Identification of a Cellular Protein That Functionally Interacts with the C2 Domain of Cytosolic Phospholipase A2α*

Cytosolic phospholipase A2(cPLA2) α plays critical roles in lipid mediator synthesis. We performed far-Western analysis and identified a 60-kDa protein (P60) that interacted with cPLA2α in a Ca2+-dependent manner. Peptide microsequencing revealed that purified P60 was identical to vimentin, a major component of the intermediate filament. The interaction occurred between the C2 domain of cPLA2α and the head domain of vimentin. Immunofluorescence microscopic analysis demonstrated that cPLA2α and vimentin colocalized around the perinuclear area in cPLA2α-overexpressing human embryonic kidney 293 cells following A23187 stimulation. Forcible expression of vimentin in vimentin-deficient SW13 cells augmented A23187-induced arachidonate release. Moreover, overexpression of the vimentin head domain in rat fibroblastic 3Y1 cells exerted a dominant inhibitory effect on arachidonate metabolism, significantly reducing A23187-induced arachidonate release and attendant prostanoid generation. These results suggest that vimentin is an adaptor for cPLA2α to function properly during the eicosanoid-biosynthetic process.

Cytosolic phospholipase A 2 (cPLA 2 ) ␣ plays critical roles in lipid mediator synthesis. We performed far-Western analysis and identified a 60-kDa protein (P60) that interacted with cPLA 2 ␣ in a Ca 2؉ -dependent manner. Peptide microsequencing revealed that purified P60 was identical to vimentin, a major component of the intermediate filament. The interaction occurred between the C2 domain of cPLA 2 ␣ and the head domain of vimentin. Immunofluorescence microscopic analysis demonstrated that cPLA 2 ␣ and vimentin colocalized around the perinuclear area in cPLA 2 ␣-overexpressing human embryonic kidney 293 cells following A23187 stimulation. Forcible expression of vimentin in vimentin-deficient SW13 cells augmented A23187-induced arachidonate release. Moreover, overexpression of the vimentin head domain in rat fibroblastic 3Y1 cells exerted a dominant inhibitory effect on arachidonate metabolism, significantly reducing A23187-induced arachidonate release and attendant prostanoid generation. These results suggest that vimentin is an adaptor for cPLA 2 ␣ to function properly during the eicosanoid-biosynthetic process.
Phospholipase A 2 (PLA 2 ) 1 hydrolyzes the ester bonds of fatty acids present at the sn-2 positions of phospholipids. PLA 2 plays crucial roles in diverse cellular responses, including phospholipid digestion and metabolism, host defense, signal transduction, and probably apoptosis. PLA 2 provides precursors for the biosynthesis of eicosanoids, such as prostaglandins (PGs) and leukotrienes, when the hydrolyzed fatty acid is arachidonic acid (AA); platelet-activating factor when the sn-1 position of phosphatidylcholine contains an alkyl ether linkage; and some bioactive lysophospholipids, such as lysophosphatidic acid. As oversynthesis of these lipid mediators causes inflammation and tissue disorders, it is important to elucidate the mechanisms that regulate the functions of PLA 2 . Mammalian tissues and cells generally contain more than one PLA 2 , each of which is regulated independently and plays distinct roles. There are three large families of mammalian PLA 2 s, cytosolic PLA 2 (cPLA 2 ), secretory PLA 2 , and Ca 2ϩ -independent PLA 2 , and the roles of type IV cPLA 2 ␣ and secretory types IIA and V sPLA 2 in lipid mediator synthesis have been studied extensively (1)(2)(3).
An 85-kDa type IV cPLA 2 ␣ appears to be one of the most important PLA 2 isozymes involved in lipid mediator synthesis following cell activation (1)(2)(3). Submicromolar concentrations of Ca 2ϩ appear to be required for cPLA 2 ␣ to exert its catalytic activity, and this enzyme preferentially hydrolyses phospholipids bearing AA (4,5). cPLA 2 ␣ is expressed in most mammalian cells, and its activation is regulated by several postreceptor signal transduction events, such as Ca 2ϩ mobilization (6,7), phosphorylation (8,9), and gene induction (10,11). After stimuli that are accompanied by an increase in the cytoplasmic Ca 2ϩ concentration, cPLA 2 ␣ translocates rapidly and often transiently to the perinuclear and endoplasmic reticular membranes (12,13), is phosphorylated by mitogen-activated protein kinases (8,9), and releases AA for immediate conversion to PGs and leukotrienes by the constitutive cyclooxygenase (COX)-1 and 5-lipoxygenase (5-LO), respectively (14 -17). Furthermore, cPLA 2 ␣ has been implicated in the inducible COX-2-dependent delayed PG generation, which lasts for hours, initiated by proinflammatory stimuli, such as interleukin-1, tumor necrosis factor ␣, and lipopolysaccharide (18 -21). cPLA 2 ␣ has several functionary distinct regions: an amino-terminal Ca 2ϩ -dependent lipid binding domain (amino acids 18 -138) called the C2 domain (22,23), a carboxyl-terminal region (amino acids 179 -749) containing the catalytic domain, a putative pleckstrin homology domain (amino acids 263-354) similar to phospholipase C-␦1 (24), and two critical serine residues (Ser 505 and Ser 727 ), which undergo activation-directed phosphorylation (25). The C2 domain is responsible for translocation of cPLA 2 ␣ from the cytosol to the membrane compartment and exhibits significant homology with the C2 domains of several proteins, such as protein kinase C, GTPase-activating protein, synaptotagmin, and phospholipase C, all of which bind to phospholipid membranes in a Ca 2ϩ -dependent manner (22,23). The carboxyl-terminal region mediates the Ca 2ϩ -independent enzymatic catalysis, which involves Ser 228 at the active site (26). Interestingly, a number of enzymes involved in AA metabolism, such as COX-1 and -2, 5-LO, and 5-LO-activating protein, are also localized in the nuclear envelope and endoplasmic reticulum (27)(28)(29).
Although the C2 domain appears to be responsible for Ca 2ϩdependent localization of cPLA 2 ␣ to its membrane substrates, whether additional interactions, such as C2 domain binding to an adaptor protein, also play roles in cPLA 2 ␣ targeting to particular membrane compartments remains to be elucidated. In this study, we carried out far-Western screening in an attempt to determine how cPLA 2 ␣ translocates specifically to the perinuclear particles and identified a protein that interacted with cPLA 2 ␣. We found that vimentin, a major component protein of intermediate filaments, interacted with the aminoterminal region of cPLA 2 ␣ in a Ca 2ϩ -dependent manner. Further experiments demonstrated that cPLA 2 ␣ and vimentin colocalized in Ca 2ϩ ionophore-stimulated, but not unstimulated, human embryonic kidney 293 cells stably transfected with cPLA 2 ␣. Introduction of vimentin into a vimentin-deficient SW13 cell line restored cPLA 2 ␣-dependent AA release in response to a Ca 2ϩ ionophore, whereas introduction of the vimentin head domain, to which cPLA 2 ␣ bound via its C2 domain, into rat fibroblastic 3Y1 cells reduced it. These results suggest that vimentin is involved in the regulation of cPLA 2 ␣ in the AA metabolic pathway.

EXPERIMENTAL PROCEDURES
Materials-Mouse cPLA 2 ␣ cDNA was provided by Dr. Tsujimoto (RIKEN Institute); the PGE 2 enzyme immunoassay kit was purchased from Cayman Chemical; and mouse monoclonal anti-glutathione Stransferase (GST), anti-human cPLA 2 ␣, and anti-human vimentin antibodies were purchased from Santa Cruz Biotechnology. Mouse monoclonal anti-porcine glial fibrillary acidic protein (GFAP), Cy3conjugated anti-human vimentin, anti-T7 tag, and anti-enhanced green fluorescent protein (EGFP) antibodies were purchased from Chemicon, Sigma, Novagen, and CLONTECH, respectively, and A23187 was purchased from Calbiochem.
Cell Culture-Adherent macrophages were prepared from the peritoneal cells of male Harlan Sprague-Dawley rats (Nippon Bio-Supply Center) 4 days after the injection of 5% (w/v) soluble starch and 5% (w/v) Bacto-peptone in saline (5 ml/100 g of body weight) as described previously (30) and cultured in RPMI 1640 medium (Nissui) supplemented with 10% (v/v) fetal calf serum. Rat fibroblastic 3Y1 cells (donated by Dr. Uehara, National Institute of Health, Tokyo, Japan), human adrenocortical carcinoma SW13 cells (donated by Dr. D. M. Marcus, Baylor College of Medicine) and human glioblastoma U251 cells (obtained from RIKEN Cell Bank) were grown in Dulbecco's modified Eagle's medium (Nissui) supplemented with 10% fetal calf serum. Mouse osteoblastic MC3T3-E1 cells (obtained from RIKEN Cell Bank) and human embryonic kidney 293 cells (obtained from the Health Science Research Resources Bank) were grown in ␣-modified Eagle's medium (ICN Biomedicals) and RPMI 1640 medium, respectively, both of which were supplemented with 10% fetal calf serum. All the above cells were cultured in a 5% CO 2 humidified atmosphere at 37°C. Sf9 cells (Invitrogen) were cultured in Grace's insect medium (Invitrogen) supplemented with 10% fetal calf serum at 27°C.
Escherichia coli JM109 cells (Takara Shuzo) expressing the appropriate constructs were cultured until they reached the late log phase, and 300 M isopropyl ␤-D-thiogalactopyranoside was added to induce GST fusion proteins at 37°C. Bacterial cell pellets were lysed by sonication (Branson Sonifier; setting 7, 25% pulse cycle, 10 min) at 0°C, and inclusion bodies were harvested and solubilized with 0.75% sodium N-lauroyl sarcosinate. The solubilized fraction was dialyzed against Tris-buffered saline (pH 7.5), Triton X-100, up to 1% (v/v), was added to the dialysate, the GST fusion proteins were adsorbed by glutathione-Sepharose CL-4B beads (Amersham Pharmacia Biotech) and eluted with Tris-buffered saline (pH 7.5) containing 20 mM glutathione.
This plasmid was transformed into E. coli strain BL21(DE3) (Stratagene), a lon mutant strain containing T7 polymerase under the control of the lacUV5 promoter, and 1 mM isopropyl ␤-D-thiogalactopyranoside was added to induce expression of T7-tagged cPLA 2 ␣ at 37°C. A crude fraction of the expressed T7-tagged cPLA 2 ␣ was obtained by solubilizing the inclusion bodies with 0.25% (v/v) Tween 20, and the solubilizate was used for far-Western analysis.
Site-directed Mutagenesis of cPLA 2 ␣ cDNA-Site-directed mutations of cPLA 2 ␣ cDNA were introduced by carrying out the mismatched primer PCR with ex Taq polymerase using truncated mouse cPLA 2 ␣ (1-138) inserted in the pCR3.1 vector as the template. In order to obtain the D43N/D93N mutant, the products obtained from the first PCR using the following mutated primer pair: D43N (5Ј-GG ACA CTC CAA ATC CTT ATG, in which Asp 43 was replaced by Asn at the underlined site) and D93N (5Ј-GTT GGC ATT CAT CAG TGT G, in which Asp 93 was replaced by Asn at the underlined site) were mixed, denatured at 95°C for 5 min, and then annealed with gradual cooling to 37°C. The second PCR was performed using the amino-terminal sense and MC-138 antisense primer pair and the annealed sample as the template. The product was subcloned into the pCR3.1 vector and the mutation of the construct was verified by DNA sequencing. A double mutant was obtained by cloning the 0.2-kilobase pair BamHI fragment of pCR3.1-cPLA 2 ␣ (1-138) D93N, which contains a single mutation, into the 5.2-kilobase pair BamHI fragment of pCR3.1-cPLA 2 ␣ (1-138) D43N. The resulting construct was subcloned into pGEX4T3 and expressed as a GST fusion protein, as described above.
Preparation of Recombinant Vimentin Proteins-Rat vimentin cDNA was amplified by performing the reverse transcription-PCR using poly(A) ϩ RNA purified from rat fibroblastic 3Y1 cells, and the primer pair, synthesized according to the rat vimentin cDNA sequence, composed of the following sequences: 5Ј-GGAATTCATGTCCACCAGG-TCCGTG and 5Ј-GGAATTCTCAAGGTCATCGTGGTGC. The rod and tail domains of vimentin protein were prepared by the PCR using fulllength vimentin cDNA as the template. The following primer pairs were designed: vimentin (95-406), 5Ј-AGGAATTCGAGTTCAAGAACACC-CGC and 5Ј-CCAAGCTTCCCTTCCAGCAGCTTCCT; and vimentin (407-466), 5Ј-AGGGATCCGAGGAGAGCAGGATTTCT and 5Ј-GGAA-TTCTCAAGGTCATCGTGGTGC. Initially, the amplified fragments were cloned into the pCR3.1 vector and sequenced as described above, and the inserts were subcloned into the prokaryotic expression vector pET21a (Novagen). The head domain of vimentin protein was released from pCR3.1-vimentin by restriction enzyme digestion (EcoRI and XhoI) and subcloned into the pET21a vector, which expressed protein as a fusion protein containing a stretch of 6 consecutive histidine residues (His-tag) at the carboxyl terminus. The plasmid was transformed into E. coli BL21(DE3), and 1 mM isopropyl ␤-D-thiogalactopyranoside was added to induce expression of the His-tagged vimentin protein, which was purified using Ni-NTA agarose (Qiagen) according to the manufacturer's instructions and used for the subsequent experiments.
SDS-Polyacrylamide Gel Electrophoresis (PAGE)/Immunoblotting-Cultured cells were washed once with 10 mM phosphate buffer (pH 7.4) containing 150 mM NaCl (phosphate-buffered saline (PBS)) and lysed by sonication. The lysates were subjected to SDS-PAGE under reducing conditions and electrotransferred to nitrocellolose membranes (Schleicher & Schuell) using a semidry blotter (milliBlot-SDE system; Millipore), according to the manufacturer's instructions. The membranes were washed once with H 2 O, blocked with 5% (w/v) skim milk in PBS containing 0.05% (v/v) Tween 20 (PBS-T) for 1 h at room temperature or overnight at 4°C, and washed with PBS-T. Then, the mouse monoclonal anti-vimentin, anti-GFAP, and anti-EGFP antibodies (diluted 1:5,000, 1:2000, and 1:1000, respectively, with PBS-T containing 1% (w/v) skim milk) were added, and the membranes were incubated for 2 h at room temperature, washed three times with PBS-T, and treated with horseradish peroxidase-conjugated goat anti-mouse IgG (Zymed Laboratories Inc.; diluted 1:3,000) in PBS-T containing 1% (w/v) skim milk for 1 h at room temperature. Finally, after six washes with PBS-T, the protein bands were visualized using the ECL Western blot analysis system (Amersham Pharmacia Biotech).
Far-Western Analysis-Samples were subjected to SDS-PAGE and electrotransferred to nitrocellulose membranes, which were blocked with 5% (w/v) skim milk in PBS-T for 1 h at room temperature or overnight at 4°C. Then, separate nitrocellulose strips with the immobilized proteins were incubated for 2 h at room temperature with 1 g/ml purified GST-cPLA 2 ␣ or GST (control). The bound proteins were immunoblotted with the anti-GST monoclonal antibody and then incubated with horseradish peroxidase-conjugated anti-mouse IgG, as described above. When T7-tagged cPLA 2 ␣ was used instead of GST-cPLA 2 ␣, the protein band was visualized using the monoclonal anti-T7 antibody (Novagen; diluted 1:2000).
Baculovirus Expression of Recombinant cPLA 2 ␣-Mouse cPLA 2 ␣ cDNA was subcloned into the baculovirus transfer vector pVL1393 (Pharmingen), and Sf9 cells were infected with the resulting baculovirus, as described elsewhere (31,32). After culture at 27°C for 4 days, the infected cells were harvested, lysed by sonication, and centrifuged at 100,000 ϫ g at 4°C for 1 h. The supernatant was introduced to a DEAE-Sephacel column (10 ϫ 50 mm) (Amersham Pharmacia Biotech) preequilibrated with Tris-buffered saline (pH 7.4), the column was washed with this buffer, and recombinant cPLA 2 ␣ was eluted with a linear gradient of 150 -500 mM NaCl at a flow rate of 15 ml/h. The fractions containing cPLA 2 ␣ as a major protein were pooled and used in the subsequent experiments.
Partial Purification of P60 -3Y1 cells (3 ϫ 10 8 ) were harvested in PBS containing trypsin/EDTA and washed with PBS, and the pelleted cells were suspended in 10 ml of ice-cold lysis buffer (10 mM Tris-HCl (pH 7.4), 10 mM NaCl, 1.5 mM MgCl 2 , 1% (v/v) Nonidet P-40, and 10 g/ml leupeptin). The Nonidet P-40-insoluble fraction was collected by centrifugation at 1200 ϫ g for 5 min at 4°C, resuspended in 1 ml of 50 mM Tris-HCl (pH 7.4) containing 25 mM KCl and 5 mM MgCl 2 , and sonicated briefly, and solid tripotassium citrate was added to produce a final concentration of 10% (w/v). Then, the sample was centrifuged at 39,000 ϫ g for 45 min at 4°C to obtain a chromatin-rich fraction (supernatant) and a nuclear matrix-rich fraction (pellet), both of which were subjected to far-Western analysis, as described above.
Sequencing of P60 -The P60 band on the first SDS-PAGE gel was visualized using Coomassie Brilliant Blue, cut out from the gel, and then partially digested in the second gel with 10 g of Achromobacter lysylendopeptidase (Wako). The digested peptides were electrotransferred to a polyvinylidene difluoride membrane (Millipore), and the two major peptide fragments obtained were subjected to amino-terminal amino acid sequencing using an Applied Biosystems 473A Protein Sequencer. Degenerate PCR primers based on the amino acid sequences were designed, and the PCR was performed (30 cycles each comprising 94°C for 30 s, 55°C for 30 s, and 72°C for 60 s) using 3Y1 cDNA as the template. The sequence of a major DNA fragment amplified was inquired into the using the BLAST program.
Coprecipitation Assays-GST-cPLA 2 ␣ (1-138)-bound glutathione-Sepharose CL-4B was incubated with purified recombinant rat vimentin for 2 h at room temperature in the presence or absence of 1 mM CaCl 2 , the glutathione-Sepharose pellet was washed with 50 mM Tris-HCl (pH 7.5) containing 1 mM MgCl 2 and 1 mM dithiothreitol, and the bound proteins were eluted by incubation with 20 mM glutathione for 1 h at room temperature. The eluted fractions were subjected to SDS-PAGE, electrotransferred to a nitrocellulose membrane, and immunoblotted with the anti-vimentin monoclonal antibody, as described above.
Immunofluorescence Microscopy-The subcellular distribution patterns of cPLA 2 ␣ and vimentin in 293 cells stably transfected with mouse cPLA 2 ␣, which were established as described previously (33), were analyzed using a double indirect immunofluorescence staining technique. The cells were grown overnight on coverslips, and when they had reached subconfluence, they were treated with 10 M A23187 or medium alone for 5 min at 37°C, washed once with PBS, and then fixed with PBS containing 3.7% (w/v) paraformaldehyde (Nakalai Tesque) for 30 min at room temperature. The cells were permeabilized in PBS containing 0.2% (v/v) Triton X-100 for 2 min, blocked with 3% (w/v) bovine serum albumin (Sigma) for 1 h at room temperature, and incubated with the mouse anti-cPLA 2 ␣ monoclonal antibody (1:100) in PBS containing 1% bovine serum albumin for 2 h at room temperature, and the unbound antibody was removed by six washes with PBS. Then, the cells were incubated with fluorescein isothiocyanate-conjugated antimouse Ig (1:100; Zymed Laboratories Inc.) in 1% bovine serum albumin-PBS for 1 h at room temperature, washed with PBS, and treated with the Cy3-conjugated mouse anti-vimentin monoclonal antibody (1:100) in 1% bovine serum albumin-PBS for 1 h at room temperature. Finally, the cells were mounted in glycerol/PBS (9:1; v/v) containing 2.5% (w/v) diazabicyclo (2.2.2)octane (Wako), and their fluorescence was visualized using a laser scanning confocal microscope (IX70; Olympus).
Transfection of Mammalian Cells with Vimentin cDNA-Full-length rat vimentin cDNA and its amino-terminal portion (corresponding to amino acids 1-125) were subcloned into pCR3.1 and pEGFP-C1 (CLON-TECH), respectively; pCR3.1-wild-type vimentin was transfected into a vimentin-deficient subclone of human carcinoma SW13 cells; and pEGFP-vimentin (1-125) was transfected into rat fibroblastic 3Y1 cells using LipofectAMINE Plus reagent (Life Technologies, Inc.), according to the manufacturer's instructions. Briefly, 1 g of required plasmid was mixed with 6 l of LipofectAMINE Plus and 4 l of LipofectAMINE in 100 l Opti-MEM (Life Technologies, Inc.) for 15 min at room temperature and then added to cells, in 800 l of Opti-MEM, that had reached 40 -60% confluence in six-well plates (Iwaki). After incubation for 6 h, the medium was replaced with 2 ml of fresh culture medium, and after culture for 2 days, the 3Y1 and SW13 cells, in culture medium supplemented with 800 and 400 g/ml G418 (Life Technologies), respectively, were each cloned by limiting dilution in 96-well plates. After culture for 1-2 weeks, wells containing a single colony were chosen, and vimentin and EGFP-vimentin (1-125) expression was assessed by both RNA blotting and immunoblotting. The established clones were expanded and used for the experiments described below.
Measurement of AA Release and PGE 2 Production-Stable transformants (10 5 cells in 1 ml of culture medium) were seeded into 24-well plates (Iwaki). In order to assess AA release, 100 nCi/ml [ 3 H]AA (NEN Life Science Products) was added to cells that were near confluence, and culture was continued for another day. The cells were washed three times with fresh medium, 250 l of medium with or without 3 or 10 M A23187 was added to each well, and the amount of free [ 3 H]AA released into the supernatant during 30 min of culture was measured by liquid scintillation counting. Portions of the supernatant were taken for PGE 2 enzyme immunoassay, which was performed using an enzyme immunoassay kit.

RESULTS
Detection of a cPLA 2 ␣-binding Protein-In an attempt to search for a protein that interacts with cPLA 2 ␣, lysates of several cells that generated PGE 2 in a cPLA 2 ␣-dependent manner (19,34,35) were subjected to far-Western analysis using epitope-tagged cPLA 2 ␣ protein as a probe. In this assay, the filters onto which the cellular proteins were transferred were incubated with either GST or GST-cPLA 2 ␣ and then blotted with an anti-GST antibody. No obvious bands were detected after incubation with GST alone, whereas a protein with a molecular mass of approximately 60 kDa was detected, after incubation with GST-cPLA 2 ␣, in the lysates of rat peritoneal macrophages and mouse osteoblastic MC3T3-E1 cells (Fig. 1A). This protein, tentatively designated P60, was also detected in the lysates of MC3T3-E1 cells and rat fibroblastic 3Y1 cells (Fig. 1B) as well as macrophages (data not shown) when T7tagged, instead of GST-tagged, cPLA 2 ␣ was used as a probe and visualized using an anti-T7 antibody. Therefore, P60 appears to be a cPLA 2 ␣-binding protein.
In order to examine the subcellular distribution of P60, 3Y1 cells were lysed with the nonionic detergent Nonidet P-40 and centrifuged at 1200 ϫ g, and the resulting supernatant and precipitate were subjected to far-Western analysis. As shown in Fig. 1F, P60 was detected exclusively in the precipitate fraction, which contained chromatin and nuclear matrix. When this Nonidet P-40-insoluble fraction was separated into chromatin-and nuclear matrix-rich fractions, P60 was detected only in the latter (Fig. 1F).
Ca 2ϩ Requirement for cPLA 2 ␣ Binding to P60 -As the C2 domain is a Ca 2ϩ -binding motif, we carried out far-Western analysis to examine the effects of Ca 2ϩ -chelators on the binding of cPLA 2 ␣ to P60. Virtually no P60 was detected in the presence of 5 mM EDTA ( Fig. 2A) or EGTA (data not shown), indicating that this binding reaction may require Ca 2ϩ , a significant amount of which was present in the skim milk used for membrane blocking.
C2 domains generally form two distinct topological folds, as exemplified by the crystal structures of these domains of synaptotagmin I (type I) and phosphoinositide-specific phospholipase C-␦1 (type II) (22,23). cPLA 2 ␣ belongs to the latter group, and the amino acid residues essential for its Ca 2ϩ binding are Asp 40 , Thr 41 , Asp 43 , Asn 65 , Asp 93 , Ala 94 , and Asn 95 (36). Therefore, we replaced two of these residues with other amino acids, D43N and D93N, as the resulting mutation has been shown recently to reduce the Ca 2ϩ sensitivity of cPLA 2 ␣ markedly (37). Under the conditions that resulted in binding of native cPLA 2 ␣ (1-138) to P60, this mutant cPLA 2 ␣ (1-138) did not bind to P60 (Fig. 2B). Recently, it was reported that in A23187-stimulated Sf9 cells, the D43N or D43N/D93N cPLA 2 ␣ mutants were unable to translocate to the nuclear envelope, whereas the D93N mutant was able to do so, revealing a distinctive role of the two aspartate residues (67). Whether mutations in either of the two aspartate residues would alter the interaction with vimentin is now under investigation.
P60 is Identical to Vimentin-In order to identify P60, the nuclear matrix-rich fraction of 3Y1 cells was separated by SDS-PAGE and electrotransferred to a polyvinylidene difluoride membrane. The P60 protein band was cut off, and peptide mapping using lysylendopeptidase by the method of Cleveland et al. (38) was performed to determine the amino-terminal amino acid sequences of two major fragments. On the basis of these sequences (ARVEVEXDNLXED and GXNEDLERQ), degenerate oligonuclotides were designed, and degenerate PCR analysis was carried out using 3Y1 cell cDNA as the template. The DNA sequence of a 1.5-kilobase pair fragment thus ob-tained was determined and found to be identical to the corresponding part of rat vimentin cDNA. A survey of the amino acid sequence of vimentin revealed that the two peptide sequences of P60 that we determined corresponded to amino acids 168 -180 and 335-343 of vimentin.
To establish whether P60 is indeed identical to vimentin, we used a mutant clone of SW13, a human adrenocortical carcinoma cell line spontaneously devoid of vimentin expression (39). P60 was detected in the wild-type SW13 cells, which expressed vimentin as an intermediate filament component, but not in the vimentin-deficient clone (Fig. 3A, left panel). Furthermore, immunoblotting with an anti-vimentin antibody showed that vimentin migrated as a 60-kDa protein, which was (like P60) detected in 3Y1, 293, and wild-type SW13 cells but not in vimentin-deficient SW13 cells. (Fig. 3A, right panel). Moreover, when the membrane shown in Fig. 1F was reprobed with an anti-vimentin antibody, a 60-kDa immunoreactive vi- The lysates (10-g protein equivalents) of several cells were subjected to Far-Western analysis using GST-tagged (A) and T7 peptide-tagged (B) cPLA 2 ␣. C and D, 3Y1 cell lysate was subjected to far-Western analysis using equivalent concentrations of GST-tagged native and truncated cPLA 2 ␣s. E, the effect of excess recombinant cPLA 2 ␣ on the interaction between GST-cPLA 2 ␣ and P60 in 3Y1 cells. The nitrocellulose membrane was incubated with GST-cPLA 2 ␣ (1-138) as a probe in the presence (ϩ) or absence (Ϫ) of more than 10 times excess recombinant cPLA 2 ␣. F, the subcellular distribution of P60 in 3Y1 cells was assessed using GST-cPLA 2 ␣ (1-138). Each fraction was prepared as described under "Experimental Procedures." Lanes 1-4, Nonidet P-40-soluble, Nonidet P-40-insoluble, chromatin-rich, and nuclear matrix-rich fractions, respectively. mentin protein was visualized in the Nonidet P-40-insoluble and nuclear matrix-rich fractions (data not shown).
We cloned full-length rat vimentin cDNA from 3Y1 cells, and recombinant His-tagged vimentin protein was prepared using a bacterial expression system. Far-Western analysis revealed that GST-cPLA 2 ␣ (1-138) interacted with recombinant vimentin (Fig. 3B). Moreover, in the coprecipitation assay, in which recombinant vimentin was incubated with GST-cPLA 2 ␣ (1-138)-bound glutathione-Sepharose beads, vimentin precipitated with the beads in the presence but not the absence of Ca 2ϩ (Fig. 3C). GST alone failed to precipitate vimentin in this assay irrespective of the presence of Ca 2ϩ .
Colocalization of cPLA 2 ␣ and Vimentin in Activated Cells-We investigated the intracellular distributions of cPLA 2 ␣ and vimentin using a 293 cell clone (293-cPLA 2 ␣) that stably expresses a high level of cPLA 2 ␣ protein (33). In unstimulated 293-cPLA 2 ␣ cells, cPLA 2 ␣ was distributed diffusely in the cytoplasm, whereas the perinuclear area was rich in vimentin (Fig. 4A). After A23187 stimulation, a significant amount of the cPLA 2 ␣ appeared to translocate around the perinuclear site and clearly colocalized with the perinuclear vimentin (Fig. 4B).
Identification of the cPLA 2 ␣-binding Domain of Vimentin-Intermediate filament proteins are usually composed of three distinct regions: the head, rod, and tail domains (40). In order to identify which domain(s) of vimentin is responsible for binding to cPLA 2 ␣, we prepared recombinant truncated vimentin mutants composed of the head (amino acids 1-125), rod (amino acids 95-406), and tail (amino acids 407-466) domains alone and assessed the ability of each to bind to GST-cPLA 2 ␣ (1-138) by performing far-Western analysis. Neither the rod nor the tail domain bound to GST-cPLA 2 ␣ (1-138), but the head domain did (Fig. 5), indicating a physical association between the cPLA 2 ␣ C2 and vimentin head domains.
To determine whether cPLA 2 ␣ binds to other intermediate filament proteins, we examined the binding of GST-cPLA 2 ␣ (1-138) to GFAP, another type of intermediate filament protein, using human glioblastoma U251 cells, which express both vimentin and GFAP. Far-Western blotting demonstrated that cPLA 2 ␣ bound to vimentin but not to GFAP (Fig. 3D). This result is consistent with the fact that the rod domains are highly conserved among several intermediate filament component proteins, whereas the head and tail domains are structurally variable, and implies rather specific interaction between cPLA 2 ␣ and vimentin.
Implications of the Interaction between cPLA 2 ␣ and Vimentin for AA Metabolism-In order to investigate the physiological roles of the interaction between cPLA 2 ␣ and vimentin, we introduced vimentin cDNA into vimentin-deficient SW13 cells. The established stable transformant SW13-G5, which expressed vimentin, and its parental cells, which did not (Fig.  6A), were prelabeled with [ 3 H]AA, and [ 3 H]AA release in response to A23187 was evaluated. SW13-G5 cells released about twice as much [ 3 H]AA as the parental cells (0.52 and 1.2% AA release by parental and G5 cells, respectively) after stimulation with 10 M A23187 for 30 min (Fig. 6B). The cPLA 2 ␣ expression levels of these cells, assessed by enzyme assay, were comparable (data not shown). These observations indicate that vimentin expression facilitates cPLA 2 ␣-mediated AA release.
Our finding that vimentin interacted with cPLA 2 ␣ via its head domain led us to formulate the hypothesis that the overexpressed vimentin head domain would compete with endogenous vimentin for binding to cPLA 2 ␣, thereby exerting a dominant-inhibitory effect on cPLA 2 ␣ function. We addressed this issue by establishing a 3Y1 cell transfectant that stably overexpressed EGFP-fused vimentin (1-125) (3Y1-vim (1-125)). The expression of vimentin (1-125) in the transfectants was verified by immunoblotting using an anti-EGFP antibody (Fig.  7A). The amount of [ 3 H]AA released (Fig. 7B) and PGE 2 produced concomitantly (Fig. 7C) after stimulation with 3 M A23187 for 30 min by 3Y1-vim (1-125) cells were significantly lower than those produced by the parent 3Y1 cells. The cPLA 2 ␣ expression levels of both cells were comparable (data not shown). Therefore, vimentin (1-125), which interacts with cPLA 2 ␣ but lacks the ability to polymerize (41), exerted a dominant-negative effect on cPLA 2 ␣ function, thereby suppressing AA release and PG generation. It should be noted that there is a significant quantitative difference between decreases in AA (2-fold) and PGE 2 (5-fold) caused by overexpressed vimentin head domain. A possible explanation for this is that a PLA 2 (s) other than cPLA 2 ␣ may also contribute to AA release, but only the AA released by cPLA 2 ␣ can be processed to the PGE 2 -biosynthetic pathway. Indeed, remaining AA release was partially suppressed by bromoenol lactone (data not shown), which inhibits iPLA 2 , a PLA 2 isozyme playing more important role in phospholipid remodeling than in eicosanoid generation (68,69). Alternatively, vimentin may function as a scaffold protein for cPLA 2 ␣ and downstream enzymes (COX and PGE 2 synthase) that assists their functional coupling, leading to efficient conversion of a subpopulation of AA to PGE 2 .

DISCUSSION
As has been shown by a number of studies, there is no doubt that cPLA 2 ␣ regulates the initial step of AA metabolism in response to stimuli that mobilize intracellular Ca 2ϩ (1-8, 12-  17,42). In vitro studies employing cPLA 2 ␣-specific inhibitors (19,34,35), antisense oligonucleotides (20,43), and cPLA 2 ␣ overexpression (33,44) have shown that this particular PLA 2 isozyme, which bears multifunctional domains, represents a rate-limiting step for the initiation of the lipid mediator-biosynthetic pathway. Furthermore, the importance of cPLA 2 ␣ has been confirmed by in vivo studies on cPLA 2 ␣ knockout mice (45,46). Following agonist-stimulated transmembrane signaling that increases the cytoplasmic Ca 2ϩ level, cPLA 2 ␣ undergoes translocation from the cytosol to the perinuclear envelope and endoplasmic reticulum (12,13), where many of the downstream eicosanoid-biosynthetic enzymes, including COX-1, COX-2, 5-LO, 5-LO-activating protein, and several terminal PG and leukotriene synthases, are located (27)(28)(29). Phosphorylation by mitogen-activated protein kinases increases the intrinsic activity of cPLA 2 ␣ in vitro (8) and synergizes with the Ca 2ϩ signaling pathway to provide the optimal AA-releasing response in vivo (6). The C2 domain of cPLA 2 ␣ is essential for Ca 2ϩ -dependent association of cPLA 2 ␣ with phospholipid vesicles (4,47,48), which is facilitated in the presence of phosphatidylinositol 4,5-bisphosphate and, therefore, consistent with the presence of a putative pleckstrin homology domain in the middle part of the enzyme (24). However, the molecular mechanism whereby cPLA 2 ␣ is directed to the intracellular membrane compartments, the perinuclear membranes in particular, in activated cells remains largely obscure. While searching for proteins that physically and functionally interact with cPLA 2 ␣, Wu et al. (49) recently identified, by means of the yeast two-hybrid system, a cellular protein, p11, that associated with the carboxyl-terminal portion of cPLA 2 ␣. Functional analysis revealed that p11 inhibited cPLA 2 ␣ activity both in vitro and in vivo, suggesting that it acts as a negative regulator of cPLA 2 ␣. In this study, we found that cPLA 2 ␣ interacted with a major intermediate filament protein, vimentin, which is expressed abundantly in the perinuclear regions of mesenchymal cells and a variety of cultured cells (40). This interaction occurred between the C2 domain of cPLA 2 ␣ and the head domain of vimentin. In agreement with the dependence of their interaction on Ca 2ϩ , confocal microscopic analysis revealed that they colocalized around the perinuclear area in A23187-stimulated but not unstimulated cells. Most importantly, overexpression of full-length vimentin augmented cPLA 2 ␣-initiated AA metabolism, whereas that of the vimentin head domain exerted a dominant-negative effect. Collectively, these results suggest that vimentin represents a functional adaptor for cPLA 2 ␣ that determines the intracellular localization and thereby modifies the function of cPLA 2 ␣, depending on the changes in the cytoplasmic Ca 2ϩ levels during cellular activation.
The C2 domain of cPLA 2 ␣, like those of protein kinase C and synaptotagmin, behaves as a Ca 2ϩ -dependent lipid-binding domain. Recent crystal structural analysis revealed that the C2 domain of cPLA 2 ␣ captured two Ca 2ϩ ions at one end of the domain between three loops, CBR1, CBR2, and CBR3 (36,37), and the Ca 2ϩ ions interacted directly with the phosphate moiety of a lipid head group. In this study, we showed that Asp 43 (which resides in CBR1) and Asp 93 (which resides in CBR3) in the C2 domain are involved in the cPLA 2 ␣-vimentin interaction. This is an important finding, because mutation of either of these two aspartates in the native C2 domain reduced Ca 2ϩ binding, eventually leading to reduce phospholipid binding and enzyme activity (37). The cPLA 2 ␣ C2 domain consists of eight ␤-sheet structures, ␤1-␤8 (36,63). Our present findings (that cPLA 2 ␣ (1-81), which contains the first four ␤-sheets (␤1-␤4) of the C2 domain, interacted with vimentin, whereas neither cPLA 2 ␣ (1-35), which contains the first ␤-sheet (␤1), nor cPLA 2 ␣ (36 -81), which contains ␤2-␤4, did so) suggest that the structural determinants that lie in the ␤1-␤4 region are critical for recognition by vimentin. Interestingly, critical residues required for phospholipid membrane binding are separated from the ␤1-␤4 region (64), raising a possibility that cPLA 2 ␣ C2 domain could bind simultaneously to both vimentin and phospholipid membranes in a Ca 2ϩ -dependent manner. If this hypothesis is correct, cPLA 2 ␣ associated with vimentin could efficiently hydrolyze phospholipids adjacent to vimentin intermediate filaments. In this regard, vimentin would act as a scaffold protein that assists appropriate interaction of cPLA 2 ␣ with perinuclear phospholipid membranes.
Several lines of evidence have suggested that vimentin intermediate filament plays a role in intracellular lipid transport (50,51,65,66). One function of the cytoskeleton is to organize the subcellular architecture and topography of the organelles (52). Microtubules maintain the organization of the Golgi apparatus and mitochondria are aligned along the microtubules, vimentin intermediate filament forms a cage around developing lipid droplets (53), lipid droplets in adrenal cells are attached to vimentin intermediate filament (54), and the nuclei of vimentin-deficient SW13 cells have an abnormally lobulated shape (55). A number of proteins that regulate cellular response and homeostasis, including kinases (56), heat shock protein (57), and transglutaminase-related antigen (58), bind to intermediate filament. Our results suggest that vimentin intermediate filament may contribute to bringing the proteins involved in AA metabolism into proximity, enabling eicosanoid generation to proceed in the microcompartments.
Our finding that the vimentin head domain alone exhibited a dominant-inhibitory effect on cPLA 2 ␣ function suggests that the intrinsic activity of cPLA 2 ␣ does not increase as a result of associating with the vimentin head domain and that the inability of the truncated vimentin to form a filament structure, which depends entirely on the rod domain (40,41), may prevent proper localization of cPLA 2 ␣. Serine/threonine residues in the head domain of vimentin are known to be phosphorylated by several protein kinases in vitro and in vivo (59,60), and such site-specific phosphorylation by various types of protein kinase contributes to regulation of the filament structure. During cytokinesis, the GTP-binding protein Rho-associated kinase phosphorylated Ser 71 on vimentin at the cleavage furrow in late mitotic cells (61). Ca 2ϩ waves in astrocytes induced global phosphorylation of vimentin by Ca 2ϩ /calmodulin-dependent kinase II (62), and a small population of vimentin intermediate filaments highly phosphorylated by Ca 2ϩ /calmodulin-dependent kinase II underwent structural alteration into short filaments (62). It remains unclear whether phosphorylation of the vimentin head domain affects the interactions and functions of cPLA 2 ␣.
Our results shed light on the regulatory mechanism of cPLA 2 ␣ in immediate AA release, which is accompanied by a sharp increase in the cytoplasmic Ca 2ϩ level. Evidence is accumulating to suggest that cPLA 2 ␣ is also essential for the delayed PG-biosynthetic response induced by proinflammatory cytokines and lipopolysaccharide, despite the relatively poor Ca 2ϩ response during this phase. We propose that spatially and temporally synchronous coupling between local elevations in the Ca 2ϩ concentration and phosphorylation may converge on the prolonged activation of cPLA 2 ␣, leading to delayed AA release. Whether vimentin participates in regulation of the delayed response is now under investigation. Nevertheless, the cooperative action of cPLA 2 ␣, an initiator of AA metabolism, and vimentin, an intermediate filament protein, implies that the assembly and/or disassembly of the cytoskeleton components not only play pivotal roles in cell morphology but also affect the cellular capacity to elicit eicosanoid-biosynthetic responses.