Cloning of a Phosphatidic Acid-preferring Phospholipase A1 from Bovine Testis*

We report the molecular cloning and expression of a phosphatidic acid-preferring phospholipase A1 from bovine testis. The open reading frame encoded an 875-amino acid protein with a calculated molecular mass of 97,576 daltons and a pI of 5.61. The sequence included a region similar to a lipase consensus sequence containing the putative active site serine and also included a potential, coiled-coil-forming region. Expression of the open reading frame in COS1 cells resulted in a 20–44-fold increase in phosphatidic acid phospholipase A1 activity over that of control cells. Mutation of the putative active site serine (amino acid 540) demonstrated that it was essential for this increase in enzyme activity. Northern blot analysis revealed at least five different messages with the highest overall message levels in mature testis, but detectable message in all tissues examined. Two possible alternately spliced regions in the open reading frame also were identified. Finally, a search of the data base identified six related proteins: a potential counterpart of the phospholipase A1 inCaenorhabditis elegans, two putative lipases in yeast, and three proteins separately encoded by the Drosophila retinal degeneration B gene and its mouse and human homologues.

We report the molecular cloning and expression of a phosphatidic acid-preferring phospholipase A 1 from bovine testis. The open reading frame encoded an 875amino acid protein with a calculated molecular mass of 97,576 daltons and a pI of 5.61. The sequence included a region similar to a lipase consensus sequence containing the putative active site serine and also included a potential, coiled-coil-forming region. Expression of the open reading frame in COS1 cells resulted in a 20 -44fold increase in phosphatidic acid phospholipase A 1 activity over that of control cells. Mutation of the putative active site serine (amino acid 540) demonstrated that it was essential for this increase in enzyme activity. Northern blot analysis revealed at least five different messages with the highest overall message levels in mature testis, but detectable message in all tissues examined. Two possible alternately spliced regions in the open reading frame also were identified. Finally, a search of the data base identified six related proteins: a potential counterpart of the phospholipase A 1 in Caenorhabditis elegans, two putative lipases in yeast, and three proteins separately encoded by the Drosophila retinal degeneration B gene and its mouse and human homologues. Cellular membranes are highly dynamic structures, and this dynamic nature plays a major role in cellular physiology. In response to extracellular or intracellular signals, a number of phospholipases can be activated, including phosphoinositidespecific phospholipase C, phosphatidylcholine-specific phospholipase C, phospholipase D, phosphatidic acid (PA) 1 phos-phohydrolase, phospholipase A 2 , and sphingomyelinase (1,2). These enzymes produce second messengers that function in cellular regulation.
One second messenger that has received increasing attention is PA. This phospholipid can be produced either by the phospholipase D-catalyzed degradation of phosphatidylcholine or by a two-step pathway involving the phospholipase C-catalyzed degradation of phosphatidylinositol 4,5-bisphosphate to diacylglycerol and the subsequent, diacylglycerol kinase-catalyzed phosphorylation of this diacylglycerol (3). PA appears to function in a number of cellular processes, including superoxide production in neutrophils (4,5) and actin polymerization (6). Other roles for PA are suggested by the facts that (a) mammalian tissues contain a large (10 isoforms) family of diacylglycerol kinases as well as several phospholipase D enzymes (7)(8)(9), and (b) phospholipase D activity can be activated by numerous factors, including phosphatidylinositol 4,5-bisphosphate, protein kinase C, p60 src , and the low molecular weight GTP-binding proteins (ADP-ribosylation factor, cdc42, and Rho A) (9). Although much is known about how PA is produced in response to cell stimulation, little is known about how the PA signal functions or is attenuated.
We recently showed that bovine testis contains a phospholipase A 1 (PLA 1 ) that preferentially catalyzes the hydrolysis of PA in mixed micelle assay systems. The activity of this PApreferring PLA 1 (PA-PLA 1 ) was present in high speed supernatant fractions from mature testis and brain, but not from newborn testis, liver, kidney, spleen, heart, or blood (10). This restricted localization implied that the PLA 1 might be taking part in a signaling pathway in brain and testis. We subsequently purified the enzyme to the point where only one major band of 110 kDa was observed on silver-stained SDS-PAGE, found evidence that the enzyme was a homotetramer in solution, and studied its enzymology (11). Like the impure PA-PLA 1 , the purified enzyme catalyzed the preferential hydrolysis of PA. In addition, PA appeared to activate the PA-PLA 1 reaction by promoting the enzyme's binding to micelles.
To learn more about the enzymology and physical role of PA-PLA 1 , we attempted to clone and sequence its cDNA. This paper describes the primary structure of PA-PLA 1 , identifies a serine residue that is critical for its activity, examines the expression pattern of its mRNA in mammalian tissues, explores the possibility that splice variants for the protein exist, and identifies six related protein sequences. With this information, a detailed study of PA-PLA 1 function can be undertaken.

EXPERIMENTAL PROCEDURES
Materials-Oligonucleotides were synthesized by Life Technologies, Inc. or by DNA International, Inc. Lipids (sn-1-alkyl-2-oleoyl phosphatidylcholine, sn-1,2-dioleoyl PA, and oleoyl lysophosphatidic acid) were * This work was supported by the Howard Hughes Medical Institute and by National Institutes of Health Grant RR00166 to the Regional Primate Research Center at the University of Washington. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The  1 The abbreviations used are: PA, phosphatidic acid; CELG, predicted sequence of an unknown protein from C. elegans; COS1 cells, SV40transformed African Green monkey kidney cells; 5Ј-RACE, 5Ј-rapid amplification of cDNA ends; kb, kilobase pair(s); MALDI, matrix-assisted laser desorption ionization; MOPS, 3-(N-morpholino)propanesulfonic acid; ORF, open reading frame; PA-PLA 1 , phosphatidic acid-preferring phospholipase A 1 ; PCR, polymerase chain reaction; PITP, phosphatidylinositol transfer protein; PLA 1 , phospholipase A 1 ; RDGBd, RDGBh, and RDGBm: sequences encoded by the retinal degeneration B gene in Drosophila and the corresponding genes in humans and mice; SCERV, predicted sequence of an unknown protein from S. cerevisiae; SPOMB, predicted sequence of an unknown protein from S. pombe; PAGE, polyacrylamide gel electrophoresis; 3Ј-UTR, 3Ј-untranslated region. purchased from Avanti Polar Lipids. Organic solvents were from J. T. Baker and were American Chemical Society grade or better. All other reagents were from Sigma, except where indicated. Human infant brain library clone R13928, which corresponded to a human expressed sequence tag, was obtained from Genome Systems, Inc.
Generation and Sequencing of Proteolytic Peptides-PA-PLA 1 (35 g), purified 2000-fold from bovine testis (11), was resolved further by 4 -15% gradient SDS-PAGE, then transferred to a polyvinylidene difluoride membrane (Millipore). A major band that had an apparent mass of 110-kDa was identified by Amido Black staining, excised, and digested with either cyanogen bromide or Endo-Lys-C (Promega), as described (12). The resulting peptides were separated by C18 reversed phase high performance liquid chromatography and sequenced by Edman degradation (12).
cDNA Library Screening-Degenerate oligonucleotides were designed from the following peptides: DWHSVDEV (23 bases, 288-fold degeneracy) and NHATHVEF (23 bases, 288-fold degeneracy). Total RNA was prepared from bovine testis using TRIzol Reagent (Life Technologies, Inc.), and poly(A) ϩ mRNA was prepared from total RNA using a Poly(A) Quick kit (Stratagene). cDNA from bovine testis mRNA was prepared by the Superscript II system (Life Technologies, Inc.). Polymerase chain reaction (PCR) using the above oligonucleotides and cDNA was carried out for 35 cycles of 60 s at 94°C, 30 s at 55°C, and 180 s at 72°C using Taq polymerase (Perkin-Elmer). The resulting 303-base fragment was cloned into the pCR II plasmid using the TA cloning system (Invitrogen). The insert was cut out of the purified plasmid with EcoRI and labeled with [ 32 P]dCTP by Random Priming (Boehringer Mannheim). This radiolabeled probe was used to screen a custom-made bovine testis cDNA library, produced with both oligo(dT) and random primers (Stratagene). The resulting 13 clones were sequenced.
5Ј-Rapid Amplification of cDNA Ends (5Ј-RACE)-Two antisense oligonucleotide primers were designed: GCACACACACCCGCTCGAT-GTT and GGTCACATCCACCTCGTAGAGG. Single-stranded cDNA was prepared from bovine testis as above. A stretch of deoxyadenosines was added to the 3Ј end of the cDNA pool using terminal deoxynucleotide transferase (Life Technologies, Inc.). PCR was carried out using this modified cDNA template, a 30-residue oligo(dT) oligonucleotide, one of the above antisense oligonucleotides, and cloned Pfu polymerase (Life Technologies, Inc.) for 35 cycles of 98°C for 30 s, 60°C for 30 s, and 72°C for 240 s. The resulting products were cloned directly into the PCR-Blunt plasmid using the Zero Blunt PCR cloning kit (Invitrogen). Three clones from the first oligonucleotide and six clones from the second ologonucleotide were fully sequenced.
DNA Sequencing-Sequencing was carried out on ABI automated sequencers at the sequencing core facilities of the Department of Biochemistry or the Department of Pharmacology at the University of Washington.
Molecular Mass Determination of Purified PA-PLA 1 by Matrix-assisted Laser Desorption Ionization (MALDI)-PA-PLA 1 (1 g), purified from bovine testis through the Superdex200 step (11), was precipitated with three volumes of acetone for 1 h on ice, followed by centrifugation at 16,000 ϫ g for 30 min at 4°C. The resulting pellet was resuspended in 10 l of water, then analyzed with a Voyager Elite Biospectrometry Research Station (PerSeptive Biosystems), following established procedures for that instrument. Control proteins used were bovine serum albumin and Escherichia coli ␤-galactosidase.
Expression of PA-PLA 1 cDNA in COS1 Cells-Oligonucleotides TT-GGCGCGCACAGCATGAATTACCCGGGCCATGG and GCTCTAGAT-CAGACTGGATCTAAACTGGGTTTCAC were designed on the basis of the region including the putative initiation ATG and the antisense sequence of the putative termination codon, respectively. The initiation oligonucleotide also contained the five nucleotides 5Ј of the ATG, as well as the mdulBssmdnmHII restriction site. The termination codon included the XbaI restriction site 5Ј of the termination TGA in the antisense direction. These two oligonucleotides were used along with singlestranded cDNA from bovine testis to amplify the PA-PLA 1 open reading frame (ORF) using cloned Pfu polymerase under the following conditions: 35 cycles of 98°C for 30 s, 65°C for 30 s, and 72°C for 5 min. Five 50-l samples of the resulting reaction mix, each containing a band of approximately 2700 bases, were precipitated with 0.1 volume of 5 M ammonium acetate, 2.5 volumes of ethanol. Each pellet was resuspended in 50 l of 1 ϫ REact 3 containing 4 units of BssHII and 10 units of XbaI (New England Biolabs), then incubated overnight at 37°C, followed by 2 h at 50°C. The digested PCR reaction was resolved on a 0.8% agarose gel, and the 2700-base band was cut out and purified using the Qiagen gel extraction kit. A 10-g sample of the pBK-CMV plasmid (Stratagene) was digested and gel-purified in the same manner. The plasmid (20 ng) and PCR product (50 ng) were ligated over-night at 16°C using 20 units of T4 DNA ligase (New England Biolabs). A 2-l sample of this ligation mix was transformed into Epicurian Coli XL-1-Blue competent cells (Stratagene) in the absence of isopropyl-1thio-␤-D-galactopyranoside, following the manufacturer's instructions. The transformed bacteria were plated onto agar plates containing Luria-Bertani broth and 50 g/ml kanamycin. Three of the resulting plasmids contained the 2700-base insert. Plasmids were purified by the Qiagen Tip500 method. These plasmids were sequenced on both strands and were found to contain identical sequence.
COS1 cells (ATCC 1650) were plated onto 100-mm polystyrene plates (Corning) at a density of 500,000 cells/plate. After growing in Dulbecco's modified Eagle's medium ϩ 10% fetal calf serum (Life Technologies) for 24 h at 37°C and 5% CO 2 , the cells were transfected with pBK-CMV alone or pBK-CMV containing the 2700-base insert by means of calcium phosphate precipitation (13). The cells were harvested 55 h after transfection in the following manner (carried out at 4°C or on ice). The growth medium was removed from the plate, and the cells were washed in 10 ml of phosphate-buffered saline. Following removal of the phosphate-buffered saline, 0.5 ml of extraction buffer (containing 25 mM MOPS, pH 7.2, 5 mM EDTA, 1 mM EGTA, 1 mM benzamidine, 1 mM phenylmethylsulfonyl fluoride, 5 g/ml leupeptin, 5 g/ml aprotinin, and 5 g/ml pepstatin) was added to the plate. The plate was scraped with a rubber policeman, and the resulting suspension was withdrawn. This procedure was repeated with a second 0.5 ml of extraction buffer, and the two solutions were pooled. The cell suspension was homogenized with the use of a 1-ml, tight-fitting Dounce homogenizer for 20 strokes, and then 10 l of 0.1 M dithiothreitol were added. A 0.5-ml aliquot of the homogenate was centrifuged at 100,000 ϫ g in a TLA 45 rotor (Beckman) for 30 min at 4°C. The resulting supernatant was transferred to an Eppendorf tube, and the pellet was resuspended in 0.5 ml of extraction buffer by 10 strokes of a Dounce homogenizer. After homogenization, 10 l of 0.1 M dithiothreitol were added.
PA-PLA Assay of the COS1 Cell Extracts-PA-PLA 1 activity in COS1 cells was determined using the Triton micelle assay, as described previously (10,11). Briefly, the 100-l assays contained 50 mM MOPS, pH 7.2, 100 mM KCl, 1 mM EGTA, 14.62 mM Triton X-100, 1.6 mM sn-1alkyl-2-oleoyl PA, and 0.08 mM 32 P-labeled sn-1,2-dioleoyl PA. The sn-1-alkyl-2-oleoyl PA and 32 P-labeled sn-1,2-dioleoyl PA were synthesized as described (10,11). Assays were conducted at 25°C and were initiated by the addition of enzyme. After 20 min, the assays were terminated by the addition of 0.9 ml of 1:1 chloroform:methanol containing 1% v/v formic acid, 25 g of oleoyl lysophosphatidic acid, 10 g of egg PA, and 2 mg/ml butylated hydroxytoluene. The resulting single phase was split by the addition of 0.65 ml of water, the upper phase was removed, and the solvent was evaporated from the lower phase in vacuo (Savant). The pellet was dissolved in 50 l of 75:25:2 chloroform:methanol:water, and 25 l were spotted onto a reversed phase C18 thin layer chromatography plate (Whatman). The plate was developed in 98:2 methanol:4 M ammonium formate for 10 min, then exposed to x-ray film (Eastman Kodak Co.). The band migrating with authentic lysophosphatidic acid was scraped and quantitated by scintillation counting in 5 ml of Ecolume (ICN).
Antibody Generation-A peptide, TKRRLREIEERLHGLKASS, corresponding to the region Thr 589 -Ser 607 of the PA-PLA 1 sequence, was synthesized, purified by high performance liquid chromatography, and coupled to keyhole limpet hemocyanin. Polyclonal antiserum to this conjugate was raised in rabbits (Research Genetics Inc.) and purified over an affinity column containing covalently bound peptide (made by Research Genetics Inc.), as described (14).
Site-directed Mutagenesis-QuickChange site-directed mutagenesis kit (Stratagene) was used for point mutation of the putative active site serine 540 or serine 730 in the PA-PLA 1 ORF. Sense and antisense oligonucleotides flanking the mutation site were used as primers. In the mutation involving serine 540, the original sequence TCC (serine) was changed to GCC (alanine); in that involving serine 730, the original sequence was changed from TCG (serine) to GCG (alanine). Native PA-PLA 1 sequence was used as template. Mutants were generated by PCR using Pfu polymerase under the following conditions: 95°C for 1 min, followed by 18 cycles at 95°C for 30 s, 45°C for 1 min, and 68°C for 15 min. After amplification, the reaction mix was digested with restriction enzyme DpnI at 37°C for 1 h to eliminate the parental plasmid. Mutant clones were then transformed in Epicurian Coli XL 1-Blue supercompetent cells (Stratagene). Plasmids were purified by the Qiagen Tip500 method. For each mutation, the mutagenized codon sequence was verified by ABI rhodamine dye sequencing.
Production and Evaluation of PA-PLA 1 Mutants-Wild type and mutant forms of PA-PLA 1 were produced in the COS1 cell expression system as described above. Cells were harvested 48 h following trans-fection, and enzyme activity assays were carried out with the cell homogenates and supernatant and pellet fractions prepared by centrifugation for 30 min at 100,000 ϫ g. The homogenates, supernatants, and pellets were analyzed for expressed protein by immunoblotting (15).
Northern Blot Analysis of PA-PLA 1 Expression-Human poly(A) ϩ RNAs were purchased from CLONTECH. Mature bovine testis and newborn calf testis poly(A) ϩ RNAs were prepared from fresh-frozen tissue as described above. Northern blots (1.2% agarose) were prepared as described (13). A 308-base fragment from base 1242 to 1550 of the cDNA in Fig. 1 was generated by PCR from one of the bovine clones, FIG. 1. cDNA and amino acid sequence of bovine testis PA-PLA 1 . A, the first and second lines indicate the nucleotide and deduced amino acid sequences, respectively. Nucleotides found to be absent in three bovine testis clones are boxed. The region corresponding to the lipase consensus sequence is indicated in white against a black background with the active site serine 540 in bold. The probable coiled-coil-forming region is underlined.
gel-purified (Qiagen), then randomly primed to about 10 9 cpm/g (Boehringer Mannheim). The Northern blots were pre-hybridized for 3 h at 62°C in 10 mM Tris, pH 7.5, 1 M NaCl, 10% dextran sulfate (Pharmacia Biotech Inc.), and 1% SDS. The probe (5 ϫ 10 7 cpm) was mixed with 10 mg/ml sheared salmon sperm DNA (0.5 ml), boiled for 5 min, then put on ice for 5 min. This solution was mixed with 50 ml of prehybridization solution that had been previously equilibrated to 62°C. The blots were transferred to this mixture and incubated at 62°C overnight. They were then washed with 0.1 ϫ SSC (5Prime-3Prime Inc.) plus 1% SDS three times for 20 min each at 25°C, followed by two 20-min washes at 62°C. The blots were then exposed to x-ray film (Kodak).
Other Methods-DNA masses were estimated by absorption of 260 nm light, using the correction factor of 50 ng/l per optical density unit. Homology searches were conducted using BLAST (National Center for Biotechnology Information). Alignments were made with CLUSTALW (16)

RESULTS
Cloning of PA-PLA 1 cDNA from Bovine Testis-Material that had an apparent molecular mass of 110-kDa, prepared from 2000-fold purified bovine testis PA-PLA 1 , was used to obtain peptide sequences (see "Experimental Procedures"). Four peptides were obtained from Endo-Lys-C digestion and two from cyanogen bromide digestion. Two of the peptide sequences were used to design oligonucleotides, which were used in the amplification by PCR of a 300-base fragment from bovine testis. This fragment, which also contained the DNA sequence for a third peptide, was used to screen a bovine testis cDNA library. Thirteen clones were obtained and partially sequenced from the 5Ј and 3Ј ends. The two clones that contained the most sequence in the 5Ј and 3Ј directions, respectively, were fully sequenced to reveal an ORF of 2293 bases with 2496 bases of 3Ј-untranslated w.ncbi.nlm.nih.gov/BLAST/). Block Maker is also available via the World Wide Web (http://blocks.fhcrc.org/). COILS is also available via the World Wide Web (http://ulrec3.unil.ch/software/COILS_form.html).  (3Ј-UTR). Analysis of the other clones revealed one other with 2496 bases of 3Ј-UTR, three clones with only the first 800 -900 bases of 3Ј-UTR, and four clones with fewer than 100 bases of 3Ј-UTR. All of the above clones contained 20 -45 adenines at their 3Ј ends. The four additional clones did not contain poly(A) sequences, and thus were probably the results of random priming. Three clones contained identical 123-base deletions in the ORF (corresponding to His 343 -Ser 382 , Fig. 1).
The 2293-base ORF, predicted to encode a polypeptide of 85,907 daltons, contained no initiation methionine or upstream stop codons. Therefore, 5Ј-RACE was used to extend the ORF to its 5Ј end. Two different gene-specific RACE primers were used, and clones from both of the RACE products were analyzed. The RACE products from both gene-specific primers added an additional 551 bases of 5Ј sequence. In the reading frame that coded for all six sequenced peptides (located at positions Ser 348 -Ser 364 , Gly 388 -Pro 405 , Ile 441 -Lys 463 , Pro 627 -Pro 639 , Lys 702 -Thr 713 , and Ile 750 -Lys 758 , respectively), the full sequence contained a methionine codon downstream of one stop codon (Fig. 1). The sequence surrounding the methionine codon, ACAGCATGA, fit reasonably well with the Kozak consensus (20). The calculated mass of the 875-residue polypeptide encoded by this ORF was 97,576 daltons. This mass was smaller than the apparent mass of the 110-kDa band observed upon analyzing the purified bovine testis PA-PLA 1 by SDS-PAGE. However, it agreed well with the mass of 97,637 Ϯ 890 daltons (n ϭ 11) obtained by MALDI (data not shown). This suggested that the full-length ORF for PA-PLA 1 had been obtained. In addition, the calculated pI of 5.61 was consistent with the purified enzyme's precipitation properties during acetic acid titration (data not shown).
Upon further inspection of the sequence, two potentially interesting regions were identified. The first region, Ser 538 -His-Ser-Leu-Gly 542 (Fig. 1, black area), resembled the consensus sequence containing the active serine nucleophile present in most lipases: GXSXG (21,22). This suggested that serine 540 in the polypeptide encoded by the ORF might correspond to the conserved serine residue in lipases. However, the Ser 538 -His-Ser-Leu-Gly 542 region of PA-PLA 1 sequence contained an initial serine instead of the highly conserved, initial glycine found upstream of the active serine in all but one of 40 lipases. The second potentially interesting region of the PA-PLA 1 sequence, Arg 580 -Ser 606 (Fig. 1, underlined), appeared to be a coiled coil-forming region. It yielded a probability score of 0.817 when analyzed with COILS version 2.1 using the non-weighted MTIDK matrix and the 28-residue window (data not shown). The significance of this region remains to be determined. However, as mentioned earlier, PA-PLA 1 appears to form a homotetramer in solution (11), and coiled-coil-forming regions have been shown to promote oligomerization including tetramerization of other proteins (23).
Expression of PA-PLA 1 in COS1 Cells-The putative PA-PLA 1 ORF was PCR-amplified from bovine testis cDNA. The PCR product was cloned into the pBK-CMV mammalian expression vector, which was then transfected into COS1 cells. Measurements of PA-PLA 1 activity in homogenates of cells transfected with plasmid containing the putative PA-PLA 1 ORF demonstrated that the homogenates contained 20 -40fold more PA-PLA activity than comparable homogenates of cells transfected with pBK-CMV alone ( Fig. 2A), and over 90% of the PA-PLA 1 activity was recovered in the supernatant after high speed centrifugation (data not shown). Consistent with the measurements of enzyme activity, analysis by SDS-PAGE and Western blotting of homogenates from cells transfected with the putative PA-PLA 1 ORF revealed the presence of a band of 110 kDa apparent molecular mass, which was indis-tinguishable from the band of 110 kDa obtained with purified bovine testis PA-PLA 1 . In contrast, analysis of homogenates from mock-transfected cells revealed no immunochemically detectable band of 110 kDa (Fig. 2B). Taken together, these results provided strong evidence that the expressed ORF coded for PA-PLA 1 . The discrepancy between the value of 110 kDa, determined by SDS-PAGE analysis of the expressed PA-PLA 1 or the purified bovine testis PA-PLA 1 , and the value of 97.6 kDa, calculated on the basis of the ORF-encoded polypeptide sequence or determined by MALDI for the purified bovine testis enzyme, most probably depended on the SDS-PAGE technique, which is well known to yield spurious values for some proteins.
Serine 540 Is Required for Catalysis-Having found that the ORF for PA-PLA 1 could be effectively expressed in COS1 cells, we used a similar approach to examine the possibility that serine 540 might be required for catalysis. It was important to do this because there is some evidence that the upstream glycine residue in the conserved GXSXG sequence in lipases may play a crucial role in catalysis (24), but a serine residue A, whole-cell homogenates were analyzed for PA-PLA activity with 32 P-labeled dioleoyl-PA in a Triton micelle assay. Results shown represent means of duplicate measurements. B, Western blot of PA-PLA 1 expression in each of the transfected cells with a polyclonal antibody to a peptide corresponding to Thr 589 -Ser 607 of the PA-PLA 1 sequence. Note that total cell homogenate and high speed supernatant and pellet fractions were analyzed and that the bulk of the PA-PLA 1 protein identified in the homogenate was recovered in the supernatant fraction. Similar results were found in two different experiments. rather than a glycine residue is present in the corresponding position in PA-PLA 1 , as mentioned earlier. Accordingly, we constructed a S540A point mutant of PA-PLA 1 and transfected it into COS1 cells. In addition, we did parallel experiments with a S730A point mutant and the wild type construct. The results of these experiments revealed that (a) cells that expressed the S540A mutant contained no more detectable PA-PLA 1 activity than the mock-transfected cells, whereas cells that expressed wild type PA-PLA 1 or the S730A construct contained high levels of PA-PLA 1 activity (Fig. 2A); and (b) cells that separately expressed the three constructs produced similar levels of soluble PA-PLA 1 protein, as determined by Western blotting (Fig. 2B). Comparable results have been reported for other lipases containing the conserved serine nucleophile in the GXSXG sequence (e.g. Refs. [25][26][27]. Therefore, the data suggest that serine 540 is the active site nucleophile in PA-PLA 1 . PA-PLA 1 Expression in Human Tissues-Northern blot analysis of poly(A) ϩ RNA from human tissues revealed five PA-PLA 1 message lengths of approximately 2.8, 3.0, 3.6, 5.3, and Ͼ10 kb (Fig. 3, A-C). Overall expression was highest in testis, which was the only tissue in which the 2.8-and 3.0-kb messages were seen, but in which the Ͼ10-kb message was absent. Consistent with this finding, the 13 bovine testis clones contained 3Ј-untranslated regions of various lengths, predicted to correspond to mRNA lengths of the four smaller messages but not to the Ͼ10-kb message (data not shown). Moreover, mature bovine testis contained high PA-PLA 1 message levels, whereas only trace amounts of message (5.3 kb) were present in newborn bovine testis (Fig. 3C) in agreement with our previous findings with regard to PA-PLA 1 activity (10).
All of the other human tissue mRNAs examined contained various amounts of the 3.6-, 5.3-, and Ͼ10-kb bands, with the highest levels in brain, spleen, and lung. When RNA from individual human brain regions was examined, the 3.6-and 5.3-kb messages were expressed relatively evenly (Fig. 3B). However, the Ͼ10-kb message was found only in cerebellum and fetal brain.
Possible PA-PLA 1 Homologues-When the translated PA-PLA 1 ORF was used to search the non-redundant protein sequence data base, only six sequences with appreciable similarity to PA-PLA 1 were identified; none of them corresponded to known lipases. Three of these sequences, CELG (ORF from Caenorhabditis elegans (C. elegans), accession no. 1208846); SPOMB (ORF from Schizosaccharomyces pombe, accession no. 2094857), and SCERV (ORF from Saccharomyces cerevisiae, accession no. 1078029) corresponded to unknown proteins identified during genome sequencing of the respective organisms. The three other sequences, RDGBd (accession no. Y08035), RDGBm (accession no. 2347141), and RDGBh (accession no. 2245317), corresponded to sequences encoded by a Drosophila gene, identified in a mutant screen for retinal degeneration (28), and its mouse and human homologues (29, 30).
The CELG sequence closely resembled the sequence of PA-PLA 1 , though it corresponded to a smaller protein (765 amino acids as compared with 875 amino acids). When the two protein sequences were compared with Align, they showed 29% identity and yielded a global alignment score of 945. When a BLAST search was done using the PA-PLA 1 sequence, the CELG sequence yielded the highest score (P (N) ϭ 1.6eϪ79). When the two sequences were compared with Block Maker ,   FIG. 3. Northern analysis of PA-PLA 1 . A, a poly(A) ϩ mRNA human multiple tissue blot was made and probed with 32 P-labeled cDNA encoding 308 bases of the bovine testis PA-PLA 1 ORF, as described under "Experimental Procedures." B, a poly(A) ϩ mRNA human blot of different brain sections was probed as in A. C, a poly(A) ϩ mRNA bovine blot of mature and newborn testis was probed as in A. Below each blot is the same blot probed with human 32 P-labeled human actin cDNA. eight regions of sequence similarity were identified (Fig. 4, regions 1, 2, 3, 5, and 8 -11). Furthermore, each sequence contained a region of identity, not identified by Block Maker (region 4); a lipase consensus region (region 6); and a probable, coiled-coil forming region (region 7). The probable coiled-coilforming region in CELG was found to have a COILS probability score of 0.94, which compares well with the COILS score of 0.817 found with a similar approach for the corresponding region in the PA-PLA 1 sequence. The striking overall similarity between the two sequences suggested that CELG might correspond to a C. elegans counterpart of PA-PLA 1 .
The two sequences from S. pombe and S. cerevisiae, which, respectively, corresponded to proteins that contained 715 amino acids and 757 amino acids, were related to the sequences of PA-PLA 1 and CELG in several respects. (a) When the sequences of PA-PLA 1 and SPOMB were compared with the aid of the Align program, 20% identity and a global alignment score of 151 were found (data not shown). (b) Analysis of the four sequences by Block Maker revealed 7 regions of similarity (Fig. 5, shaded sections of regions 1-7). (c) Region 4 contained lipase consensus sequences. This suggested that SCERV and SPOMB might correspond to lipases in a fourmember family.
However, the two yeast sequences seemed to be more closely related to each other than they were to the sequences of PA-PLA 1 or CELG. When the two were compared with the aid of the Align program, 27% identity and a global alignment score of 741 were found. When they were compared with Block Maker, large stretches of sequence similarity were found in regions 1, 2, and 6 ( Fig. 5). Most importantly, analysis with the COILS program provided evidence that neither of the putative yeast lipases contained a coiled-coil-forming region. Therefore, it appeared that PA-PLA 1 and CELG might correspond to one type of lipase in the putative, four-member family of lipases, while SPOMB and SCERV might correspond to another.
Unlike the sequences of the putative lipases from C. elegans and yeast, the sequences of RDGBd, RDGBm, and RDGBh, which encode membrane-associated phosphatidylinositol transfer proteins (PITP), showed only a limited similarity to that of PA-PLA 1 . For example, when we used the Block Maker program to compare the sequences of RDGBd, RDGBm, and RDGBh with those of PA-PLA 1 , CELG, SCERV, and SPOMB, only two unambiguous regions of similarity were identified (Fig. 6). Both were located in the probable membrane-association domain of the RDGB proteins, downstream of the phosphatidylinositol transfer protein domain, and corresponded to regions of similarity, identified in PA-PLA 1 and the putative lipases (Fig. 5, regions 5 and 6). The functional significance of these regions remains to be determined. It may be noteworthy that a sequence in RDGBd, Gly 642 -Asp-Ser-Met-Gly, not identified by the Block Maker search, resembled the lipase consensus sequence but contained an aspartic acid residue that could potentially place it in a subgroup of lipases distinct from that of PA-PLA 1 and the three putative lipases from C. elegans and yeast (31). However, the corresponding sequences in RDGBm and RDGBh, e.g. Gly 563 -Asp-Gly-Val-Gly in RDGBh, lacked a corresponding serine residue. Therefore, the functional importance of the putative lipase sequence in RDGBd has to be questioned.

DISCUSSION
In this study, we cloned and sequenced the cDNA for bovine testis PA-PLA 1 . In support of this conclusion, the sequences of six regions in the putative ORF corresponded to those of peptides isolated from digests of the purified bovine testis enzyme, the calculated molecular mass of the ORF (97,576 daltons) agreed well with the molecular mass of purified bovine testis PA-PLA 1 determined by MALDI, expression of the ORF in COS1 cells was accompanied by a 20 -40-fold increase in PA-PLA 1 activity, and serine 540, which is located in a region of the ORF that resembles a conserved sequence in lipases, was shown to be required for PA-PLA 1 activity.
With knowledge of the sequence of PA-PLA 1 in hand, several important questions can now be addressed. For example, one set of questions concerns the structural basis of PA-PLA 1 activity. Are other specific amino acids in addition to serine 540 required for catalysis? What is the basis for the enzyme's substrate specificity? What regions of the PA-PLA 1 sequence influence the enzyme's association with the membrane lipid bilayer? It may be possible to address these questions by taking advantage of some of the other observations made in the course of this study.
Catalysis by several known lipases has been shown to require the presence of a "catalytic triad" of amino acids: the conserved serine nucleophile, a histidine residue, and an aspartic acid residue (see, for example, Ref. 32). The results of the data base search with Block Maker that are shown in Fig. 5 may provide clues concerning the location of potential catalytic triad histidine and aspartic acid residues in PA-PLA 1 and the putative lipases from C. elegans and yeast. Several of the regions of similarity that are shown contained aligned residues of histidine and aspartic acid that could be candidates for FIG. 4. Sequence similarities between PA-PLA 1 and an unidentified protein encoded by a gene from C. elegans. Eight regions of similarity between the sequences of PA-PLA 1 and the C. elegans protein, CELG, identified by Block Maker are shown (regions 1-3, 5, and 8 -11; the number of the first amino acid in a region, indicated by a preceding superscript, identifies the region's location). Region 4, was not identified by Block Maker. Region 6 contains sequences that correspond to a conserved sequence in lipases (shown in black with bold letters). Region 7 (underlined), also not identified by Block Maker, corresponds to predicted, coiled-coil-forming regions. Note that the order of regions in PA-PLA 1 differs somewhat from that in CELG.
catalytic triad function, and it should be possible to examine the functional significance of these residues by experimentation with point mutants of PA-PLA 1 .
As mentioned earlier, experiments with mixed micelles have suggested that PA-PLA 1 may have both a substrate-binding site for PA, involved in catalysis, and multiple binding sites for PA involved in enzyme activation (10,11). Experiments with well defined, unilamellar liposomes, currently under way in  our laboratory, have supported this possibility. 3 However, the amino acids or domains that contribute to the PA-binding sites have yet to be identified. If expression studies of the ORFs from C. elegans and yeast confirm that they encode lipases, it might be possible to identify the relevant PA-binding sites in PA-PLA 1 by carefully comparing the properties and sequences of the four enzymes.
The experiments with unilamellar liposomes that have done so far have also provided evidence that non-substrate lipids influence the ability of PA-PLA 1 to interact with PA. Therefore, it is possible that the enzyme may bind to liposome surfaces in additional, yet-to-be-identified ways. Domains that might be responsible for this binding remain to be identified, but the regions of sequence similarity, regions 5 and 6, shown in Fig. 5, might be good candidates for study because of their similarity to two regions identified in the putative membrane association domains of the RDGB proteins (Fig. 6).
Another set of questions concerns the relation between PA-PLA 1 and other lipases: Does PA-PLA 1 belong to a special family of lipases? What defining structural and functional characteristics do members of this family share? How do members of the family differ from one another? Evidence related to the first of these questions is already beginning to accumulate.
The sequence of PA-PLA 1 definitely differs from the sequences of many other lipases, even though the region that surrounds serine 540 in PA-PLA 1 resembles a conserved region in many of these lipases. For example, PA-PLA 1 lacks further sequence similarity to types I-IV phospholipase A 2 (33, 34), a phosphatidylserine-specific PLA 1 from rat platelets (35), lecithin:cholesterol acyltransferase (36), lysophospholipases (27,37), and triacylglycerol lipases (38,39). However, the sequence of PA-PLA 1 does resemble that of CELG, SCERV, and SPOMB, as shown in Fig. 5; and SCERV and SPOMB appear to correspond to different, but closely related family members. We are currently attempting to prepare and express the cDNA that corresponds to SCERV to examine the catalytic properties of the putative SCERV lipase by direct experimentation.
If PA-PLA 1 does indeed belong to a special family of lipases, splice variants of PA-PLA 1 might well be included. The three bovine testis cDNAs that contained the 123-base deletion shown in Fig. 1 may correspond to such splice variants. The 40 amino acids that would be eliminated by the deletion would normally be located between regions 2 and 3 identified by Block Maker (Fig. 4), but their functional significance is unknown. We have also obtained evidence for the existence of a second type of PA-PLA 1 splice variant. 4 We sequenced a cDNA clone from human infant brain, which corresponded to an expressed sequence tag (accession no. R13928), and found that it contained a 64-base deletion in another region of the ORF. Further work with these variants might suggest functions for the deleted regions.
If splice variants of PA-PLA 1 that have different substrate specificities are present in mammalian cells, an apparent discrepancy between the results of an earlier study (10) and the expression patterns of human PA-PLA 1 mRNA that are shown in Fig. 3 (A and C) might conceivably be explained. We previously found high PA-PLA 1 activity in high speed supernatant fractions from homogenates of mature bovine testis and brain, but little or no PA-PLA 1 activity in corresponding fractions from newborn testis, liver, spleen, heart, kidney, and blood (10). In good agreement with the measurements of enzyme activity in bovine tissues, expression of PA-PLA 1 mRNA was highest by far in human testis, strong in brain, and very low in liver. However, unexpectedly, there also appeared to be expression of human PA-PLA 1 mRNA in the spleen and kidney (Fig.  3A) as well as in the heart (data not shown). One possible explanation of this unexpected result could be that the majority of the message detected in the spleen, kidney, and heart was due to PA-PLA 1 splice variants, and that the enzymes that corresponded to these splice variants had substrate specificities that differed from that of PA-PLA 1 and so escaped detection in our assay system.
The presence of splice variants of PA-PLA 1 might also account, at least in part, for the multiple message sizes for PA-PLA 1 observed in this study. The Ͼ10-kb message is of particular interest because it was absent from testis but strongly present in lung, spleen, cerebellum, and fetal brain (Fig. 3, A  and B). Do these different transcripts correspond to particular splice variants, such as the ones described above? Another possibility, not mutually exclusive with the first, is that differences in message lengths depend on differences in the 3Јuntranslated regions of the messages, which confer different properties to the transcripts, such as variations in RNA halflife, translational control, or cellular localization (40,41).
A different set of questions concerns the possible biological functions of PA-PLA 1 and its close relatives. Answers to many of these questions will have to await the accumulation of further information about the PA-PLA 1 lipase family. However, it may be possible to address questions related to the biological function of PA-PLA 1 itself in the near future. As mentioned earlier, we have been focusing attention on PA-PLA 1 because experiments with mixed micelle systems and unilamellar liposomes have shown that it hydrolyzes and is activated by PA, and because it is present in high levels in the mature testis and brains. This has suggested that it may function in phospholipase D-or diacylglycerol kinase-dependent signaling systems required for spermatogenesis and neuronal interactions. Now that the bovine testis PA-PLA 1 has been sequenced and an antibody to the enzyme has been prepared, it may be possible to obtain further clues concerning the biological role of the enzyme using the tools of molecular biology and immunocytochemistry.
In summary, this paper announces the cloning of the cDNA for a bovine testis PA-PLA 1 . Knowledge of the primary structure of PA-PLA 1 will be of great utility in the characterization of its enzymatic properties and physiologic function. In addition, several different splice variants of the protein appear to exist, suggesting modulations of its function in different cells or sub-cellular regions. Finally, the PA-PLA 1 ORF contains regions that are similar to regions in six other ORFs, suggesting interesting possibilities for further experimentation.