Effects of protein kinase CK2, extracellular signal-regulated kinase 2, and protein phosphatase 2A on a phosphatidic acid-preferring phospholipase A1.

A soluble, phosphatidic acid-preferring phospholipase A1, expressed in mature bovine testes but not in newborn calf testes, may contribute to the formation or function of sperm. Here we incubated a recombinant preparation of the phospholipase in vitro with several enzymes including protein kinase CK2 (CK2), extracellular signal-regulated kinase 2 (ERK2), and protein phosphatase 2A (PP2A) to identify effects that might be of regulatory importance in vivo. Major findings were that 1) CK2 phosphorylated the phospholipase on serines 93, 105, and 716; 2) ERK2 phosphorylated the enzyme on serine 730; 3) there was cross-antagonism between the reactions that phosphorylated serines 716 and 730; 4) PP2A selectively hydrolyzed phosphate groups that were esterified to serines 716 and 730; 5) CK2alpha formed a stable, MgATP/MgGTP-dependent complex with the phospholipase by a novel mechanism; and 6) the complex showed reduced phospholipase activity and resembled a complex identified in homogenates of macaque testis. These results provide the first available information about the effects of reactions of phosphorylation and dephosphorylation on the behavior of the phospholipase, shed light on properties of CK2alpha that may be required for the formation of complexes with its substrates, and raise the possibility that a complex containing CK2alpha and the phospholipase may play a special biological role in the testis.

Mammalian tissues contain a soluble phospholipase A1 that can catalyze the preferential hydrolysis of PA 1 in assays using mixed micelles (1) or unilamellar vesicles (2). The enzyme has been purified to homogeneity from bovine testes (3) and shown to have a molecular mass of 97.6 kDa, as determined by matrixassisted laser desorption/ionization (4). Its cDNA has been cloned and sequenced and shown to encode an 875-amino acid protein that resembles other phospholipases only in so far as it contains a five-amino acid lipase consensus domain that includes a central serine residue (serine 540) required for catalysis (4). Moreover, analyses of the distribution of the human enzyme's mRNA have provided evidence that this enzyme and one of its splice variants are expressed selectively in human tissues. 2 However, the enzyme's biological role remains to be determined, and little is known about the regulation of its activity inside cells.
The aim of the present investigation was to explore the possibility that protein kinases and phosphatases might affect the behavior of the first identified (bovine) splice variant of the enzyme, which we now call PA-PLA1␣. We expressed an affinity-tagged, recombinant form of this enzyme in Sf9 cells, purified it, and examined the ability of several protein kinases and phosphatases to phosphorylate or dephosphorylate it in vitro. But only CK2 and ERK2 phosphorylated the phospholipase with significant stoichiometry, and only PP2A could catalyze the hydrolysis of the phosphate esters. We used mass spectrometry to identify the amino acids that were phosphorylated or dephosphorylated and used several other approaches, including immunoprecipitation, quantitative densitometry, size exclusion chromatography, and enzyme activity analysis to characterize complexes of the enzyme that were formed in vitro or were identified in homogenates of the macaque testis and cerebral cortex. ERK1 (rat) was from Upstate Biotechnology, Inc. (Lake Placid, NY). PP2A (catalytic subunit), sequencing grade modified trypsin, and AspN were from Promega. Protein standards for size exclusion chromatography were from Bio-Rad. Sephadex G50, ATP␥S, AMP-PNP, and the mixture of protease inhibitors used ("TM Complete Protease Inhibitor Mixture," which contained antipain, bestatin, chymostatin, E-64, leupeptin, pepstatin, phosphoramidon, pefabloc, EDTA, and aprotinin) were from Roche Molecular Biochemicals. Thesit was from ICN Biochemicals. C18 microcolumn packing was from Michrom Bioresources, Inc. High purity acetonitrile for high pressure liquid chromatography was from Burdick and Jackson. Glutathione-Sepharose 4B beads, Hi-Load Superdex 200 HR 10/30 columns, and HiLoad Superdex 200 26/60 columns were from Amersham Pharmacia Biotech. PVDF membranes were from Millipore Corp. All other reagents were from Sigma or J.T. Baker Inc. unless mentioned otherwise.
Antibodies-Antibodies to CK2␣, CK2␣Ј, and CK2␤, which had been prepared by David Litchfield, were gifts from Dr. Dongxia Li. The antibody to ERK2 was from Calbiochem. Polyclonal antibodies against two peptides from PA-PLA1␣, TKRRLREIEERLHGLKASS (corresponding to a putative coiled-coil-forming region, Thr 589 -Ser 607 ) and KHEHDNNVKPSLDPV (corresponding to the C-terminal region, Lys 861 -Val 875 ) were prepared by Research Genetics Inc. and subsequently affinity-purified on peptide columns, as described (4). Horseradish peroxidase-coupled anti-rabbit IgG antibody was from Amersham Pharmacia Biotech.
Expression and Purification of PA-PLA1␣ from Sf9 Cells-The open reading frame of PA-PLA1␣, attached at its 5Ј end to a sequence of nucleotides that corresponded to the FLAG peptide (DYKDDDDK) followed by hexahistidine (HHHHHH), was cloned into pFASTBAC-HTc vector (Life Technologies), and recombinant virus was prepared according to the manufacturer's instructions. Sf9 cells growing in TNM-FH medium (Grace's insect medium, Life Technologies) containing 10% fetal calf serum, 100 units/ml penicillin, 100 g/ml streptomycin were infected for 66 h with the recombinant virus at a multiplicity of infection of 5 and a density of 2.5 ϫ 10 6 cells/ml. The cells were then harvested and homogenized in 3 volumes of homogenization buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, protease inhibitors) at 4°C using a Dounce homogenizer. The homogenate was centrifuged for 15 min at 600 ϫ g. The resulting low speed supernatant was centrifuged for 1 h at 235,000 ϫ g. The recombinant enzyme in the final, high speed supernatant was adsorbed on a column of FLAGM2 affinity beads. The column was washed three times with 5 column volumes of homogenization buffer, and the enzyme was eluted from the column by competitive replacement with FLAG peptide according to the manufacturer's instructions. Typically, 1 liter of Sf9 culture medium yielded 2-5 mg of purified protein, which appeared as a single band of about 110 kDa when analyzed by SDS-PAGE and had a specific activity of 300 -360 pmol/min/g of protein when analyzed by the mixed micelle enzyme assay described below.
Dephosphorylation of the Purified Recombinant Enzyme-Before studying the phosphorylation of recombinant PA-PLA1␣ (in most experiments), we removed esterified phosphate groups that Sf9 cells had introduced into it by attaching the enzyme (ϳ1 mg of protein) to FLAGM2 affinity beads and incubating it for 30 min at 30°C with 20,000 units of protein phosphatase in the presence of 0.1 mM Thesit (to stabilize the PA-PLA1␣). After the incubations, the enzyme-containing beads were washed three times with 5 column volumes of homogenization buffer, and the enzyme was eluted as described above.
Measurements of Phosphorylation Reaction Stoichiometry after Incubations Involving Soluble Enzymes-Recombinant PA-PLA1␣ (0.5 g) was incubated for 10 -60 min at 30°C in 25 l of phosphorylation buffer (20 mM HEPES, pH 7.4, 20 mM MgCl 2 , 0.55 mM MgATP, 100 M [␥-32 P]ATP) that contained 500 units of CK2␣ 2 ␤ 2 , CK2␣, or CK2␣Ј or the following amounts of activated MAP kinases: ERK2 (120 units), ERK1 (50 units), c-Jun N-terminal kinase (30 units), p38 S6 kinase (1 g), and p34 cdc2 (0.5 g). The reactions were stopped by the addition of SDS sample buffer (5), the mixtures were boiled for 5 min, and the phosphorylated enzymes were separated by SDS-PAGE and stained with Coomassie Blue R-250. The gels were then dried and examined by autoradiography using Eastman Kodak Co. BIOMAX MS film. The identified bands were excised and soaked in 5 ml of Ecolume (ICN Biochemicals). The radioactivity was measured with a Beckman scintillation counter. The molar ratio of incorporated phosphate/PA-PLA1␣ was calculated on the basis of the combined specific radioactivity of the ATP and the total amount of phospholipase that had been used in the assays. This amount was determined by analysis using SDS-PAGE, Western blotting with the antibody to the coiled-coil-forming region, and quantitative densitometry (see below). Control incubation experiments using ␤ casein or myelin basic protein as a substrate demonstrated that each of the kinases used in the above incubations was active.
Identification of Phosphorylated Sites by ESI-LC-MS/MS Analysis-After phosphorylating PA-PLA1␣ with CK2 or ERK2, we used SDS-PAGE to purify the 32 P-labeled enzyme and then digested aliquots of the purified enzyme separately with trypsin and AspN as described by Shevchenko et al. (6). Briefly, the gel band containing the phosphorylated phospholipase was excised, dehydrated for 10 min in CH 3 CN, and dried in a Speedvac (Savant). Pieces of dried gel were incubated for 45 min at 4°C in 100 l of 50 mM NH 4 HCO 3 that contained trypsin (12.5 g/ml) or AspN (12.5 g/ml). Then the temperature was increased to 37°C, and the incubation was continued overnight. After the incubations, the digested PA-PLA1␣ was extracted from the gel pieces, first with the use of 100 l of 20 mM NH 4 HCO 3 and then with 100 l of 50% CH 3 CN, 5% formic acid, 45% water. The extracted peptides were pooled and dried and then dissolved in solvent A, which contained 5% CH 3 CN, 0.4% acetic acid, 0.005% heptafluorobutyric acid in water. Approximately 200 fmol of the digested protein sample were loaded onto a homemade, 75-m inner diameter microcolumn of C18, which had been prepared as described (7). Capillary LC was performed with the use of Applied Biosystems 149B dual syringe pumps at a flow rate of 100 l/min and a precolumn flow splitting ratio of 50:1, which resulted in a final flow rate through the column of 200 nl/min. After the sample was loaded, the column was washed for 5 min with 100% solvent A. Then the peptides were eluted over a 60-min time period with a linear (0 -80%) gradient of solvent B (80% CH 3 CN, 0.4% acetic acid, 0.005% heptafluorobutyric acid). The eluted peptides were analyzed by on-line ESI-MS/MS using a Finnigan LCQ ion trap mass spectrometer (Finnigan MAT LCQ, San Jose, CA) (8). ESI was performed using a needle voltage set at 1.8 kV. The heated capillary temperature was set at 170°C. The scan range was 400 -1800 m/z. The computer algorithm SEQUEST (9) was used to compare tandem mass spectra directly with amino acid sequence data bases and the PA-PLA1␣ sequence. Peptides that contained phosphorylated serine or threonine residues were identified by searching protein sequences that contained the serine or threonine residues corrected for the presence of phosphate ester groups (m ϩ80). Results obtained from automated sequence data base searching were manually confirmed.
Analysis of Cross-antagonism between CK2␣ 2 ␤ 2 and ERK2-protein phosphatase-pretreated PA-PLA1␣ (0.5 g) was incubated for 30 min at 30°C with CK2␣ 2 ␤ 2 (500 units) in 25 l of phosphorylation buffer that contained unlabeled MgATP (0.55 mM). Then ERK2 (120 units) and [ 32 P]ATP (100 M) were added in 10 l of phosphorylation buffer and the incubation was continued for an additional 60 min. Alternatively, the phospholipase was incubated for 30 min with ERK2 plus unlabeled ATP. Then CK2␣ 2 ␤ 2 and radioactive ATP were added, and the incubation was continued for an additional 60 min. In either case, the reaction was stopped by the addition of SDS sample buffer, the mixture was boiled for 5 min, radioactive PA-PLA1␣ was isolated by SDS-PAGE, the gel was stained with Coomassie Blue R-250, the band of radioactive phospholipase was excised and counted in a scintillation counter, and the number of moles of phosphorus that had been incorporated per mole of phospholipase in the incubations with radioactive ATP was calculated on the basis of the combined specific radioactivity of ATP in the incubations.
Dephosphorylation of PA-PLA1␣ by PP2A-PA-PLA1␣ that had been phosphorylated by either CK2 or ERK2, as mentioned above, was brought up to a volume of 100 l by the addition of 50 mM Tris-HCl, pH 8.5, 20 mM MgCl 2 , 1 mM DTT, 0.01% ␤-mercaptoethanol, 0.1 mg/ml BSA; passed through a column of Sephadex G50 to remove free nucle-otides; and incubated for 30 min at 30°C with PP2A. The reaction mixture was then boiled in SDS sample buffer, PA-PLA1␣ from the reaction mixture was purified by SDS-PAGE, and the amount of radioactive phosphate that remained associated with the phospholipase was calculated as described above.
Binding of PA-PLA1␣ to CK2␣ 2 ␤ 2 or Its Subunits-Recombinant, GST-tagged preparations of CK2 ␣ 2 ␤ 2 (4 g), CK2 ␣Ј 2 ␤ 2 (4 g), CK2␣ (2 g), CK2␣Ј(2 g), or CK2␤ (2 g) were bound separately to beads of glutathione-Sepharose 4B and then incubated for 30 min at 30°C with dephosphorylated, recombinant PA-PLA1␣ (5 g) in 500 l of phosphorylation buffer. After the incubations, the beads were washed three times with 1-ml portions of 50 mM Tris-HCl, pH 7.5, 150 mM NaCl and then extracted with SDS sample buffer. The PA-PLA1␣ in the extracts was purified by SDS-PAGE and transferred to PVDF membranes; then the enzyme protein on the membranes was identified by Western blotting using the antibody to the putative coiled-coil-forming region of PA-PLA1␣.
Relation between Phosphorylation of PA-PLA1␣ by CK2␣ and Complex Formation by the Two Enzymesprotein phosphatase-pretreated, recombinant PA-PLA1␣ (0.2 pmol) was incubated for 30 min at 4°C with FLAGM2 affinity beads (40 l of a 50% slurry) in homogenization buffer (500 l) that contained 0.1 M Thesit. After the incubation, aliquots of the beads were washed three times with 1-ml portions of homogenization buffer (to remove unbound PA-PLA1␣) and then incubated separately for 5-120-min periods at 30°C with 5 mol of GST-CK2␣ in phosphorylation buffer (25 l) containing 32 P-labeled ATP (for phosphorylation studies) or unlabeled MgATP (for studies of complex formation). After each incubation, the beads were washed three times with 1 ml of homogenization buffer, the reactions were stopped by the addition of SDS sample buffer, the mixtures were boiled for 5 min, and the enzymes they contained were purified by SDS-PAGE and stained with Coomassie Blue R-250. After this, phosphate incorporation into the enzymes was measured as described under "Measurements of Phosphorylation Reaction Stoichiometry after Incubations Involving Soluble Enzymes," or the time course and stoichiometry of complex formation by PA-PLA1␣ and CK2␣ was determined by transferring the enzymes to PVDF membranes and probing them with the antibody to CK2␣ or the coiled-coil-forming region of PA-PLA1␣ followed by the horseradish peroxidase-coupled antibody to rabbit IgG. The response of each enzyme was visualized by enhanced chemiluminescence (2), quantitated with the use of a Bio-Rad model GS-700 imaging densitometer (2), and converted into a molar concentration by comparison with signals from standards containing recombinant GST-CK2␣. Different amounts of sample were analyzed to ensure that the amount of GST-CK2␣ measured would fall within the linear range of the standard curve.
Stability of the Complex between PA-PLA1␣ and CK2␣-PA-PLA1␣ that had been immobilized on FLAGM2 beads was incubated for 30 min at 30°C with CK2␣ and MgATP in phosphorylation buffer. Then the beads were washed with homogenization buffer to remove unbound CK2␣ and either extracted directly with SDS sample buffer or incubated separately for 30 min at 30°C in control buffer (50 mM Tris-HCl, pH 8.5, 20 mM MgCl 2 , 1 mM DTT, 0.01% ␤-mercaptoethanol, 0.1 mg/ml BSA) or in this buffer plus PP2A (5 units), 1% Triton X-100, 150 mM NaCl, or 350 mM KCl. After the incubation, the amounts of CK2␣ that had dissociated from the beads were determined by SDS-PAGE, Western blotting, and quantitative densitometry.
Size Exclusion Chromatography of the Complex between PA-PLA1␣ and CK2␣-Recombinant forms of PA-PLA1␣ (50 g), GST-tagged CK2␣ (50 g), and a complex of PA-PLA1␣ and GST-tagged CK2␣ (which had been prepared by incubating 50 g of the phospholipase with 31 g of the kinase for 30 min at 30°C in phosphorylation buffer that contained 0.55 mM MgATP plus 0.1 mM Thesit) were chromatographed separately at 4°C on a column of HiLoad Superdex 200 HR 10/30 that was connected to a BioLogic HR Chromatography System (Bio-Rad). The column had been preequilibrated with a "cytosolic" buffer that contained 1) 10 mM PIPES, pH 7.2, 150 mM potassium glutamate, 5 mM nitrilotriacetic acid, 0.5 mM EGTA, 2 mM MgATP, 1 mM DTT (a mixture of ingredients reported to support the metabolism of permeabilized mammalian cells (10)) and 2) the following mixture of protease inhibitors: 1 mM benzamidine, 1 mM phenylmethanesulfonyl fluoride, 2 g/ml each of leupeptin, pepstatin, and aprotinin. A similar buffer was used for the size exclusion chromatography at a flow rate of 0.4 ml/min. An aliquot (50 l) of each fraction (0.5 ml) from the column was blotted onto an Immobilon-P PVDF membrane (Millipore) using a Bio-Dot apparatus (Bio-Rad) and probed with either an antibody to the predicted coiled coil-forming region of PA-PLA1␣ as described (2) or an antibody to CK2␣. The chemiluminescent response also was measured, as described (2) Analysis of Complexes Containing PA-PLA1␣ in Homogenates of Macaque Testis or Cerebral Cortex-Testes or brain prefrontal cortical regions were removed from adult male macaques shortly before death and minced by hand. The minced tissues were washed three times with ice-cold cytosolic buffer and then homogenized in 3 volumes of this buffer using a Potter-Elvehjem homogenizer. The homogenate was centrifuged for 10 min at 800 ϫ g, and the low speed supernatant was collected and centrifuged for 1 h at 235,000 ϫ g in a Beckman Ti 45 rotor. The resulting high speed supernatant was flash-frozen in 10-ml aliquots. Subsequently, the aliquots were thawed separately and loaded at a flow rate of 2.6 ml/min onto a column of HighLoad Superdex 200 26/60 that had been preequilibrated with cytosolic buffer as described above for the size exclusion chromatography of recombinant proteins. Fractions were collected at 1.8-ml intervals beginning at 107 ml, aliquots of the fractions were analyzed by Western blotting and quantitative densitometry using antibody to the C-terminal region of PAPLA1␣, and peaks containing the enzyme were pooled and concentrated to a final volume of about 1 ml using a Centricon concentrator (Millipore). Finally, the concentrated peaks containing PA-PLA1␣ from the testes or brains were precleared by treatment for 1 h at 4°C with 2 g of nonspecific rabbit IgG and 40 l of protein A-Sepharose beads (50% slurry). The remaining, unabsorbed material was incubated for 3 h at 4°C with antibody to the C-terminal region of PA-PLA1␣ (2 g) plus protein A-Sepharose beads (40 l). The beads were then washed three times with 1 ml of cytosolic buffer, extracted with SDS sample buffer, boiled for 5 min, and analyzed by Western blotting using antibodies to PA-PLA1␣, CK2␣, or CK2␤.
Analysis of Complex Formation between PA-PLA1␣ and ERK2-Recombinant PA-PLA1␣ (0.5 g) that had been pretreated with protein phosphatase was immobilized on FLAGM2 beads and then incubated for 30 min at 30°C with recombinant ERK2 (2 g) in 500 l of phosphorylation buffer in the presence or absence of 0.55 mM MgATP. After the incubations, the beads were washed three times with 1-ml portions of 50 mM Tris-HCl, pH 7.5, plus 150 mM NaCl and then extracted with SDS sample buffer. The extract was analyzed by SDS-PAGE and transferred to PVDF membranes. Then an antibody to ERK2 was used to probe the membranes.
Other Methods-Proteins were measured by the micro-BCA method (Bio-Rad). Phosphorylation sites were predicted with the use of the protein data base search software program PROSITE (available on the World Wide Web) for protein functional regions and post-translational modifications. The molecular masses of recombinant, epitope-tagged PA-PLA1␣ and GST-CK2␣ were determined with the software program PeptideMass (available on the World Wide Web). Statistical analyses were done with Microsoft Excel. Molecular modeling studies of the predicted coiled-coil-forming region were done with Rasmol.

CK2␣ 2 ␤ 2 , CK2␣, CK2␣Ј, and a Putative Sf9 Cell MAP Kinase
Phosphorylate PA-PLA1␣-The sequence of PA-PLA1␣ contains predicted phosphorylation sites for several protein kinases including CK2␣ 2 ␤ 2 (not shown). To investigate the possibility that CK2␣ 2 ␤ 2 might catalyze the phosphorylation of the PA-PLA1␣ Phosphorylation and Dephosphorylation phospholipase in vitro, we incubated purified recombinant preparations of the two enzymes together for 60 min in the presence of radioactive ATP and then measured the stoichiometry of PA-PLA1␣ phosphorylation (see "Experimental Procedures"). The results of six experiments demonstrated that 2 mol of phosphorus were incorporated per mole of the phospholipase (Table I). To identify the sites that were phosphorylated, we analyzed digests of the phospholipase by ESI-LC-MS/MS (see "Experimental Procedures"). Unexpectedly, the analysis identified four phosphorylated peptides, not two, as would have been predicted from the measurements of phosphorylation stoichiometry that were made after the in vitro incubations. Moreover, each peptide had a molecular mass that exceeded the value predicted from cDNA sequence analysis by m ϩ80 and therefore contained a single esterified phosphate group.
Three of the peptides contained serine residues that preceded nearby glutamates and were predicted CK2␣ 2 ␤ 2 phosphorylation sites. 1) Peptide 91 D . . . R 102 was phosphorylated on serine 93 (Fig. 1A). 2) Peptide 103 Y . . . R 143 , which contained eight serines, appeared to be phosphorylated only on serine 105 (Fig. 1B). We identified this phosphorylation site by exclusion on the basis of the combined results of the b and y ion series. We detected a mass of ϩ80 in the b ion series b 1-14 ( 103 Y . . . S 116 ) but were unable to assign the phosphate from this series. On the other hand, we detected no mass of ϩ80 in y 1-37 ( 143 R . . . G 107 ), which contained seven out of the eight serines in the sequence, serines 109, 114, 115, 116, 117, 128, and 130. 3) Peptide 711 D . . . R 735 was phosphorylated on serine 716 because there was a loss of mass of 80 between b 5 and b 7 (Ile 715 and Glu 717 ; Fig. 1C).
The fourth phosphorylated peptide, 786 D . . . L 812 , contained an esterified phosphate group on serine 793, which was not a predicted CK2␣ 2 ␤ 2 phosphorylation site (Fig. 1D). Instead, proline residues were present nearby, consistent with the possibility that serine 793 might have been phosphorylated by a proline-directed kinase in Sf9 cells. Note that others have shown that Sf9 cells contain a MAP kinase that can phosphorylate a recombinant form of the arachidonoyl-specific phospholipase A2 (11) but also that the phosphorylated peptide 786 D . . . L 812 , which we detected, comprised only a minority of the total 786 D . . . L 812 peptide identified in the digests of recombinant PA-PLA1␣.
The results of other experiments using a preparation of the phospholipase that had been treated with protein phosphatase before being incubated with radioactive ATP plus either CK2␣ 2 ␤ 2 , CK2␣, or CK2␣Ј showed that each of the three preparations of CK2 could catalyze the incorporation of a maximum of 3 mol of phosphorus into the phospholipase (Table I; also see Fig. 5A). Therefore, it appeared that all three forms of CK2 could phosphorylate the phosphatase-pretreated PA-PLA1␣ on serine 93, serine 105, and serine 716.
The fact that incubation with CK2 in vitro caused two phos-phate groups to be incorporated into the untreated phospholipase but three to be incorporated into the treated phospholipase suggested that Sf9 cells might have partially phosphorylated the phospholipase on CK2-dependent sites during its expression. In support of this possibility, the mobility of a preparation of untreated phospholipase that had not been incubated with CK2 in vitro increased when it was incubated with protein phosphatase, as determined by SDS-PAGE (data not shown), and analysis of the resulting product by mass spectrometry provided evidence that it contained no residual esterified phosphate (data not shown). However, we were unable to obtain further proof that Sf9 cells had phosphorylated the enzyme on CK2-dependent sites by examination of a digest of a phospholipase preparation that had been exposed to neither the kinase nor the phosphatase. The only phosphorylated peptide that we could detect was one containing serine 793 (data not shown). The MS analysis was not quantitative, so we may well have missed peptides that were phosphorylated on CK2 sites in low stoichiometry. But conclusive proof that Sf9 cells phosphorylated the phospholipase on CK2 sites remains to be obtained. ERK2 Phosphorylates PA-PLA1␣-In subsequent in vitro incubation experiments, we attempted to identify a MAP kinase that could catalyze the phosphorylation of serine 793, as the putative proline-directed kinase in Sf9 cells did. We incubated phosphatase-pretreated, recombinant PA-PLA1␣ with radioactive ATP plus a constitutively activated form of either ERK1, ERK2, c-Jun N-terminal kinase 1, p38, or p34 cdc2 but found that only ERK2 could phosphorylate the PA-PLA1␣ with significant stoichiometry. It catalyzed the incorporation of 1 mol of phosphate/mol of the phospholipase (Table II), and we identified only one phosphorylated peptide, 711 D . . . R 735 upon examining digests of the ERK2-phosphorylated phospholipase using LC-MS/MS (Fig. 2). Identification of the phosphorylated amino acid in this peptide was difficult at first because of the peptide's fragmentation pattern and its content of four prolines, six serines, and three threonines. But we scanned through a narrow window of m/z 2500 -3000 using alternate MS-MS/MS in a second run and were able to identify serine 730 as the only phosphorylated residue. Thus, ERK2 phosphorylated the enzyme in vitro but on a different site than the putative prolinedirected protein kinase in Sf9 cells did. Therefore, the identity of this Sf9 cell protein kinase remains to be determined.
Selective, Cross-antagonism of Phosphorylation by CK2 and ERK2-The proximity of the CK2 phosphorylation site, serine 716, and the ERK2 phosphorylation site, serine 730, raised the possibility that phosphorylation reactions involving the two sites might influence each other. To examine this possibility, we first incubated a phosphatase-pretreated preparation of recombinant PA-PLA1␣ for 30 min at 30°C with CK2␣ 2 ␤ 2 plus unlabeled ATP under conditions that could cause the incorporation of 3 mol of phosphate into each mole of the phospholipase and then added ERK2 plus radioactive ATP to the mixture and continued the incubation for an additional 60 min. The results revealed that less than 0.1 mol of radioactive phosphate became incorporated into the phospholipase during the incubation with ERK2 (Table II). Moreover, when we incubated recombinant PA-PLA1␣ with ERK2 plus unlabeled ATP in a parallel experiment and then added CK2␣ 2 ␤ 2 plus radioactive ATP and continued the incubation, 2 mol of radioactive phosphate, not 3 mol, were incorporated per mole of the phospholipase. Because incubation experiments with mixtures of the phospholipase plus radioactive ATP plus either ERK2 or CK2␣ 2 ␤ 2 without pretreatment with the other kinase showed significantly higher amounts of incorporated radioactive phosphate (Table II), these results provided evidence that phospho-

PA-PLA1␣ Phosphorylation and Dephosphorylation
rylation of the phospholipase by CK2␣ 2 ␤ 2 inhibited the subsequent phosphorylation of the phospholipase by ERK2 and vice versa. Furthermore, analysis by LC/MS/MS showed that the inhibitory effect of phosphorylation by ERK2 on phosphorylation by CK2␣ 2 ␤ 2 specifically involved the CK2␣ 2 ␤ 2 phosphorylation site, serine 716. Thus, upon phosphorylating the enzyme successively with ERK2 and CK2␣ 2 ␤ 2 , we could identify only three phosphorylated amino acid residues: the CK2␣ 2 ␤ 2 phosphorylation sites, serine 93 and serine 105, and the ERK2 phosphorylation site, serine 730 (scans not shown). PP2A Dephosphorylates PA-PLA1␣ Selectively-To identify a physiologically relevant protein phosphatase that could dephosphorylate PA-PLA1␣, we first incubated the phospholipase in the presence of radioactive ATP and CK2␣ 2 ␤ 2 and then removed the remaining radioactive nucleotides and incubated the phosphorylated phospholipase with calcineurin, protein phosphatase 1, or the catalytic subunit of PP2A (see "Experimental Procedures"). Upon reisolating the PA-PLA1␣ and measuring its content of radioactive phosphorus, we found that only PP2A could dephosphorylate the phospholipase and that it removed about one-third of the total radioactivity (Table III). Furthermore, we obtained a similar result when we incubated the phospholipase with MgATP and CK2␣ 2 ␤ 2 , treated the phosphorylated phospholipase with 350 mM KCl, washed out the KCl, and then incubated the phospholipase with PP2A (data not shown). This control experiment ruled out the possibility that a complex formed between CK2␣ 2 ␤ 2 and PA-PLA1␣ (see below) might have prevented the PP2A from catalyzing the hydrolysis of the remaining esterified phosphate groups on the phospholipase, because complexes between the enzyme and A-C, CK2 phosphorylation sites; D, presumptive proline-directed protein kinase phosphorylation site. Sequences of identified phosphopeptides are shown above the mass spectra; phosphorylation sites are indicated by asterisks; numbers in superscript on the amino and carboxyl termini of each peptide denote the location of the peptide within the PA-PLA1␣ sequence; b n denotes the ion generated by cleavage of the peptide bond after the nth amino acid from the amino terminus; y n denotes the ions generated from the carboxyl terminus; identified b or y ions are shown in boldface letters; and values of m/z (mass/charge) for ions are indicated in the mass spectra.
CK2␣ or CK2␣ 2 ␤ 2 dissociate when they are treated with 350 mM KCl (Table IV and data not shown).
To determine whether PP2A catalyzed the complete removal of phosphate from a single site on the phospholipase or catalyzed the partial removal of phosphates from several sites, we used LC-MS/MS to analyze digests of phospholipase that had been successively phosphorylated by CK2␣ 2 ␤ 2 and then dephosphorylated by PP2A. The results of the analysis demonstrated that serine 716 had lost 80 mass units, whereas serine 93 and 105 were fully phosphorylated ( Fig. 3 and data not shown). Therefore, the action of the phosphatase was site-specific. When we did a comparable set of incubation experiments using PA-PLA1␣ that had been phosphorylated by ERK2, we found that PP2A could hydrolyze almost all of the phosphate that was esterified to the phospholipase, whereas both protein phosphatase 1 and calcineurin were without effect (Table  III and data not shown). Since ERK2 could phosphorylate only one site, serine 730, this site was clearly the one that was dephosphorylated. Therefore, PP2A dephosphorylated both serine 716 and serine 730, the same two serines that showed cross-antagonism of phosphorylation by CK2␣ 2 ␤ 2 and ERK2. Whether this was coincidental or reflected a special structural feature of the PA-PLA1␣ remains to be determined (but see "Discussion").
PA-PLA1␣ Forms Stable Complexes with CK2␣ 2 ␤ 2 , CK2␣, and CK2␣Ј in Vitro-CK2 has been reported to form stable complexes with several of its substrates through interactions that involve either its ␣ subunit or ␤ subunit (12,13). To determine whether CK2␣ 2 ␤ 2 and its subunits form stable complexes with PA-PLA1␣, we incubated protein phosphatase-pretreated PA-PLA1␣ with or without MgATP in the presence of beads of glutathione-Sepharose 4B that contained immobilized GST-tagged CK2␣ 2 ␤ 2 , CK2␣Ј 2 ␤ 2 , CK2␣, CK2␣Ј, or CK2␤ and then washed the beads and measured the amount of phospholipase that bound to them (see "Experimental Procedures"). The results demonstrated that PA-PLA1␣ could bind to GST-tagged CK2␣ 2 ␤ 2 , CK2␣, or CK2␣Ј in the presence of MgATP but not in its absence. Furthermore, they showed that the phospholipase could not bind to GST-tagged CK2␤ or to GST alone in the presence or absence of MgATP (Fig. 4 and data not shown). Therefore, PA-PLA1␣ appears to belong to the group of CK2 substrates that bind directly to CK2␣ or CK2␣Ј. between ERK2 and CK2␣ 2 ␤ 2 Recombinant PA-PLA1␣ that had been pretreated with protein phosphatase was incubated for 60 min with 32 P-labeled ATP plus ERK2 or CK2␣ 2 ␤ 2 (as a control). Alternatively, the phospholipase was first incubated for 30 min with CK2␣ 2 ␤ 2 plus unlabeled ATP and then incubated for an additional 60 min with ERK2 plus 32 P-labeled ATP or incubated for 30 min with ERK2 plus unlabeled ATP and then incubated for 60 min with CK2 plus 32 P-labeled ATP. Following the incubations, radioactive phosphorus (P) incorporated into the phospholipase was determined on the basis of the combined specific radioactivity of MgATP in the incubations (see "Experimental Procedures"). Results represent means Ϯ S.E. of duplicate measurements from two different sets of experiments.

FIG. 2. Identification of phosphorylated site after incubation of phosphatase-pretreated PA-PLA1␣ with ERK2 and MgATP.
The pretreated phospholipase was incubated for 30 min at 30°C with ERK2 and MgATP and then purified, digested, and analyzed by LC/ MS/MS (see "Experimental Procedures"). The sequence of the phosphorylated peptide, the phosphorylation site, the identified b or y ions, and the m/z value are shown as in Fig. 1.

TABLE III
PP2A dephosphorylates PA-PLA1␣ selectively Samples containing protein phosphatase-pretreated recombinant PA-PLA1␣ were incubated for 30 min at 30°C in the presence of radioactive 32 P-labeled ATP plus either CK2␣ 2 ␤ 2 or ERK2 and then treated for 30 min with PP2A at 30°C (see "Experimental Procedures"). Alternatively, control samples of the phospholipase were incubated for 30 min with radioactive 32 P-labeled ATP and CK2␣ 2 ␤ 2 or ERK2 but were not subsequently treated with PP2A. After the incubations, radioactive phosphorus (P) in PA-PLA1␣ was measured as described (see "Experimental Procedures"  In further studies, we focused attention on the complex that contained PA-PLA1␣ and CK2␣. To examine the relation between the phosphorylation of the phospholipase by CK2␣ and the formation of this complex, we incubated the two enzymes together for periods of 5-120 min in the presence of unlabeled MgATP or radioactive ATP and compared the time courses and reaction stoichiometries of these processes (see "Experimental Procedures"). The results of two independent experiments demonstrated that 1) the time courses of PA-PLA1␣ phosphorylation and complex formation were similar, although not completely identical, 2) 3 mol of phosphorus were ultimately incorporated into each 97.6-kDa molecule of PA-PLA1␣, 3) the complex ultimately contained 0.8 mol of CK2␣/mol PA-PLA1␣, and 4) no detectable radioactive phosphate was incorporated into CK2␣ (Fig. 5, A and B). This provided evidence that the phosphorylation of PA-PLA1␣ by CK2␣ promoted the formation of a stable, 1:1 complex between the two enzymes.
We next examined the stability of the complex and found that it was relatively resistant to extraction with 1% Triton X-100 or 150 mM KCl and remained essentially intact after approximately one-third of the phosphate esterified to PA-PLA1␣ was removed by treatment with PP2A (Table IV). This indicated that the stability of the complex did not depend on the PP2A-sensitive phosphate group that was esterified to serine 716 (see Table III and Fig. 3). Importantly, however, the complex dissociated when it was treated with 350 mM KCl (Table IV), which suggested that the stability of the complex depended on electrostatic interactions involving other components of the two enzymes.
Molecular Basis of Complex Formation-To examine the possibility that these electrostatic interactions might have involved the phosphate groups that were esterified to serines 93 and 105, we took advantage of the results of the above studies to prepare phosphorylated, kinase-free forms of the phospholipase that contained esterified phosphate groups on serines 93 and 105 or serines 93, 105, and 716 (see "Experimental Procedures"). Upon incubating these preparations separately with fresh CK2␣, we were surprised to find that even the maximally phosphorylated form of PA-PLA1␣ did not bind the CK2␣ except in the presence of MgATP (Fig. 6A). Furthermore, MgGTP could substitute for the MgATP, but (Mg)ATP␥S, (Mg)AMP-PNP, MgADP, MgUTP, and MgCTP were without effect (Fig. 6, A and B). These results left open the possibility that formation of the complex might have depended on the MgATP-dependent phosphorylation of FIG. 3. PA-PLA1␣ that has been phosphorylated by CK2␣ 2 ␤ 2 can be selectively dephosphorylated by PP2A. PA-PLA1␣ that had been pretreated with protein phosphatase was incubated for 30 min at 30°C with CK2␣ 2 ␤ 2 and MgATP and then isolated by size exclusion chromatography and incubated with PP2A as described under "Experimental Procedures." Afterward, the PA-PLA1␣ was purified by SDS-PAGE, and peptides obtained by in-gel digestion were analyzed by LC/MS/MS (see "Experimental Procedures"). The sequences of the phosphorus-containing peptides, identified b or y ions, and m/z values are shown as in Fig. 1.   FIG. 4. PA-PLA1␣ forms stable complexes with CK2. Samples of GST-linked CK2␣ 2 ␤ 2 , CK2␣, CK2␣Ј, CK2␤, or GST alone, which had been immobilized on glutathione-Sepharose 4B beads, were incubated separately with PA-PLA1␣ plus MgATP for 30 min at 30°C. The beads were washed with homogenization buffer, extracted with SDS sample buffer, and analyzed by SDS-PAGE followed by Western blotting with the antibody to the coiled-coil-forming region of PA-PLA1␣. Similar results were obtained in two additional experiments.

FIG. 5. Relation between the phosphorylation of PA-PLA1␣ by CK2␣ and the formation of a complex between the two enzymes.
PA-PLA1␣ that had been immobilized on FLAGM2 affinity beads was incubated with GST-CK2␣ for periods of 5-120 min in phosphorylation buffer that contained 32 P-labeled ATP or unlabeled MgATP. After each incubation period, the beads were washed with homogenization buffer and extracted with SDS-sample buffer. Then the two enzymes were purified by SDS-PAGE, and the time courses of PA-PLA1␣ phosphorylation (A) and complex formation (B) were determined as described under "Experimental Procedures." Note that no incorporation of radioactive phosphorus into CK2␣ was detected, that the results represent duplicate analyses from two different experiments, and that similar results were obtained when time course experiments were performed with enzymes in solution.
serines 93 and 105 but showed that an additional, highly specific, MgATP/MgGTP-dependent mechanism was involved. Because CK2␣ can accommodate either MgATP or MgGTP in its Mg-trinucleotide-substrate-binding site (14) and could phosphorylate PA-PLA1␣ in the presence of either one (data not shown), it seemed likely that MgATP or MgGTP contributed to the formation of the complex by binding to this site (see "Discussion").
Evidence That the Complex of PA-PLA1␣ and CK2␣ Contains Four Molecules of Each Protein-We used size exclusion chromatography to estimate the molecular masses of soluble preparations of recombinant PA-PLA1␣ and GST-tagged CK2␣ and to compare them with that of a complex of the two enzymes that had been formed in the presence of MgATP (Fig. 7). When the recombinant PA-PLA1␣ was chromatographed by itself, it had an apparent molecular mass of 220 kDa, which provided evidence that it was a dimer. On the other hand, the GST-tagged CK2␣ was a monomer because it had a molecular mass of 62 kDa, as compared with the expected mass of 68 kDa (26-kDa GST plus 42-kDa CK2␣). In view of these results, we expected that a 1:1 complex of the two enzymes that was formed in the presence of MgATP would have an apparent molecular mass of about 340 kDa. However, upon incubating the enzymes together under the appropriate conditions and examining the resulting complex by size exclusion chromatography, we found that the complex had an apparent molecular mass of about 650 kDa, which suggested that it was a heterooctamer formed from four molecules of PA-PLA1␣ and four molecules of GST-tagged CK2␣. These results suggested that at least three types of binding sites contributed to the complex: 1) a binding site on the phospholipase that promotes the formation of an enzyme dimer, 2) a binding site for CK2␣ on each subunit of the dimer, and 3) a binding site that promotes the conversion of a presumptive, intermediate enzyme heterotetramer into a heterooctamer. Further structural work will be required to clarify the molecular basis of these binding phenomena.
The Effects of Phosphorylation and Complex Formation on the Phospholipase Activity of PA-PLA1␣-To examine the effect of phosphorylation by CK2␣ on the catalytic properties of PA-PLA1␣, we incubated the kinase with the protein phosphatase-pretreated phospholipase in the presence of unlabeled MgATP under conditions that could cause complex formation (control experiments using radioactive ATP confirmed that 3 mol of phosphorus were incorporated into the phospholipase) and then used a mixed micelle assay or a unilamellar vesicle assay to measure the activity of the phospholipase (see "Experimental Procedures"). The results of two independent experiments with each type of assay indicated that the incubation caused a 50% loss of phospholipase activity (Table V). These results differed from those of corresponding incubation experiments with PA-PLA1␣ plus ERK2 and MgATP (see "Experimental Procedures"), which provided no evidence for complex formation or phosphorylation-dependent loss of phospholipase activity (data not shown and Table V).
Macaque Testis Contains a Complex of CK2␣ and PA-PLA1, but Macaque Cerebral Cortex Does Not-After characterizing the complex that PA-PLA1␣ formed with CK2␣ in vitro, we investigated the possibility that homogenates of freshly obtained macaque testes might contain a similar complex. We centrifuged the homogenates to prepare a high speed supernatant fraction from each one and then analyzed aliquots of this fraction by size exclusion chromatography (see "Experimental Procedures"). Immunoanalysis of the effluent from the chromatography columns using the C-terminal antibody to PA-PLA1␣ revealed the presence of cross-reactive material in both the void volume peak and a complex, second peak that emerged immediately afterward (Fig. 8A). Note that the molecular mass of the material in each peak was equal to or greater than 670 kDa. To analyze the material in the peaks in more detail, we pooled the fractions that comprised each peak, concentrated the pools, and then used the antibody to FIG. 6. Specificity of the nucleotide requirement for complex formation between PA-PLA1␣ and CK2␣. Aliquots of GST-CK2␣ were incubated separately for 30 min at 30°C in phosphorylation buffer that contained one of several nucleotides plus one of the following preparations of recombinant PA-PLA1␣ that were bound to FLAGM2beads: 1) a preparation that had been dephosphorylated by phosphatase; 2) a similar preparation that had first been dephosphorylated by phosphatase and then phosphorylated for 30 min at 30°C with CK2␣ and subsequently treated with 350 mM KCl; or 3) a similar preparation of successively dephosphorylated and phosphosphorylated PA-PLA1␣ that had been treated with PP2A before being treated with 350 mM KCl. After the incubations, the beads were washed with buffer, and the amounts of CK2␣ that had bound to PA-PLA1␣ were determined (see "Experimental Procedures"). In A, the nucleotides used were MgATP, ATP␥S, or AMP-PNP; in B, the nucleotides were MgATP, MgGTP, MgADP, MgCTP, or MgUTP. Similar results were obtained in two experiments. the C-terminal region of PA-PLA1␣ to precipitate the phospholipase (see "Experimental Procedures"). Analysis of the PA-PLA1 activity in the precipitates using mixed micelle assays demonstrated that each one contained the enzyme (data not shown). Moreover, analysis by SDS-PAGE, silver staining, and Western blotting using antibodies to the predicted coiled coil-forming region or C-terminal region of PA-PLA1␣ showed that each precipitate contained material that had an apparent molecular mass of 115 kDa 3 (data not shown). On the other hand, analysis by SDS-PAGE, silver staining, and Western blotting using the antibodies to CK2␣, CK2␣Ј, or CK2␤ showed that only the precipitate from pool 1 contained CK2␣, whereas neither of the precipitates contained CK2␣Ј, CK2␤, or other proteins that were detectable by silver staining (Fig. 8C and data not shown).
To confirm that the concentrated material from pool 1 contained a complex of PA-PLA1␣ plus CK2␣, we used the antibody to CK2␣ to precipitate the kinase from the concentrate and then analyzed the material in the precipitate by SDS-PAGE and Western blotting using the antibodies to the coiledcoil-forming region and the C-terminal region of PA-PLA1␣. As anticipated, the results demonstrated that PA-PLA1␣-like material that had an apparent molecular mass of 115 kDa was present in the immune precipitate (data not shown).
To measure the relative distribution of PA-PLA1-like material in the two peaks obtained by size exclusion chromatography, we used the antibody to the C-terminal region of PA-PLA1␣ to precipitate enzyme-containing material from each of the concentrated pools and then examined the precipitates by SDS-PAGE and quantitative densitometry using additional aliquots of the antibody. The results of two independent experiments revealed that pool 1 contained about 40% of the total recovered enzyme and that about one-fourth of this was present as a complex with CK2␣ (data not shown).
We used a similar approach to determine whether high speed supernatant fractions from homogenates of the frontal cerebral cortex of macaque brains also contain a complex of PA-PLA1␣ plus CK2␣. Two separate size exclusion chromatography experiments showed that PA-PLA1␣-like material emerged as a single peak of Ͼ670 kDa (Fig. 8B). Moreover, in each case, material precipitated from a concentrate of the corresponding fractions (with the use of the antibody to the C-terminal region of PA-PLA1␣) could be shown to contain both phospholipase activity, as determined by the mixed micelle assay, and a protein that had an apparent molecular mass of 115 kDa, as determined by SDS-PAGE and Western blotting using antibodies to the coiled-coil-forming region or C-terminal region of PA-PLA1␣ (data not shown). Interestingly, however, no CK2␣, CK2␣Ј, or CK2␤ could be detected by Western blotting with the respective antibodies (data not shown). The combined results of these experiments support the conclusions that 1) the testes of macaques contain large complexes of PA-PLA1 including a complex that also contains CK2␣, and 2) the macaque cerebral cortex also contains a large complex of PA-PLA1-like material but no detectable complex that also contains CK2␣. Further work will be required to determine the basis and functional significance of this apparent difference in tissue distribution.

DISCUSSION
The results of this study provide new information not only about the ability of PA-PLA1␣ to be phosphorylated and dephosphorylated in vitro but also about the enzyme's ability to form complexes with CK2␣ 2 ␤ 2 and its subunits, about the mechanism and effects of complex formation, and about the distribution of PA-PLA1␣-containing complexes in macaque tissues. Furthermore, they identify a number of questions that warrant attention. One of these questions relates to the region of the phospholipase that contains serines 716 and 730. The fact that CK2 can phosphorylate serine 716, ERK2 can phosphorylate serine 730, there is cross-antagonism between these phosphorylation reactions, and PP2A can catalyze the selective hydrolysis of phosphate groups that are esterified to the two serines raises the possibility that the native counterparts of these enzymes might coordinately regulate the phosphorylation state of this region in vivo. This possibility is of interest although the phosphorylation of serine 730 by ERK2 had no effect on the activity of PA-PLA1␣ in vitro, as mentioned previously. An alternate possibility that has to be considered is that the phosphorylation state of the two serines might affect some other property of the enzyme, such as the ability to bind to another cell protein. There is precedent for this type of regulation, because the CK2-dependent phosphorylation of serines 26 and 73 in caldesmon has been shown to reduce the ability of this substrate to bind myosin (15), the ERK2-dependent phosphorylation of serine 702 has been reported to reduce the ability of caldesmon to bind actin-tropomyosin (16), and the ERK2-dependent phosphorylation of serine 64 in another protein, PHAS-I, has been shown to reduce the ability of this substrate to bind to initiation factor 4E (17,18).
Another question that warrants attention relates to the acidic, 28-amino acid region of the phospholipase that contains serines 93 and 105. Our results show that CK2 phosphorylates both of these serines (Fig. 1, A and B) and raise the possibility that the phosphorylation reactions might influence the formation of a stable complex between the phospholipase and CK2␣ (Fig. 5, A and B; Table IV). But whether the phosphate groups that become esterified to serines 93 and 105 interact directly with the phospholipase remains to be determined. One way to explore this possibility might be to do incubation experiments with a synthetic, 28-amino acid peptide that corresponds to the region of the phospholipase that contains the two serines. If this peptide turns out to be a good substrate for CK2␣, the possibility that it might form a stable complex with the kinase could be tested by direct experimentation. Why MgATP or MgGTP is required for the formation of stable complexes between CK2␣ and prephosphorylated forms of PA-PLA1␣ also remains to be determined. One possibility is that MgATP or MgGTP might bind to the trinucleotide substrate-binding site on CK2␣ and have an allosteric effect on the kinase that promotes the exposure of one or more regions of the kinase that are required for the formation of a stable complex with the phosphorylated phospholipase. If a phosphorylated form of the above mentioned synthetic peptide forms a MgATP/ MgGTP-dependent complex with the kinase, it might be possible to examine this potential mechanism directly with the use of NMR. Experimentation of this type would appear to be justified because similar ATP/GTP-dependent mechanisms might contribute to the formation of stable complexes between CK2␣ and its other protein substrates.
Our observation that the complex containing CK2␣ and PA-PLA1␣ behaves like a heterooctamer when it is analyzed by size-exclusion chromatography (Fig. 7) raises other questions.
While PA-PLA␣ was being purified from bovine testis, size exclusion experiments provided evidence that it was a homotetramer (3). On the other hand, size exclusion experiments done in the present investigation showed that the dephosphorylated preparation of recombinant PA-PLA␣ that we used was a homodimer (Fig. 7). The phospholipase contains a 28-amino acid, putative coiled-coil-forming region (4) that may influence the enzyme's ability to form homooligomers because model peptides that correspond to this region form amphipathic helices that interact hydrophobically (data not shown). But it is not clear why the enzyme forms homotetramers in some instances and homodimers in others or why the MgATP/MgGTP-dependent formation of a complex between CK2␣ and the phospholipase results in the apparent formation of a heterooctamer. The latter question deserves to be studied further because the presumptive heterooctamer showed reduced phospholipase activity (Table V), and the possibility has to be considered that the enzyme's ability to form large complexes with CK2␣ might FIG. 8. Size exclusion chromatography and immunoprecipitation of native PA-PLA1 from homogenates of macaque testis and brain. A, a 10-ml aliquot of the high speed supernatant fraction of a macaque testis homogenate was chromatographed on a column of Superdex 200 26/60, which had been preequilibrated with cytosolic buffer (see "Experimental Procedures"). The absorbance of the effluent was monitored at 280 nm, and aliquots from alternate fractions of the effluent were analyzed by SDS-PAGE, Western blotting, and quantitative densitometry using the antibody to the C-terminal region of PA-PLA1␣ (see "Experimental Procedures"). The dashed lines indicate the positions of cross-reactive material; the horizontal bars indicate fractions of this material that were subsequently pooled; and the arrows indicate the positions of protein standards, as in Fig. 7. Similar results were obtained in a second, independent experiment. B, a 10-ml aliquot of the high speed supernatant fraction of a macaque brain homogenate was chromatographed on a column of Superdex 200 26/60, and the column effluent was analyzed as described for A. The dashed lines, horizontal bar, and arrows indicate material that reacted with the antibody to the C-terminal region of PA-PLA1␣, pooled fractions of this material, and protein standards, as in A. Similar results were obtained in a second, independent experiment. C, analysis of pooled fractions of material from the column effluents shown in A and B that cross-reacted with the antibody to the Cterminal antibody. Each pooled fraction was concentrated and then precipitated using the C-terminal antibody (see "Experimental Procedures"). The precipitates were extracted separately in SDS sample buffer and then analyzed by SDS-PAGE and Western blotting using either the Cterminal antibody or antibodies to CK2␣, CK2␣Ј, or CK2␤. The figure shows results obtained using the C-terminal antibody or the antibody to CK2␣ and also shows results obtained in parallel control experiments using rabbit IgG. Similar results were obtained in two independent experiments. Results obtained using the antibodies to CK2␣Ј and -␤ were negative in each case and are not shown. be of regulatory importance in vivo (see below).
The large complexes of PA-PLA1 that contain no CK2␣ (Fig.  8, A and B) also warrant further study, though it is not yet clear that they have reduced activity. Some members of the phospholipase A2 superfamily also form soluble complexes (19 -21). Furthermore, it has been proposed that clusters of phospholipase A2 in the cytoplasm of fibroblasts may form a localized, inactive pool from which active monomers can be recruited (21).
Whether the complex of CK2␣ and PA-PLA1 that we detected in homogenates of the macaque testis plays a special role in this organ is unclear. All that is known at present is that the mRNA for PA-PLA1 is expressed selectively in the spermatids of macaque testes and that the enzyme is present in bovine sperm. 4 Therefore, it is reasonable to suppose that the complex between PA-PLA1 and CK2 may play a role in the differentiation of spermatids or in the function of sperm.
The differentiation of spermatids (spermiogenesis) is a fascinating process that involves the regulated translation of mRNA (22), the formation and storage of proteins for later use in assembly of the sperm tail (23), and impressive processes of polarization that are associated with major changes in the character and distribution of cytoplasmic organelles (24). These processes occur over a period of about 2 weeks, while the spermatids are bound to Sertoli cells. Therefore, it might be a challenging task to determine whether or how a complex of CK2␣ and PA-PLA1 might contribute to the specific molecular events that are involved. On the other hand, a study of the role and regulation of PA-PLA1 in sperm might be feasible. The biochemistry and cell biology of sperm have been studied extensively by others, evidence has been obtained that they contain CK2 and ERK2, and protein phosphorylation reactions have been shown to occur during sperm capacitation and the acrosome reaction (25,26).