD (50 M) for 2 h) and then cultured for an additional 4 h in plates precoated with or without laminin-10/11. fmk, fluoromethyl ketone. ␤ 1 Integrin Is Involved in Laminin-10/11-induced Cleavage of iPLA 2 -␤ 1 and ␤ 4 integrins are the major receptors for laminins, and ovarian cancer cells express high levels of ␤ 1 integrin (21)(22)(23). To test whether integrin(s) is involved in the laminin-10/11-induced proteolysis of iPLA 2 , we pretreated iPLA 2 (aa 1-806) overexpressing cells with blocking antibodies against ␤ 1 or ␤ 4 integrins. Although ␤ 1 integrin blocking antibody completely inhibited laminin-10/11-induced iPLA 2 cleavage, the ␤ 4 blocking antibody had only a minimal effect (Fig. 6A). ␣ 6 ␤ 4 integrin is reported to be the receptor for laminin 5 (24,25). Interestingly, we found that laminin 5 did not trigger cleavage of iPLA 2 (data not shown), demonstrating the specificity of iPLA 2 activation. In addition, an activating antibody against ␤ 1 integrin stimulated iPLA 2 cleavage in cells not in contact with laminin-10/11 (Fig. 6A, last lane), confirming the role of ␤ 1 integrin and suggesting that ␤ 1 integrin activation is necessary and sufficient to induce iPLA 2 activation.
In an attempt to determine which ␣ integrin was involved, we pretreated cells with blocking antibodies against ␣ 1 , ␣ 2 , ␣ 3 , and ␣ 6 . However, none of these antibodies blocked the cleavage (Fig. 6B). ␣ 7 ␤ 1 is a specific integrin receptor for laminin. The lack of a commercially available neutralizing antibody against ␣ 7 integrin makes it difficult to test its effect directly and remains to be investigated.
Laminin-10/11-mediated Caspase-3 Activation Does Not Induce Apoptosis, and Akt Plays an Important Role in Blocking Apoptosis-As shown above, laminin-10/11-induced cleavage and activation of iPLA 2 is caspase-3dependent. Caspase-3 is an effector enzyme for apoptosis (26). To examine whether iPLA 2 -transfected HEY cells exposed to laminin-10/11 undergo apoptosis, we used Western blotting to assess cleavage of PARP, which is one of the targets of caspase-3 and a hallmark of apoptosis (27). As shown in Fig. 7A, no PARP cleavage was observed in parental, vector-, or iPLA 2 -transfected HEY cells exposed to laminin-10/11 (left panel). In contrast, both paclitaxel, a chemotherapeutic agent for the treatment of advanced breast and ovarian carcinoma (28), and sphingosylphosphorylcholine induced PARP cleavage (right panel) in HEY cells. A, wild type iPLA 2 -, iPLA 2 D513A-, and iPLA 2 D733A-transfected HEY cells were cultured for 4 h in the presence or absence of laminin-10/11. Collected cellular lysates were analyzed by Western blot with antibodies against iPLA 2 . B, HEY cell iPLA 2 D513A mutant, but not D733A, shows decreased migratory potential to laminin-10/11. Migration to laminin-10/11 was performed as described for Fig. 2. C, comparison of PLA 2 activity between the two point mutants and wild type iPLA 2 . Cell based activity assays were performed as described for Fig. 3B. Cells were cultured with or without contacting laminin-10/11 as described in Fig. 3B. Representative pictures are presented here, and statistical analysis of fluorescence intensity was performed as described for Fig. 3B. Ln-10/11, laminin 10-/11. These results have raised an intriguing concept; although caspase-3 is activated, resulting in iPLA 2 cleavage and activation, it appears that there is a cellular mechanism to control the consequences, so that the apoptotic pathway(s) is not activated. We have shown previously that iPLA 2 activation leads to LPA production, which activates Akt (13,29), an important anti-apoptotic signaling molecule. It is possible that the pathophysiologic role of laminin-10/11 is to induce cell migration, in which caspase-3-mediated activation of iPLA 2 is required; meanwhile, down-stream signals of iPLA 2 activation concomitantly activate Akt as a feedback loop to prevent cell apoptosis. To test this hypothesis, we pretreated iPLA 2 -tranfected HEY cells with an Akt inhibitor before exposure to laminin-10/11. We found that inhibition of Akt resulted in PARP cleavage (Fig. 7B, upper left  panel). The role of Akt was further confirmed by overexpressing a dominant negative version of Akt (DN-Akt) in this system, which resulted in a similar effect in PARP cleavage (Fig. 7B,  upper right panel). Moreover, when cells were treated with HELSS to block LPA production (and, thus, the predicted con-  (right panel). siRNA against GFP (100 nM) served as a negative control, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used for the loading control for reverse transcription-PCR analyses. For immunoblotting assay, HEY cells were treated with siRNA against iPLA 2 or glyceraldehyde-3-phosphate dehydrogenase (100 nM, 48 h), then Western blot analyses were performed, and ␤-actin was used for checking equal loading. B, down-regulation of iPLA 2 by siRNA decreased migration of HEY cells to laminin-10/11, whereas no ECM-coated Transwell was used as a control. Cells were pretreated with siRNA against iPLA 2 or GFP for 48 h before migration assays. Data are the average of two independent tests with error bars. C, caspase-3-dependent cleavage of endogenous iPLA 2 (left panel) and cell migration (right panel). HEY cells pretreated with or without caspase-3 selective inhibitor DEVD-CHO (2 M, 2h) were further incubated for 4 h in plates coated with laminin-10/11. Collected lysates were analyzed by Western blotting to detect the cleavage of endogenous iPLA 2 (left panel). For migration cells were pretreated with solvent or the indicated caspase inhibitors as described for Fig. 3C before migration assays. Representative results were presented here from two or three similar experiments. ***, p Ͻ 0.001 (Student's t test) versus HEY-Ln control. Ln, laminin-10/11. D, the effects of a caspase-3 siRNA (100 nM, 48 h) and a control siRNA (against GFP) on caspase-3 expression and cell migration to laminin-10/11. FIGURE 6. The role of integrin subunits in cleavage of iPLA 2 . A, ␤ 1 , but not ␤ 4 integrin is required for cleavage of iPLA 2 . Equal numbers of full-length iPLA 2 -transfected HEY cells were starved from serum for 16 h, pretreated with a neutralizing or an activating anti-integrin antibody (Ab, 5 g antibody/10 6 cells/ml were mixed and rotated for 1 h), and then cultured for an additional 4 h in plates precoated with or without laminin-10/11 (Ln). B, role of ␣ 1 , ␣ 2 , ␣ 3 , and ␣ 6 integrin subunits in laminin-10/11-induced iPLA 2 cleavage. Neutralizing antibodies against integrin subunits were used as described above. Data presented here are representative of two or more independent experiments.  To test whether another  caspase-3 substrate and apoptotic marker, DFF (30, 31), could also be regulated in a similar manner, we examined the cleavage of DFF. Similar to PARP, DFF was not cleaved in HEY cells contacting with laminin even though we showed that caspase-3 is modestly activated (see Fig. 7C). However, inhibiting Akt resulted in a complete cleavage of DFF (Fig. 7B, middle panel).
Together, these findings strongly support the hypothesis that LPA produced as a consequence of laminin-10/11-induced iPLA 2 cleavage activates Akt, which is important for preventing apoptosis in cells with activated caspase-3. We, thus, hypothesize that caspase-3 is only modestly activated in HEY cells contacting laminin-10/11, and apoptosis is not initiated under these circumstances due to protection from Akt. We directly compared the caspase-3 activities induced by contacting laminin-10/11 and paclitaxel in parental HEY cells. As shown in Fig. 7C, laminin-10/11 induced a significant, but modest increase in caspase-3 activity in parental HEY cells. In contrast, caspase-3 activation induced by paclitaxel was much stronger (Fig. 7C, left panel). In full-length iPLA 2 overexpressing cells, laminin-10/11 also induced caspase-3 activation, which was not significantly affected by Akt inhibitor (Fig. 7C,  right panel), indicating that caspase-3 is upstream of the action of Akt. Taken together, these data demonstrated that unlike the apoptotic pathway induced by paclitaxel, laminin-10/11 induces moderate caspase-3 activity, which activates iPLA 2 but not apoptosis. Other factors, such as subcellular locations and co-factors associated under different conditions may also be involved in the differential target cleavage of caspase 3.
In summary, we have revealed novel signaling pathways/ loops initiated by laminin-10/11 in iPLA 2 -expressing ovarian cancer cells (Fig. 8). Laminin-10/11, via ␤ 1 integrin-mediated signals(s), leads to caspase-3 activation. Consequently, iPLA 2 is cleaved and activated to produce LPA, possibly leading to cPLA 2 activation and production of AA, a key mediator for cell migration (part of these actions are shown in our previous publication, Tang et al. (32)). Concomitantly, LPA-activated Akt sends a feedback signal preventing PARP cleavage and apoptosis (Fig. 8).

DISCUSSION
We have demonstrated a novel mechanism of iPLA 2 activation stimulated by ECM protein, laminin-10/11. Activation of iPLA 2 in this way plays a critical role in cell migration, which is involved in many important biological processes, such as development, immunological and inflammatory responses, and tumor biology. Laminin-10/11-induced activation of iPLA 2 involves caspase-3-dependent cleavage of iPLA 2 . Although such activation has been previously reported under apoptotic conditions (7,11), the current work is novel and significant in the following aspects. 1) We show for the first time that caspase-3 can be activated by laminin-10/11 via an ␤ 1 integrin; 2) we have shown that caspase-3 activation induced by laminin-10/11 is moderate and under tight control, leading to iPLA 2 cleavage at Asp 513 and activation but not apoptosis; 3) we identify a feedback loop involving LPA production and Akt activation to control the consequences of caspase-3 activation; 4) arachidonic acid has a strong chemokinetic activity in ovarian cancer cells. These findings not only expand our understanding of the cellular functions and regulatory mechanisms of iPLA 2 but also reveal a novel function of the classical apoptotic caspase under non-apoptotic conditions. In addition, we have linked these activities to cell migratory activity, a critical component of tumor metastasis.
The role(s) of the ankyrin repeats in iPLA 2 have been studied in previous work. Tang et al. (32) have shown that when truncated iPLA 2 isoforms (removal of the ankyrin repeats) were expressed in COS cells, the iPLA 2 activity was lost using the micelles iPLA 2 assays and, thus, concluded that the ankyrin repeats are required for activity. However, results suggesting that the ankyrin repeats play a potential negative role have been shown by several groups (7,11,18). Larsson et al. (18) have shown that certain human cells express multiple splice variants of iPLA 2 , including two forms (ankyrin-iPLA 2 -1 and ankyrin iPLA 2 -2) without exons encoding the C-terminal half of iPLA 2 . No activity above background was observed when ankyrin-iPLA 2 -1 cDNA was transfected into COS cells. However, cotransfection of ankyrin-iPLA 2 -1 and iPLA 2 cDNAs resulted in a 2-fold reduction in activity compared with iPLA 2 alone. Atsumi et al. (7) showed that caspase-3 cleaves iPLA 2 . The truncated iPLA 2 lacking most of the first ankyrin repeat increases iPLA 2 functions, including regulation of cell growth and apoptosis. Lauber at el. (11) have demonstrated that removal of ankyrin repeat overexpression iPLA 2 (aa 514 -806) and iPLA 2 (aa 514 -733)) without overexpression of the ankyrin repeat domain) increases iPLA 2 -mediated release of chemotactic activity. In addition, Manguikian and Barbour (34) have shown that a splice variant iPLA 2 (ankyrin-iPLA 2 -1) mRNAs are preferentially expressed during the later G 1 and S phase cells, which can Laminin-10/11, through activating ␤ 1 integrin, triggers the activation of caspase-3, leading to cleavage and activation of iPLA 2 , which produces LPA. LPA activates its own receptor and downstream signaling pathways, including phosphatidylinositol 3-kinase (PI3K)-extracellular signal-regulated kinase-p38 mitogen-activated protein kinase-dependent Akt activation (29) and PI3K-dependent cPLA 2 activation (13). cPLA 2 is responsible for LPA-induced production of AA, which is the major mediator of cell migration. Akt activation forms a feedback loop to block PARP cleavage by caspase-3 and prevent cells from apoptosis. be co-immunoprecipitated with the full-length iPLA 2 and inhibits its activity.
Consistent with work by Ackermann et al. (33), Larsson et al. (18) have also suggested that the ankyrin repeats may be required in oligomerization of iPLA 2 . Ackermann et al. (33) have provided evidence to show that activating iPLA 2 exists as a large oligomeric complex either through self-aggregation or association of the enzyme with other proteins. However, the authors did not directly address the role of the ankyrin repeats and did not identify the protein(s) present in the complex.
Nevertheless, all of the published data and our data presented in this work may be complementary rather than contradictory. None of the published data can rule out the potential positive role and requirement of the endogenous ankyrin repeat domain in iPLA 2 activation, although overexpression of these motifs may have a negative impact on iPLA 2 activity. These issues remain to be further addressed.
We have found that the enzymatic activity of iPLA 2 is required for its migratory effect using both a truncated iPLA 2 form without the catalytic site and a point mutation at the site. However, we did not observe a simple linear relationship between the expression levels of iPLA 2 , its activity (as measured by either the micelles-based or fluorescent cell-based methods), and migratory activity. It is possible that there is a threshold, and even low levels of PLA 2 activity are sufficient to induce cell migration. There are several possibilities why increased PLA 2 activity may not be linearly proportional to migratory activity. 1) AA, which is produced after PLA 2 activation and is a major component responsible for cell migration, does not induce cell migration in a linear fashion; 2) different PLA 2 forms may also differentially induce or alter the expression/secretion of other factors, which may positively or negatively modulate migratory activity; 3) different PLA 2 forms may also have differential effects in iPLA 2 complex formation and may play different structural modulating roles. In another words, although our data, particularly the data derived from the S519A catalytic site mutant, strongly support the notion that the catalytic activity of iPLA 2 is required for its migratory activity, other potential effect(s) of iPLA 2 unrelated to its catalytic activity may also be involved in modulating the migratory activity.
Laminin-5 and -10 are the most abundant laminin molecules in several human cancers including ovarian carcinomas (35). The roles of laminins in ovarian cancer development and in tumor metastasis and proliferation in particular have been demonstrated (36,37). In addition, ovarian cancer cells express high levels of ␤ 1 integrin (21)(22)(23). We find that activation of iPLA 2 in non-apoptotic ovarian cancer cells is highly specific. The laminin used in this and the previous study (13) was obtained from Chemicon International and is mainly composed of laminin-10/11. Laminin-10/11 has been reported to be specifically recognized by integrin ␣ 3 ␤ 1 , ␣ 6 ␤ 1 , ␣ 6 ␤ 4 , and ␣ 7 ␤ 1 (38,39). We also tested laminin-5, which uses ␣ 6 ␤ 4 integrin as its main receptor and found that it was ineffective in inducing iPLA 2 activation. These results suggest that the integrin receptor(s) determines the specificity of laminin-10/11-induced iPLA 2 activation. Although it is clear that the ␤ 1 integrin subunit is involved in cell migration to laminin-10/11 (13), to determine which ␣ integrin subunit is involved in the process may be more complicated and need more extensive studies.
Proteolytic digestion by caspase-3 to remove the N-terminal portion of iPLA 2 under apoptotic conditions is an intriguing mechanism to activate iPLA 2 (7,11). However, iPLA 2 activation appears to be involved in many conditions unrelated to apoptosis as well (5,9). Several studies reveal a role for ECM proteins and integrins in protecting cells from apoptosis (40,41). In fact, a specific form of apoptosis, termed anoikis, is associated with integrin inactivation and/or disruption of cell attachment (42). To our knowledge, an ECM-and/or integrin-mediated apoptotic caspase activation has not been previously reported. In this work, we have found that non-apoptotic cells employ a similar mechanism of proteolytic processing by caspase-3 to activate iPLA 2 . This observation raises an important biological question of how cells regulate the consequences of caspase-3 activation and prevent apoptosis under certain conditions. In one case it was reported that when caspases are mildly activated, the partial cleavage of a caspase substrate, RasGAP, protects cells from apoptosis (43). Data presented here provide some clues regarding the control of laminin-induced caspase-3 activity. First, we demonstrate that the major product of iPLA 2 activation induced by laminin-10/11 is a truncated iPLA 2 (aa 514 -806) that is different from the major form produced under apoptotic conditions (iPLA 2 aa 514 -733) (11). We show here that both truncated forms are active, and that the shorter form (aa 514 -733) is generally more active than the longer (aa 514 -806) ( Figs. 1 and 2). This suggests that fully activated caspase-3 generated under apoptotic conditions is required to produce the shorter and more active form of iPLA 2 . Laminin-10/11 only induces modest activation of caspase-3, which leads to cleavage of iPLA 2 only at Asp 513 and does not result in apoptosis. Second, one of the concomitant consequences of iPLA 2 activation induced by laminin-10/11 is LPA production and Akt activation (our previous work in Refs. 13 and 29 and this work). The anti-apoptotic and cell survival activity of Akt are well established (44 -48). We show here that activation of Akt by LPA composes an elegant feedback loop to prevent PARP cleavage and apoptosis. Thus, our results show that "classical" apoptotic caspases (such as caspase-3) may also be involved in non-apoptotic cellular processes. This is consistent with recent findings showing that classical apoptotic caspases are involved in nonapoptotic biological processes such as development (49). How laminin-10/11/␤ 1 integrin activates caspase-3 remains to be investigated.
As reported previously, the major lysophospholipid produced under apoptotic conditions is LPC, not LPA (11). In contrast, laminin-10/11 mainly induces LPA production rather than LPC. The molecular mechanisms controlling this selective production of lysophospholipids remain to be further investigated.
We have established here that the enzymatic activity of iPLA 2 is required for ovarian cancer cell migration and that AA, instead of LPA, accounts for this activity. This appears to be contradictory to our previous report showing that LPA stimulates cell migration (13). However, there is a major difference between the assays performed in the two studies. In the previous study, endogenous iPLA 2 is not activated until the cells contact laminin facing the lower chamber, and thus, LPA is produced in the lower chamber. As shown previously, LPA is a strong chemotactic factor but poor activator for chemokinetics in ovarian cancer cells (13). In contrast, in the current study we have expressed activated iPLA 2 in cells in the upper chamber, resulting in production of AA, which is a much stronger activator for chemokinetics than chemoattractant (see Fig. 2D). AA may be produced by cPLA 2 , a specific PLA 2 for AA liberation (50). As we reported previously, activation of iPLA 2 triggers cPLA 2 activation (13). iPLA 2 may also directly contribute to AA production.
In summary, our data reveal a novel inducer of iPLA 2 activation and its role in ovarian cancer cell migration. Laminin-10/ 11, through integrin ␤ 1 -mediated signals, triggers the cleavage of iPLA 2 via a caspase-3 dependent pathway, resulting in the formation of truncated active iPLA 2 . The activation of iPLA 2 increases the production of LPA and AA, with the former playing an important role in preventing apoptosis, and the latter playing a more direct role in cell migration (Fig. 8). These findings not only expand our understanding of the molecular mechanisms of iPLA 2 activation but also provide critical information on the mechanisms of cell migration and the metastatic potential of ovarian cancer cells.