Functional Association of Type IIA Secretory Phospholipase A2 with the Glycosylphosphatidylinositol-anchored Heparan Sulfate Proteoglycan in the Cyclooxygenase-2-mediated Delayed Prostanoid-biosynthetic Pathway*

An emerging body of evidence suggests that type IIA secretory phospholipase A2(sPLA2-IIA) participates in the amplification of the stimulus-induced cyclooxygenase (COX)-2-dependent delayed prostaglandin (PG)-biosynthetic response in several cell types. However, the biological importance of the ability of sPLA2-IIA to bind to heparan sulfate proteoglycan (HSPG) on cell surfaces has remained controversial. Here we show that glypican, a glycosylphosphatidylinositol (GPI)-anchored HSPG, acts as a physical and functional adaptor for sPLA2-IIA. sPLA2-IIA-dependent PGE2 generation by interleukin-1-stimulated cells was markedly attenuated by treatment of the cells with heparin, heparinase or GPI-specific phospholipase C, which solubilized the cell surface-associated sPLA2-IIA. Overexpression of glypican-1 increased the association of sPLA2-IIA with the cell membrane, and glypican-1 was coimmunoprecipitated by the antibody against sPLA2-IIA. Glypican-1 overexpression led to marked augmentation of sPLA2-IIA-mediated arachidonic acid release, PGE2 generation, and COX-2 induction in interleukin-1-stimulated cells, particularly when the sPLA2-IIA expression level was suboptimal. Immunofluorescent microscopic analyses of cytokine-stimulated cells revealed that sPLA2-IIA was present in the caveolae, a microdomain in which GPI-anchored proteins reside, and also appeared in the perinuclear area in proximity to COX-2. We therefore propose that a GPI-anchored HSPG glypican facilitates the trafficking of sPLA2-IIA into particular subcellular compartments, and arachidonic acid thus released from the compartments may link efficiently to the downstream COX-2-mediated PG biosynthesis.

linked with the downstream cyclooxygenase (COX) and lipoxygenase pathways for eicosanoid biosynthesis, is a highly regulated cellular response that requires gene induction and/or posttranslational modification of a group of regulatory enzymes, namely phospholipase A 2 (PLA 2 ) (1). An expanding recognition of the structural and functional diversity of mammalian PLA 2 enzymes has revealed that the two major classes of Ca 2ϩ -dependent PLA 2 s, namely 85-kDa cytosolic PLA 2 ␣ (cPLA 2 ; type IV) and 14-kDa secretory PLA 2 (sPLA 2 ) isozymes (types IIA and V), act as "signaling" PLA 2 s, which contribute to the release of AA from agonist-stimulated cells, depending upon the phase of cell activation (2,3). Among them, cPLA 2 has received much attention as a key regulator of stimulus-initiated eicosanoid biosynthesis, because it selectively releases AA, shows submicromolar Ca 2ϩ sensitivity, and is activated by mitogen-activated protein kinase-directed phosphorylation (4,5). cPLA 2 undergoes Ca 2ϩ -dependent translocation from the cytosol to perinuclear and endoplasmic reticular membranes (6,7), where several downstream eicosanoid-generating enzymes, including two COX isozymes, are localized (8). Studies on cPLA 2 -deficient mice have confirmed its critical role in lipid mediator generation during the acute allergic response, parturition, and postischemic brain injury (9,10).
Among several members of the sPLA 2 family, sPLA 2 -IIA is the most widely distributed isozyme in humans and rats (11). The expression of sPLA 2 -IIA is often dramatically up-regulated by proinflammatory stimuli, such as bacterial endotoxin, interleukin (IL)-1, and tumor necrosis factor (TNF) (12)(13)(14)(15), and is down-regulated by glucocorticoids (16). Raised sPLA 2 -IIA levels at inflamed sites suggest that it plays a crucial role in the propagation of inflammatory responses (17)(18)(19), which has been further supported by recent in vivo studies (20,21). Current in vitro studies suggest that sPLA 2 -IIA can amplify stimulus-initiated AA metabolism, particularly the delayed prostaglandin (PG)-biosynthetic response, which is accompanied by de novo synthesis of sPLA 2 -IIA and COX-2 (2,3,12,14,22,23). In the mouse, sPLA 2 -V, a close relative of sPLA 2 -IIA, may replace sPLA 2 -IIA under certain conditions (2,3,24,25). However, the molecular mechanisms whereby these sPLA 2 s regulate AA metabolism are still poorly understood. sPLA 2 s-IIA and -V have high affinities for heparanoids (2), and significant portions of these isozymes are associated with the cell surface, most likely through binding to heparan sulfate proteoglycans (HSPGs), which are expressed in most mammalian cells. We (2,14,22,23,26,27) and others (28,29) have shown that some of the cellular functions of these heparinbinding sPLA 2 s depend on their cell surface HSPG-binding abilities. Association of sPLA 2 -IIA with heparan or chondroitin sulfate chains increases the hydrolytic rate of phosphatidylcholine present in lipoprotein particles modestly (30,31). On the other hand, some reports have indicated that the actions of exogenous sPLA 2 -IIA on cells depend only on its interfacial interaction with substrate phospholipids rather than on its association with HSPGs (32).
The cell surface HSPGs fall into two families of molecules that differ in their core protein domain structures (33). The syndecans have core proteins with a transmembrane and a cytoplasmic domain, and they possess heparan and/or chondroitin sulfate chains near the N terminus distal to the plasma membrane (34). The glypicans, by contrast, lack a membranespanning domain, are anchored to the external surface of the plasma membrane via glycosylphosphatidylinositol (GPI), and have three heparan sulfate chains near the C terminus, which are close to the plasma membrane (35). Consistent with a GPI-anchored moiety, glypicans are mobile in the cell membrane and exhibit both apical and basolateral distributions, whereas syndecans are distributed basolaterally to be attached to extracellular matrix proteins (36). Interestingly, recent immunohistochemical studies have revealed that significant portion of glypican translocates to the nucleus in cells undergoing cell division and activation (37). There is extensive literature concerning glycosaminoglycans in the nuclear compartment (38 -40). These findings appear to be compatible with the observations that several extracellular heparin-binding growth factors, such as fibroblast growth factor (FGF) and angiogenin, translocate into the nucleus via a HSPG-dependent route (41)(42)(43)(44).
In an effort to clarify the role of sPLA 2 -IIA in the regulation of the PG-biosynthetic pathway, we have identified the cellular component that is functionally associated with sPLA 2 -IIA. We found that a GPI-anchored HSPG glypican acts as a cellular sPLA 2 -IIA-binding partner that contributed to enhancement of the sPLA 2 -IIA-mediated, cytokine-induced, delayed PG-biosynthetic response. In agreement with the emerging notion that GPI-anchored proteins reside in the microdomains called caveolae (45)(46)(47)(48)53), which are vesicular invaginations of the plasma membrane that are rich in signal-transducing molecules and implicated in vesicular transport and potocytosis between the plasma and intracellular membranes (49,50), sPLA 2 -IIA was found to accumulate in the caveolae and perinuclear sites, rather than being distributed uniformly on the cell surface as we had previously thought.

EXPERIMENTAL PROCEDURES
Materials-Human embryonic kidney (HEK) 293 cells were obtained from the Health Science Research Resources Bank, rat liver-derived BRL-3A cells were from RIKEN Cell Bank, and rat fibroblastic 3Y1 cells were from Dr. Y. Uehara (National Institute of Health, Tokyo). The culture conditions for these cell lines have been described previously (2,3,14,27). The cDNAs for mouse sPLA 2 -IIA and its heparin non-binding mutant KE4 (22), mouse cPLA 2 , human COX-1, and human COX-2 were described previously (2,3). The cDNA for rat glypican-1 was provided by Dr. R. Margolis (New York University Medical Center, New York, NY). The rabbit anti-human cPLA 2 antibody was provided by Dr. R. M. Kramer (Lilly Research). Preparation of the rabbit anti-rat sPLA 2 -IIA antibody and its conjugation with cyanogen bromide-activated Sepharose (Amersham Pharmacia Biotech) were described previously (51). The goat anti-human COX-2 antibody and rabbit anti-human caveolin-2 antibody were purchased from Santa Cruz Biotechnology. The rabbit anti-human COX-1 antibody was provided by Dr. W. L. Smith (Michigan State University, Ann Arbor, MI). The PGE 2 enzyme immunoassay kit was purchased from Cayman Chemical. Human TNF␣ was provided by Dr. H. Ishimaru (Asahi Chemical Industry). Human and mouse IL-1␤s were purchased from Genzyme. Lipo-fectAMINE PLUS reagent, Opti-MEM medium, and TRIzol reagent were obtained from Life Technologies, Inc. RPMI 1640 medium was purchased from Nissui Pharmaceutical. Bacillus cereus GPI-specific phospholipase C (GPI-PLC) was purchased from Roche Molecular Bio-chemicals. Heparin and Flavobacterium heparinum heparinase III were purchased from Sigma. Fluorescein isothiocyanate (FITC)-conjugated goat anti-mouse IgG, FITC-rabbit anti-goat IgG ,and FITC-goat anti-rabbit IgG antibodies were purchased from Zymed Laboratories Inc. Cy3-conjugated donkey anti-rabbit IgG antibody was from Chemicon.
Establishment of Transfectants-Establishment of 293 cell transformants that stably expressed sPLA 2 -IIA, cPLA 2 , COX-1, and COX-2 was described previously (2,3). Briefly, 1 g of each cDNA subcloned into pcDNA3.1 (Invitrogen) was mixed with 5 l of LipofectAMINE PLUS in 200 l of Opti-MEM medium for 30 min and then added to cells that had attained 40 -60% confluence in six-well plates (Iwaki) containing 1 ml of Opti-MEM. After incubation for 6 h, the medium was replaced with 2 ml of fresh culture medium comprising RPMI 1640 containing 10% (v/v) fetal calf serum (FCS). After overnight culture, the medium was replaced again with 2 ml of fresh medium and culture was continued at 37°C in a CO 2 incubator flushed with 5% CO 2 in humidified air. In order to establish stable transfectants, cells transfected with each cDNA were cloned by limiting dilution in 96-well plates in culture medium supplemented with 800 g/ml Geneticin (Life Technologies, Inc.). After culture for 3-4 weeks, wells containing a single colony were chosen and the expression of each protein was assessed by immunoblotting. The established clones were expanded and used for the experiments as described below.
In order to establish sPLA 2 -IIA/glypican-1 double transformants, 293 transformants expressing sPLA 2 -IIA were subjected to a second transfection with glypican-1 cDNA, which had been subcloned into pcDNA3.1/Zeo(ϩ) (Invitrogen) at the EcoRI site. Three days after transfection, the cells were used for the experiments or seeded into 96-well plates to be cloned by culture in the presence of 50 g/ml zeocin (Invitrogen) in order to establish stable transformants overexpressing both sPLA 2 -IIA and glypican-1. The expression of each was examined by immunoblotting, RNA blotting, and, in the case of sPLA 2 -IIA, by measuring the PLA 2 activity of the supernatants.
RNA Blotting-Approximately equal amounts (ϳ10 g) of the total RNAs obtained from the transfected cells were applied to separate lanes of 1.2% (w/v) formaldehyde-agarose gels, electrophoresed, and transferred to Immobilon-N membranes (Millipore). The resulting blots were then probed with the respective cDNA probes that had been labeled with [ 32 P]dCTP (Amersham Pharmacia Biotech) by random priming (Takara Shuzo). All hybridizations were carried out as described previously (22).
Cell Activation-293 cells (5 ϫ 10 4 /ml) were seeded into each well of 24-or 48-well plates. To assess AA release, 0.1 Ci/ml [ 3 H]AA (Amersham Pharmacia Biotech) was added to the cells in each well on day 3, when they had nearly reached confluence, and culture was continued for another day. After three washes with fresh medium, 250 l (24-well plate) or 100 l (48-well plate) of RPMI 1640 with or without 1 ng/ml IL-1␤ and/or 10% FCS was added to each well and the amount of free [ 3 H]AA released into the supernatant during culture for 4 h was measured. The percentage release of AA was calculated using the formula [S/(S ϩ P)] ϫ 100, where S and P are the radioactivities measured in equal portions of the supernatant and cell pellet, respectively. The supernatants from replicate cells were subjected to the PGE 2 enzyme immunoassay. AA release and PG generation by [ 3 H]AA-prelabeled cells were also assessed by thin layer chromatography. Among the radiolabeled products released, Ͼ90% were AA and the rest (Ͻ10%) corresponded to PGs. Among PGs, PGE 2 was the major product, fol-lowed by modest production of PGD 2 and PGF 2␣ . Therefore it is likely that the radioactivity released into the supernatants largely reflects [ 3 H]AA release.
Culture and cytokine stimulation of 3Y1 (14) and BRL-3A (27) cells were performed according to our previous studies with slight modifications. In brief, 3Y1 or BRL-3A cells that had attained 60 -80% confluence in 12-well plates (Iwaki) were replaced with Dulbecco's modified Eagle's medium (Nissui) containing 2% FCS. After overnight culture, the cells were stimulated with 1 ng/ml mouse IL-1␤ and 100 units/ml human TNF␣ for 24 h in the medium containing 10% FCS.
Immunoaffinity Column Chromatography Using Anti-sPLA 2 -IIA Antibody-HEK293 cells coexpressing sPLA 2 -IIA and glypican-1 were grown in a 150-mm diameter dish, washed once with phosphate-buffered saline (PBS), and lysed in 10 ml of PBS containing 1% Nonidet P-40 (Nakalai Tesque), 50 g/ml leupeptin (Sigma), 1.5 M pepstatin (Peptide Institute), 1 mM phenylmethanesulfonyl fluoride (Wako), and 5 mM EDTA (cell lysis buffer). After incubation for 30 min at 4°C, the crude nuclear fraction was obtained by low spin centrifugation at 450 ϫ g, as reported previously (52). The remaining supernatants were centrifuged for 1 h at 100,000 ϫ g at 4°C, and the resulting supernatants were applied to a rabbit anti-sPLA 2 -IIA antibody-conjugated Sepharose column. After applying the samples, the column was washed with the cell lysis buffer. The bound proteins were eluted with glycine-HCl buffer (pH 2). In separate experiments, the column was washed with the cell lysis buffer containing 1 M NaCl, followed by elution with glycine-HCl buffer.
Statistical Analysis-Data were analyzed by Student's t test. Results are expressed as the mean Ϯ S.E., with p ϭ 0.05 as the limit of significance.

Binding of sPLA 2 -IIA to GPI-Anchored HSPG Is Essential for
Its PG-Biosynthetic Activity-We have previously shown that HEK293 cells transfected with sPLA 2 -IIA cDNA mainly express a cell membrane-associated form of sPLA 2 -IIA, which appears to play an important role in the promotion of IL-1induced, COX-2-dependent delayed PGE 2 generation (2, 3). To verify whether sPLA 2 -IIA is indeed associated with the HSPG moiety on 293 cell surfaces, we treated the cells with heparinase or exogenous heparin and then looked for the appearance of sPLA 2 -IIA in their supernatants (Fig. 1, A and B). There was modest sPLA 2 -IIA release, which we detected by enzyme assay (Fig. 1A) and very faintly by immunoblotting (Fig. 1B), from the sPLA 2 -IIA-transfected, but not control, cells. sPLA 2 -IIA activity in the supernatant increased time-dependently when the sPLA 2 -IIA-expressing, but not control, cells were cultured in the presence of heparinase, which degrades heparan sulfate chains (27,28,(35)(36)(37) (Fig. 1A). Pretreatment of the sPLA 2 -IIA-expressing, but not control, cells with heparin, which competes with cell surface HSPG for heparin-binding proteins (14, 22, 26 -29), also caused an increase in sPLA 2 -IIA activity in the culture media (Fig. 1A), the result essentially consistent with our previous observations (2). This heparinase (data not shown) or heparin (see Fig. 5) treatment solubilized the cellassociated sPLA 2 -IIA almost completely. The increased release of sPLA 2 -IIA into the supernatants of the sPLA 2 -IIA-express-ing cells after treatment with heparinase or heparin was confirmed by immunoblotting (Fig. 1B). sPLA 2 -IIA mRNA expression in the sPLA 2 -IIA-transfected cells was unaffected after treatment with heparinase or heparin (data not shown). sPLA 2 -IIA-expressing, but not control, cells treated with IL-1 produced a significant amount of PGE 2 (Fig. 1C), the production of which depended upon inducible COX-2, as reported previously (2,3). Treatment of the sPLA 2 -IIA-expressing cells with heparinase suppressed this PGE 2 generation markedly; the kinetics indicated an inverse relationship between augmented sPLA 2 -IIA release (Fig. 1A) and inhibition of PGE 2 generation (Fig. 1C). Treatment of replicate cells with heparin also led to a nearly 80% reduction of IL-1-induced PGE 2 generation. These results, together with our previous findings that a heparin-nonbinding sPLA 2 -IIA mutant KE4, in which a cluster of four lysine residues in the C-terminal domain is replaced by glutamic acid, is not associated with the cell surface and does not augment IL-1-induced AA metabolism (2,3,22), strongly suggest that cellular HSPG is required for the full action of sPLA 2 -IIA in this experimental system.
As there are two families of cellular HSPGs, the integral syndecans and the GPI-anchored glypicans (33)(34)(35), we wanted to know which HSPG species is the major sPLA 2 -IIA-binding target. To address this issue, we used GPI-PLC, which cleaves the GPI linkage and is thereby capable of solubilizing GPIanchored plasma membrane proteins and their associated proteins. We found that treatment of the sPLA 2 -IIA-expressing, but not control, 293 cells with GPI-PLC markedly increased the amount of soluble sPLA 2 -IIA, as assessed by both enzyme assay ( Fig. 2A) and immunoblotting ( Fig. 2A, inset), and reduced sPLA 2 -IIA-mediated PGE 2 generation in response to IL-1 by nearly 80% (Fig. 2B). These results suggest that binding to a GPI-anchored HSPG glypican is required for sPLA 2 -IIA to exert its PG-biosynthetic function.
Binding of sPLA 2 -IIA to Glypican-To obtain more convincing evidence for the interaction between sPLA 2 -IIA and the GPI-anchored HSPG glypican, we attempted to establish sPLA 2 -IIA/glypican-1 double transfectants. Fig. 3A depicts the expression levels of transcripts for sPLA 2 -IIA and glypican-1 in HEK293 cells transfected with their cDNAs alone or in combination. Endogenous glypican-1 protein was faintly detected in 293 cells (Fig. 3A, clones 1, 3, and 5), and was significantly elevated in the glypican-1 transfectants (clones 2, 4, and 6). Several clones with different expression levels of sPLA 2 -IIA were selected, including those expressing sPLA 2 -IIA at a low level (clones 3 and 4), a very high level (clones 5 and 6), and without expression (clones 1 and 2).
The culture supernatants of 293 clones with high sPLA 2 -IIA expression (clones 5 and 6) were collected after 3 days of culture, and the cells were then washed for 15 min with culture medium containing 1 M NaCl to solubilize the cell surfaceassociated sPLA 2 -IIA, by a procedure that was reported earlier (2). The ratio of sPLA 2 -IIA released into the supernatants to that associated with the cell surface increased from 1:3 in cells overexpressing sPLA 2 -IIA alone (clone 5) to 1:10 in cells overexpressing both sPLA 2 -IIA and glypican (clone 6), as assessed by enzyme assay. Hence, glypican-1 overexpression revealed a more than 3-fold increase in the cells' capacity to capture sPLA 2 -IIA.
The 100,000 ϫ g supernatant of Nonidet P-40-solubilized cells coexpressing sPLA 2 -IIA and glypican-1 (clone 6) was analyzed by immunoprecipitation using an anti-sPLA 2 -IIA antibody-conjugated Sepharose column, followed by immunoblot-ting with anti-sPLA 2 -IIA and anti-glypican-1 antibodies (Fig.  4A). After washing the column with cell lysis buffer, proteins bound to the column were eluted with an acidic buffer (pH 2). As anticipated, a 14-kDa protein band corresponding to sPLA 2 -IIA was detected in fractions eluted from the column, but not in the flow-through and washing fractions (Fig. 4A). Notably, a 64-kDa glypican-1 protein band was detected in the acid-eluted fractions in which sPLA 2 -IIA was also recovered, whereas the flow-through and washing fractions contained only a trace level of glypican-1 (Fig. 4A). When the replicate immunoprecipitate was washed with the cell lysis buffer containing 1 M NaCl, which was expected to facilitate the dissociation of sPLA 2 -IIA from the heparan sulfate chains of glypican-1 without disturbing the interaction between sPLA 2 -IIA and the anti-sPLA 2 -IIA antibody, and then the antibody-bound protein was eluted with the acidic buffer, most of the glypican-1 was recovered from the NaCl-eluted fraction with only a minor portion being retained until elution with the acidic buffer, whereas sPLA 2 -IIA was detected exclusively in the acidic buffer fractions (Fig. 4B). These results support the idea that sPLA 2 -IIA physically interacts with glypican-1, most likely through its heparan sulfate chains, in 293 cells.
Functional Interaction between sPLA 2 -IIA and Glypican-To assess the functional significance of the sPLA 2 -IIA/glypican interaction, the established transfectants (Fig. 3A, clones 1-6) were prelabeled with [ 3 H]AA and delayed [ 3 H]AA release in response to IL-1 was investigated (Fig. 3B, upper panel). Cells expressing a high level of sPLA 2 -IIA showed increased AA release up to 2.7% (clone 5) relative to the control cells, which released no more than 0.7% AA (clone 1). AA release by the sPLA 2 -IIA high expression cells (clone 5) was augmented to 4.4% by glypican-1 coexpression (clone 6), whereas the expression of glypican-1 alone increased AA release only minimally (clone 2), revealing a synergistic action between the two. The augmentative role of glypican-1 in sPLA 2 -IIA-mediated AA release was more obvious when the sPLA 2 -IIA expression level was low (clone 3); AA release by these cells was comparable to that of the control cells (clone 1), indicating that this level of sPLA 2 -IIA was insufficient to modulate IL-1-induced AA release, yet introduction of glypican-1 into these cells dramatically enhanced the AA release to a maximal and saturable level (clone 4).
Comparison of the increases in AA release and PGE 2 generation in the clones with high sPLA 2 -IIA expression (clones 5 and 6) revealed an approximately 1.8-fold increase in AA and a 4-fold increase in PGE 2 following glypican-1 coexpression, reflecting greater sensitivity of PGE 2 generation than of AA release to the action of sPLA 2 -IIA. This fact led us to examine whether overexpression of sPLA 2 -IIA and/or glypican affected the expression of endogenous COX-2, a COX isozyme predominantly involved in the conversion of AA released by sPLA 2 -IIA to PGE 2 during the delayed response (2,3,14,22). COX-2 expression, which was induced modestly in the control cells by IL-1 stimulation, was significantly augmented in cells transfected with sPLA 2 -IIA alone (Fig. 3C). Cells transfected with glypican-1 alone expressed COX-2 even before IL-1 stimulation at a level comparable to that observed in IL-1-stimulated control 293 cells, and IL-1 stimulation increased its expression further in the glypican-1-expressing cells. Most importantly, coexpression of sPLA 2 -IIA and glypican-1 increased COX-2 expression in a synergistic manner in the presence, but not in the absence, of IL-1 stimulation (Fig. 3C). These results are reminiscent of our recent finding that sPLA 2 -V, another heparinbinding sPLA 2 isozyme, augments IL-1 induction of COX-2 expression in a heparin-sensitive manner, whereas the heparin-nonbinding sPLA 2 -IIA mutant KE4 and the heparin-nonbinding isozyme sPLA 2 -X fail to do so (53).
Collectively, these results suggest that (i) glypican acts as an adaptor for sPLA 2 -IIA, probably by bringing it close to the cellular membrane through heparan sulfate chains near the C terminus; (ii) glypican facilitates sPLA 2 -IIA-mediated AA release from IL-1-stimulated cells; (iii) glypican and sPLA 2 -IIA act in synergy to enhance IL-1-induced COX-2 expression; and (iv) increased AA release and COX-2 induction by the coordinated actions of glypican and sPLA 2 -IIA lead to marked elevation of PGE 2 generation. Subcellular Localization of sPLA 2 -IIA Overexpressed in HEK293 Cells--Immunostaining of permeabilized sPLA 2 -IIAexpressing 293 cells showed that sPLA 2 -IIA accumulated in punctate domains and in the perinuclear area (Fig. 5, A and B). The punctate domains were evident, whereas the perinuclear signal was barely found, in non-permeabilized cells (Fig. 5C). Incubation of sPLA 2 -IIA-expressing cells with heparin abolished the punctate sPLA 2 -IIA signal, whereas some perinuclear staining remained in permeabilized (Fig. 5D), but not in non-permeabilized (Fig. 5E), cells. Cells expressing the heparin-nonbinding sPLA 2 -IIA mutant KE4, which was exclusively released into the supernatant (2, 22), were not stained appreciably by the anti-sPLA 2 -IIA antibody (Fig. 5F), although RNA blotting showed that the expression level of KE4 was comparable with that of the native enzyme in the respective transfectants (2). Based on these observations, we hypothesize that sPLA 2 -IIA is sorted predominantly into punctate microdomains on the plasma membrane that are sensitive to washing with exogenous heparin, and that a small portion is transported into the perinuclear area, possibly by the vesicular transport machinery. Importantly, double antibody staining of sPLA 2 -IIA-expressing cells revealed that sPLA 2 -IIA was colocalized with caveolin-2 (Fig. 5G). This result implies that the punctate domains in which sPLA 2 -IIA resided are caveolae, organella that contain GPI-anchored proteins and may be responsible for potocytotic vesicular transport (49,50). In support of this speculation, subcellular fractionation studies showed the presence of both immunoreactive sPLA 2 -IIA and glypican-1 in the nucleus-enriched fraction (Fig. 3A).
The subcellular localization of cPLA 2 and two COX isoforms, COX-1 and COX-2, in the respective HEK293 transfectants was also investigated. cPLA 2 was detected throughout the cytoplasm in unstimulated cells (Fig. 5H), and IL-1 stimulation promoted the translocation of cPLA 2 to the perinuclear region FIG. 4. Glypican-1 binding to sPLA 2 -IIA. A, the 100,000 ϫ g supernatant of Nonidet P-40-solubilized 293 cells that coexpressed sPLA 2 -IIA and glypican-1 was applied to an anti-sPLA 2 -IIA antibody affinity column. Bound proteins were eluted with an acidic buffer (pH 2), and each fraction was analyzed by immunoblotting using antibodies against glypican-1 or sPLA 2 -IIA. Right, a crude nuclear fraction obtained from 293 cells that coexpressed sPLA 2 -IIA and glypican-1 was immunoblotted using the same antibodies. B, Anti-sPLA 2 -IIA antibody-conjugated beads were incubated with the high speed supernatant of Nonidet P-40-lyzed 293 cells that coexpressed sPLA 2 -IIA and glypican-1 and then washed with a buffer containing 1 M NaCl followed by an acidic buffer. Fractions were analyzed by immunoblotting using the same antibodies. A result representative of three independent experiments is shown. within 10 min (Fig. 5I). This perinuclear redistribution of cPLA 2 continued over 4 h of culture with IL-1 (data not shown), during which cPLA 2 -mediated AA release continued to take place (2,3). Both COX-1 (Fig. 5J) and COX-2 (Fig. 5K) were found in the perinuclear envelope and endoplasmic reticulum. These staining patterns were essentially consistent with current reports (6 -8).
Subcellular Localization of Cytokine-inducible Endogenous sPLA 2 -IIA-To ensure that the preferential distribution of sPLA 2 -IIA in the caveolae and perinuclear location was not a peculiarity of the 293 transfectants, we next examined the subcellular distribution of endogenous sPLA 2 -IIA, which was induced by proinflammatory stimuli in several cell types. For this purpose, we used rat fibroblastic 3Y1 cells because the expression of sPLA 2 -IIA and COX-2 and attendant delayed generation of PGE 2 are known to be strongly induced by stimulation with IL-1 and TNF in these cells (14). Whereas sPLA 2 -IIA immunoreactivity was weak in the punctate domains in unstimulated 3Y1 cells (Fig. 6, A and B), it increased markedly in the punctate and perinuclear compartments in replicate cells activated with IL-1 and TNF for 24 h (Fig. 6C), the period in which delayed PGE 2 generation occurs at the maximal rate (14). sPLA 2 -IIA staining in IL-1/TNF-stimulated cells was markedly reduced when the cells were cultured in the presence of heparin (Fig. 6D), which solubilizes the cell surface-associated sPLA 2 -IIA without affecting its expression level and eventually causes a reduction in PGE 2 generation (14). COX-2 was distributed in the perinuclear envelope and endoplasmic reticulum in IL-1/TNF-stimulated cells (Fig. 6E), but not in control cells (data not shown). Double antibody staining for sPLA 2 -IIA and caveolin-2 demonstrated their colocalization (Fig. 6F), providing further support for the accumulation of endogenous sPLA 2 -IIA in the caveolae of cytokine-stimulated 3Y1 cells.
The other cell type we examined was rat liver-derived BRL-3A cells, in which a heparin-sensitive, membrane-associated, cytokine-inducible sPLA 2 -IIA has been reported to be involved in cytokine-stimulated delayed PGE 2 generation (27). Immunocytostaining of BRL-3A cells revealed again that sPLA 2 -IIA was localized in the punctate and perinuclear domains, in which caveolin-2 coexisted, in IL-1/TNF-stimulated, but not in unstimulated, cells (Fig. 7). As in the 3Y1 cells, COX-2 was found in the perinuclear envelope and endoplasmic reticulum in IL-1/TNF-stimulated BRL-3A cells (data not shown).

DISCUSSION
Several previous studies have demonstrated the dependence of cellular functions of sPLA 2 -IIA on HSPGs, particularly when delayed PG generation is accompanied by its concomitant expression in cells exposed to proinflammatory stimuli (2,3,14,22,26,27). In a proposed model of this, sPLA 2 -IIA released from cytokine-stimulated cells is captured by the heparan sulfate chains of a HSPG and thus accumulates on the plasma membrane, from which the enzyme liberates AA. sPLA 2 -IIA bound to a HSPG on the surface of fibroblasts greatly enhances the PG-biosythetic response in adjacent mast cells through a juxtacrine route (23). Furthermore, deposition of sPLA 2 -IIA into the matrix proteoglycan biglycan increases its hydrolytic efficiency of lipoprotein particles severalfold, an event that is implicated in the exacerbation of atherosclerosis (30,31). As opposed to these positive regulatory effects of HSPG on sPLA 2 -IIA functions, however, there are also some contradictory reports that sPLA 2 -IIA acts independently of HSPG (32), which suggests that the relation between proteoglycans and sPLA 2 -IIA action is more complicated than previously thought. To reconcile this discrepancy, it seemed necessary to clarify what kinds of HSPG bind sPLA 2 -IIA and how they affect sPLA 2 -IIA After fixation with paraformaldehyde, cells were treated with (A, B, D, and F) or without (C and E) saponin, followed by rabbit anti-sPLA 2 -IIA antibody and FITC-conjugated anti-rabbit IgG antibody. In D and E, the cells were cultured in the presence of 1 mg/ml heparin before fixation. G, double staining of sPLA 2 -IIA-expressing cells with rabbit anti-sPLA 2 -IIA and mouse anti-caveolin-2 antibodies, which were visualized using Cy3-conjugated anti-rabbit IgG and FITC-conjugated antimouse IgG antibodies, respectively. H and I, immunostaining for cPLA 2 . Cells incubated with (I) or without (H) IL-1 for 10 min were fixed, permeabilized, and then treated with rabbit anti-cPLA 2 antibody. J and K, immunostaining for two COX isoforms. Transfectants expressing either COX-1 or COX-2 were fixed, permeabilized, and then treated with rabbit anti-COX-1 antibody and goat anti-COX-2 antibody, respectively. cPLA 2 and COX-1 were each visualized using FITC-conjugated anti-rabbit IgG antibody and COX-2 by FITC-conjugated anti-goat IgG antibody.
functions in cells undergoing delayed PG generation.
The findings of the present study suggest that endogenously expressed sPLA 2 -IIA associates with the GPI-anchored form of HSPG, glypican. Removal of GPI-anchored HSPG from cell surfaces markedly attenuated sPLA 2 -IIA-mediated PGE 2 production ( Figs. 1 and 2). sPLA 2 -IIA and glypican were coimmunoprecipitated from the cells, implying their association in vivo (Fig. 4). Moreover, the ability of sPLA 2 -IIA to enhance IL-1induced AA metabolism was markedly enhanced by coexpression with glypican (Fig. 3). This effect was particularly evident when the expression of sPLA 2 -IIA was suboptimal. Importantly, the maximal response was produced even by low concentrations of sPLA 2 -IIA (ng/ml as estimated by enzymatic activity and the intensity of immunoblot bands) when combined with overexpressed glypican that were about 1,000 times lower than that required for exogenously added sPLA 2 -IIA to exhibit AA-releasing function (23,28,32,54,55). It is therefore likely that the role of glypican is to amplify the function of sPLA 2 -IIA expressed at physiological levels. As demonstrated by Gelb and co-workers (32), high concentrations of exogenous sPLA 2 -IIA could act on cells independently of HSPG to elicit rapid and transient AA release.
GPI-anchored proteins generally occur in microdomains of the cell membrane called caveolae or the caveolae-related domain (45)(46)(47)(48). It has been shown recently that there are dynamic changes in the subcellular distribution of glypican, which moves to the nucleus and punctate caveolae-like domains, depending upon the cell cycle (37). Based on our present finding that sPLA 2 -IIA is functionally associated with the GPI-anchored HSPG glypican, it is tempting to speculate that glypican plays a specific role in the sorting of sPLA 2 -IIA into specific subcellular compartments; AA released by sPLA 2 -IIA in these compartments will be more accessible to the COX-2-dependent PGE 2 -biosynthetic pathway than if it was released randomly from the plasma membrane surface as was thought previously. Indeed, sPLA 2 -IIA appears to be localized in the caveolae in three different cell lines that exhibit cytokine-stimulated delayed PGE 2 generation (Figs. 5-7), although final conclusion for this localization must be awaited until more detailed electron microscopic studies will be perfomed. In this regard, an earlier work using electron microscopic analysis showed a punctate staining pattern of sPLA 2 -IIA in TNF-stimulated rat vascular smooth muscle cells, where the surface of the caveshaped compartments on the plasma membrane gave a strong signal for sPLA 2 -IIA (56). We expect that the cave-shaped compartments observed in the vascular smooth muscle cells also correspond to caveolae.
Caveolae are known to form a unique endocytic and exocytic compartment at the surface of most cells, capable of importing molecules, delivering them to specific locations within the cell, and compartmentalizing a variety of signaling activities (49,50). They are not simply an endocytic device with a peculiar membrane shape but constitute an entire membrane system with multiple functions essential for the cell. Our detection of both sPLA 2 -IIA and caveolin-2 in the perinuclear region (Fig. 6) suggests that the caveolae-mediated endocytotic event called potocytosis occurs in cytokine-stimulated cells. Several previous subcellular fractionation studies demonstrated the pres- ence of significant sPLA 2 -IIA in some intracellular organelle fractions including the nucleus (52,57). The perinuclear localization of sPLA 2 -IIA is also supported by the fact that glypican-1, an sPLA 2 -IIA-adaptor protein, has a bipartite motif for nuclear localization signals (37). Since considerable evidence has indicated that caveolae are a site of Ca 2ϩ storage and entry into the cell (50), it is likely that sPLA 2 -IIA, a Ca 2ϩ -dependent enzyme, present inside caveola particles retains its enzyme activity even after internalization and translocation to the perinuclear domain. Moreover, sPLA 2 -IIA has been reported to be active in the presence of only micromolar concentrations of Ca 2ϩ under certain assay conditions (58), raising the possibility that it is active even within the cell. The caveolae membrane is enriched in sphingomyelin, which inhibits the enzymatic activity of sPLA 2 -IIA in vitro (59). The high packing density of the bilayer leaflet enriched in this lipid hinders the penetration of sPLA 2 -IIA into the plasma membrane, which may explain why quiescent cells are fairly resistant to sPLA 2 -IIA. In keeping with the present finding that sPLA 2 -IIA resides in caveolae, a sphingomyelin-rich microdomain (49, 50), a decrease in the cellular sphingomyelin content caused by sphingomyelinase in response to cytokines (60) may allow the otherwise silent sPLA 2 -IIA to become active toward the caveola membrane. Cholesterol, which is also abundant in caveolae (49,50,61), counteracts the effect of sphingomyelin-based inhibition of sPLA 2 -IIA (62) and may contribute to the temporal and spatial regulation of this enzyme during cell activation. Rather selective release of [ 3 H]AA by sPLA 2 -IIA, which we observed in sPLA 2 -transfected cells (2), may be a reflection of the fact that [ 3 H]AA is preferentially incorporated into the caveola and perinuclear membranes (63). It should be noted, however, that we have not yet conclusively shown that AA release by sPLA 2 -IIA for PG production is due to enzyme on the inside of the cell. The enhanced AA metabolism by glypican expression could be due to enhanced internalization of sPLA 2 -IIA or due to enhanced capture on the outside of the cell for AA release from caveolae on the plasma membrane. This point should be clarified in a future study.
Enhancement of cytokine-induced COX-2 expression is another intriguing feature of sPLA 2 -IIA, which was synergistically augmented by coexpression with glypican (Fig. 3C). This result is consistent with our recent finding that COX-2 induction by sPLA 2 -IIA and sPLA 2 -V in 293 cells depends on their binding to HSPG as well as on their enzymatic activities (53). Thus, increased production of AA metabolites by the concerted actions of sPLA 2 -IIA and glypican may lead to further amplification of COX-2 expression through a positive feedback route in this occasion. Alternatively, the binding of sPLA 2 -IIA to the heparan sulfate chains of glypican facilitates its presentation to the putative sPLA 2 -IIA receptor, which transduces signals leading to increased COX-2 expression. In line with this hypothesis, COX-2 induction by sPLA 2 -IIA in rat serosal mast cells appears to involve a receptor-mediated pathway (23). This type of receptor system, utilizing both HSPG and a signaltransducing receptor subunit, has been described for the FGF receptor system in which the ligand, its tyrosine kinase receptor, and HSPG form a stable ternary complex on the cell surface, thereby potentiating transmembrane signals (41,42,64). The participation of a receptor-like molecule(s) in this sPLA 2 -IIA action should be examined because many receptor molecules as well as intracellular signal-transducing molecules are known to be associated with the caveolae (49,50). Note that the M-type receptor, the only sPLA 2 receptor cloned to date, possesses a sequence stretch for receptor internalization in its short cytoplasmic domain (65,66), and that sPLA 2 -IB, a ligand for the M-type receptor, translocates to the perinuclear compartment through the receptor-dependent process (67).
It is likely that sPLA 2 -IIA is able to associate with proteoglycans other than glypican. The binding of sPLA 2 -IIA to extracellular matrix HSPGs, such as perlecan and biglycan, may prevent its movement to the plasma membrane, eventually leading to inhibition of its cellular functions. This situation is reminiscent of FGF, which is stored in the extracellular matrix as a latent form; active FGF is released by the degradation of matrix HSPG by heparinase and proteases, and then interacts with a signal-transducing receptor complex composed of the FGF receptor and HSPG (glypican or syndecan) on the target cells (64). sPLA 2 -IIA associated with matrix HSPG has a specific role in the development of atherosclerosis (19). Whether the syndecans, integral cellular HSPGs, regulate sPLA 2 -IIA functions remains to be elucidated. In hematopoietic cells, such as platelets and mast cells, sPLA 2 -IIA is stored in secretory granules rather than binding to a cell surface HSPG (68 -70), implying that the subcellular distribution of sPLA 2 -IIA varies according to the cell type and HSPG molecular species. The secretory granules of mast cells contain serglycin, a soluble FIG. 7. Subcellular distribution of endogenous sPLA 2 -IIA in rat liver BRL-3A cells. BRL-3A cells cultured for 24 h with (B-D) or without (A) IL-1 and TNF were fixed, permeabilized, and double-immunostained using a rabbit anti-sPLA 2 -IIA antibody (A and B), a mouse anti-caveolin-2 antibody (C), or both (D). sPLA 2 -IIA and caveolin-2 were visualized using Cy3-conjugated anti-rabbit IgG and FITC-conjugated anti-mouse IgG antibodies, respectively. proteoglycan (71), which may trap sPLA 2 -IIA inside the proteoglycan-formed particles and may inhibit its activity to prevent unregulated hydrolysis of granule membranes. Activation of mast cells causes rapid exocytosis of sPLA 2 -IIA, which may be released from serglycin to become active on target cells.
In summary, we have provided new insight into the mechanisms that regulate delayed PG biosynthesis via sPLA 2 -IIA. sPLA 2 -IIA produced by cytokine-stimulated cells (autocrine route) or from extracellular microenvironments (paracrine or juxtacrine route) binds to GPI-anchored HSPG, glypican, which plays a crucial role in sorting of sPLA 2 -IIA into the caveola signalsomes. A significant portion of sPLA 2 -IIA may enter the cells through potocytosis and reach the perinuclear area, where the upstream (cPLA 2 ) and downstream (COX-2) PG-biosynthetic enzymes are located. Considering that prior activation of cPLA 2 is crucial for sPLA 2 to act (72), cPLA 2 may play a role in the sensitization of the caveola and/or perinuclear membrane to sPLA 2 -IIA. The AA thus released from the particular compartments by sPLA 2 -IIA may be readily supplied to COX-2, allowing their efficient functional coupling. sPLA 2 -IIA can also has the ability to increase COX-2 expression, which contributes to further amplification of PGE 2 production. sPLA 2 -IIA secreted from the cells mediates the transcellular PG-biosynthetic response (3). Continuous production and supply of sPLA 2 -IIA are crucial for it to function fully in the delayed PG-biosynthetic response. It is likely that sPLA 2 -V utilizes the same regulatory machinery in some cells, because it also binds to a cellular HSPG and elicits delayed PG biosynthesis in a manner similar to sPLA 2 -IIA under certain (2,3,(72)(73)(74), if not all (75), conditions. Whether sPLA 2 -IID, a recently identified sPLA 2 -IIA homolog (76,77), also utilizes the same machinery is now under investigation. A heparin-nonbinding isozyme, sPLA 2 -X, may hydrolyze phosphatidylcholine in the outer leaflet of the plasma membrane randomly and thereby exhibit unique fatty acid-releasing properties different from sPLA 2 -IIA (53).