Molecular Cloning and Functional Expression of aCaenorhabditis elegans Aminopeptidase Structurally Related to Mammalian Leukotriene A4 Hydrolases*

In a search of the Caenorhabditis elegans DNA data base, an expressed sequence tag of 327 base pairs (termed cm01c7) with strong homology to the human leukotriene A4 (LTA4) hydrolase was found. The use of cm01c7 as a probe, together with conventional hybridization screening and anchored polymerase chain reaction techniques resulted in the cloning of the full-length 2.1 kilobase pair C. elegansLTA4 hydrolase-like homologue, termed aminopeptidase-1 (AP-1). The AP-1 cDNA was expressed transiently as an epitope-tagged recombinant protein in COS-7 mammalian cells, purified using an anti-epitope antibody affinity resin, and tested for LTA4 hydrolase and aminopeptidase activities. Despite the strong homology between the human LTA4 hydrolase andC. elegans AP-1(63% similarity and 45% identity at the amino acid level), reverse-phase high pressure liquid chromatography and radioimmunoassay for LTB4 production revealed the inability of the C. elegans AP-1 to use LTA4 as a substrate. In contrast, the C. elegans AP-1 was an efficient aminopeptidase, as demonstrated by its ability to hydrolyze a variety of amino acid p-nitroanilide derivatives. The aminopeptidase activity of C. elegans AP-1 resembled that of the human LTA4 hydrolase/aminopeptidase enzyme with a preference for arginyl-p-nitroanilide as a substrate. Hydrolysis of the amide bond of arginyl-p-nitroanilide was inhibited by bestatin with an IC50 of 2.6 ± 1.2 μm. The bifunctionality of the mammalian LTA4hydrolase is still poorly understood, as the physiological substrate for its aminopeptidase activity is yet to be discovered. Our results support the idea that the enzyme originally functioned as an aminopeptidase in lower metazoa and then developed LTA4hydrolase activity in more evolved organisms.

In a search of the Caenorhabditis elegans DNA data base, an expressed sequence tag of 327 base pairs (termed cm01c7) with strong homology to the human leukotriene A 4 (LTA 4 ) hydrolase was found. The use of cm01c7 as a probe, together with conventional hybridization screening and anchored polymerase chain reaction techniques resulted in the cloning of the full-length 2.1 kilobase pair C. elegans LTA 4 hydrolase-like homologue, termed aminopeptidase-1 (AP-1). The AP-1 cDNA was expressed transiently as an epitope-tagged recombinant protein in COS-7 mammalian cells, purified using an anti-epitope antibody affinity resin, and tested for LTA 4 hydrolase and aminopeptidase activities. Despite the strong homology between the human LTA 4 hydrolase and C. elegans AP-1(63% similarity and 45% identity at the amino acid level), reverse-phase high pressure liquid chromatography and radioimmunoassay for LTB 4 production revealed the inability of the C. elegans AP-1 to use LTA 4 as a substrate. In contrast, the C. elegans AP-1 was an efficient aminopeptidase, as demonstrated by its ability to hydrolyze a variety of amino acid pnitroanilide derivatives. The aminopeptidase activity of C. elegans AP-1 resembled that of the human LTA 4 hydrolase/aminopeptidase enzyme with a preference for arginyl-p-nitroanilide as a substrate. Hydrolysis of the amide bond of arginyl-p-nitroanilide was inhibited by bestatin with an IC 50 of 2.6 ؎ 1.2 M. The bifunctionality of the mammalian LTA 4 hydrolase is still poorly understood, as the physiological substrate for its aminopeptidase activity is yet to be discovered. Our results support the idea that the enzyme originally functioned as an aminopeptidase in lower metazoa and then developed LTA 4 hydrolase activity in more evolved organisms.
Leukotriene A 4 (LTA 4 ) 1 hydrolase (EC 3.3.2.6) is the rate-limiting enzyme in the lipoxygenase cascade of arachidonic acid metabolism leading to the biosynthesis of the proinflammatory substance leukotriene B 4 (LTB 4 ) from the epoxide intermediate LTA 4 (1,2). At nanomolar concentrations, LTB 4 elicits chemotaxis and adherence of leukocytes, and in higher doses it also triggers degranulation and generation of superoxide anions (3). Due to these biological properties, LTB 4 is regarded as an important chemical mediator in a variety of inflammatory diseases (4). Sequence comparison of LTA 4 hydrolase with other zinc metalloenzymes, e.g. aminopeptidase M and thermolysin, led to the identification of a zinc binding motif in the primary structure of the enzyme (5)(6)(7). Further studies verified that LTA 4 hydrolase contained one catalytic zinc atom coordinated by His 295 , His 299 , and Glu 318 (8). Subsequently, the enzyme was shown to exhibit a previously unknown zinc-dependant peptidase/amidase activity toward synthetic substrates (9, 10) that was specifically stimulated by monovalent anions, e.g. chloride ions (11), and also by albumin (12). Although a physiological peptide substrate for the aminopeptidase activity of the enzyme has not yet been found, LTA 4 hydrolase has been shown to efficiently hydrolyze several arginyl tri-and dipeptides, leading to its identification as an arginine aminopeptidase (13). Both the aminopeptidase and the LTA 4 hydrolase activity of the enzyme are inhibited by the aminopeptidase inhibitor bestatin (10) and the angiotensin converting enzyme inhibitor captopril (14), suggesting that the active sites corresponding to the two activities are overlapping (15). Important questions regarding the dual activity of the mammalian LTA 4 hydrolase/aminopeptidase remain unanswered. For example, does the enzyme demonstrate both LTA 4 hydrolase and aminopeptidase activities in other species? Which function originated first in evolution? What is the significance of this bifunctionality? LTA 4 hydrolase/aminopeptidase is a soluble monomeric protein (M r Ϸ 69,000) (16,17) that has been cloned from human (18), mouse (19), rat (20), and guinea pig (21). Recently a partial sequence from the slime mold Dictyostelium discoideum and a gene from the yeast Saccharomyces cerevisiae (22) have been deposited into the Gen-Bank TM data base as putative LTA 4 hydrolases (accession numbers U27538 and X94547, respectively). Both sequences encode proteins similar in their primary amino acid sequences to the mammalian LTA 4 hydrolase, but neither of them has been expressed or characterized. In addition, an enzyme from the pathogenic yeast Candida albicans with 41% homology to the mammalian LTA 4 hydrolase exhibited mainly aminopeptidase activity, whereas its hydrolase activity converted the ma-* 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) AF068200 and AF068201.

MATERIALS AND METHODS
Cloning of a LTA 4 Hydrolase-like cDNA Homologue from C. elegans-The cm01c7 phage clone from a C. elegans mixed stage hermaphrodite cDNA library (made by Chris Martin) containing the LTA 4 hydrolase-like EST in SHLX2 phage vector (23) was obtained from Dr. R. H. Waterston (24). MC1061 recA Ϫ tet R (used for plating SHLX2) and the pop-out Escherichia coli cam R Kan R strain (used to convert SHLX2 clones to plasmid clones) were also generously provided by Dr. R. H. Waterston. The pop-out strain was infected with the SHLX2 phage containing cm01c7 EST using standard protocols (25). Five colonies were picked, and plasmid DNA was prepared using either the Wizard plus kit (Promega, Madison, WI) or Qiagen tip-500 (Qiagen Inc, Santa Clarita, CA). DNA was then used to transform XL-1 blue E. coli strain (Stratagene, La Jolla, CA) followed by DNA preparation and verification by restriction analysis. The resulting 0.95-kb C. elegans fragment in the pRAT II plasmid was then sequenced using T7 and SP6 primers and automated DNA sequencing on an Applied Biosystems model 386 DNA sequencer utilizing T7 DNA polymerase and internal labeling with fluorescein-15-dATP (26).
Approximately 2 ϫ 10 6 phage from a mixed stage C. elegans cDNA library in bacteriophage vector UNI-ZAP XR (Stratagene) were plated and screened by hybridization as described previously (27) using the [␣-32 P]dCTP-labeled (Boehringer Mannheim) 0.95-kb ApaI/ScaI C. elegans fragment obtained from the cm01c7 clone as a probe. Hybridization was performed in 50% deionized formamide, 0.1% SDS, 5ϫ SSC, 5ϫ Denhardt's solution, and 100 g ml Ϫ1 denatured calf thymus DNA at 42°C. After overnight hybridization, filters were washed three times for 10 min each at room temperature in 2ϫ SSC, 0.1% SDS, two times for 30 min each at 65°C in 1ϫ SSC, 0.1% SDS, and exposed to X-OMAT AR film (Eastman Kodak Co). Positive plaques were rescreened twice with the same probe, and the size of positive inserts was determined by PCR amplification using the pBluescript SK phagemid-based primers T3 and T7. Five positive plaques were isolated and confirmed to be related to the cm01c7 EST by PCR using C. elegans cm01c7 EST-based primers. Following in vivo excision of the pBluescript phagemid from the UNI-ZAP vector, plasmid DNA was prepared, and the positive inserts were sequenced.
The longest LTA 4 hydrolase-like clone obtained, termed C5 (Ϸ1.4 kb), lacked the 5Ј-end of the coding region (as predicted from the size of cDNA of all previously cloned LTA 4 hydrolases). To isolate the 5Ј-end of the C. elegans cDNA, several anchored PCR amplifications using the phagemid-based primers T3, SK, and pSK (Stratagene) and a series of antisense primers based on the most 5Ј sequences in clone C5 were performed using the PCR Core kit (Boehringer Mannheim) and a Perkin-Elmer thermal cycler. PCRs were carried out in a buffer containing 10 mM Tris/HCl, pH 8.3, 50 mM KCl, 2.5 mM MgCl 2 , 0.2 mM deoxynucleotide triphosphates, 0.5 M primers and 1 l, of the C. elegans cDNA library or 1 l of the primary or secondary PCR amplification products as templates and cycling conditions of 35 cycles of 1 min at 94°C, 1 min at 55-62°C, and 1 min at 72°C. PCR products were separated by electrophoresis in 1% agarose gels, visualized by ethidium bromide staining, and Southern blotted to Hybond-N ϩ nylon membranes (Amersham Pharmacia Biotech) by overnight capillary transfer using 0.4 M NaOH. The PCR-amplified fragments were screened by overnight hybridization using the nested [␥-32 P]ATP-labeled primer C5T3 (5Ј-CCA GAC GGC GCA TCT TTC GCT-3Ј) (based on the most 5Ј sequence in clone C5) and T4 polynucleotide kinase (Boehringer Mannheim) in 6ϫ SSC, 20 mM NaH 2 PO 4 , 0.4% SDS, 5ϫ Denhardt's solution, 500 g ml Ϫ1 denatured calf thymus DNA at 42°C. PCR products of Ϸ 600 -800 bp were identified by their hybridization, isolated on 1.9% agarose gel, purified using Qiaquick spin columns (Qiagen), and subcloned into the TA vector (Invitrogen, Co. San Diego, CA). The ligation mixtures of PCR-generated fragments were transformed into DH5␣ E. coli, and inserts were characterized by restriction analysis, PCR, and DNA sequencing. From this cloning approach, several clones encoding the missing 5Ј coding region of clone C5 were identified. The full-length C. elegans cDNA clone (2.1 kb) was then reconstructed using a common EcoRI restriction site at the 5Ј-end of clone C5 and the 3Ј-end of the PCR-amplified NH 2 -terminal sequence.
Expression of the Recombinant C. elegans AP-1 Protein-A 1.8-kb NotI/XbaI fragment representing the entire AP-1 coding sequence was amplified by PCR using Expand high fidelity Taq Polymerase (Boehringer Mannheim) and the primers flag-1 (5Ј-CAT GCA TGC ATG GCG GCC GCG GCA CCT CCA CAT CCG AGA GAT CCC-3Ј) and flag-2 (5Ј-CAT GCA TGC ATG TCT AGA TTA TTT GAG AAG ACT TTG GAT TGC-3Ј). The flag-1 primer introduces an NH 2 -terminal NotI site (the C. elegans translation initiation codon was abolished to force translation to start from the ATG, supplied by the pFLAG CMV2 expression vector), and the flag-2 primer introduces a COOH-terminal XbaI site immediately after the stop codon (thus eliminating the 3Ј-untranslated region). The AP-1 NotI/XbaI fragment was then subcloned into the NotI/XbaI restricted mammalian expression vector pFLAG CMV2 (Kodak), and the resulting clone, pFLAG.celAP-1, was verified by sequencing.
Cell Culture and Transfection-The African green monkey SV40 transformed kidney cell line (COS-7 cells), obtained from the American Type Culture Collection, was grown in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal calf serum (Sigma), 50 units/ml penicillin, 50 g/ml streptomycin (Flow Laboratories, McLean, VA), and 2 mM glutamine (Flow Laboratories) at 37°C under an atmosphere of 6% CO 2 . 10 ϫ 10 6 cells per 600-cm 2 culture dish were seeded in 100 ml of media and transiently transfected at 80% confluence with 94 g of pFLAG.celAP-1 or pFLAG control plasmid and 283 l of LipofectAMINE reagent (Life Technologies, Inc.), following the recommendations of the manufacturer. Two days after transfection, cells were harvested in phosphate-buffered saline, centrifuged at 1100 ϫ g, resuspended in TBS (50 mM Tris/HCl, pH 7.4, 150 mM NaCl), and recentrifuged at 10,000 ϫ g for 10 min. Both the 10,000 ϫ g pellet and supernatant were assayed for recombinant expression of AP-1 protein by immunoblot analysis.
Preparation of C. elegans Extracts-Frozen mixed stage hermaphrodite C. elegans worms were a generous gift from J. McGhee (University of Calgary, Alberta, Canada). 5 ml of wet worms (resuspended in phosphate-buffered saline) were homogenized under liquid nitrogen (using a mortar and pestle) and resuspended in 5 ml of 0.1 M Tris, pH 7.0, and then in 10 ml of TBS (50 mM Tris pH 7.4, 150 mM NaCl) to a total volume of 20 ml. The homogenate was then sonicated at 4°C (three times, 20 s each). The suspension was first centrifuged at 2000 ϫ g for 10 min at 4°C to yield a large membrane fraction, followed by centrifugation of the resultant supernatant at 200,000 ϫ g for 60 min at 4°C to prepare microsomal and cytosolic fractions. Protein concentrations were determined using a protein assay kit (Bio-Rad).
Affinity Chromatography Purification of the Recombinant C. elegans AP-1 Protein-Chromatography columns were packed with 3 ml each of anti-FLAG M2 affinity resin (Kodak), equilibrated three times with 3 ml of TBS, and activated by washing three times with 3 ml of glycine/ HCl at pH 3.5, followed by washing three times with 3 ml of TBS. A total of 30 mg (10 ml of 3 mg/ml) of the 10,000 ϫ g supernatants of COS-7 cells transfected with either pFLAG vector or pFLAG.celAP-1 construct were incubated with 3 ml of the activated anti-FLAG M2 affinity gel in 15-ml polypropylene tubes and left to rotate at 4°C overnight. Each slurry was transferred back to chromatography columns, and the flow-through samples from the columns were drained the next day, followed by washing three times with 3 ml of TBS. Columns were then eluted using 11 ml (0.5 ml/fraction) of the FLAG octapeptide (0.5 mg/ml in TBS; NH 2 -Asp-Tyr-Lys-Asp-Asp-Asp-Asp-Lys-COOH) (Kodak). Fractions containing the C. elegans FLAG fusion protein (as assessed by immunoblot analysis using the anti-FLAG M2 monoclonal antibody) were pooled. Aliquots of the columns flowthrough, combined washes, and the different eluted fractions were kept for immunoblot analysis, and the rest of the samples were frozen at Ϫ80°C when not used immediately for functional assays.
Immunoblot Analysis-The 10,000 ϫ g COS-7 cell supernatants (lysates of COS-7 cells transfected with either pFLAG vector or pFLAG.cel AP-1 construct), the anti-FLAG M2 affinity columns flow-through, combined washes, and the different FLAG peptide eluted fractions, as well as the NH 2 -terminal FLAG fusion protein of E. coli bacterial alkaline phosphatase control (Kodak) were separated electrophoretically on 10% polyacrylamide gels according to the method of Laemmli (28). This was followed by electrophoretic transfer to nitrocellulose membranes using a Novex immunoblot system according to the manufacturer's instructions (Novex). The nitrocellulose membranes were developed using a 1:300 dilution of mouse anti-FLAG M2 monoclonal antibody (Kodak). The secondary horse radish peroxidase-linked donkey anti-mouse IgG antibody (Amersham Pharmacia Biotech) was used at a dilution of 1:1000. Immunodetection was performed using enhanced chemiluminescence according to the manufacturer's instructions (Amersham Pharmacia Biotech). Autoradiographs for chemiluminescence detection were exposed to Kodak X-OMAT x-ray films for 3 min and then developed. A polyclonal anti-rabbit human LTA 4 hydrolase antiserum (raised against the entire protein) (29) was also used for immunoblot analysis (Merck Frosst, Pointe Claire-Dorval, Quebec, Canada).
LTB 4 Assays-LTA 4 ethyl ester was synthesized at the Merck Frosst Center for Therapeutic Research (Montreal, Quebec, Canada). LTB 4 , prostaglandin B 2 , 6-trans-LTB 4 , 6-trans-12-epi-LTB 4 , and (5S,6S)-di-HETE, (5S,6R)-diHETE standards were purchased from Cayman Chemical Co., Inc. (Ann Arbor, MI). The soluble supernatant fraction following a 100,000 ϫ g centrifugation (100S fraction) of Sf9 cells infected with a recombinant human LTA 4 hydrolase construct (29) was used as a positive control (Merck Frosst Center for Therapeutic Research). Alkaline hydrolysis of LTA 4 ethyl ester was carried out as described (30). LTA 4 hydrolase assays on C. elegans cytosolic fraction (125 g), anti-FLAG M2 purified COS-7 lysates (200 l of 10,000 ϫ g supernatant of COS-7 cells transfected with pFLAG vector and 0.8 g of FLAG-tagged C. elegans AP-1 protein), or recombinant human LTA 4 hydrolase (25 g) were performed as described previously (31). Samples in 250-l reactions were incubated with 25 M LTA 4 in 0.1 M Tris/HCl, pH 8.0, 1 mg/ml BSA for 10 min at room temperature. Reactions were terminated with the addition of an equal volume of methanol containing 1 nmol/ml prostaglandin B 2 standard. Eicosanoid products were extracted using an equal volume of chloroform, evaporated under nitrogen and resuspended in 100 l of the HPLC solvent methanol/water/acetic acid (75:25:0.01). Eicosanoid products were analyzed by reverse-phase HPLC on a 3.9 ϫ 150 mm NovaPak C18 column (Waters). The MeOH/ H 2 O/acetic acid solvent was pumped isocratically at a flow rate of 1 ml/min. The effluent was monitored at 270 nm by a photodiode array detector. Products were compared with the retention times and spectra of known eicosanoid standards.
Radioimmunoassay detection of LTB 4 production was performed by incubating 100 ng of either purified C. elegans AP-1(FLAG) or protein eluted from affinity columns loaded with mock-transfected cell extracts with 25 M LTA 4 in 0.1 M Tris/HCl, pH 8.0, 1 mg/ml BSA for 10 min at room temperature. Reactions were stopped by the addition of 20% methanol followed by a brief centrifugation at 10,000 ϫ g for 20 s. LTB 4 production was assayed using a radioimmunoassay kit ([ 3 H]leukotriene B 4 assay system) (Amersham Pharmacia Biotech). The human recombinant LTA 4 hydrolase was used as a positive control as described above.
Aminopeptidase Assays-Amino acid p-nitroanilides (Sigma or Bachem Bioscience Inc.) were incubated (final concentration, 0.05-5 mM) at room temperature with 0.17 g of purified C. elegans AP-1 FLAG-fusion protein, anti-FLAG M2 affinity gel-purified fractions of lysates of COS-7 cells transfected with pFLAG vector, or 0.17 g of purified human recombinant LTA 4 hydrolase/aminopeptidase in 250 l of buffer containing 0.1 M Tris, pH 8, 200 mM NaCl, 1 mg/ml BSA. The assays were performed in 96-well microtiter plates (path length, 0.7 cm), and the formation of the product (p-nitroaniline, ⑀ ϭ 10, 800 M Ϫ1 cm Ϫ1 ) was monitored for 60 min at 405 nm using a kinetic microplate reader spectrophotometer (Molecular Devices). Spontaneous hydrolysis of the substrate (Ϸ 0.03 milli-absorbance units/min) was corrected for by subtracting the absorbance of control incubations without enzyme. Kinetic constants (K m and k cat ) were determined by nonlinear regression (Kaleidagraph software) of the data to the Michaelis-Menten equation.
Inhibition of Aminopeptidase Activity-Bestatin, an inhibitor of aminopeptidases (Sigma), was evaluated as an inhibitor of the C. elegans aminopeptidase activity. Bestatin (final concentration, 0 -50 M) in 250 l of buffer containing 0.1 M Tris, pH 8, 200 mM NaCl, 1 mg/ml BSA was incubated at room temperature with 0.17 g of purified C. elegans AP-1(FLAG) enzyme for 10 min. L-Arginine-p-nitroanilide (1 mM) was then added, and substrate hydrolysis was monitored as described above. IC 50 values were estimated from a log dose-response curve of initial velocity versus inhibitor concentration. 4 Hydrolase-like cDNA Homologue from C. elegans-A search of the annotated files in Gen-Bank using the key words "leukotriene hydrolase" yielded an EST (GenBank accession number M88793) clone termed cm01c7 (327 bp) from a C. elegans mixed stage hermaphrodite cDNA library. When translated, the EST revealed strong homology (51%) to the human LTA 4 hydrolase. The complete sequence of a 0.95-kb ApaI/Sca I restriction fragment containing the C. elegans LTA 4 hydrolase-like EST was obtained from the phage clone cm01c7 (see under "Materials and Methods" for details) and was confirmed as a mammalian LTA 4 hydrolaselike sequence by sequence homology comparisons using the BLASTN and TBLASTN algorithms (National Center for Biotechnology Information) (32) and FASTA and TFASTA programs (33). The 0.95-kb fragment was radiolabeled with 32 P and used as a probe to screen a C. elegans phage cDNA library, resulting in the isolation of clone C5 (Ϸ 1.4 kb), which showed a strong sequence homology (Ϸ 45%) to the human LTA 4 hydrolase but was missing the 5Ј part of the gene (as predicted from the sizes of cDNAs encoding previously cloned LTA 4 hydrolases). A series of anchored PCRs with the C. elegans cDNA library as template and using pBluescript SK phagemid-based primers, as well as several antisense primers based on the most 5Ј sequence of clone C5, was carried out to isolate the missing 5Ј-end of clone C5. The full-length C. elegans cDNA sequence and the deduced amino acid sequence, named AP-1, are shown in Fig. 1. The C. elegans AP-1 cDNA sequence is 2152 bp long, consisting of a short 15-base 5Јuntranslated region, an open reading frame encoding a 609amino acid protein, a 282-bp-long 3Ј-untranslated region, and a 28-bp-long poly(A ϩ ) tail. No consensus N-glycosylation sites, targeting signals, or putative phosphorylation sites were found in the sequence. The first ATG triplet (starting at nucleotide 1 in Fig. 1) in the sequence was considered to be the initiation codon of protein translation because 1) it matches the location of the translation initiation codon from both the human and the mouse LTA 4 hydrolase cDNAs (18,19) with only one extra codon in the C. elegans sequence (Fig. 2), 2) the nucleotide sequence flanking it (AATATGG) is in agreement with Kozak's rule for translational initiation consensus sequence (34), and 3) the open reading frame starting from this methionine and ending at the TAA terminator codon (nucleotide 1828) encodes a 68,248-kDa protein, corresponding to the molecular mass of the recombinant C. elegans AP-1 protein estimated by SDS-polyacrylamide gel electrophoresis (see below). The length of the AP-1 open reading frame (1827 bp), the number of the deduced amino acids (609 residues), and the molecular mass of the encoded protein (68,248 kDa) are comparable to the human (1830 bp, 610 residues, 69,140 kDa) and the mouse (1830 bp, 610 residues, 68,917 kDa) LTA 4 hydrolases (18,19).

Molecular Cloning of a LTA
Amino Acid Sequence Comparison of the Human and the Mouse LTA 4 Hydrolases to Their C. elegans Homologue-Data base searches identified the human LTA 4 hydrolase as the most closely related protein to the C. elegans AP-1 translation product (63% similarity and 45% identity) (results not shown). His 297 , His 301 , and Glu 320 in the C. elegans amino acid sequence (the underlined residues in Fig. 1) conform to a catalytic zinc site of mammalian LTA 4 hydrolases and zinc metallopeptidases (5,35) and match His 295 , His 299 , and Glu 318 in both the human and the mouse LTA 4 hydrolase sequences. These three residues are likely to be involved in the coordination of the zinc atom as described previously for the mouse LTA 4 hydrolase (8) and for certain peptidases and neutral proteases (5). A multiple alignment of the amino acid sequences of the human LTA 4 hydrolase, the mouse LTA 4 hydrolase, and their putative C. elegans homologue and the conserved residues in the three sequences are shown in a consensus (Fig. 2). A high degree of homology is seen in the region between amino acid residues 246 and 320 in the C. elegans sequence, the homology then decreases toward either the amino or the carboxyl terminus of the protein. A typical consensus zinc binding motif HEXXHX 18 E is indicated starting from amino acid 298 to 321 in the C. elegans AP-1 sequence. This motif is found in several reported metallopeptidases and allows the classification of the C. elegans enzyme under the M1 family of metalloexopeptidases (36). Members of this family also include aminopeptidase A, aminopeptidase N, cysteine aminopeptidase, and LTA 4 hydrolase. The tyrosine residue Tyr 383 in both the human and the mouse sequences, which is essential for the peptidase activity of the mammalian LTA 4 hydrolase/aminopeptidase enzyme and may act as a proton donor in a general base mechanism (37), is conserved in the C. elegans AP-1 (Fig. 2, Tyr 387 in the C. elegans sequence). Glu 296 in both human and mouse sequences, the mutation of which to Gln 296 abolishes the aminopeptidase activity of the mammalian enzyme (38), is also conserved in the C. elegans sequence (Fig. 2, Glu 298 in the C. elegans sequence). On the other hand, Tyr 378 in both the human and the mouse LTA 4 hydrolase sequences, which is known to be involved in the covalent binding of LTA 4 (39), is replaced by a phenylalanine (Phe 382 ) in the C. elegans sequence.
The Structure of the C. elegans Aminopeptidase Gene-The cloned C. elegans AP-1 cDNA was compared with sequences in the genomic section of the Sanger Center C. elegans data base, and two cosmid clones were identified (cosmids C42C1and Y39C12) that showed a 100% match to the cDNA sequence using a BLASTN search. A map of the structure of the C. elegans aminopeptidase gene was then constructed (Fig. 3A).  2. A multiple alignment of amino acid sequences of the human leukotriene A 4 , the mouse leukotriene A 4 hydrolase, and the C. elegans AP-1. The alignment was made using the Pretty Plot function of the GCG program (33). Amino acids are numbered beginning with the first methionine residue in the C. elegans sequence. Conserved residues in all three sequences are shown in the consensus. Both the zinc-binding motif, common among members of the M1 family of metallopeptidases (HEXXHX 18 E), and the tyrosine residue (number 383 in both the human and the mouse sequences), which is essential for the peptidase activity of the human and mouse LTA 4 hydrolase/aminopeptidase, are indicated on a black background. The conserved glutamate residue necessary for peptidolysis of mammalian LTA 4 hydrolases (Glu 296 in both human and mouse sequences) is underlined. The tyrosine residue Tyr 378 (in both the human and the mouse LTA 4 hydrolase sequences) involved in the covalent binding of LTA 4 to the human LTA 4 hydrolase is indicated by an asterisk and is replaced by a phenylalanine (Phe 382 in the C. elegans sequence). size of the introns is expected as most introns in the nematode C. elegans are very short (40). The sequences of exon-intron boundaries were determined by comparing the cDNA sequence and the genomic sequence (Fig. 3B). The exon-intron junction in intron 2 follows the GT/AG rule and agrees with consensus sequences for the donor and acceptor sites (41). On the other hand, introns 1 and 3 lack an AG at the 3Ј splice acceptor site, which agrees with the finding that splicing in C. elegans does not require this AG (42). As shown for over 98% of C. elegans introns, all three introns have an elevated A-U content just upstream of the 3Ј splice site with a U present at position -5 relative to the cleavage site (40). The proposed zinc-binding histidine residues (His 297 and His 301 ) and glutamate residue (Glu 320 ), which constitute the zinc-binding domain (HEXXHX 18 E), are located on one exon (Fig. 3, exon 3), unlike the structure of the human LTA 4 hydrolase/aminopeptidase gene (43), in which the two essential zinc-binding histidine residues (His 295 and His 299 ) are present on exon 10, whereas the third zinc-binding ligand glutamate (Glu 318 ) is located on another exon (exon 11).
Expression of Recombinant C. elegans AP-1 and Immunoblot Analysis-The C. elegans AP-1 open reading frame (1.8 kb) was subcloned into the mammalian expression vector pFLAG CMV2, which provides a translation initiation codon ATG, followed by an amino-terminal FLAG epitope (NH 2 -Asp-Tyr-Lys-Asp-Asp-Asp-Asp-Lys-COOH). COS-7 cells were transiently transfected with the pFLAG.cel AP-1 or the vector control DNA and harvested, followed by preparation of membrane and soluble protein fractions and analysis by SDS-polyacrylamide gel electrophoresis and Western blotting using anti-FLAG M2 antibody. As expected from its calculated molecular mass, a 69-kDa immunoreactive protein was detected using the anti-FLAG antibody in the supernatant of COS-7 cells transfected with pFLAG.cel AP-1 cDNA (Fig. 4, lane 2) but not in the supernatant of mock-transfected cells (results not shown). As described previously for the mammalian LTA 4 hydrolase/aminopeptidase (16), the C. elegans AP-1 is a soluble protein expressed in the cell cytosol with minimal detection in either the microsomal or the membrane fractions (100,000 ϫ g and 2000 ϫ g pellets, respectively) (results not shown). The expressed FLAG-tagged C. elegans AP-1 protein was partially purified (Ϸ 30% purity) using anti-FLAG M2 affinity chroma-tography, eluted with the FLAG octapeptide, resolved by 10% SDS-polyacrylamide gel electrophoresis, and detected by the monoclonal antibody anti-FLAG M2. All of the expressed C. elegans AP-1 protein was bound to the anti-FLAG M2 affinity resin. No protein was detected in either the flow-through from the column or the column washes with the C. elegans FLAGtagged AP-1 eluting mainly in fractions 4 -7 (Fig. 4, lanes  8 -11). Despite the significant amino acid sequence homology  5-15). The samples were separated by SDS-polyacrylamide gel electrophoresis, transferred to nitrocellulose, and immunoblotted using anti-FLAG M2 antiserum (1:300 dilution). Enhanced chemiluminescence was used for detection. The positions of molecular mass markers are indicated. between the C. elegans AP-1 and the human LTA 4 hydrolase/ aminopeptidase (45% identity), a rabbit anti-human LTA 4 hydrolase polyclonal antiserum (29) failed to detect any LTA 4 hydrolase specific immunoreactive proteins in either the supernatant of COS-7 cells transfected with pFLAG.cel AP-1 or the cytosolic extract from C. elegans worms (results not shown).
Aminopeptidase Activity of the C. elegans AP-1 Enzyme and Its Inhibition by Bestatin-Based on the conservation of the catalytic zinc binding motif HEXXHX 18 E in the primary structure of C. elegans AP-1 protein and several other zinc proteases and peptidases, the C. elegans AP-1 FLAG fusion protein was assayed for aminopeptidase activity using 11 different amino acid p-nitroanalide derivatives as chromogenic amide substrates. These compounds represent acidic, basic, and neutral amino acids, as well as amino acids with NH 2 -terminal substitutions and D-stereochemistry. The purified C. elegans AP-1 FLAG fusion protein contained an intrinsic aminopeptidase activity that was absent in the anti-FLAG M2 affinity gel purified fractions of supernatant from mock-transfected COS-7 cells (Table I). The rate of hydrolysis of L-arginine p-nitroanilide was dependent on protein and substrate concentrations with a K m of 0.43 Ϯ 0.01 mM and a V max of 0.18 Ϯ 0.01 mol/min/mg enzyme. These values can be compared with a K m of 0.09 Ϯ 0.01 mM and a V max of 0.47 Ϯ 0.01 mol/min/mg obtained for the human LTA 4 hydrolase/aminopeptidase (Fig.  5). The K m and k cat values for the hydrolysis of the amino acid p-nitroanilides by C. elegans AP-1 enzyme were dependent on the amino acid substituent (Table I). Comparison of the specificity constant k cat /K m for all 11 compounds tested reveals that the recombinant C. elegans AP-1 preferentially hydrolyzed the L-arginine derivative. Acidic amino acids, amino acids with NH 2 -terminal substitutions, and amino acids with D-stereochemistry were poor substrates. The human recombinant LTA 4 hydrolase/aminopeptidase enzyme had a similar substrate specificity for the selected p-nitroanilides. The human enzyme is considered an arginine aminopeptidase, despite its wide cleavage specificity, because it preferentially hydrolyzes tripeptides with L-arginine at the NH 2 -terminal position (13). In the absence of a physiological substrate for the aminopeptidase activity of the human enzyme and its high catalytic efficiency for several synthetic tripeptides (exceeding the k cat /K m for LTA 4 by 10-fold), the enzyme was suggested to be involved in the metabolism of dietary peptides and neuropeptides (13). This role can also be proposed for the C. elegans AP-1 enzyme. Bestatin, a potent inhibitor of human LTA 4 hydrolase/aminopeptidase (10), as well as other aminopeptidases, inhibited the hydrolysis of L-arginine p-nitroanilide by AP-1. The concentration for half-maximal inhibition (IC 50 ) of p-nitroaniline formation was 2.6 Ϯ 1.2 M (results not shown).
During the cloning of the 5Ј-end of the C. elegans AP-1 cDNA, a PCR error introduced a point mutation at amino acid position 117, changing an alanine residue to a valine. When clones containing the Ala 117 to Val 117 PCR mutation were analyzed for aminopeptidase activity, they failed to hydrolyze the amide bond of any amino acid p-nitroanilide tested. This raises the possibility that certain conserved residues other than the previously documented Tyr 383 and Glu 296 may be important for the aminopeptidase activity of the mammalian LTA 4 hydrolase/aminopeptidase enzyme. It is also interesting to note that this alanine residue (Ala 114 in the human sequence) is conserved evolutionary as it is found in the C. albicans LTA 4 hydrolase (which mainly exhibits aminopeptidase activity), the S. cerevisiae proposed LTA 4 hydrolase (which is yet to be characterized), all cloned mammalian LTA 4 hydrolases, including human, mouse, rat, and guinea pig, and the C. elegans AP-1 (data not shown).
Measurement of LTA 4 Hydrolase Activity of C. elegans AP-1 Enzyme-The C. elegans AP-1 protein was analyzed for epoxide hydrolase activity using LTA 4 as a substrate. Reverse-phase HPLC analysis of products formed when the purified FLAGtagged C. elegans AP-1 enzyme or the cytosolic extract of C. elegans worms was incubated with LTA 4 revealed no production of LTB 4 (Fig. 6, peak 4, tracings 6 and 2, respectively). In contrast, the human LTA 4 hydrolase (used as a positive con- The different amino acid p-nitroanilides (0.05-5 mM) were incubated at room temperature with 0.17 g of either purified C. elegans AP-1 (FLAG) enzyme or human LTA 4 hydrolase/aminopeptidase enzyme in 250 l of 0.1 M Tris, pH 8.0, 200 mM sodium chloride containing BSA (1 mg/ml). p-Nitroaniline formation was monitored spectrophotometrically for 60 min at 405 nm. Kinetic constants (K m and k cat ) were determined by nonlinear regression analysis. Activities of purified vector construct with each substrate were less than 10% of values obtained with purified C. elegans AP-1 enzyme and were comparable to the observed nonenzymatic hydrolysis rates. trol) produced mainly peak 4 (Fig. 6, tracing 1), which eluted with the retention time of the expected enzymatic product LTB 4 . Small amounts of the other peaks (peaks 2, 3, 5, and 6) represent the nonenzymatic hydrolysis of LTA 4 . In the absence of LTA 4 hydrolase activity (Fig. 6, all tracings except tracing 1), LTA 4 was mostly converted to the nonenzymatic hydrolysis products, the all-trans-LTB 4 epimers at C12 (6-trans-LTB 4 and 12-epi-6-trans-LTB 4 , peaks 2 and 3, respectively), as well as (5S,6R)-diHETE (peak 5), and (5S,6S)-diHETE (peak 6). These species are normally formed in small and equal amounts in aqueous solutions by spontaneous hydrolysis (44), and their peaks eluted with the retention times and spectra of known eicosanoid standards. Peaks 2, 3, 5, and 6 were also produced by boiled C. elegans cytosol (Fig. 6, tracing 4), as well as boiled FLAG-tagged AP-1 (data not shown), confirming their nonenzymatic origin. C. elegans cytosol pretreated with 100 M bestatin (an inhibitor of both LTA 4 hydrolase and aminopeptidase activities of the human enzyme (10)), showed the same chromatographic profile (Fig. 6, tracing 3) as the untreated or the boiled C. elegans cytosol. The inability of the C. elegans AP-1 enzyme to synthesize LTB 4 from LTA 4 was also confirmed by the absence of any LTB 4 production (data not shown) using a LTB 4 radioimmunoassay (lower limit of detection is 16 pg of LTB 4 ml Ϫ1 ).
Although C. elegans AP-1 does not appear to hydrolyze LTA 4 , the high degree of identity between the active sites of AP-1 and mammalian LTA 4 hydrolases (Fig. 2) suggests that LTA 4 may bind in the active site of AP-1 without being a substrate for catalysis. To test this possibility, we examined the effect of LTA 4 -ethyl ester on the aminopeptidase activity of AP-1 using L-arginine-p-nitroanilide as a substrate. We chose the LTA 4ethyl ester as it is a suicide inactivator of the mammalian LTA 4 hydrolases and is much more resistant than LTA 4 to nonenzy-matic hydration. The LTA 4 -ethyl ester had no effect on the aminopeptidase activity of AP-1 at concentrations up to 100 M, 10 times the K m of mammalian LTA 4 hydrolase for LTA 4ethyl ester (results not shown).
The bifunctional human LTA 4 hydrolase/aminopeptidase enzyme is suicide-inactivated during catalysis via an apparently mechanism-based irreversible binding of LTA 4 to the protein (45), with tyrosine at position 378 identified as the site for covalent binding of LTA 4 . Interestingly, the mutation of Tyr 378 to Phe 378 in the human LTA 4 hydrolase yielded an enzyme with increased turnover and resistance to mechanism-based inactivation (39), thus dissociating catalysis and covalent modification/inactivation events. This tyrosine residue is a phenylalanine in the C. elegans AP-1 enzyme sequence (Phe 382 ), but the C. elegans enzyme does not hydrolyze LTA 4 , indicating that other residues (lacking in the C. elegans AP-1 sequence) must also be important for LTA 4 binding and catalysis.
That the cloned C. elegans AP-1 enzyme functions as an aminopeptidase with no LTA 4 hydrolase activity is interesting, as its primary structure resembles LTA 4 hydrolases more than it does aminopeptidases (Table II). Comparison of the C. elegans AP-1 enzyme and other proteins in the SwissProt data base revealed 45% identity to the human, mouse, rat, and guinea pig LTA 4 hydrolases at the amino acid level, with lower identity to the human, rat, and pig aminopeptidase-N (28 -30%). Moreover, the identity between C. elegans AP-1 and mammalian LTA 4 hydrolases extends over their entire primary structures, with some divergence in the N and C termini. In contrast, C. elegans AP-1 only overlaps a limited region of about 300 amino acids with other aminopeptidase enzymes (a region that contains the canonical zinc-binding motif HEXXHX 18 E). It is interesting to note that the same identity (an average of 30%) that is shown between C. elegans AP-1 and  5), and 0.8 g of FLAG-tagged C. elegans AP-1 protein purified using anti-FLAG M2 affinity gel (tracing 6). LTA 4 hydrolase assay and analysis of eicosanoid products were carried out as described previously (29). Peaks were identified by elution with co-chromatographed standards and their characteristic absorbance spectrum. Peak 1, prostaglandin B 2 (internal standard); peak 2, 6-trans-LTB 4 ; peak 3, 6-trans-12-epi-LTB 4 ; peak 4, LTB 4 ; peak 5, (5S,6R)-diHETE; peak 6, (5S,6S)-diHETE. The chromatograms are representative of three experiments with identical results. other aminopeptidases is also seen between mammalian LTA 4 hydrolases and any given aminopeptidase enzyme (Table II). The structural similarity between C. elegans AP-1 and mammalian LTA 4 hydrolases suggests an evolutionary relationship. Recently, three other proteins that are structurally related to mammalian LTA 4 hydrolases have been identified in lower invertebrates. These proteins include a C. albicans LTA 4 hydrolase-related enzyme (41% identity to human LTA 4 hydrolase) that functions mainly as an aminopeptidase but fails to hydrolyze LTA 4 to LTB 4 (47), a gene from the yeast S. cerevisiae (39% identity to human LTA 4 hydrolase) (22), and a partial amino acid sequence (316 residues) of a D. discoideum cDNA (GenBank accession number U27538) 2 with 38% identity to human LTA 4 hydrolase. Biochemical studies to clarify the enzymatic activity (or activities) of both the yeast and the Dictyostelium LTA 4 hydrolase-like proteins await their expression and characterization. LTB 4 production has not been reported in either C. albicans or D. discoideum, and our analysis of mixed stage C. elegans worms failed to detect any LTA 4 hydrolase activity. The high primary sequence identity of the C. elegans AP-1 to mammalian LTA 4 hydrolases suggests that AP-1 may represent an evolutionary precursor of the mammalian LTA 4 hydrolases. Thus, mammalian LTA 4 hydrolases may have originated from aminopeptidases like AP-1, retaining their aminopeptidase activity and developing a LTA 4 hydrolase function in higher eukaryotes. In support of this hypothesis, an aminopeptidase B has recently been cloned from rat testes (46) that shows highest homology to mammalian leukotriene A 4 hydrolases (44%), intermediate homology to C. elegans AP-1 (38%), and lowest homology to mammalian N-type aminopeptidases (21-24%) (Table II) and can catalyze the conversion of LTA 4 to LTB 4 (46).
In conclusion, we have cloned and functionally expressed a 69-kDa protein from C. elegans, the primary structure of which is more homologous to mammalian LTA 4 hydrolases than other zinc aminopeptidases. This protein functions as an aminopeptidase with broad substrate specificity but lacks any LTA 4 hydrolase activity. The primary sequence identity of C. elegans AP-1 enzyme to mammalian LTA 4 hydrolases and rat aminopeptidase B suggests that these enzymes are evolutionarily related.

hydrolases, and aminopeptidases
A BLASTP search of SwissProt data base (release 96) identified LTA 4 hydrolases followed by aminopeptidases as the most closely related proteins to the C. elegans AP-1. The Bestfit algorithm in the GCG sequence analysis software (gap weight of 12, length weight of 4, gap creation penalty of 12, and gap extension penalty of 4) was used to determine percentage sequence identity. hLTA 4 , human LTA 4 hydrolase; mLTA 4 , mouse LTA 4 hydrolase; rLTA 4 , rat LTA 4 hydrolase; gpLTA 4 , guinea pig LTA 4 hydrolase; rAP-B, rat aminopeptidase B; hAP-N, human microsomal aminopeptidase N; rAP-N, rat microsomal aminopeptidase N; pgAP-N, pig microsomal aminopeptidase N.