Cloning and Characterization of a Bifunctional Leukotriene A 4 Hydrolase from Saccharomyces cerevisiae *

In mammals, leukotriene A 4 hydrolase is a bifunc- tional zinc metalloenzyme that catalyzes hydrolysis of leukotriene A 4 into the proinflammatory leukotriene B 4 and also possesses an arginyl aminopeptidase activity. We have cloned, expressed, and characterized a protein from Saccharomyces cerevisiae that is 42% identical to human leukotriene A 4 hydrolase. The purified protein is an anion-activated leucyl aminopeptidase, as assessed by p -nitroanilide substrates, and does not hydrolyze leukotriene A 4 into detectable amounts of leukotriene B 4 . However, the S. cerevisiae enzyme can utilize leuko- triene A 4 as substrate to produce a compound identified as 5 S ,6 S -dihydroxy-7,9- trans -11,14- cis -eicosatetraenoic acid. Both catalytic activities are inhibited by 3-(4-ben-zyloxyphenyl)-2-( R )-amino-1-propanethiol (thioamine), a competitive inhibitor of human leukotriene A 4 hydro- lase. Furthermore, the peptide cleaving activity of the S. cerevisiae enzyme was stimulated approximately 10-fold by leukotriene A 4 with kinetics indicating the presence of a lipid binding site. Nonenzymatic hydrolysis products of leukotriene A 4 , leukotriene B 4 , arachidonic acid, or phosphatidylcholine were without effect. Moreover, leukotriene A 4 could displace the inhibitor thioamine and restore maximal aminopeptidase activity, S. the Reverse transcription total RNA m chain C 18 EC, Macherey Nagel) and finally analyzed by reverse-phase high performance liquid chroma- tography (HPLC). The column (Nova-Pak C 18 , 4 m m Radial-Pak car-tridge, 5 3 100 mm, Waters) was eluted with a mixture of acetonitrile/ methanol/water/acetic acid (30:35:35:0.01, by volume) at a flow rate of 1.0 ml/min. The absorbance of the eluate was monitored at 270 nm. The products were identified by their chromatographic mobility relative standard compounds, as well as by UV spectrophotometry. Protein Analyses— Protein concentrations were determined according to the method of Bradford, using the Bio-Rad protein-assay reagent and bovine serum albumin as standard. SDS-polyacrylamide gel elec-trophoresis (PAGE) was performed on a Phast system (Amersham Pharmacia Biotech) using 10–15% gradient gels. Bands of protein were visualized by staining with Coomassie Brilliant Blue. N-terminal amino acid sequencing was carried out with an Applied Biosystems 477A instrument. Phenylthiohydantoin derivatives were analyzed with on- line HPLC systems (Protein Analysis Center, Karolinska Institutet).

In mammals, leukotriene A 4 hydrolase is a bifunctional zinc metalloenzyme that catalyzes hydrolysis of leukotriene A 4 into the proinflammatory leukotriene B 4 and also possesses an arginyl aminopeptidase activity. We have cloned, expressed, and characterized a protein from Saccharomyces cerevisiae that is 42% identical to human leukotriene A 4 hydrolase. The purified protein is an anion-activated leucyl aminopeptidase, as assessed by p-nitroanilide substrates, and does not hydrolyze leukotriene A 4 into detectable amounts of leukotriene B 4 . However, the S. cerevisiae enzyme can utilize leukotriene A 4 as substrate to produce a compound identified as 5S,6S-dihydroxy-7,9-trans-11,14-cis-eicosatetraenoic acid. Both catalytic activities are inhibited by 3-(4-benzyloxyphenyl)-2-(R)-amino-1-propanethiol (thioamine), a competitive inhibitor of human leukotriene A 4 hydrolase. Furthermore, the peptide cleaving activity of the S. cerevisiae enzyme was stimulated approximately 10-fold by leukotriene A 4 with kinetics indicating the presence of a lipid binding site. Nonenzymatic hydrolysis products of leukotriene A 4 , leukotriene B 4 , arachidonic acid, or phosphatidylcholine were without effect. Moreover, leukotriene A 4 could displace the inhibitor thioamine and restore maximal aminopeptidase activity, indicating that the leukotriene A 4 binding site is located at the active center of the enzyme. Hence, the S. cerevisiae leukotriene A 4 hydrolase is a bifunctional enzyme and appears to be an early ancestor to mammalian leukotriene A 4
The mammalian LTA 4 hydrolase is a metalloenzyme containing 1 mol of zinc per mol of protein. In addition to the epoxide hydrolase activity, i.e., the hydrolysis of LTA 4 into LTB 4 , the enzyme also possesses an anion-dependent arginylaminopeptidase activity, the physiological role of which is presently unknown (6 -9). The zinc atom is required for both catalytic activities and is bound to His-295, His-299, and Glu-318 (10). Because of its zinc binding motif and aminopeptidase activity, LTA 4 hydrolase is homologous to a multitude of other zinc peptidases present in a variety of species spanning from mammals to bacteria, in particular those belonging to the M1 family (11). On the other hand, the epoxide hydrolase activity, i.e. the production of LTB 4 , has only been detected in vertebrates, including birds, fish, and frogs (12)(13)(14)(15), and a nonmammalian form of LTA 4 hydrolase was recently purified from the African claw toad, Xenopus laevis. The toad enzyme contained zinc and exhibited both epoxide hydrolase and peptidase activity (16). In fact, formation of LTB 4 has never been convincingly demonstrated in any lower animal species, bacteria, or plants. Thus, an aminopeptidase-1 was recently cloned and characterized from Caenorhabditis elegans, that was 45% identical (63% similar) at the amino acid level to mammalian LTA 4 hydrolase (17) and exhibited an arginyl aminopeptidase activity (Fig. 1). Despite this high level of sequence identity, the C. elegans enzyme failed to hydrolyze LTA 4 into LTB 4 , and no other functional link to LTA 4 hydrolase was reported. Apparently, very little is known about the evolution of LTA 4 hydrolase and the phylogenetic relationship between its two catalytic activities.
In the course of sequencing the genome of Saccharomyces cerevisiae, an open reading frame was identified as an LTA 4 hydrolase homologue, with 42% identity (53% similarity) to human LTA 4 hydrolase (18). In the present study, we have cloned, expressed, and characterized the corresponding gene product (Fig. 1). We show that it is a bifunctional enzyme possessing an anion-activated leucyl aminopeptidase activity as well as an epoxide hydrolase activity toward LTA 4 . Moreover, the aminopeptidase activity is strongly stimulated by LTA 4 in a fashion suggesting the presence of a lipid binding pocket located at the active center of the enzyme and presumably overlapping with the catalytic site(s). Hence, the homologue in S. cerevisiae appears to be an early ancestral gene to the vertebrate forms of LTA 4 hydrolase.

EXPERIMENTAL PROCEDURES
Materials-Restriction endonucleases, T4 DNA ligase, and T7 sequencing kit were purchased from Amersham Pharmacia Biotech.
Isolation of RNA from S. cerevisiae-Cells from S. cerevisiae were grown in 10 ml of yeast medium (0.1 g of yeast extract, 0.1 g Bactopeptone, 0.2 g of glucose, and 0.2 mg of adenine) at 30°C to midexponential phase (A 600 ϭ 1), pelleted by centrifugation for 3 min at 1,500 ϫ g at 4°C, and resuspended in ice-cold water followed by a short centrifugation at 14,000 ϫ g. Total RNA was isolated according to the hot acid phenol method (21). The RNA was precipitated with ethanol, dissolved in RNase-free water, and stored at Ϫ80°C until use.
Cloning of the S. cerevisiae LTA 4 Hydrolase-Reverse transcription of total yeast RNA (2 g/20 l of reaction mix) was performed using the Perkin-Elmer GeneAmp® RNA polymerase chain reaction (PCR) kit, using oligo(dT) 16 as primer. From the genomic sequence (AC X94547) of the S. cerevisiae LTA 4 hydrolase, two PCR primers were designed containing SalI and SphI restriction enzyme sites: FK22, covering the start codon (5ЈGTG TTC GTC GAC ATG TTC TTG CTT CCA TTT GTC, SalI site underlined) and FK21 over the stop codon (5Ј GTT ACG ACG CAT GCC TTT TCA AAG ACC TAA ATC, SphI site underlined). Using the cDNA as template and FK21 and FK22 (30 pmol) as primers, a PCR was performed to amplify the S. cerevisiae LTA 4 hydrolase gene using the Expand TM high fidelity PCR system (Roche Molecular Biochemicals). A 30-l reaction contained 1 ϫ reaction buffer, 200 M dNTP, 1 unit of Taq and Pwo DNA polymerase mix, and 2 l cDNA. The sample was first denatured for 2 min at 95°C followed by 35 cycles composed of denaturation for 30 s at 94°C, annealing for 30 s at 50°C, and elongation for 2 min at 72°C. The PCR was terminated by an elongation at 72°C for 7 min. The PCR product, representing the entire coding sequence of the S. cerevisiae LTA 4 hydrolase, was digested with SalI and SphI, purified by low melting agarose (0.7%) electrophoresis, and cloned into an expression vector.
Expression of S. cerevisiae LTA 4 Hydrolase in Escherichia coli-For expression in E. coli, the plasmid pT3-12LO was used (22). The 12lipoxygenase insert was removed by cleavage with SalI and SphI, and the opened vector was purified by agarose gel electrophoresis. The PCR product was ligated into the plasmid to produce pT3-yLTAh, which was transformed into competent E. coli (JM101) cells. Expression and preparation of crude protein extracts was performed as described previously (23). The proteins were precipitated with 80% ammonium sulfate and stored at 4°C until further purification.
Expression of S. cerevisiae LTA 4 Hydrolase in Spodoptera frugiperda (Sf9) cells-Expression of the S. cerevisiae LTA 4 hydrolase gene was performed using the Bac-to-Bac® baculovirus expression system (Life Technologies, Inc.). The cDNA was removed from pT3-yLTAh by cleavage with SalI and SphI, purified by agarose gel electrophoresis, and ligated into the recombinant donor plasmid pFastBac1 opened with SalI and SphI. This plasmid, carrying the S. cerevisiae LTA 4 hydrolase gene, was in turn transformed into competent E. coli cells (JM101). Correctly ligated plasmids, selected by PCR and restriction enzyme cleavage, were purified using the Wizard® Plus minipreps DNA purification system (Promega, Madison, WI) and transformed into competent DH10Bac cells (Life Technologies, Inc.) containing the bacmid DNA. The preparation of the bacmid DNA, transfection of Sf9 cells, and harvesting of baculovirus were performed as described in the instruction manual of the Bac-to-Bac® baculovirus expression system. The S. cerevisiae LTA 4 hydrolase cDNA was sequenced by the dideoxy chain termination method to verify that no changes in the nucleotide sequence had occurred during the cloning procedure (24).
For protein expression, the Sf9 cells were grown to a density of 1.5-2 ϫ 10 6 cells/ml in Sf-900 II serum-free medium (Life Technologies, Inc.) and transfected with baculovirus containing the S. cerevisiae LTA 4 hydrolase. The infected cells were grown for approximately 72 h, pelleted by centrifugation at 600 ϫ g for 10 min, and resuspended in lysis buffer (50 mM Tris-Cl, pH 8.5, 5 mM ␤-mercaptoethanol, 100 mM KCl, 1 mM phenylmethylsulfonyl fluoride, 1% Nonidet P-40). The cells were disrupted by sonication (3 ϫ 10 s) and centrifuged at 10,000 ϫ g for 10 min. The resulting supernatant was centrifuged at 100,000 ϫ g for 1 h to remove membranes and subjected to streptomycin sulfate precipitation. The proteins were precipitated with 80% ammonium sulfate and stored at 4°C until purification.
Protein Purification-The following protocol was used for purification of recombinant enzymes from both E. coli and Sf9 cells. For anionexchange chromatography on fast protein liquid chromatography (Am-ersham Pharmacia Biotech), a Mono-Q HR10/10 column, or a column packed with Q Sepharose Fast Flow, was equilibrated with 10 mM Tris-Cl, pH 8. The ammonium sulfate precipitate was desalted by gel filtration (PD10 column; Amersham Pharmacia Biotech) and applied to the column. Adsorbed proteins were eluted with a linear gradient of KCl (0 -500 mM), and the enzyme activity was recovered at approximately 150 mM. For hydrophobic interaction chromatography, active fractions were supplemented with 20% (w/v) ammonium sulfate and applied to a Phenyl Superose HR 5/5 column (Amersham Pharmacia Biotech) equilibrated with 20 mM Tris-Cl, pH 8, containing 1.5 M ammonium sulfate. A linear gradient of decreasing ammonium sulfate (1.5-0 M) was applied and the enzyme activity was eluted at approximately 660 mM salt. For hydroxyapatite chromatography, a TSKgel HA-1000 column (Tosohaas) was equilibrated in 10 mM potassium phosphate buffer, pH 7.1, supplemented with 0.2 mM CaCl 2 . Pooled active fractions from the previous step were applied in the column equilibration buffer. A linear gradient of increasing phosphate (10 -400 mM) was developed and active fractions were eluted at approximately 220 mM. For chromatofocusing, we used a Mono-P HR 5/20 column (Amersham Pharmacia Biotech) preequilibrated with 25 mM Bis Tris, pH adjusted to 7.0 with iminodiacetic acid. Active fractions from the hydroxyapatite chromatography in 10 mM Tris-Cl, pH 8, were applied to the column. Adsorbed proteins were eluted with a pH gradient (7.0 -4.5) by changing the buffer to Polybuffer 74 (Amersham Pharmacia Biotech), diluted 10 times in water, and pH adjusted to 4.5 with iminodiacetic acid. The pH was measured in collected fractions (500 l), and 20 l of 1 M Tris-Cl, pH 8.0, was added to restore an alkaline pH. Protein activity was recovered in fractions between pH 5.4 and 5.9. A final step of anionexchange chromatography was performed on a Mono-Q HR 5/5 column as described above and the purified enzyme eluted at 180 mM KCl. Purification of recombinant human LTA 4 hydrolase was carried out essentially as described (25).
Enzyme Activity Determinations-The peptidase activity was determined with a spectrophotometric assay in the wells of a microtiterplate, essentially as described (25). Briefly, the enzyme (0.7-1 g) was incubated with p-nitroanilide derivatives of Ala, Arg, Glu, Gly, Leu, Met, Pro, and Val at a concentration of 1 mM in 50 mM Tris-Cl, pH 7.5, containing 100 mM KCl. Formation of the product (p-nitroaniline) was measured at room temperature as the increase in A 405 using a multiscan spectrophotometer, MCC/340 (Labsystems).
For determination of the epoxide hydrolase activity, incubations were performed with enzyme (2-10 g) in 100 l of 10 mM Tris-Cl, pH 8, with 30 M LTA 4 for 60 s at room temperature. The reaction was stopped with 1-2 volumes of MeOH, and 400 -700 pmol of prostaglandin B 1 was added as an internal standard. The samples were subjected to solid-phase extraction (Chromabond C 18 EC, Macherey Nagel) and finally analyzed by reverse-phase high performance liquid chromatography (HPLC). The column (Nova-Pak C 18 , 4 m Radial-Pak cartridge, 5 ϫ 100 mm, Waters) was eluted with a mixture of acetonitrile/ methanol/water/acetic acid (30:35:35:0.01, by volume) at a flow rate of 1.0 ml/min. The absorbance of the eluate was monitored at 270 nm. The products were identified by their chromatographic mobility relative standard compounds, as well as by UV spectrophotometry.
Protein Analyses-Protein concentrations were determined according to the method of Bradford, using the Bio-Rad protein-assay reagent and bovine serum albumin as standard. SDS-polyacrylamide gel electrophoresis (PAGE) was performed on a Phast system (Amersham Pharmacia Biotech) using 10 -15% gradient gels. Bands of protein were visualized by staining with Coomassie Brilliant Blue. N-terminal amino acid sequencing was carried out with an Applied Biosystems 477A instrument. Phenylthiohydantoin derivatives were analyzed with online HPLC systems (Protein Analysis Center, Karolinska Institutet).

Cloning, Expression, and Purification of S. cerevisiae LTA 4
Hydrolase-The LTA 4 hydrolase homologue was cloned by PCR from reverse transcribed total RNA isolated from S. cerevisiae (w303-1b), using two primers raised against the genomic sequence. The cDNA (2016 bp) was initially expressed in E. coli (JM101) and purified to apparent homogeneity in five chromatographic steps by fast protein liquid chromatography, using anion exchange, hydroxyapatite, hydrophobic interaction, and chromatofocusing resins. The level of expression was very low, and from 6 liters of medium, only 50 g of protein was recovered. However, the identity of the isolated protein could be established by N-terminal amino acid sequencing, which also showed that the recombinant protein lacked the initial 40 amino acids (cf. Fig. 1).
To increase the yield of recombinant protein, the S. cerevisiae LTA 4 hydrolase was also expressed in Sf9 cells, using the baculovirus system, and was recovered in the cell pellet rather than in the medium. The protein was purified to homogeneity using the same procedure as described above, and the molecular mass was determined to 72 kDa by SDS-PAGE (Fig. 2). Typically, from 450 ml of infected Sf9 cell culture, 500 g of purified enzyme was recovered, and unless otherwise stated, this preparation was used for the enzyme characterization. During the chromatofocusing step, four active peaks were eluted at pH 5.9, 5.7, 5.6 and 5.4. The material under the three large peaks (I, III, and IV) were collected separately and either treated as different variants of the protein or pooled together. The molecular mass of the four variants did not differ, as judged by SDS-PAGE (Fig. 2). N-terminal amino acid sequencing was performed on the two major components (peaks I and IV), which revealed that they were indeed the expected protein but with different N termini. Thus, the first detected residue in variant I was Ile-32, whereas the other protein (variant IV) started at Met-40 (Fig. 1).
Aminopeptidase Activity-The aminopeptidase activity of the S. cerevisiae LTA 4 hydrolase was assayed using p-NA derivatives of different amino acids. For all variants tested (I, III and IV), the highest specific activity (190, 140, and 210 nmol/ mg/min, respectively), under standard assay conditions, was obtained with Leu-p-NA. For I and III, Met-p-NA was the second best substrate (100 and 82 nmol/mg/min) followed by Ala-(58 and 52 nmol/mg/min), and Val-p-NA (19 and 16 nmol/ mg/min). For variant IV, however, the second best substrate was Pro-p-NA (200 nmol/mg/min), followed by Met-(114 nmol/ mg/min) and Ala-p-NA (110 nmol/mg/min). Arg-, Glu-, and Gly-p-NA were very poor substrates for all three variants of the S. cerevisiae LTA 4 hydrolase. The protein obtained from E. coli displayed an almost identical substrate specificity with Leup-NA as the preferred substrate, followed by Met-p-NA and Ala-p-NA.
Kinetic Constants-Apparent kinetic constants were determined using Leu-, Met-, and Ala-p-NA as substrate in 50 mM phosphate buffer, pH 7.5, containing 100 mM KCl and did not differ significantly between the three variants of the protein.
Using Leu-p-NA as the substrate, the mean value of V max for all three variants was calculated to 520 Ϯ 110 nmol/mg/min (mean ϮS.D.; n ϭ 6), and the value for K m was determined to 1.5 Ϯ 0.4 mM (n ϭ 6) ( Table I). The values of V max , using Metand Ala-p-NA as substrates, were calculated to 360 Ϯ 110 (n ϭ 3) and 240 Ϯ 170 nmol/mg/min (n ϭ 3), respectively, whereas the corresponding values of K m were determined to be 1.8 Ϯ 0.45 (n ϭ 3) and 2.0 Ϯ 1.0 mM (n ϭ 3), respectively. As judged by values of k cat /K m , Leu-p-NA was the best substrate (420 Ϯ 100 s Ϫ1 M Ϫ1 ; n ϭ 6) followed by Met-pNA (230 Ϯ 10 s Ϫ1 M Ϫ1 , n ϭ 3) and Ala-p-NA (130 Ϯ 30 s Ϫ1 M Ϫ1 ; n ϭ 3). In all subsequent experiments, Leu-p-NA was used as the standard substrate. To allow a direct comparison, recombinant human LTA 4 hydrolase (0.7 g) was assayed in a parallel experiment with Leu-p-NA, and the kinetic constants V max and K m were calculated to 310 nmol/mg/min and 0.15 mM, respectively (Table I).
Effects of pH and Anions-The pH optimum for the aminopeptidase activity of the S. cerevisiae LTA 4 hydrolase (1 g of enzyme, 1 mM Leu-p-NA) in both Tris-Cl and phosphate buffer containing 100 mM KCl, was determined to 7.3 (data not shown). Furthermore, the enzyme was dose-dependently stimulated by monovalent anions. As described previously for LTA 4 hydrolase, thiocyanate was a very potent activator of all three  1. A multiple alignment of amino acid sequences of the human LTA 4 hydrolase, the S. cerevisiae LTA 4 hydrolase, and the C. elegans AP1. The alignment was made using the GAP alignment program in the GCG package (30). The zinc binding ligands are indicated by shaded boxes. Amino acids that have been shown to be important for the peptidase activity of human LTA 4 hydrolase, i.e. Glu-296 and Tyr-383, are indicated by open boxes (31,32). Tyr-378, the residue in human LTA 4 hydrolase to which LTA 4 binds covalently during suicide inactivation, is indicated by an asterisk and is replaced by a phenylalanine in both S. cerevisiae LTA 4 hydrolase (Phe-424) and C. elegans AP1 (Phe-382). I and IV designate the N termini of variants I and IV, respectively. The E. coli protein has the same N terminus as variant IV. protein variants. The effect peaked at approximately 100 mM, corresponding to Ͼ100-fold stimulation. At higher concentrations (Ͼ100 mM), this anion had an inhibitory effect on the aminopeptidase activity. For chloride ions, the pattern of stimulation seemed to obey saturation kinetics. At 100 mM, a 40fold activation was observed, and a plateau was reached at approximately 600 mM salt. From Eadie-Hofstee plots of the kinetic data, an apparent affinity constant (K A ) for chloride was calculated to 120 mM. Similarly, the apparent affinity constant for thiocyanate (in the range 25-100 mM) was calculated to 28 mM.
Effects of Inhibitors-Three competitive inhibitors of LTA 4 hydrolase, i.e. the general aminopeptidase inhibitor bestatin and two specific LTA 4 hydrolase inhibitors, a thioamine and a hydroxamic acid, were tested for their inhibitory effect on the S. cerevisiae LTA 4 hydrolase. The inhibitors were tested at concentrations ranging between 0.01 and 100 M, using Leup-NA as substrate. For all three variants of the protein, the thioamine was shown to be the most potent inhibitor with an IC 50 of 0.06 Ϯ 0.01 M (mean Ϯ S.D.; n ϭ 4). The hydroxamic acid was the second best inhibitor, with an IC 50 of 0.24 Ϯ 0.09 M (n ϭ 4), whereas bestatin was a relatively poor inhibitor of this protein, with an IC 50 of 3.2 Ϯ 0.6 M (n ϭ 3). Parallel experiments were performed with recombinant human LTA 4 hydrolase, using Leu-p-NA as substrate, which showed a similar pattern of inhibition with IC 50 values for the three inhibitors of 0.02, 0.03, and 0.5 M for the thioamine, the hydroxamic acid, and bestatin, respectively.
The mode of inhibition for bestatin and the thioamine was determined from Eadie-Hofstee plots of kinetic data obtained with untreated and inhibited enzyme. For both compounds, a mixed type of inhibition was observed, suggesting that the inhibitors bind at the active site but can not be fully displaced by the substrate (Fig. 3).
Epoxide Hydrolase Activity-The S. cerevisiae LTA 4 hydrolase was assayed for its ability to convert LTA 4 into LTB 4 . Aliquots of a pool of all three protein variants (2-10 g) were incubated in 100 l of 10 mM Tris-Cl, pH 8, with 30 M LTA 4 for 60 s at room temperature. Under these experimental conditions, production of LTB 4 could not be detected in any of the incubations with the S. cerevisiae protein. Similar incubations were performed with the purified recombinant protein from E. coli, as well as crude yeast cell extracts, none of which generated detectable amounts of LTB 4 .
Further analysis of the product profile revealed that the S. cerevisiae LTA 4 hydrolase converted LTA 4 into a less polar compound eluting late in the reverse phase HPLC chromatogram (Fig. 4). Comparison with synthetic standards showed that this peak coeluted with 5S,6S-dihydroxy-7,9-trans-11,14cis-eicosatetraenoic acid (5S,6S-DHETE). Furthermore, the material under this peak exhibited a UV spectrum typical of a conjugated triene moiety, with max at 274 nm, in agreement with previously published data for 5S,6S-DHETE (26). By peak area measurements, the specific activity of this activity was calculated to 20 Ϯ 2.1 nmol/mg/min (mean Ϯ S.D.; n ϭ 6) using prostaglandin B 1 as the internal standard. Heating the S. cerevisiae LTA 4 hydrolase at 90°C for 5 min completely abolished the formation of 5S,6S-DHETE demonstrating the enzymatic nature of its formation. Moreover, preincubation of the enzyme with the competitive inhibitor thioamine (30 M) also abolished this enzyme activity (Fig. 4), suggesting that epoxide hydrolysis occurs at a site close to or identical to the aminopeptidase active site.
Effects of LTA 4 on the Peptidase Activity-To test whether the peptidase activity was inactivated by LTA 4 , as is the case with all LTA 4 hydrolases described thus far (27), the S. cerevisiae protein (0.7 g) was incubated with LTA 4 added three times at 10-or 20-min intervals, to a final concentration of 80 M. Subsequent measurement of the enzyme activity, using 1 mM Leu-p-NA as substrate, revealed that LTA 4 stimulated the specific peptidase activity from 150 to 1800 nmol/mg/min, corresponding to an 11-fold stimulation. When Ala-, Arg-, Pro-, and Met-p-NA were tested, Ala-p-NA was the only other substrate that was hydrolyzed with increased efficiency. Thus, the specific activity was increased from 120 to 500 nmol/mg/min, corresponding to a 4-fold stimulation.
The stimulatory effect was not reversible, and 70% of the initial effect was still present in a sample stored for a month in the refrigerator. Furthermore, stimulated enzyme could be chromatographed on a Mono-Q column without significant loss of activity, indicating a tight binding between lipid and protein (data not shown).
Further kinetic experiments showed that the stimulation was both time-and dose-dependent (Fig. 5). Hence, addition of 30 M LTA 4 to a standard incubation of enzyme and substrate (1 g of protein, 1 mM Leu-p-NA) caused a rapid increase in reaction velocity over the first 5 min of incubation and then gradually leveled off during the following 15 min (Fig. 5A). When the S. cerevisiae enzyme was preincubated with increasing concentrations of LTA 4 , the peptidase activity toward Leup-NA and Ala-p-NA was stimulated in a dose-dependent fash-  ion, reaching a maximum at about 80 and 40 M, respectively (Fig. 5B). Moreover, the stimulatory effect appeared to obey saturation kinetics, and when the data were plotted according to the Eadie-Hofstee method, linear relationships (r 2 ϭ 0.83 and 0.94) were observed. From the slope of the lines, apparent affinity constants (K A ) were calculated to 19 and 6 M for Leuand Ala-p-NA, respectively. From the same plots, maximal catalytic efficiencies (at 1 mM substrate) were calculated to 2200 and 450 nmol/mg/min for Leu-and Ala-p-NA, respectively.
The apparent kinetic constants for the stimulated S. cerevisiae LTA 4 hydrolase (0.7 g of protein; 70 M LTA 4 ) were determined for Leu-p-NA as substrate in 50 mM phosphate buffer, pH 7.5, containing 100 mM KCl. Thus, values of both K m and V max had increased to 2.1 mM and 4400 nmol/mg/min, respectively. Consequently, the value of k cat /K m increased from 430 to 2400 s Ϫ1 ϫ M Ϫ1 (Table I).
To study the specificity of the stimulatory effect, the S. cerevisiae enzyme (0.7 g) was incubated with LTB 4 (70 M), arachidonic acid (60 M), or phosphatidylcholine (80 g/ml) prior to addition of the substrate, Leu-p-NA. None of these lipids had any significant stimulatory effect on the aminopeptidase activity. Furthermore, incubations were performed in which LTA 4 was allowed to undergo spontaneous hydrolysis before addition of the enzyme. Under these conditions, no activation of the peptidolysis could be observed, indicating that the nonenzymatic hydrolysis products of LTA 4 , primarily 6-trans-LTB 4 and 12-epi-6-trans-LTB 4 , are without effect and that the intact epoxide moiety of LTA 4 is necessary for the stimulation. This result also shows that other minor components of the substrate mix, e.g. small amounts of LiOH, have no stimulatory effect. For comparison, similar experiments were performed in parallel with recombinant human LTA 4 hydrolase, which, as expected, was inhibited (65-70%) by 80 M LTA 4 .
Competition between LTA 4 and Inhibitors-The S. cerevisiae protein was preincubated with 0.06 M thioamine or 4 M bestatin, corresponding to 60% inhibition, and then stimulated with increasing doses of LTA 4 . Again, the aminopeptidase activity was stimulated in a saturable manner, and from a linear plot of the kinetic data K A for LTA 4 was calculated to be approximately 70 and 40 M in the presence of the thioamine and bestatin, respectively (Fig. 6A). Interestingly, LTA 4 could displace the thioamine completely, and a maximal catalytic efficiency (1 mM Leu-p-NA) was calculated to 2.0 mol/mg/min, very close to the value for enzyme not treated with the inhibitor (2.2 mol/mg/min). In contrast, LTA 4 competed only partially with bestatin, and the maximal catalytic efficiency for the inhibited protein reached 1.1 mol/mg/min (Fig. 6A). These differences between the two inhibitors regarding their competition with LTA 4 may be explained by their structures and proposed binding modes to the active sites of LTA 4 hydrolase (Fig. 6B).
In an opposite experiment, the thioamine was tested for its ability to inhibit the S. cerevisiae enzyme prestimulated with LTA 4 . Thus, the S. cerevisiae protein (10 g in 2 ml of 25 mM Hepes buffer, pH 7.5, supplemented with 100 mM KCl) was preincubated with 100 M LTA 4 , added in two doses at 10-min intervals, before the inhibitor was added. In the presence of LTA 4 , the potency of the inhibitor was drastically reduced (approximately 500 times) with an IC 50 of 25 M (results not shown), in agreement with a tight binding between LTA 4 and the protein. DISCUSSION LTA 4 hydrolase has been characterized from several mammals, but little, if anything, is known about its evolution and properties in lower animal species. The enzyme is distantly related to many bacterial and yeast proteases and aminopeptidases, e.g. aminopeptidase N and thermolysin, by virtue of its zinc binding site and aminopeptidase activity. Hence, LTA 4 hydrolase has been classified as a member of the M1 family of metallopeptidases (11). However, the degree of identity or similarity at the amino acid level between LTA 4 hydrolase and other members of this family is usually low and confined to the zinc binding region. Using bioinformatics, several proteins with higher degree of homology have been identified. Thus, a protein with 45% identity (63% similarity) to LTA 4 hydrolase was found in C. elegans (17). However, expression and characterization of this gene revealed that it was yet another aminopeptidase without functional links to LTA 4 hydrolase.
In the present work, we have cloned and characterized another homologue of LTA 4 hydrolase present in S. cerevisiae, which was discovered serendipitously via the sequencing of the yeast genome (18). The S. cerevisiae enzyme was initially expressed in E. coli, but the recovery of recombinant protein was very poor. Edman degradation showed that it lacked the initial 40 amino acids at the N terminus, suggesting that translation is initiated at the second methionine (Met-40) of the open reading frame or that the protein undergoes N-terminal proteolytic processing. To get a higher yield of protein we turned to expression in a baculovirus insect cell (Sf9) system that generated four different products in the final step of chromatofocusing chromatography. The molecular masses of these four protein species were indistinguishable by SDS-PAGE (Fig. 2), and all exhibited similar catalytic properties. However, upon Edman degradation, two of the variants differed in their N termini, again suggesting different translational initiation sites or N-terminal proteolytic processing (Fig. 1). The S. cerevisiae protein is larger than the human and C. elegans proteins. It contains 672 amino acids and has an N-terminal extension of about 50 residues. In view of the sequencing data, it is tempting to speculate that a segment of the N terminus is not present in the mature S. cerevisiae protein.
The S. cerevisiae LTA 4 Hydrolase Is an Anion-stimulated Leucyl Aminopeptidase-The S. cerevisiae protein is an aminopeptidase, as assessed by enzyme activity determinations with synthetic chromogenic substrates of recombinant proteins from both E. coli and Sf9 cells. The best substrate was Leu-p-NA, a substrate specificity different from that of human LTA 4 hydrolase, which prefers Arg-or Ala-p-NA (28). As previously reported for LTA 4 hydrolase, the aminopeptidase activity of the S. cerevisiae homologue was greatly stimulated by thiocyanate and chloride in an allosteric fashion, with apparent affinity constants of approximately 28 and 120 mM, respectively. These data are in good agreement with the corresponding values for human LTA 4 hydrolase (28).
The S. cerevisiae LTA 4 Hydrolase Can Convert LTA 4 into 5S,6S-DHETE-The S. cerevisiae enzyme was tested for LTA 4 hydrolase activity. Under our experimental conditions (10 g of enzyme, 30 M LTA 4 ), we failed to observe any significant conversion of LTA 4 into LTB 4 . Interestingly, the S. cerevisiae enzyme was capable of hydrolyzing LTA 4 into another product, identified as 5S,6S-DHETE by its chromatographic mobility in reverse-phase HPLC and UV spectrometry (Fig. 4). Notably, enzymatic formation of this particular metabolite of LTA 4 has FIG. 6. A, LTA 4 activation of S. cerevisiae LTA 4 hydrolase preinhibited by thioamine or bestatin. The enzyme (0.7 g) was pretreated with 0.06 M thioamine or 4 M bestatin for 30 min. Untreated enzyme (E) and enzyme inhibited with thioamine (OE) or bestatin (q) were stimulated with different doses of LTA 4 (10 -160 M), prior to incubation with 1 mM Leu-p-NA in 50 mM Tris-Cl, pH 7.5, containing 100 mM KCl. The dependence of the initial reaction velocity on the concentration of LTA 4 was plotted in a linear fashion according to the Eadie-Hofstee method. Each data point represents mean of duplicate samples. B, model for the binding of bestatin and thioamine to the active site of LTA 4 hydrolase. The leucyl moiety of bestatin interacts with a peptide recognition site and the catalytic metal (zinc in mammalian enzymes), whereas the bicyclic portion of the thioamine is thought to interact primarily with a hydrophobic pocket accommodating the fatty acid backbone of LTA 4 . previously only been described for a mutated form of LTA 4 hydrolase (26). Thus, a point mutation of Tyr-383 in human LTA 4 hydrolase generates a recombinant enzyme capable of converting LTA 4 into large amounts of 5S,6S-DHETE. In the S. cerevisiae protein, the corresponding amino acid is a tyrosine (Tyr-429) (Fig. 1). Nevertheless, this epoxide hydrolase activity and ability to turn over LTA 4 into 5S,6S-DHETE represents a functional link between the evolutionary distant S. cerevisiae enzyme and mammalian LTA 4 hydrolases.
LTA 4 Stimulates the Aminopeptidase Activity of S. cerevisiae LTA 4 Hydrolase-We also wanted to see whether the yeast homologue could be inhibited and covalently modified by LTA 4 , as is the case with all LTA 4 hydrolases described thus far. To our surprise, we found that the aminopeptidase activity was time-and dose-dependently stimulated by LTA 4 in a fashion suggesting the presence of a lipid binding site. An apparent K A for LTA 4 was calculated as 18 and 6 M using Leu-and Ala-p-NA, respectively. The stimulatory effect was specific for LTA 4 , because LTB 4 , nonenzymatic hydrolysis products of LTA 4 , arachidonic acid, or phosphatidylcholine had no effect. Several lines of evidence indicated that LTA 4 binds tightly to the S. cerevisiae protein, although we could not demonstrate covalent bond formation. In this context, it is interesting to note that during suicide inactivation, LTA 4 binds covalently to the phenolic hydroxyl group of Tyr-378 in the mammalian LTA 4 hydrolase (29), which in the S. cerevisiae homologue corresponds to Phe-424 (Fig. 1).
LTA 4 Stimulation Occurs via a Lipid Binding Pocket Located at the Active Center of the Enzyme-Further kinetic studies with active site-directed inhibitors allowed us to locate the LTA 4 binding site. Thus, stimulation with LTA 4 competed with bestatin and the thioamine (Fig. 6A). In the case of bestatin, the competition with LTA 4 was partial, whereas for the thioamine it was complete. This discrepancy between the two inhibitors agrees well with their proposed binding modes to the active site of LTA 4 hydrolase (Fig. 6B). Thus, bestatin is a derivative of leucine and is expected to occupy the peptide binding site, whereas the thioamine carries a bicyclic hydrophobic structure believed to mimic the fatty acid backbone of LTA 4 . Furthermore, the K A values for LTA 4 were substrate-dependent and differed between Leu-and Ala-p-NA, suggesting some competition or interaction between the activator LTA 4 and the aminopeptidase substrates at the active site (Fig. 5B). In addition, the site for LTA 4 stimulation could be linked to the site for epoxide hydrolysis because the conversion of LTA 4 into 5S,6S-DHETE could be blocked with the thioamine inhibitor (Fig. 4), which in turn could be displaced by LTA 4 . Hence, the binding site for LTA 4 appears to be located at the active center of the enzyme, presumably overlapping with the site(s) responsible for peptide and epoxide hydrolysis. A model for the active center of the S. cerevisiae LTA 4 hydrolase and its relation to the active center of the mammalian LTA 4 hydrolase is depicted in Fig. 7. To the best of our knowledge, leukotriene biosynthesis and formation of LTA 4 has never been described in yeast. Therefore, it appears likely that the LTA 4 binding site and/or the site for epoxide hydrolysis in the S. cerevisiae enzyme accommodates some other lipid substrate structurally related to LTA 4 . Nevertheless, the S. cerevisiae homologue appears to be an early ancestral gene to LTA 4 hydrolase, thus sharing its unique ability to combine lipid and peptide metabolism in a single protein.