Group IVC cytosolic phospholipase A₂γ is farnesylated and palmitoylated in mammalian cells

Abstract

Cytosolic phospholipase A₂γ (cPLA₂γ) is a member of the group IV family of intracellular phospholipase A₂ enzymes, but unlike the well-studied cPLA₂α, it is constitutively bound to membrane and is calcium independent. cPLA₂γ contains a C-terminal CaaX sequence and is radiolabeled by mevalonic acid when expressed in cPLA₂α-deficient immortalized lung fibroblasts (IMLF^−/−). The radiolabel associated with cPLA₂γ was identified as the farnesyl group. The protein farnesyltransferase inhibitor BMS-214662 prevented the incorporation of [³H]mevalonic acid into cPLA₂γ and partially suppressed serum-stimulated arachidonic acid release from IMLF^−/− and undifferentiated human skeletal muscle (SkMc) cells overexpressing cPLA₂γ, but not from cells overexpressing cPLA₂α. However, BMS-214662 did not alter the amount of cPLA₂γ associated with membrane. These results were consistent in COS cells expressing the C538S cPLA₂γ prenylation mutant. cPLA₂γ also contains a classic myristoylation site and several potential palmitoylation sites and was found to be acylated with oleic and palmitic acids but not myristoylated. Immunofluorescence microscopy revealed that cPLA₂γ is associated with mitochondria in IMLF^−/−, SkMc cells, and COS cells.

Phospholipases A₂ (PLA₂s) hydrolyze the sn-2 ester bond of membrane phospholipids and function in the production of lipid mediators, phospholipid acyl-chain remodeling, dietary lipid breakdown, and host defense. Convention divides this family into three main types based on their subcellular localizations and structural differences: the secreted PLA₂s, the intracellular group IV cytosolic phospholipase A₂s (cPLA₂s), and the group VI calcium-independent PLA₂s (1). The mammalian secreted PLA₂s constitute a large group of enzymes that have low molecular mass, require calcium for activity, and use a His/Asp dyad in the catalytic mechanism (2, 3). The intracellular PLA₂s, calcium-independent PLA₂s, and cPLA₂s have larger molecular masses, use an active site serine, and play a variety of roles in signaling, inflammation, and membrane remodeling, depending on the tissue and cell type (4–9).

One member of the group IV PLA₂ family, group IVA cPLA₂α (cPLA₂α), has been well studied because of its specificity for releasing sn-2 arachidonic acid from membrane phospholipids for the production of eicosanoids (6, 8, 10). The oxygenated metabolites of arachidonic acid, prostaglandins, and leukotrienes, produced through the cyclooxygenase and lipoxygenase pathways, promote acute inflammatory responses and are also important in regulating many physiological processes (11). cPLA₂α is subject to posttranslational regulatory controls via phosphorylation and Ca²⁺ (12–16). Calcium functions by binding to an N-terminal C2 domain and inducing the translocation of cPLA₂α from the cytosol to the Golgi, endoplasmic reticulum, and nuclear envelope (17–20).

Two additional group IV enzymes have been identified. Group IVC cPLA₂γ (cPLA₂γ) and group IVB cPLA₂β (cPLA₂β); both have ~30% homology to cPLA₂α (21–23). The active site residues found in cPLA₂α are conserved in these paralogs, suggesting that they have a similar catalytic mechanism (22). cPLA₂β contains an N-terminal C2 domain that confers calcium sensitivity and an additional N-terminal extension containing a JmjC domain, the function of which is unknown (21–24). In contrast, cPLA₂γ does not contain a C2 domain, consistent with its lack of regulation by Ca²⁺ (21, 22). The regulatory phosphoryla tion sites used by cPLA₂α are not present in cPLA₂γ, although it possesses multiple putative PKC phosphorylation sites, the use of which has yet to be investigated. Unlike cPLA₂α and cPLA₂β, which are widely distributed in mammalian tissues, cPLA₂γ message is abundantly expressed in skeletal muscle (SkMc), brain, and heart (21, 22). In another significant departure from cPLA₂α, cPLA₂γ and cPLA₂β do not exhibit strong sn-2 acyl chain specificity (25, 26). cPLA₂γ exhibits relatively high lysophospholipase activity, as reported previously for cPLA₂α (25, 27).

A unique property of cPLA₂γ is that it is constitutively bound to cell membrane and contains putative acylation sites and a C-terminal prenylation site that may regulate its membrane association (21). The C-terminal sequence CCLA on cPLA₂γ fits the consensus sequence of a CaaX box (a is usually, but not necessarily, an aliphatic residue), a motif that is recognized by protein prenyltransferases for the attachment of either a 15 carbon farnesyl or a 20 carbon geranylgeranyl to the cysteine sulfhydryl (SH) group (28, 29). Indeed, cPLA₂γ becomes radiolabeled when expressed in COS cells grown in the presence of [³H]mevalonic acid, the precursor of prenyl groups in mammalian cells (21). The structure of the prenylated C terminus cannot be inferred from the cPLA₂γ CaaX sequence. The CCLA sequence could be recognized by protein farnesyltransferase, resulting in the attachment of a farnesyl group to the N-terminal-most cysteine SH.

cPLA₂γ has been shown to be farnesylated when expressed in insect cells (30); however, the structure of the prenyl group on cPLA₂γ in mammalian cells has not been investigated. cPLA₂γ could also be a substrate for protein geranylgeranyltransferase type I, resulting in geranylgeranylation of the N-terminal-most cysteine SH. Finally, it may be noted that a subset of Rab GTPases contain the C-terminal sequence CCXX, and the enzyme protein geranylgeranyltransferase type II attaches a geranylgeranyl group to each of the two cysteine SH groups (31). Thus, cPLA₂γ could be a doubly geranylgeranylated protein in mammalian cells. cPLA₂γ also contains a putative myristoylation site as well as several potential fatty acylation sites.

Both myristoylation and palmitoylation are widespread fatty acid modifications on membrane-associated proteins. In addition to facilitating high-affinity membrane association on dually lipidated proteins, these modifications can aid in protein trafficking of enzymes to specific compartments or subdomains (32, 33). Palmitoylation has also been implicated in the regulation of enzymatic activity in proteins, notably multiple mitochondrial enzymes (34–36). In the present study, we have found that cPLA₂γ is farnesylated and acylated in mammalian cells. Prenylation of cPLA₂γ was also found to be important for the function of cPLA₂γ in intact cells. In addition, we have found that cPLA₂γ is localized to the mitochondria when expressed in mammalian cells.

METHODS

Materials

The protein farnesyltransferase inhibitor BMS-214662 was prepared as described (37). An analog of BMS-214662 containing a methyl group at N1 of the imidazol ring (Me-BMS-214662) was prepared similarly. The compounds were purified by HPLC on a C18 reverse-phase column and shown to have the correct structure by ¹H-NMR and electrospray ionization mass spectrometry. Simvastatin lactone (a gift from Merck, Rahway, NJ) was saponified by dissolving 25.5 mg in 520 μl of 0.1 M NaOH, adding an equal volume of 150 mM NaCl, and incubating at 60°C for 3 h until the simvastatin was dissolved. The pH was adjusted to 7.0 with 0.1 N HCl, and the solution was filter-sterilized and stored at −20°C. [5,6,8,9,11,12,14,15-³H]arachidonic acid (specific activity, 100 Ci/mmol), [³H]mevalonic acid (specific activity, 40 Ci/mmol), [³H]palmitic acid (specific activity, 45 Ci/mmol), and [³H]myristic acid (specific activity, 60 Ci/mmol) were from Perkin-Elmer Life Sciences. Amplify, used for fluorography of tritiated bands on polyacrylamide gels, was from Amersham. Mouse serum was from Atlanta Biologicals. Clonetics^™ SkMc cells and growth media were from Cambrex (Rutherford, NJ). HEK293 (QBI-293A) cells and adenovirus containing the gene for green fluorescent protein (GFP) were obtained from Qbiogene (Carlsbad, CA). Production of baculovirus and adenovirus for expression of cPLA₂γ in Spodoptera frugiperda (Sf9) and mammalian cells, respectively, was carried out as described previously (25). The mitochondrion-specific monoclonal antibody to oxidative phosphorylation complex V, subunit b, and the secondary antibodies used for immunofluorescence were from Molecular Probes.

cPLA₂γ polyclonal antibody production and Western blot analysis

Human cPLA₂γ was cloned as described previously (25). To produce glutathione S-transferase (GST)-cPLA₂γ, cPLA₂γ was cloned into the EcoRI/PstI sites of pAcGHLT baculovirus transfer vector, and baculovirus was generated as described previously (25). Sf9 cells grown in suspension (500 ml) were infected with baculovirus for 48 h and lysed in 50 mM Hepes buffer, pH 7.4, 150 mM NaCl, 10% glycerol, 1% Triton X-100, 1 mM EGTA, 1 mM EDTA, and one complete protease inhibitor cocktail tablet per 50 ml (Roche). GST-cPLA₂γ was affinity-purified using glutathione agarose beads by standard protocols. GST-cPLA₂γ (50–100 μg) was added to an equal volume of complete Freund’s adjuvant and injected subcutaneously into rabbits. Subsequent booster injections were carried out every 3 weeks using Freund’s incomplete adjuvant. Antiserum was obtained 10 days after each injection and analyzed for reactivity to cPLA₂γ by Western blotting.

Cell homogenates for Western blot analysis of cytosol and membrane fractions were prepared by sonicating cells in 50 mM Hepes, pH 7.4, containing 0.34 M sucrose, 1 mM EGTA, 10% glycerol, 10 μg/ml leupeptin, 10 μg/ml aprotinin, and 1 mM phenylmethylsulfonyl fluoride. Homogenates were centrifuged at 100,000 g for 45 min to obtain the cytosol and membrane fractions. The protein concentration of the cell fractions was determined using the bicinchoninic acid reagent. Lysates were diluted in Laemmli buffer, and proteins were separated on 10% polyacrylamide gels, transferred to nitrocellulose, and blocked for 1 h in Tris-buffered saline containing 0.25% Tween 20 and 5% nonfat dry milk. Nitrocellulose membranes were incubated overnight in a 1:1,000 dilution of anti-cPLA₂γ antiserum, and immunoreactive protein was detected using the Amersham Biosciences ECL system.

Cell culture, production of DNA constructs, and recombinant adenovirus

Immortalized mouse lung fibroblasts lacking group IVA cPLA₂α (IMLF^−/−) were isolated and immortalized with SV40, as described previously (38). The replication-deficient recombinant adenoviruses carrying the cDNA for untagged cPLA₂γ (Ad-cPLA₂γ) and for GFP-tagged cPLA₂α (Ad-GFPcPLA₂α) were generated using the AdEasy vector system (Qbiogene) and titered, and expression levels were determined as described previously (25). Freshly isolated human SkMc cells obtained from Cambrex were grown in complete human SkMc growth medium for two passages and then frozen. Cells were thawed and passaged once by trypsinization according to the manufacturer’s protocol (Cambrex) before use in experiments. The C538 in the CaaX box of cPLA₂γ was mutated to a serine (C538S) using PCR-based site-directed mutagenesis (Stratagene) and the primer 5′-CTA TGC CAA GCA GCT ACT TCG GGC ACT-3′. The putative N-myristoylation site (G2) was mutated to alanine (G2A) by the same method. The primer used was 5′-TTC GGA CCG CAG TGC ACC ATG GCA AGC TCT GAA GTT-3′. The C-terminal truncated cPLA₂γ was created by mutating K490 into a TAG stop codon using the primer 5′-GAC ACA TAC GAC ACA TTC TAG CTT GCT GAC-3′. Wild-type and mutant cPLA₂γ were cloned into the pcDNA3.0 mammalian expression vector. All constructs were confirmed by sequencing. COS and HEK293 cells were transiently transfected using FuGENE transfection reagent according to the manufacturer’s instructions (Roche Diagnostics, Indianapolis, IN). Expression of wild-type and mutant cPLA₂γ was confirmed by Western blotting. RT-PCR of endogenous cPLA₂γ in SkMc cells was conducted using 5′-GCT CAC ATT GCC TGC CTT GGG GTC CTG-3′ and 5′-AGT GCC CGA AGT TGC TGC TTG GCA TAG-3′ and RNA isolated with the RNeasy mini kit (Qiagen).

Structural analysis of the cPLA₂γ prenyl group

IMLF^−/− were plated at 1 × 10⁶ cells/100 mm dish and incubated for 10 h in DMEM containing 2% FBS. The cells were washed and incubated in 4 ml of serum-free DMEM containing 0.1% BSA and Ad-cPLA₂γ as described previously (25). After incubation for 90 min, additional DMEM containing 0.1% BSA, 1 mCi of [³H]mevalonic acid, and 10 μM simvastatin was added to the cells to prevent the mevalonate from becoming metabolized to cholesterol. For some experiments, the medium also included 1 μM of the farnesyltransferase inhibitor BMS-214662 or Me-BMS-214662. After incubation for 26 h, cells were rinsed with PBS and solubilized in ice-cold lysis buffer (50 mM Hepes, pH 7.4, 150 mM sodium chloride, 1% Nonidet P-40, 200 μM sodium orthovanadate, 10 mM tetrasodium pyrophosphate, 100 mM sodium fluoride, 10 μg/ml leupeptin, 10 μg/ml aprotinin, 100 μM phenylmethylsulfonyl fluoride, and 300 nM p-nitrophenyl phosphate). The lysate was incubated at 0°C for 15 min, then centrifuged for 10 min at 4°C at full speed in a microfuge. Total protein in the supernatant was quantified using the bicinchoninic acid method. Immunoprecipitation of cPLA₂γ was carried out by incubating the lysates at 4°C with protein A-Sepharose beads and anti-cPLA₂γ antiserum (1:25) overnight while rotating. The beads were washed five times in lysis buffer and then boiled in Laemmli buffer for 10 min. One-third of the labeled proteins was separated by SDS-PAGE, and the gel was fixed in isopropanol-water-acetic acid (25:65:10) for 10 min. The gel was incubated for 30 min in Amplify at room temperature before drying and then exposed to film for 8 days for visualization of the bands.

For samples to be extracted for HPLC prenylation analysis, the remaining two-thirds of the immunoprecipitate was electrophoresed and dried without fixation or Amplify treatment. The pieces of dried gel corresponding to the location of radiolabeled cPLA₂γ (61 kDa) were excised. Protein in the gel slices was extracted in the presence of 100 mg of BSA as carrier (39), and half of the extract was further processed. Extracted protein was precipitated with trichloroacetic acid, and the precipitate was washed with cold acetone as described (39). The dried protein was dissolved in guanidine-HCl/sodium phosphate buffer, N-acetyl-cysteine(S-farnesyl) and N-acetyl-cysteine(S-geranylgeranyl) (20 mg each; Bachem, Inc.) were added, and the sample was treated with Raney-nickel as described (39). The pentane extract was concentrated and submitted to HPLC analysis, and radioactivity in the column fractions was determined by scintillation counting (39).

Analysis of cPLA₂γ fatty acylation

Adenoviral infection and immunoprecipitation were modified from the protocol for the farnesylation experiments described above. IMLF^−/− cells were cultured in serum-free DMEM containing 0.1% fatty acid-free BSA for 22 h and then incubated for 4 h in the same medium containing either 0.3 mCi/ml [³H]palmitate or 0.1 mCi/ml [³H]myristate. After immunoprecipitation of cPLA₂γ (as described above), proteins were separated by SDS-PAGE and the gels were either submitted to fluorography or incubated briefly in water and dried. For analysis of fatty acylation, bands corresponding to cPLA₂γ were excised. The gel slice was mixed in 1 ml of 50% aqueous methanol at room temperature in a polypropylene tube. After 2–3 h, the liquid was removed and the gel slice was washed again as above. The gel slice was dried with a Speed-Vac (Savant Instruments), and 0.7 ml of 1.5 M aqueous NaOH was added to the tube. After incubation for 2 h at 30°C, the solution was brought to pH ~2 by the addition of 6 M HCl (monitored with pH paper). The gel slice and solution were transferred to a glass tube, and 2.5 ml of Dole solvent (24 ml of isopropanol, 0.6 ml of 0.5 M H₂SO₄, and 6 ml of n-heptane) was added (using a portion to rinse the polypropylene tube). The sample was vortexed, 1 ml of water and 1.5 ml of n-heptane were added, and the sample was vortexed again.

The upper organic phase was transferred to a glass tube, and solvent was removed from the organic and aqueous phases with a Speed-Vac. To the tube containing the residue from the aqueous phase and gel slice was added 0.7 ml of 6 M HCl, the top of the tube was sealed with a glassblower torch, and the tube was heated in an incubator for 4 h at 100°C. After cooling, the tube was opened with the aid of a scoring file, and solvent was removed with a Speed-Vac. Water was added to the residue (0.7 ml), and the sample was extracted by the addition of Dole solvent, water, and n-heptane as described above. The upper organic phase was transferred to a glass tube, and the solvent was removed from both tubes with a Speed-Vac. Methanol (1 ml) was added to each tube, and the radioactivity in 0.1 ml aliquots was determined by scintillation counting.

Methanol in the tube containing the organic extract of the NaOH-treated and HCl-acidified sample (see above) was transferred to a 1.5 ml vial with a Teflon-septum screw cap, 1 mmol each of myristic acid, palmitic acid, and oleic acid were added (from 50 mM stock solutions in methanol), and solvent was removed with a Speed-Vac. Diisopropylethylamine (0.5 ml of 10%, v/v; Aldrich) and 1% (v/v) pentafluorobenzyl bromide (Pierce) in CH₃CN was added to the residue, and the capped vial was heated at 60°C for 15 min. An overnight Speed-Vac treatment removed the solvent and excess reagents. The residue was dissolved in CH₃CN and injected onto a C18 reverse-phase HPLC column (1 × 25 cm; Vydac 218TP1010) equilibrated previously with 70% CH₃CN in water at a flow rate of 1.5 ml/min. The column was developed with a solvent gradient of 70% CH₃CN in water to 100% CH₃CN over 20 min and then held at 100% CH₃CN for 60 min. Absorbance at 254 nm was monitored, and 2 min fractions were collected into scintillation vials. Solvent was removed with a Speed-Vac, and the residues were dissolved in scintillation fluid and counted.

Arachidonic acid release assays

Cells were plated at 1.25 × 10⁴ cells/cm² on 24-well plates in DMEM containing 2% FBS for IMLF^−/− or in complete SkMc medium for growing SkMc cells and incubated overnight. COS cells were plated at the same density and transfected as described above. The cells were washed and then incubated in 150 μl of DMEM containing 0.1% BSA with Ad-cPLA₂γ or Ad-GFP control virus. After incubation for 90 min, DMEM (500 μl) containing 0.1% BSA and 0.2 μCi/ml [³H]arachidonic acid was added. In some experiments 1 μM BMS-214662 or Me-BMS-214662 was added. After a 26 h incubation, cells were washed two times in medium containing 0.1% BSA to remove unincorporated arachidonic acid. Cells in this medium were then stimulated with 10% mouse serum for 3 h (25). The amount of radioactivity released into the medium was determined and expressed as a percentage of the total counts incorporated into the cells.

Immunofluorescence microscopy

IMLF⁻_/⁻ or SkMc cells were plated on 35 mm glass-bottomed MatTek plates at 1.25 × 10⁴ cells/cm² and infected with Ad-cPLA₂γ as described above. The cells were rinsed once with PBS and incubated for 15 min in ice-cold fixative containing 3.2% paraformaldehyde and 3% sucrose in PBS. After fixation, cells were rinsed five times with cold PBS and incubated for 15 min in 0.1% Triton X-100 in PBS. Cells were then rinsed with PBS and blocked for 1 h in PBS containing 10% FBS. Fixed cells were incubated with anti-cPLA₂γ polyclonal antiserum (1:50) overnight, followed by a 2 h incubation with goat anti-rabbit Texas Red-conjugated secondary antibody (1:100). Cells were costained with mitochondrion-specific anti-oxidative phosphorylation complex V monoclonal antibody (1:20) and an AlexaFluor 488-conjugated anti-mouse secondary antibody (1:100). Control experiments with Texas Red anti-rabbit secondary antibody only and with both primary antibodies and both secondary antibodies were included to identify any nonspecific reactions. After incubation with each antibody, the cells were washed five times in PBS containing 10% FBS. All antibodies were diluted into blocking solution and centrifuged at full speed in a microfuge before use. Controls using secondary antibodies alone were included and revealed only a low level of background fluorescence (data not shown). IMLF^−/− were visualized using a Nikon diaphot inverted microscope with a 60×, 1.4 numerical aperture oil-immersion lens and a Photometrics charge-coupled device camera using FITC and Tetra-methyl Rhodamine Iso-Thiocyanate filters. Images were acquired with IP Labs software (Scanalytics, Inc.). Immunofluorescence microscopy of SkMc cells was carried out using an Olympus inverted microscope with a 60×, 1.25 numerical aperture oil-immersion objective, and images were collected with a charge-coupled device camera using Chroma dichroic mirrors fitted with emission filters for FITC and Texas Red detection. Image acquisition and analysis were performed using TILL visION software (TILL Photonics).

RESULTS

cPLA₂γ is farnesylated in mammalian cells

IMLF^−/− infected with Ad-cPLA₂γ were grown in the presence of [³H]mevalonic acid followed by immunoprecipitation of cPLA₂γ. One-third of the total volume of the immunoprecipitate was resolved by SDS-PAGE, and the gel was submitted to fluorography. As shown in Fig. 1A, a major radioactive band was seen with an apparent molecular mass of ~60 kDa, which is the predicted molecular mass of cPLA₂γ, demonstrating that cPLA₂γ is labeled by mevalonic acid, consistent with previous results (21). The remaining two-thirds of the immunoprecipitate was resolved by SDS-PAGE but not submitted to fluorography, and a gel piece corresponding to the cPLA₂γ region was excised. Protein was eluted from the gel slice and treated with Raney-nickel, which cleaves the carbon-sulfur bond of protein prenyl groups (31, 39). HPLC analysis of the released radiolabeled material clearly shows that it comigrates with the 15 carbon trimethyl dodecatriene (fraction 32) and that no 20 carbon material was detected (Fig. 1B). The results demonstrate that cPLA₂γ is a farnesylated protein in mammalian cells.

Fig. 1.

Cytosolic phospholipase A₂γ (cPLA₂γ) is farnesylated when expressed in immortalized mouse lung fibroblasts lacking group IVA cPLA₂α (IMLF^−/−). A: cPLA₂γ was immunoprecipitated from [³H]mevalonic acid-labeled IMLF^−/−, separated by SDS-PAGE, and subjected to fluorography. B: Reverse-phase HPLC analysis of the radiolabeled hydrocarbon material released from immunoprecipitated cPLA₂γ that was treated with Raney-nickel. Fractions were collected every 2 min and submitted to scintillation counting to provide dpm of radioactive material, which is plotted. The arrows indicate the elution positions of the 15 carbon all-trans-2,6,10-tri-methyl-2,6,10-dodecatriene (derived from the farnesyl group) and the 20 carbon all-trans-2,6,10,14-tetramethyl-2,6,10,14-hexadecateratraene (derived from the geranylgeranyl group). These retention times are derived by monitoring the absorbance at 210 nm (not shown) to detect the authentic hydrocarbons derived from Raney-nickel cleavage of the N-acetyl-cysteine(S-prenyl) standards that were added to the reaction mixture (see Methods).

Functional effects of farnesylation inhibition

The tetrahydrobenzodiazepine analog BMS-214662 is a potent inhibitor of mammalian protein farnesyltransferase, with an IC₅₀ of 1.35 nM (37). BMS-214662 is also cell-permeable, as 100 nM causes 60% reversion of the transformation of Rat-1 fibroblasts that express oncogenic Ras GTPase (37). Thus, we tested BMS-214662 for its ability to block the incorporation of [³H]mevalonic acid into cPLA₂γ expressed in IMLF^−/−. cPLA₂γ was immunoprecipitated from both untreated cells and cells exposed to 1 μM BMS-214662. A radiolabeled band at ~60 kDa was present in untreated cells. As shown in Fig. 2, incorporation of [³H]mevalonic acid into the 60 kDa cPLA₂γ band was almost completely blocked by the farnesyltransferase inhibitor BMS-214662. The inhibitor was added to the cells at the time of adenovirus infection to prevent the prenylation of cPLA₂γ as it was being expressed. Prenylation is thought to be an irreversible modification; therefore, the inhibitor would not cause the loss of the prenyl group from an existing pool of cPLA₂γ protein. BMS-214662 did not affect the levels of cPLA₂γ expression (see Fig. 5).

Fig. 2.

Inhibition of [³H]mevalonic acid labeling of cPLA₂γ with the protein farnesyltransferase inhibitor BMS-214662. IMLF^−/− were infected with Ad-cPLA₂γ and incubated with or without 1 μM BMS-214662. cPLA₂γ was immunoprecipitated, separated by SDS-PAGE, and subjected to fluorography.

Fig. 5.

Membrane association of cPLA₂γ is not determined by the C-terminal region or farnesylation. A: Cell homogenates of IMLF^−/− and SkMc cells infected with Ad-cPLA₂γ and incubated with or without BMS-214662 were prepared as described in Methods. The relative amount of cPLA₂γ in the 100,000 g soluble (cytosol; C) and particulate (membrane; M) fractions was determined by Western blot analysis of 30 μg of total protein per lane. B: Cell homogenates of COS cells overexpressing cPLA₂γ with the C-terminal CaaX sequence mutated were prepared as described in Methods. Amounts of mutated cPLA₂γ in the soluble (C) and particulate (M) fractions were determined by Western blotting. C: Cell homogenates of HEK cells overexpressing an ~55 kDa cPLA₂γ with the C terminus truncated were prepared as described in Methods. Amounts of mutated cPLA₂γ in the soluble (C) and particulate (M) fractions were determined by Western blotting.

To analyze the potential role of prenylation on cPLA₂γ function, the effect of BMS-214662 on [³H]arachidonic acid release from IMLF^−/− expressing cPLA₂γ was determined. We have previously reported that cPLA₂γ expressed in IMLF^−/− is activated by serum, resulting in the release of arachidonic acid and other fatty acids (25). IMLF^−/− were infected with Ad-cPLA₂γ, and with Ad-GFP as a control, and the effect of BMS-214662 on serum-induced arachidonic acid release was determined. As shown in Fig. 3 (upper panel), there was increased arachidonic acid release from IMLF^−/− expressing cPLA₂γ compared with control cells infected with Ad-GFP, and BMS-214662 inhibited cPLA₂γ-mediated [³H]arachidonic acid release by ~60%. A recent study has shown that mutation of the CaaX sequence also suppresses A23187-stimulated arachidonic acid release in HEK293 cells overexpressing cPLA₂γ (40).

Fig. 3.

BMS-214662 inhibits cPLA₂γ-mediated arachidonic acid release from IMLF⁻_/⁻. IMLF^−/− were infected with Ad-cPLA₂γ and Ad-green fluorescent protein (GFP) control virus (upper panel) or Ad-GFPcPLA₂α and Ad-GFP control virus (lower panel) and incubated with or without 1 μM BMS-214662 in medium containing [³H]arachidonic acid. The cells were washed and then stimulated with 10% mouse serum for 3 h. The amount of radioactivity released into the medium is expressed as a percentage of the total incorporated radioactivity. Data shown are the average ±SEM of three independent experiments.

To determine the specificity of the inhibitor, the effect of BMS-214662 on cPLA₂α-mediated arachidonic acid release was tested (Fig. 3, lower panel). Serum stimulation of IMLF^−/− expressing cPLA₂α resulted in a large increase in arachidonic acid release compared with control cells, as we reported previously, and this response was unaffected by the protein farnesyltransferase inhibitor. Similar experiments were carried out in primary cultures of SkMc cells to determine whether prenylation of cPLA₂γ plays a functional role (Fig. 4). SkMc cells overexpressing cPLA₂γ released more arachidonic acid in response to mouse serum than control cells infected with Ad-GFP, and cPLA₂γ-mediated arachidonic acid release was inhibited to near basal levels by BMS-214662. Another inhibitor, Me-BMS-214662, which inhibits mammalian protein farnesyltransferase with a similar potency as BMS-214662 (M. Gelb, unpublished observation), similarly suppressed arachidonic acid release from SkMc cells expressing cPLA₂γ. Arachidonic acid release experiments were also conducted using transiently transfected COS cells overexpressing both the wild type and the cPLA₂γ C538S prenylation mutant. Consistent with the inhibitor experiments, the mutation partially prevented arachidonic acid release by cPLA₂γ in COS cells (Fig. 4B).

Fig. 4.

A: Protein farnesyltransferase inhibitors suppress serum-stimulated arachidonic acid release from skeletal muscle (SkMc) cells overexpressing cPLA₂γ. SkMc cells were infected with Ad-cPLA₂γ and Ad-GFP control virus and incubated with or without 1 μM BMS-214662 or Me-BMS-214662 in medium containing [³H]arachidonic acid. The cells were washed and then stimulated with 10% mouse serum for 3 h. The amount of radioactivity released into the medium is expressed as a percentage of the total incorporated radioactivity. B: Inhibition of farnesylation of cPLA₂γ by mutating the C-terminal CaaX box cysteine to a serine reduces fatty acid release by cPLA₂γ in transiently transfected COS cells. COS cells were transiently transfected with both wild-type and mutant cPLA₂γ using FuGENE transfection reagent. Expression of cPLA₂γ protein was confirmed by Western blotting (data not shown). Data shown are averages ± SEM of three independent experiments.

Effect of prenylation on membrane association of cPLA₂γ

To determine whether the farnesylation of cPLA₂γ is necessary for its constitutive association with membrane, the effects of BMS-214662 and the C538S prenylation mutant on the relative amounts of cPLA₂γ in the cytosol and on the membrane were investigated. IMLF^−/− and SkMc cells were infected with Ad-cPLA₂γ in the presence and absence of 1 μM BMS-214662, and the relative levels of cPLA₂γ in the 100,000 g soluble and particulate fractions were determined by Western blotting. Infection of both cell types with Ad-cPLA₂γ resulted in increased levels of cPLA₂γ that were predominantly associated with the membrane, although a lower level was observed in the soluble fraction, as reported previously in CHO and Sf9 cells overexpressing cPLA₂γ (Fig. 5) (21, 25). Inhibition of prenylation, by treatment of the cells with BMS-214662 (Fig. 5A) or expression of the C538S prenylation mutant (Fig. 5B), had no effect on the distribution of cPLA₂γ, which remained primarily membrane-associated. This is consistent with previous findings showing that mutation of the CaaX sequence and the putative myristoylation site does not affect the membrane binding of cPLA₂γ expressed in CHO cells (21). In contrast, a recent study has suggested that mutation of the CaaX sequence partially decreases the affinity of cPLA₂γ for membrane in HEK293 cells based on enzymatic assays, although the relative amount of cPLA₂γ present in the membrane by Western blot analysis was not determined (40). In our experiments, transfection of C538S mutant cPLA₂γ into HEK293 cells did not result in a change in membrane distribution of the cPLA₂γ according to Western blotting (data not shown). In addition, when membranes (100,000 g pellet) from COS cells expressing cPLA₂γ mutated at the prenylation site (C538S) were treated with 1.0 M NaCl, cPLA₂γ remained associated with the membrane (data not shown).

To determine whether the hydrophobic and basic regions in the C terminus of cPLA₂γ play a role in the membrane association of the enzyme, the C terminus of cPLA₂γ was truncated. The addition of a stop codon replacing a lysine (K490) allowed the expression of an ~55 kDa truncated version of cPLA₂γ. The truncation also removed the CaaX sequence, thus preventing farnesylation of the enzyme, in addition to removing the hydrophobic and basic regions of the enzyme. Overexpression of the truncated cPLA₂γ and separation of the membrane fraction (100,000 g) in HEK293 cells indicated that without these regions, cPLA₂γ remained associated with the membrane (Fig. 5C).

Fatty acylation of cPLA₂γ

IMLF^−/− infected with Ad-cPLA₂γ were incubated in medium containing [³H]palmitate or [³H]myristate for 4 h, followed by immunoprecipitation of cPLA₂γ, separation by SDS-PAGE, and fluorography to visualize tritiated bands. As shown in Fig. 6A, major radioactive bands were seen with an apparent molecular mass of ~60 kDa, the predicted molecular mass of cPLA₂γ, demonstrating that cPLA₂γ is labeled by [³H]fatty acids. Samples from separate experiments were treated identically but not submitted to fluorography, and a gel piece corresponding to the cPLA₂γ region was excised. When the gel slice containing immunoprecipitated cPLA₂γ obtained from [³H]myristate-or [³H]palmitate-labeled cells was washed three times with 50% aqueous methanol, <1% of the total radioactivity in the gel slice was obtained in the combined washes. This shows that all of the radiolabel is protein bound (i.e., that there are no free radiolabeled fatty acids in the gel slice). The washed gel slice was treated with 1.5 M aqueous NaOH at 30°C for 2 h, conditions known to hydrolyze thiolester- or oxyester-linked fatty acid groups from proteins (41). It was found that 35% and 38% of the total radioactivity in the gel slice was obtained in the organic phase of the Dole solvent extract from [³H]myristate- and [³H]palmitate-labeled cells, respectively.

Fig. 6.

cPLA₂γ is acylated with palmitic and oleic acids but it is not myristoylated. A: cPLA₂γ was immunoprecipitated from cell homogenates of IMLF^−/− cells infected with Ad-cPLA₂γ using anti-cPLA₂γ antiserum (lane 1) or preimmune serum (lane 3) and analyzed by Western blotting using anti-cPLA₂γ antiserum. The Western blot of homogenate, before immunoprecipitation, is shown in lane 2. B: cPLA₂γ was immunoprecipitated from [³H]palmitic (P) or [³H]myristic (M) acid-labeled IMLF^−/−, separated by SDS-PAGE, and subjected to fluorography. C: HPLC analysis of fatty acid esters released from cPLA₂γ. Fatty acids released from cPLA₂γ by treatment with aqueous NaOH were converted to their pentafluorobenzyl esters. The dashed line represents the absorbance at 254 nm and reveals the position of the pentafluorobenzyl esters of the nonradiolabeled fatty acid standards added to the reaction mixture: myristic acid (14:0), oleic acid (18:1), and palmitic acid (16:0) (indicated by arrows). The absorbance peak marked with an asterisk is derived from esterification reagents, because it was seen in a blank run containing only reagents. The absorbance peaks eluting earlier than 20 min are attributable to unknown substances, presumably hydrophilic material derived from impurities present in the cPLA₂γ sample. The dotted line indicates the amount of tritium (dpm) in each column fraction. The total dpm of tritium eluting from the column is 85% of that injected.

When the remaining gel slice and residue in the water layer was hydrolyzed in 6 M HCl at 100°C for 4 h, conditions that lead to the hydrolysis of amide-linked fatty acyl groups, 8% and 6% of the total gel slice radioactivity was recovered in the Dole solvent extract from [³H]myristate-and [³H]palmitate-labeled cells, respectively. Fifty-seven percent and 56% of the total gel slice radioactivity remained in the water layer from [³H]myristate- and [³H]palmitate-labeled cells, respectively. These results indicate that the fatty acyl groups attached to cPLA₂γ are ester-linked rather than amide-linked. The radioactivity remaining in the water layer after hydrolysis in 6 M HCl is presumably attributable to radiolabeled amino acids that resulted from metabolic β-oxidation of the radiolabeled fatty acids into acetyl-CoA and then into amino acids in the fibroblasts, but we did not further investigate this water-soluble radio-labeled material.

The material resulting from Dole extraction of the aqueous NaOH-treated sample was treated with pentafluorobenzyl bromide in the presence of the carrier nonradiolabeled fatty acids myristic, palmitic, and oleic acids, and the sample was analyzed by reverse-phase HPLC using ultraviolet absorbance at 254 nm to monitor the carrier fatty acid esters and by scintillation counting to monitor the radiolabel. For [³H]palmitate-labeled cells, two peaks of radioactivity coeluted precisely with the nonradiolabeled oleic acid and palmitic acid ester standards, and no radioactivity eluted with the myristic acid ester standard (Fig. 6B). The HPLC retention times for the myristate and oleate esters are reproducible to within 10 s, a time much shorter than the difference in retention times for the two esters. The radioactivity clearly comigrates with the 18:1 ester without a discernible shoulder at a shorter retention time. These results show that cPLA₂γ contains ester-linked oleoyl and palmitoyl groups. The same HPLC pattern was seen for [³H]myristate-labeled cells (data not shown), indicating that the 14 carbon fatty acid was converted to the 16 and 18 carbon fatty acids in fibroblasts and that cPLA₂γ is not myristoylated. The formation of the cPLA₂γ-linked radiolabeled oleoyl group is presumably the result of elongation of the radiolabeled fatty acyl-CoA to stearoyl-CoA followed by the action of stearoyl-CoA desaturase to form oleoyl-CoAs.

Subcellular localization of cPLA₂γ

The localization of expressed untagged cPLA₂γ in fixed IMLF^−/−, SkMc cells, and COS cells was carried out using a rabbit polyclonal antibody generated against full-length cPLA₂γ. A comparison of cPLA₂γ localization and a variety of organelle markers demonstrated that cPLA₂γ localized primarily to mitochondria. The morphology of mitochondria is heterogeneous in cells, but they often appear as small ovoid structures or branched “threads” (42). Mitochondria often concentrate around the cell nucleus, where they are closely apposed to the endoplasmic reticulum. However, mitochondria can be clearly resolved in the cell periphery, where they are not as closely associated with the endoplasmic reticulum (42).

In IMLF^−/−, the localization of cPLA₂γ was concentrated in the central, thickest region of the cell around the nucleus, where little structural detail could be observed (Fig. 7). However, farther out in the cell extensions, cPLA₂γ was clearly localized to branched, thread-like structures that stained with antibodies to the mitochondrial marker oxidative phosphorylation complex V, sub-unit b. The mitochondrial localization of cPLA₂γ was also clearly apparent when it was expressed in SkMc cells (Fig. 8). cPLA₂γ localized on branched, tubular structures that wrapped around the nucleus and extended toward the periphery of the cell, and these structures costained with antibodies to the mitochondrial marker oxidative phosphorylation complex V. Although overexpressed cPLA₂γ predominately costained with the mitochondrial markers, there was a small amount of cPLA₂γ in these cells that did not colocalize with the endoplasmic reticulum, lysosomes, nucleus, or mitochondria (data not shown). Control experiments with secondary antibody alone (Fig. 7F) as well as each secondary antibody with the opposite primary antibody were negative (Fig. 7G). Additional experiments demonstrated that BMS-214662 had no effect on the localization of cPLA₂γ on the mitochondria in SkMc cells (data not shown).

Fig. 7.

cPLA₂γ localizes to mitochondria in IMLF^−/−. IMLF^−/− were infected with Ad-cPLA₂γ, fixed, and probed with polyclonal anti-serum to anti-cPLA₂γ and monoclonal antibodies to anti-oxidative phosphorylation complex V. Secondary antibodies used were conjugated to Texas Red and AlexaFluor 488, respectively. Immunofluorescence of cPLA₂γ (A, C) and of the mitochondrial marker oxidative phosphorylation complex V (B, D) is shown. An overlay of cPLA₂γ fluorescence (red) and mitochondrial marker fluorescence (green) is shown (E). IMLF^−/− overexpressing cPLA₂γ were probed with Texas Red secondary antibody only (F), and probed with anti-oxidative phosphorylation complex V primary monoclonal antibodies and anti-rabbit secondary Texas Red antibodies as controls (G).

Fig. 8.

cPLA₂γ localizes to mitochondria in SkMc cells. SkMc cells were infected with Ad-cPLA₂γ, fixed, and probed with polyclonal anti-serum to anti-cPLA₂γ and monoclonal antibodies to anti-oxidative phosphorylation complex V. Secondary antibodies used were conjugated to Texas Red and AlexaFluor 488, respectively. Immunofluorescence of cPLA₂γ (A, C, F) and of the mitochondrial marker oxidative phosphorylation complex V (B, D, G) is shown. Overlays of cPLA₂γ fluorescence (red) and mitochondrial marker fluorescence (green) are shown (E, H).

It was suggested recently that N-terminal FLAG-tagged cPLA₂γ localizes to the endoplasmic reticulum when expressed in HEK293 cells, based on a cytoplasmic reticular staining pattern, although this was not confirmed with organelle markers (40). In HEK293 cells, FLAG-tagged cPLA₂γ was found enriched in the perinuclear region, but no signal was seen on the nuclear envelope, which is an extension of the endoplasmic reticulum (40). In another study, it was reported that cPLA₂γ with an N-terminal GFP tag localized at the Golgi, endoplasmic reticulum, and nuclear envelope when expressed in CHO cells but did not colocalize with a lysosomal or mitochondrial marker (26).

The reports suggesting that N-terminal-tagged cPLA₂γ primarily localizes to the endoplasmic reticulum and that mutation of the prenylation site affects the localization of FLAG-tagged cPLA₂γ are distinct from our results with untagged cPLA₂γ (26, 40). Although it is possible that the localization is cell type-dependent, it is also possible that expression of cPLA₂γ with an N-terminal tag has an effect on its localization and conformation on the membrane. We have found that N-terminal GFP and 6× histidine (His) tags negatively affect the enzymatic activity of cPLA₂γ, resulting in 80% and 60% less activity than in the wild-type enzyme, respectively (25). The N-terminal tags also suppressed the amount of arachidonic acid released by cPLA₂γ when expressed in Sf9 cells (Fig. 9). We reported previously that Sf9 cells can be used as a model to study the function of cPLA₂α, which is activated in Sf9 cells by A23187, resulting in the release of arachidonic acid (19, 43). As shown in Fig. 9, A23187 also induced a large increase in arachidonic acid release from Sf9 cells expressing cPLA₂γ compared with vector control cells. In contrast to cells expressing untagged cPLA₂γ, A23187-induced arachidonic acid release from Sf9 cells expressing GFP- or His-tagged cPLA₂γ was dramatically reduced to near basal levels, suggesting that the N terminus of the enzyme is critical for the function of cPLA₂γ. The N-terminal-tagged cPLA₂γ enzymes were expressed at similar levels as the wild-type enzyme in Sf 9 cells and were constitutively bound to membrane (25). It is important to note that the ability of A23187 to activate cPLA₂γ is cell type-specific and is also observed when cPLA₂γ is overexpressed in HEK293 cells (40), but A23187 does not stimulate arachidonic acid release from IMLF^−/− or SkMc cells overexpressing cPLA₂γ. The ability of A23187 to activate cPLA₂γ is unexpected, because cPLA₂γ is calcium-independent; however, an increase in intracellular calcium may trigger unique signals in some cells that play a role in regulating cPLA₂γ. There is a pronounced lag phase preceding cPLA₂γ-mediated arachidonic acid release in Sf 9 cells treated with A23187, consistent with an indirect mechanism. In contrast, arachidonic acid release from Sf9 cells expressing cPLA₂α occurs rapidly in response to A23187, as reported previously, as a result of the direct effect of Ca²⁺ binding to the C2 domain of cPLA₂α and inducing translocation to the membrane (19, 43). We have also observed similar differences in the time course of serum-induced arachidonic acid release from IMLF^−/− expressing cPLA₂γ and cPLA₂α (25).

Fig. 9.

Effect of N-terminal tags on A23187-stimulated [³H]arachidonic acid release from Spodoptera frugiperda (Sf9) cells expressing cPLA₂γ. [³H]arachidonic acid-labeled Sf9 cells expressing cPLA₂γ (circles), 6× histidine-tagged cPLA₂γ (diamonds), GFP-cPLA₂γ (triangles), or empty vector (squares) were treated with either vehicle (DMSO; open symbols) or 2 μg/ml A23187 (closed symbols) for the times shown. The amount of [³H]arachidonic acid released into the medium is expressed as a percentage of the total incorporated radioactivity. Data shown are averages ± SD of triplicate samples of one experiment that is representative of three independent experiments.

DISCUSSION

The results of this study demonstrate that cPLA₂γ is farnesylated in mammalian cells and document that prenylation plays a role in cPLA₂γ-induced fatty acid hydrolysis. The C-terminal residue of cPLA₂γ is alanine, and it may be noted that yeast α-factor also has a CaaX box with an X residue of alanine and is farnesylated. It is currently thought that the most preferred substrates for protein geranylgeranyltransferase type I have a CaaX box in which X is leucine or phenylalanine (28). Our studies with cPLA₂γ show that not all proteins with a CCXX sequence are doubly geranylgeranylated. Rab proteins have been shown to contain sequences in addition to the double cysteine motif that are recognized by the Rab escort protein that delivers the Rab protein to protein geranylgeranyltransferase type II (44).

The mechanism involved in regulating cPLA₂γ function by farnesylation is not known, but our results suggest that it alone is not responsible for membrane binding. For membrane targeting of other farnesylated proteins such as Ras, the farnesylated C terminus is not sufficient but requires either a polybasic domain or protein fatty acylation (45). In addition to membrane anchoring, prenylation can play a role in the heterodimeric protein interaction that is proposed to involve a two-site recognition (46). cPLA₂γ also has a cluster of basic residues adjacent to a hydrophobic region in the C terminus that may be significant in its association with the membrane (30). Because cPLA₂γ mutated at the prenylation site does not separate to the soluble fraction in the presence of 1.0 M NaCl, electrostatic interactions and prenylation alone are unlikely to determine the membrane binding of cPLA₂γ. To investigate the role that the hydrophobic and basic regions in the cPLA₂γ C terminus may play in membrane binding more thoroughly, a truncated cPLA₂γ, missing these regions, was created. However, our results indicate no change in membrane association of cPLA₂γ attributable to this mutation, suggesting that these regions do not play a predominant role in membrane association. In addition to experiments to determine the possible mechanism of farnesylation, fatty acid labeling experiments were conducted using cells expressing wild-type cPLA₂γ and the C538S mutant. Farnesylation could be required for the fatty acylation of cPLA₂γ. However, our data indicate that this is not the case for cPLA₂γ, because the fatty acid labeling is not significantly changed when the prenylation site is mutated (data not shown).

cPLA₂γ contains a putative N-myristoylation site; however, our results demonstrate that cPLA₂γ is acylated with palmitic and oleic acids, but we saw no evidence that myristoylation of the enzyme occurs. Consistent with this observation, we found that cPLA₂γ mutated at both the N-terminal glycine (to block possible myristoylation) and the prenylation site remains bound to the membrane (data not shown). Although cPLA₂γ is radiolabeled with [³H]myristic acid, it is first metabolized into [³H]palmitic and [³H]oleic acids, which subsequently associate with cPLA₂γ. Thus, although our data show that N-terminal tags on cPLA₂γ negatively affect function, this is not attributable to blocking myristoylation but may have an affect on enzyme conformation, acylation, or membrane trafficking. Although these results indicate that cPLA₂γ is fatty acylated, the locations of the oleic and palmitic acyl groups on the protein chain remain to be determined. It is possible that both types of fatty acyl groups could be linked to the same residue or that the enzyme could be acylated on multiple residues. Investigation of mutations of the individual cysteines is under way. However, the more rare palmitoylation of serine or threonine residues via oxyester bonds can also occur and must be considered.

The observation that cPLA₂γ primarily associates with mitochondria suggests that it may play a role in the function of this organelle. Although cPLA₂γ message is abundant in human SkMc, to date, primarily immortalized cells have been used to study cPLA₂γ (21, 23, 25, 26, 40). By RT-PCR, we have confirmed that cPLA₂γ message is produced in both myotubes and myoblasts of SkMc cells (data not shown). However, despite the presence of message, only very low levels of endogenous cPLA₂γ protein were detectable in SkMc cells, making the direct study of endogenous cPLA₂γ difficult. The regulation of membrane binding and localization of cPLA₂γ to mitochondria are very different from the known properties of cPLA₂α, which has not been shown to associate with mitochondria. Recently, an increase in the group IVC cPLA₂ in apoptotic macrophage cells was observed (47). This observation suggests a potential role in the mitochondrial induction of apoptosis for cPLA₂γ. Both calcium-dependent and -independent PLA₂ activities have been found associated with isolated mitochondria, including the group VI calcium-independent PLA₂ and the group IIA sPLA₂ (48–51). The presence of diverse PLA₂ enzymes in mitochondria suggests that there are multiple, independently regulated pathways in this organelle for the hydrolysis of membrane phospholipid.

Abbreviations

cPLA₂

cytosolic phospholipase A₂

GFP

green fluorescent protein

GST

glutathione S-transferase

His

6× histidine tag

IMLF

immortalized mouse lung fibroblasts

PLA₂

phospholipase A₂

Sf9

Spodoptera frugiperda

SH

sulfhydryl

SkMc

skeletal muscle