Cloning and Expression of cDNA Encoding Rat Liver 60-kDa Lysophospholipase Containing an Asparaginase-like Region and Ankyrin Repeat*

Mammalian tissues contain small form and large form lysophospholipases. Here we report the cloning, sequence, and expression of cDNA encoding the latter form of lysophospholipase using antibody raised against the enzyme purified from rat liver supernatant (Sugimoto, H., and Yamashita, S. (1994) J. Biol. Chem. 269, 6252–6258). The 2,539-base pair cDNA encoded 564 amino acid residues with a calculated M r of 60,794. The amino-terminal two-thirds of the deduced amino acid sequence significantly resembled Escherichia coliasparaginase I with the putative asparaginase catalytic triad Thr-Asp-Lys and was followed by leucine zipper motif. The carboxyl-terminal region carried ankyrin repeat. When the cDNA was transfected into HEK293 cells, not only lysophospholipase activity but also asparaginase and platelet-activating factor acetylhydrolase activities were expressed. Reverse transcription-polymerase chain reaction revealed that the transcript occurred at high levels in liver and kidney but was hardly detectable in lung and heart from which large form lysophospholipases had been purified, suggesting the presence of multiple forms of large form lysophospholipase in mammalian tissues.

Large form lysophospholipases with molecular masses of 50 -63 kDa were purified from beef liver (60 kDa) (11), rabbit myocardium (63 kDa) (24), and rat liver (60 kDa) (25). Partially purified enzymes were also obtained from lung (50-kDa) (26) and stomach (27). Most lysophospholipases of this type were shown to exhibit transacylase activity as well, producing one molecule of disaturated phosphatidylcholine from two molecules of lyso-PC releasing GPC. In addition, the beef liver enzyme was reported to exhibit platelet-activating factor (PAF) acetylhydrolase activity (28). However, the molecular structure of the large form lysophospholipase was totally unknown. In this paper, we report the cDNA cloning of a large form lysophospholipase from rat liver using antibody against the purified enzyme (25). The cloned cDNA encoded 564 amino acids with a calculated molecular mass of 60,794 and showed no sequence similarity to the small form lysophospholipases of rat liver and macrophage, indicating that the large form lysophospholipase is evolutionarily apart from the small form enzyme. The large form enzyme was predicted to be composed of two domains, the amino-terminal putative catalytic domain significantly resembling Escherichia coli asparaginase I and the carboxyl-terminal ankyrin repeat. When the cDNA was transfected into HEK293 cells, not only lysophospholipase activity but also asparaginase and PAF acetylhydrolase activities were expressed.

EXPERIMENTAL PROCEDURES
Materials-1-[ 14 C]Palmitoyl-GPC (55 mCi/mmol), [␣-32 P]dCTP, and [␣-32 P]UTP were obtained from Amersham Pharmacia Biotech. 1-Hexadecyl-2-acetyl-sn-glycero-3-phospho-[ 14 C]choline (55 mCi/mmol) and * This work was supported in part by grants-in-aid for scientific research from the Ministry of Education, Science and Culture, Japan. 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 nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AB009372.
Enzyme Purification and Partial Amino Acid Sequence-60-kDa lysophospholipase was purified as described previously (25) except that the G-butyl column step was omitted and 10% glycerol was used instead of dimethyl sulfoxide during S-200 column chromatography. Peptides produced by digestion of the purified enzyme with lysyl endopeptidase were separated by an octylsilane column (2 ϫ 150 mm, Capcell Pak C8, Shiseido) using a 120-min linear gradient of acetonitrile concentration from 1 to 45% (v/v) in 0.05% (w/v) trifluoroacetic acid, at a flow rate of 0.15 ml/min and then sequenced as described previously (29) using the Applied Biosystems model 476A gas phase protein sequence analyzer (Foster City, CA).
Preparation of Antiserum-Two rabbits were anesthetized, injected into popliteal lymph nodes with 30 g of purified lysophospholipase mixed with 5 mg of killed Mycobacterium butyricum (Difco), and emulsified in complete Freund's adjuvant (Difco) and then injected intramuscularly with 2 ϫ 10 10 killed cells of Bordetella pertussis. Every 4 and 6 weeks, booster injections were performed subcutaneously in the backs of the animals with 50 g of enzyme emulsified with 5 mg of killed M. butyricum in complete Freund's adjuvant. Antiserum was obtained 2 weeks after the final booster injection.
Construction of Rat Liver cDNA Libraries-Total RNA was extracted from male rat liver (1 g) using Isogen (Nippon Gene, Toyama, Japan) according to the manufacturer's instruction. Poly(A) ϩ RNA was fractionated from total RNA using Oligotex-dT30 Super (Nippon Roche, Tokyo, Japan). Random-primed and oligo(dT)-primed cDNA libraries were prepared using a cDNA synthesis kit (Amersham Pharmacia Biotech), ligated to the gt11 and gt10 arms (Promega, Madison, WI), respectively, and then packaged using Giga Pack II Gold packaging extracts (Stratagene, La Jolla, CA).
Cloning of Lysophospholipase cDNA and Nucleotide Sequence-The rat liver random-primed gt11 cDNA library was screened with antilysophospholipase serum diluted 1:500 as described by Huynh et al. (30) with some modifications. Briefly, plaques from E. coli Y1090r Ϫ infected with the recombinant gt11 phages were imprinted on isopropyl-␤-Dthiogalactopyranoside-impregnated nitrocellulose filters (Amersham Pharmacia Biotech), treated with lysophospholipase antiserum and goat anti-rabbit IgG-alkaline phosphatase conjugate, and then stained with the ProtoBlot Immunoscreening System (Promega). A positive plaque was amplified in Y1090r Ϫ , and its insert was subcloned into pBluescript II SK(Ϫ) (Stratagene). Clone H-1 thus obtained was linearized and used for preparing a radiolabeled RNA probe with the T7 RNA synthesis kit (Nippon gene) and [␣-32 P]UTP. The probe was hybridized with the oligo(dT)-primed gt10 cDNA library on Hybond-N ϩ filters (Amersham Pharmacia Biotech) in a solution containing 48% (v/v) formamide, 4.8 ϫ SSC (0.72 M NaCl in 0.072 M sodium citrate, pH 7.0), 1 ϫ Denhardt's solution (31), 10% (w/v) dextran sulfate, and 0.1% (w/v) SDS at 42°C overnight. The filters were washed twice with 2 ϫ SSC containing 0.1% SDS and twice with 0.2 ϫ SSC containing 0.1% SDS for 15 min at 65°C and then exposed to x-ray films (Fuji Photo Film, Kanagawa, Japan). Phages exhibiting positive signals were saved, and their inserts were subcloned into pBluescript II SK(Ϫ). Nucleotide sequences were determined on both strands using a DNA sequencing kit (Perkin Elmer, Foster City, CA) with the 373A DNA sequencer (Perkin-Elmer).
Expression of Lysophospholipase cDNA in HEK293 Cells-In order to obtain the open reading frame, clone I-2, containing the entire open reading frame with the shortest 5Ј-noncoding sequence, was cleaved with NotI and cloned into pREP9 (Invitrogen, San Diego, CA). The resulting plasmid carrying the insert in the correct orientation (pREX-Ly) was selected, transfected into HEK293 cells by the calcium phosphate precipitation method (32), and cultured in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) containing 5% (v/v) fetal calf serum (Commonwealth Serum Laboratories, Melbourne, Australia). After 44 h, the cells were scraped, harvested by centrifugation at 200 ϫ g for 10 min, and then suspended in 0.2 ml of ice-cold lysis buffer (20 mM Tris-HCl, pH 7.5, 0.1 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride, 1 g/ml pepstatin A, 2 g/ml antipain, 1 g/ml chymostatin, 10 g/ml benzamide, 2 mM ␤-mercaptoethanol, 1 mM dithiothreitol, and 10% (v/v) glycerol). The collected cells were disrupted by sonication for 20 s at an output of 75 watts, centrifuged at 200 ϫ g to remove cell debris, and centrifuged at 10,000 ϫ g for 20 min. The pellet was suspended in 0.2 ml of the lysis buffer and then used for immunoblot analysis and enzyme assays.

RESULTS
Cloning and Sequencing of 60-kDa Lysophospholipase cDNA-Cloning of the complete cDNA for 60-kDa lysophospholipase was carried out as follows. Plaque replicas (1 ϫ 10 6 plaques) from a rat liver random-primed gt11 cDNA library were screened using two separate batches of antibody raised against the purified rat liver enzyme (25). One positive clone recognized by both batches of antibody was saved, and its insert was cloned into pBluescript II SK(Ϫ). The resulting plasmid, H-1 (Fig. 1), was used for the preparation of the radiolabeled RNA probe, which was then used for screening an oligo-dT-primed gt10 library (1 ϫ 10 6 plaques). Out of the 28 positive clones obtained, two 5Ј-extended clones were recloned into pBluescript II SK(Ϫ) to obtain clones I-1 and I-2 (Fig. 1). The two sequences were identical except that I-1 was slightly more 5Ј-extended than I-2. The 2,539-base pair sequence of clone I-1 is shown in Fig. 2. Within the sequence, there was a large open reading frame encoding 564 amino acid residues with a calculated molecular mass of 60,794 Da. We assumed the ATG at positions 80 -82 as the initiation codon for the following reasons. First, the predicted molecular mass was consistent with the size of the enzyme determined on SDSpolyacrylamide gel electrophoresis. Second, the assumed open reading frame expressed lysophospholipase activity in HEK293 cells (see below). Third, the surrounding sequence, CCAGC-CAUGG, conformed to the consensus sequence for the translational initiation in higher eukaryotes (37). The deduced amino acid sequence contained the partial peptide sequences, SHLV-VHSN (amino acid positions 226 -233) and DYSGQTPLH-VAAR (positions 428 -440), determined with the lysyl endopeptidase-digested enzyme. Hydropathy analysis (38) predicted the presence of a hydrophobic stretch of 20 amino acid residues with an average hydropathy of ϩ1.74 at positions 157-176, consistent with the membrane localization of the expressed enzyme in HEK293 cells (see below) and a requirement for Triton X-100 during enzyme purification (25). The 765-base pair 3Ј-noncoding region contained a putative polyadenylation signal, AATAAA (39), at positions 2,502-2,507 and ended with a poly(A) ϩ tail.

60-kDa Lysophospholipase Contains an E. coli Asparaginase I-like Region and Ankyrin Repeat-
The enzyme showed no sequence similarity to the previously cloned lysophospholipidhydrolyzing enzymes, such as pancreatic lysophospholipase/ cholesterol esterase (7,8), Charcot-Leyden crystal protein (23), 24-kDa rat liver lysophospholipase (12), mouse lysophospholipase I (20), cytosolic phospholipase A 2 (1-4), and calciumindependent phospholipase A 2 (5). We searched the protein data bases for similar sequences using the BLAST program (40) and found that the N-terminal region of the enzyme (residues 8 -361) resembled E. coli L-asparaginase I encoded by the ansA gene (41). The C-terminal sequence (residues 405-470) was similar to the ankyrin repeat region of cytoskeleton-associated protein, ankyrin. Thus, the enzyme protein comprised two regions, the asparaginase I-like region and the ankyrin repeat (Fig. 2). Furthermore, there was a leucine zipper motif at the end of the asparaginase I-like region (residues 331-352).
In Fig. 3A, the asparaginase I-like region (residues 8 -361) was aligned to the entire sequence of E. coli asparaginase I with 45.5% amino acid identity. The same region also showed significant sequence similarity to other microbial asparaginases, such as E. coli asparaginase II (42) (24.0% identity as determined by the maximum matching program (43) (44) were also present in the 60-kDa lysophospholipase sequence at residues 16 -21 (TGGTIG) and residues 109 -118 (GFVVIHGTDT). It is known that the active site of microbial asparaginases is structurally highly conserved and includes three conserved amino acids called the "asparaginase triad," Thr-Asp-Lys (50,51). From crystallographic and biochemical studies, the triad was assigned to Thr 89 -Asp 90 -Lys 162 in E. coli asparaginase II (52), Thr 100 -Asp 101 -Lys 173 in Pseudomonas 7A asparaginase (50), and Thr 93 -Asp 94 -Lys 166 in W. succinogenes asparaginase (51). As shown in Fig. 3B, comparison of the equivalent region of 60-kDa lysophospholipase with those of the three asparaginase suggests that the asparaginase triad of the lysophospholipase is Thr 116 -Asp 117 -Lys 188 .
Expression of 60-kDa Lysophospholipase cDNA in HEK293 Cells-Attempts to express 60-kDa lysophospholipase as a ␤-galactosidase fusion in E. coli cells were unsuccessful. Therefore, we decided to use mammalian cells as the expression system. cDNA clone I-2 was cleaved with NotI, and the fragment obtained was cloned into the pREP9 expression vector carrying the Rouse sarcoma virus promoter. The resulting plasmid, pREX-Ly, contained nucleotide positions 75-2,538 encoding the entire open reading frame. HEK293 cells were transfected with the plasmid by calcium phosphate precipitation method (32), cultured for 44 h, and then disrupted by sonication. The membrane fraction thus obtained was used for immunoblot analysis and lysophospholipase assay, since a major portion of the expressed enzyme was localized in the 10,000 ϫ g pellet. As shown in Fig. 4A, the transfectant produced a protein band recognizable by antiserum against the enzyme. The size was indistinguishable from that of the purified enzyme. The vector control did not produce the enzyme protein.
As shown in Fig. 4B, the transformant exhibited about 2.5 times higher lysophospholipase activity than the control. These results indicate that the cloned cDNA indeed encoded 60-kDa lysophospholipase. Furthermore, a low, but significant increase in transacylase activity was seen in the transformant (data not shown), consistent with the data obtained with the purified enzyme (25).
Since the enzyme bears significant sequence similarity to microbial asparaginases, we were interested to examine whether or not the transformant contained asparaginase activity. The homogenate of the transformant was incubated with [G-3 H]asparagine, and the aspartic acid formed was separated by DEAE-cellulose paper chromatography and counted. As shown in Fig. 4C, the asparaginase activity of the transformant was 4.3 times as high as that of the vector control, indicating that the expressed enzyme indeed exhibited asparaginase activity. Aarsman et al. (28) previously showed that the 60-kDa lysophospholipase from bovine liver (lysophospholipase II according to their nomenclature (11)) contained PAF acetylhydrolase activity. Thus, we were interested in examining whether or not the expressed enzyme exhibited PAF acetylhydrolase activity. As shown in Fig. 4D, the transformant exhibited PAF acetylhydrolase activity. The activity was 2.1 times higher than that of the vector control, confirming that the enzyme contains PAF acetylhydrolase activity as an intrinsic activity. It is interesting to note that calcium-independent phospholipase A 2 , which seemingly resembled the present enzyme in carrying the ankyrin repeat, also exhibits PAF acetylhydrolase (6). The increases in the activities of lysophospholipase (2.5-fold), asparaginase (4.3-fold), and PAF acetylhydrolase (2.1-fold) by transformation appeared to be somewhat lower than expected from the immunoblot data (Fig. 4A), but this was partly due to the endogenous activities present in HEK293 cells. A possibility also exists that the enzyme required an as yet unidentified activator protein, deficient in HEK293 cells (see "Discussion").
In agreement with the above results, the purified enzyme showed asparaginase and PAF acetylhydrolase activities with the relative hydrolytic rates of 100:6:1 for lyso-PC, asparagine, and PAF. Although this value somewhat differed from that obtained with the expressed enzyme (100:38:32), the data confirm that the 60-kDa lysophospholipase contains asparaginase and PAF acetylhydrolase activities.
Tissue Distribution of 60-kDa Lysophospholipase mRNA-To determine the size of 60-kDa lysophospholipase mRNA, we isolated poly(A) ϩ RNA from rat liver and performed Northern blot analysis using the 32 P-labeled BalI fragment of clone I-1 (nucleotide positions 116 -573) as the probe. As shown in Fig.  5A, a 2.3-kilobase band hybridized to the probe under the high stringency condition. Thus, the size of the transcript was fairly consistent with that of the cloned cDNA. Although clearly observed in liver, the transcript was not detectable in heart and lung (data not shown), suggesting that the transcript was expressed in a tissue-specific manner. Thus, we examined the tissue distribution of 60-kDa lysophospholipase mRNA by RT-PCR using a pair of specific primers (Fig. 2). As shown in Fig.  5B, 60-kDa lysophospholipase mRNA was abundantly expressed in liver and kidney and at a low level in stomach, but it was hardly seen in spleen, lung, heart, and brain. Thymus showed a weak, slowly moving band, whose entity was unknown. The results obtained here indicate that 60-kDa lysophospholipase mRNA is distributed in limited tissues.

DISCUSSION
This paper reports the initial cloning of large form lysophospholipase in mammalian tissues. The sequence bears little sequence similarity to that of small form lysophospholipase (12,20). Small form lysophospholipase resembles the esterases of Pseudomonas fluorescens (21) and Spirulina platensis (22) and contains the esterase/lipase consensus, Gly-X-Ser-X-Gly (12,20). The serine in the consensus is part of the esterase catalytic triad, Ser-Asp-His, and plays a role of nucleophile in the hydrolytic reaction of the enzyme. In P388 macrophage lysophospholipase, the catalytic triad was assigned to Ser 119 -Asp 174 -His 208 (67). In contrast to the small form lysophospholipase, large form lysophospholipase resembles microbial asparaginases. Indeed, the large form lysophospholipase cloned here catalyzed the hydrolysis of asparagine. The mechanism of the asparaginase reaction is reminiscent of that of serine esterase. However, threonine, rather than serine, has been postulated to play a role of nucleophile (52), and lysine, instead of histidine, is thought to enhance the nucleophilicity of the threonine (51). Crystallographic and biochemical studies identified the asparaginase catalytic triad as Thr 89 -Asp 90 -Lys 162 for E. coli asparaginase II (52) asparaginase (50), and Thr 93 -Asp 94 -Lys 166 for W. succinogenes asparaginase (51). Sequence alignment of 60-kDa lysophospholipase with these bacterial asparaginases suggests that the catalytic triad of the enzyme might be Thr 116 -Asp 117 -Lys 188 , which had well conserved surrounding sequences. It is tempting to speculate that these amino acids play an essential role in the catalytic activity of 60-kDa lysophospholipase and that the Thr 116 is the nucleophile involved in the formation of the acylenzyme intermediate.
From the results of RT-PCR, mRNA encoding 60-kDa lysophospholipase is expressed at high levels in liver and kidney and a low level in stomach but at negligible levels in other tissues. Lysophospholipase-transacylase partially purified from gastric mucosa (27) showed similar properties to the rat liver enzyme. It had a pH optimum at 6.0 with the K m value of 0.25 mM for lyso-PC, but it was only 10% active at pH 8.5. The isoelectric point was determined to be 5.4, while the value for or those conforming to the ankyrin repeat consensus (53) (C). In panel A, the sequence of 60-kDa lysophospholipase was compared with that of E. coli asparaginase I using the MACAW program (70). Lyso, lysophospholipase; AspI, E. coli asparaginase I. In panel B, the sequence of 60-kDa lysophospholipase is aligned to those of the bacterial asparaginases whose catalytic triads had been identified as indicated by the arrowheads. Alignment was made using the multiple alignment program (71). Lyso, lysophospholipase; EcoII, E. coli asparaginase II; Pse, Pseudomonas 7A asparaginase; Wol, W. succinogenes asparaginase. In panel C, each of the two ankyrin repeat units (repeat 1, 405-437; repeat 2, 438 -470) of 60-kDa lysophospholipase was compared with the protein data bases using the BLAST algorithm (39), and the highest scoring sequences were aligned to the ankyrin repeat consensus (53). The star in the cactus sequence denotes dipeptide, Gly-Lys. Lyso, lysophospholipase; CDK6I, human cyclin-dependent kinase 6 inhibitor; CDK4I, human cyclin-dependent kinase 4 inhibitor D; rSMPP, rat smooth muscle protein phosphatase 1M (chain M110); cSMPP, chicken smooth muscle protein phosphatase (large chain, long form); IBRP, human IB-related protein; cactus, D. melanogaster ankyrin repeat acidic protein cactus; notch 3, mouse notch 3 protein; DGK, eye-specific diacylglycerol kinase; ANK3, human long form ankyrin 3; ANK2, human brain ankyrin 2; p53BP, human p53-binding protein 2; ANK1, human erythrocyte ankyrin 1; DAPK, human death-associated protein kinase; res2, S. pombe cell cycle regulator; NRF2, human nuclear respiratory factor 2 (␤2-chain). the liver enzyme was calculated to be 5.39 from the deduced amino acid sequence. Thus, the gastric enzyme described by Lin et al. (27) is probably the same as the present enzyme. de Jong et al. (11) reported the purification of a large form lysophospholipase called lysophospholipase II with a molecular mass of 60 kDa from bovine liver. The bovine enzyme was shown to be localized in the microsomal fraction (68). To examine the relationship of 60-kDa lysophospholipase to the enzyme of de Jong et al. (11), we performed Western blot analysis of rat liver microsomes using the antibody against the 60-kDa enzyme. The antibody recognized a 60-kDa protein in microsomes. Furthermore, we could solubilize lysophospholipasetransacylase activity with Triton X-100 and purify it by a procedure similar to that used for the enzyme in the soluble fraction. 2 Moreover, lysophospholipase II catalyzed the deacetylation of PAF (28), and the 60-kDa enzyme also exhibited the same activity. Thus, lysophospholipase II described by de Jong et al. (11) and the 60-kDa enzyme reported here are very likely the same enzyme. The only difference reported between lysophospholipase II and the 60-kDa enzyme is that the former was reported to be deficient in transacylase activity, but this could be due to the different conditions used for transacylase assay, e.g. pH of the reaction mixture. Lysophospholipase-transacylase was also purified from other sources, such as heart (24) and lung (26), clearly showing that heart and lung are rich sources of large form lysophospholipase. However, as shown above, the transcript encoding hepatic 60-kDa lysophospholipase could not be detected in heart and lung. These results are consistent with the view that a different isoform of large form lysophospholipase is present in heart and lung.
An interesting feature of 60-kDa lysophospholipase is that it contains the ankyrin repeat in the C-terminal region. Various forms of the ankyrin repeat have been found in many proteins involved in tissue differentiation and cell cycle control. However, the motif was recently shown to be also contained in a class of phospholipase, calcium-independent phospholipase A 2 (6). The enzyme activity was shown to be lost when the ankyrin repeat was deleted, although the active site of the enzyme still remained (6). The ankyrin repeat is often regarded as a motif for protein-protein interaction (69), and the ankyrin repeat in calcium-independent phospholipase A 2 was proposed to be involved in enzyme activity through oligomeric complex formation. 60-kDa lysophospholipase contains not only the ankyrin repeat but also leucine zipper motif. Probably, the enzyme exists in a complexed form with some other protein that may regulate the activities of the enzyme. Identification and characterization of such a binding protein would provide a useful information on the regulation and physiological function of 60-kDa lysophospholipase.