Purification and molecular cloning of an 8R-lipoxygenase from the coral Plexaura homomalla reveal the related primary structures of R- and S-lipoxygenases.

Lipoxygenases that form S configuration fatty acid hydroperoxides have been purified or cloned from plant and mammalian sources. Our objectives were to characterize one of the lipoxygenases with R stereospecificity, many of which are described in marine and freshwater invertebrates. Characterization of the primary structure of an R-specific enzyme should help provide a new perspective to consider the enzyme-substrate interactions that are the basis of the specificity of all lipoxygenases. We purified an 8R-lipoxygenase of the prostaglandin-containing coral Plexaura homomalla by cation and anion exchange chromatography. This yielded a colorless enzyme preparation, a band of ∼100 kDa on SDS-polyacrylamide gel electrophoresis, and turnover numbers of 4000 min−1 of 8R-lipoxygenase activity in peak chromatographic fractions. The full-length cDNA was cloned by PCR using peptide sequence from the purified protein and by 5′- and 3′-rapid amplification of cDNA ends. The cDNA encodes a polypeptide of 715 amino acids, including over 70 amino acids identified by peptide microsequencing. A peptide presequence of 52 amino acids is cleaved to give the mature protein of 76 kDa; the difference from the estimated size by SDS-PAGE implies a post-translational modification of the P. homomalla enzyme. All of the iron-binding histidines of S-lipoxygenases are conserved in the 8R-lipoxygenase. However, the C-terminal amino acid is a threonine, as opposed to the isoleucine that provides the carboxylate ligand to the iron in all known S-lipoxygenases. These results establish that the 8R-lipoxygenase is related in primary structure to the S-lipoxygenases. A model of the basis of R and S stereospecificity is described.

Lipoxygenases are nonheme iron dioxygenases that catalyze the oxygenation of polyunsaturated fatty acids to specific hydroperoxide products (1,2). The enzymes occur widely in plants and animals, where they function in the biosynthesis of signaling molecules and bioactive mediators (1)(2)(3)(4). Plant and mammalian lipoxygenases are related in primary structure and contain certain absolutely conserved amino acids that are critical for catalytic activity (5). From the crystal structures of the soybean lipoxygenases L-1 and L-2 it is known that three histidine residues and the C-terminal carboxyl of the protein (invariably an isoleucine) are ligands of the active site iron (6 -9). One of the soybean lipoxygenase L-1 crystal structures has an additional iron ligand (7), represented as an asparagine in plants and an asparagine or histidine in the mammalian enzymes. These iron ligands are among the conserved residues of all lipoxygenases and, as indicated by site-directed mutagenesis, are essential to the function of these enzymes (10 -12).
All the lipoxygenases characterized so far form hydroperoxides of the S stereoconfiguration. There exists, however, a group of enzymes in several species of invertebrate that catalyze the oxygenation of their polyunsaturated fatty acid substrates with R stereochemistry. These R lipoxygenases are reported in coral, sea urchin eggs, oocytes of starfish and clams, 1 crabs, barnacles, and marine and freshwater hydroids (13-18, 20 -23). They catalyze oxygenation in the 5R, 8R, 11R, or 12R configurations. These enzymes, in crude cell extracts, show the expected features of a typical lipoxygenase family member: (i) the primary products are specific hydroperoxides; (ii) the catalysis proceeds with an initial hydrogen abstraction (reflected in a strong primary isotope effect when the hydrogen is replaced with tritium) (17,24); and (iii) there is an antarafacial relationship between the hydrogen abstraction and the insertion of molecular oxygen (17,24). Notwithstanding this evidence, none of the R-specific enzymes have been purified, and there is no evidence to establish whether they constitute related family members or a completely different group of enzymes. The mammalian cyclooxygenases, as an example, can show similar catalytic features to lipoxygenases in crude extracts, yet these are hemoproteins and completely unrelated in structure.
One source of R-lipoxygenase, the coral Plexaura homomalla, has a distinguished history in the eicosanoid field. It contains 2-3% by weight of prostaglandin esters and for many years served as a commercial source of prostaglandins (25,26). In 1987, we reported that extracts of P. homomalla avidly metabolize arachidonic acid to the 8R-hydroperoxide (8R-HPETE) 2 (16). Based on preliminary investigations of the feasibility of purifying an R-lipoxygenase from a number of marine sources, we selected P. homomalla as a particularly rich source of enzyme. Here we report the purification, peptide microsequencing, and molecular cloning of a P. homomalla 8R-lipoxygenase. The results contribute toward our long term goal of an understanding of the basis of specificity of the S-and R-lipoxygenase enzymes. * This work was supported by National Institutes of Health Grants GM-15431 and GM-49502 and the Prostaglandin Core and Protein Chemistry Core Laboratories of Grant HD-05797. 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) U59223.
‡ To whom correspondence should be addressed.  (27), and 8R-HETE was synthesized using the 8R-lipoxygenase of P. homomalla (16). P. homomalla was collected in the Florida Keys and immediately placed on dry ice and stored at Ϫ80°C.

8R-Lipoxygenase Purification Protocol
The 8R-lipoxygenase activity was solubilized from 2 g of coral acetone powder (equivalent to about 10 -20 g of coral) by stirring for 30 min at 4°C in 200 ml of 40 mM Tris, 2 mM EGTA, pH 8. The suspension was then centrifuged for 30 min at 10,000 ϫ g at 4°C, the supernatant was collected, and Emulgen 911 (E911) detergent was added to a final concentration of 0.1%. After stirring for 30 min at 4°C, the blackcolored solution was loaded on an open bed Q-Sepharose column (10 ϫ 2.5 cm, 50-ml bed volume) equilibrated in 40 mM Tris, 2 mM EGTA, pH 8, with 0.1% E911. The column was washed with loading buffer until the eluant was almost colorless (ϳ200 ml), and then the lipoxygenase activity was eluted with the same buffer containing 0.4 M NaCl. Fractions of 10 ml were collected and assayed for lipoxygenase activity by measurement of the rate of increase in absorbance at 235 nm on conversion of arachidonic acid to 8R-HPETE; an aliquot of each fraction (50 l) was diluted to 1 ml with 40 mM Tris, pH 8, and then arachidonic acid (10 g in 2 l of ethanol) was added, and the increase in absorbance at 235 nm was recorded for 1-2 min. Protein was assayed with the BCA (bicinchoninic acid) protein assay (Pierce) against standard curves of albumin in equivalent concentrations of E911 detergent).
Fractions with lipoxygenase activity were pooled (100 -150 ml total) and dialyzed for 4 -5 h with 2 liters of 2.5 mM sodium phosphate, pH 7, 0.1% E911 and then overnight using another 2 liters of fresh dialysis buffer. The sample was then loaded onto an open bed column of 10 ml of hydroxyapatite Bio-Gel-HT (Bio-Rad) and eluted with 100 mM sodium phosphate, pH 7, 0.1% E911. Fractions of 5 ml were collected, and active fractions were pooled (one or two fractions). Immediately prior to cation exchange chromatography, the sample was acidified to pH 5 by careful addition of 4 N phosphoric acid (ϳ2.5 l/ml of sample) and then injected on a Mono-S HR 5/5 column (Pharmacia Biotech Inc.) equilibrated in 50 mM sodium phosphate (pH 5.0), 0.1% E911. The column was eluted with equilibration buffer for 15 min at a flow rate of 0.9 ml/min (pressure limitations), and then programmed using a linear gradient to 1.3 M NaCl, 50 mM sodium phosphate, 0.5% E911 over 45 min with on-line UV detection at 280 nm. Fractions of 1 min were collected in tubes containing 0.1 ml of 1 M Tris pH 8, and aliquots were assayed for lipoxygenase using the UV assay. Fractions containing lipoxygenase activity were pooled and dialyzed overnight with 2 liters of 40 mM Tris (pH 8.0), 2 mM EGTA, and 0.1% E911.
The sample was then loaded on a Mono-Q HR 5/5 column (Pharmacia) equilibrated in 40 mM Tris, pH 8.0, 2 mM EGTA, and 0.1% E911. The column was eluted with equilibration buffer for 20 min at 0.75 ml/min and then programmed from 20 to 60 min with a linear gradient to 0.4 M NaCl, 40 mM Tris, pH 8.0, 2 mM EGTA, and 0.3% E911. To remove the E911 detergent, active fractions were pooled, dialyzed into 2.5 mM sodium phosphate (pH 7), and loaded on a hydroxyapatite Biogel-HT column (5 ϫ 0.5 cm, 1 ml). After washing with equilibration buffer (2.5 mM sodium phosphate) to remove the detergent, the buffer was changed to 200 mM sodium phosphate, and the active lipoxygenase was collected in 1-2 ml.
The 8R-lipoxygenase was further purified in denatured form by RP-HPLC on a C4 Vydac column (25 ϫ 0.46 cm) using a water/acetonitrile gradient in the presence of 0.1% trifluoroacetic acid. The proteins were detected by on-line UV monitoring at 214 nm, and aliquots of the UV-absorbing fractions were analyzed by SDS-PAGE to identify the ϳ100-kDa protein; it represented the main UV-absorbing peak by RP-HPLC. Semipreparative SDS-PAGE was used as an alternative method for further purification of the ϳ100-kDa protein.

CNBr Cleavage/HPLC of Peptides
The purified 8R-lipoxygenase was taken to dryness and redissolved in 100 l of 70% formic acid, and one or two crystals of cyanogen bromide were added. The reaction was allowed to proceed for 24 h in the dark at room temperature under nitrogen. Water (1 ml) was added, and the sample was taken to dryness under vacuum using a Speedvac (Savant). To eliminate poor chromatographic performance attributed to formylation of peptides (28), the sample was treated with 20 l of ethanolamine for 5 min at room temperature and then evaporated again to dryness. The peptides were then dissolved in 5 M guanidine, 0.1 M Tris, pH 8.5, 0.1 mM dithiothreitol, warmed for 5 min at 50°C to reduce dithiols, and then separated by RP-HPLC using a C4 Vydac column (25 ϫ 0.46 cm) and a water/acetonitrile gradient in the presence of 0.1% trifluoroacetic acid.

HPLC Analysis of Lipoxygenase Metabolism
The lipoxygenase metabolism of [1-14 C]arachidonic acid was evaluated essentially as described previously (16,24). Following incubation with substrate (50 M or 100 M [1-14 C]arachidonic acid), products were extracted using the Bligh and Dyer procedure (29), and the extracts were analyzed by RP-HPLC, SP-HPLC, and chiral column analysis. Prior to chiral column analysis, an aliquot of the sample was run on RP-HPLC, and then the main sample was injected and the products were collected. Following methylation with diazomethane and treatment with triphenylphosphine to reduce the hydroperoxides, the HETE methyl ester was collected on SP-HPLC, and then the stereochemistry was analyzed using a Chiralcel OD column (30).

Quantitation and HPLC of PCR Primers
The molar extinction coefficient of primers was calculated (31), the concentration was established by UV spectroscopy, and the working solutions were prepared at 4 M. For PCR reactions involving the production of full-length lipoxygenase clones, the PCR primers were ordered with the dimethoxytrityl protecting group on. The dimethoxytrityl-protected primers were purified by RP-HPLC using a Hamilton 5-m PRP-1 column (15 ϫ 0.41 cm) run at a 0.7 ml/min flow rate using an initial solvent of acetonitrile, 0.1 M TEAA buffer (triethylamine acetate, pH 6.5) (95:5, v/v) and programmed to acetonitrile, 0.1 M TEAA buffer in the proportions 40:60 (v/v) over 40 min at a flow rate of 0.7 ml/min. The failed sequence products (which lack a dimethoxytrityl group) eluted at 18 -22 min, and the desired product, a single main peak, eluted at 35-37 min. The main peak was collected, evaporated to approximately half volume, and then deprotected on a reversed-phase C18 Poly-Pak resin (Glen Research, Sterling, VA) by treatment on column with 2% trifluoroacetic acid according to the manufacturer's instructions. The deprotected primer was eluted with water/acetonitrile (80:20, v/v), evaporated to near dryness to remove the acetonitrile, and quantified by UV spectroscopy.

Preparation of Total RNA
In our experience it is very difficult to obtain clean RNA from P. homomalla. Simple extractions such as the Chomczynski and Sacchi method (32), or the CsCl method of Chirgwin et al. (33) give browncolored pellets that cannot be converted to cDNA. The following procedure was developed by a slight modification of methods designed for RNA preparation from "difficult" sources, bark of yew tree (of interest for taxol synthesis) (34) and marine algae (35); it gave an almost colorless pellet of RNA that was used successfully for cDNA synthesis. Approximately 10 g of P. homomalla stored at Ϫ80°C was pulverized to a fine powder in liquid nitrogen. The pulverized coral was immediately homogenized in 30 ml of lysis buffer (150 mM Tris borate, pH 7.5, 2% SDS, 1% ␤-mercaptoethanol, 50 mM EDTA, with 0.4 g of polyvinylpolypyrrolidone (Sigma)) using four short full-speed bursts with a Polytron. Absolute ethanol ( 1 ⁄4 volume) and 5 M potassium acetate ( 1 ⁄9 volume) were added, and the mixture was vortexed for 1 min. The solution was then extracted with 1 volume of chloroform/isoamyl alcohol (49:1, v/v), and the phases were separated by centrifugation at 5000 ϫ g for 10 min. The aqueous (top) phase was collected and reextracted with one volume of phenol/chloroform/isoamyl alcohol (24:23:1, by volume), and the sample was then centrifuged again at 5000 ϫ g for 10 min. The top phase was collected. Precipitation of the total RNA was carried out by addition of 1 ⁄3 volume of 12 M LiCl and ␤-mercaptoethanol (final concentration of 1%, v/v) at Ϫ20°C for 48 h. The total RNA was pelleted by centrifugation at 20,000 ϫ g for 90 min and resuspended with 4 ml of cesium chloride (400 mg/ml). Further purification of the P. homomalla total RNA was carried out by layering the sample on a cushion of cesium chloride and centrifugation at 31,000 rpm using a SW 41 rotor for 20 h at 10°C (34). The pellet of RNA was dissolved in water (0.5 ml) and quantified by UV spectroscopy. Approximately 100 g of total RNA is recovered using this protocol.
cDNA Synthesis cDNA reactions were run as described previously (36) using either random hexamer primers or an oligo-dT sequence linked at the 5Ј-end to an adaptor sequence (5Ј-ATG-AAT-TCG-GTA-CCC-GGG-ATC-C(T) 17 -3Ј). cDNA synthesis for 5Ј-RACE is described in the next section.

PCR Experiments
Initial PCR Clone-Upstream degenerate primers were based on the peptide sequences GFPAKIETI, 5Ј-GGi-TTY-CCi-GCi-AAR-ATi-GAR-ACi-AT-3Ј (where i is inosine, R is A or G, and Y is C or T), and for the second round nested PCR reaction (E/Q)DLIDVV, 5Ј-AR-GAY-YTi-ATi-GAY-GTi-GT-3Ј. Downstream primers were based on conserved sequences at the C terminus of plant lipoxygenases (discussed later): NSISI-stop, 5Ј-ACC-TCA-GAT-GGA-GAT-RCT-RTT-3Ј, and for the second round nested PCR reaction, GIPNSI, 5Ј-AT-RCT-RTT-iGG-DAT-iCC-3Ј (where D is A, G, or T). We noticed that the first of the two serines in these downstream sequences invariably is encoded by AGY (in sense code) in all known plant and mammalian lipoxygenases, thus allowing some reduction in the degeneracy of these primers. The first round PCR reaction was primed with P. homomalla cDNA prepared from 1 g of total RNA/50-l PCR reaction and using 10 mM Tris, pH 8.3, 50 mM KCl, 3 mM MgCl 2 with 0.2 mM of each dNTP and 0.25 l (1.25 units) of AmpliTaq DNA polymerase (Perkin-Elmer) in a Perkin-Elmer 480 thermocycler. After the addition of the cDNA at 80°C (hot start), the PCR was programmed in a touchdown protocol (37) as follows: 94°C for 2 min, 1 cycle; 55°C for 1 min, 72°C for 15 s, 94°C for 1 min, 3 cycles; this was repeated in another 3 cycles, except with the annealing temperature lowered by 1°C to 54°C, and so on in 1°C increments until reaching 45°C (and a total of 33 cycles). The protocol was completed with one cycle of 72°C for 10 min, and then the block temperature was held at 4°C. The second round reaction was primed with the equivalent of 0.1 l of the first round reaction products (added as a 10-fold dilution). The protocol was 94°C for 2 min, 1 cycle; 42°C for 1 min, 72°C for 15 s, 94°C for 1 min for 30 cycles; the protocol was completed with one cycle at 72°C for 10 min, and then the block temperature was held at 4°C. As a technical comment on this experiment, we note that sequencing of the 406-bp PCR product described under "Results" revealed that the most 3Ј inosine in the downstream primer used in the second round PCR appeared to be absent (for unknown reasons), and this apparently facilitated annealing of the primer designed to encode the sequence glycine-isoleucine-proline (GIP . . . ) to the actual 8R-lipoxygenase sequence encoding tryptophanmethionine-proline (3Ј-CC(missing i)-TAD-GGi . . . -5Ј, in antisense code).
Sequence to the N Terminus-Upstream degenerate primers were designed based on the N-terminal peptide sequences LAPKEEP, 5Ј-CTG-GCG-CCi-AAR-GAR-GAR-CC-3Ј and PKEEPGD, 5Ј-CCi-AAR-GAR-GAR-CCi-GGi-GA-3Ј. The downstream primer was based on part of a CNBr peptide sequence ANVGGV, 5Ј-TC-GAC-iCC-iCC-iAC-RTT-iGC-3Ј. The first round PCR used the LAPKEEP and ANVGGV primers. After a hot start at 80°C using 1 l of random primed cDNA, the PCR was programmed as follows in a touchdown protocol (37): 94°C for 2 min, 1 cycle; 60°C for 1 min, 72°C for 30 s, 94°C for 1 min, 3 cycles; this was repeated in another 3 cycles, except with the annealing temperature lowered by 1°C to 59°C, and so on in 1°C increments until reaching 50°C (and a total of 33 cycles). The protocol was completed with one cycle of 72°C for 10 min, and then the block temperature was held at 4°C. The second round PCR used the PKEEPGD upstream primer and the original downstream primer (ANVGGV); after a hot start at 80°C by spiking with the equivalent of 0.1 l of the first round PCR products, the PCR was programmed as follows: 94°C for 2 min, 1 cycle; 55°C for 1 min, 72°C for 30 s, 94°C for 1 min, 30 cycles; 72°C for 10 min, and then the block temperature was held at 4°C.
Additional Internal Sequence-The sequence between the N-terminal and C-terminal clones was completed by PCR using random primed cDNA and the upstream primer 5Ј-A-CGA-GAA-ATG-AGA-GGA-TCA-CGA-G-3Ј and the downstream primer 5Ј-CAC-AGG-ATA-GTT-GAT-AGC-GTG-3Ј. The PCR protocol was based on 94°C, 2 min, 1 cycle and then 30 cycles of 62°C for 1 min, 72°C for 1 min 30 s, and 94°C for 1 min and completed by a 10-min extension at 72°C, and then the block temperature was held at 4°C.
3Ј-RACE-This was accomplished using first strand cDNA prepared using the adaptor-linked oligo(dT) primer. The upstream primer for the first round PCR was 5Ј-C-AAG-CGA-ATT-GTA-TTT-ATT-CCC-GT-3Ј, and the downstream primer was the adaptor sequence 5Ј-TTCGGTAC-CCGGGATCC-3Ј with a PCR program based on 94°C, 2 min, 1 cycle and then 30 cycles of 55°C for 1 min, 72°C for 1 min, and 94°C for 1 min. The nested PCR reaction used the upstream primer 5Ј-TTT-GTG-CCA-AAT-TTA-CCC-3Ј and a second adaptor primer, 5Ј-ATGAATTCG-GTACCCGGG-3Ј, with the same PCR program except the annealing temperature was lowered to 50°C.
5Ј-RACE Giving an ϳ450-bp PCR Product-This was accomplished using the Marathon cDNA amplification kit (Clontech). We modified the first strand cDNA synthesis by using 5 g of total RNA (1 g of mRNA or total RNA is recommended) and otherwise followed the manufacturer's instructions and guidelines. The final preparation of doublestranded cDNA ligated 5Ј and 3Ј with adaptor sequences was diluted to 100 l, and 5 l was used per 50-l PCR reaction. For the 5Ј-RACE reactions, the upstream primers for the first and second round reactions were specific for the ligated adaptor sequence, 5Ј-CCATCCTAATAC-GACTCACTATAGGGC-3Ј and 5Ј-ACTCACTATAGGGCTCGAGCGGC-3Ј, respectively. The downstream primer for the first round reaction was immediately downstream of the most 3Ј conserved histidine, 5Ј-CAC-AGG-ATA-GTT-GAT-AGC-GTG-3Ј. The PCR program was 94°C for 2 min, 1 cycle; 68°C for 3 min, 94°C for 30 s, 30 cycles; 68°C for 7 min, 1 cycle; hold at 4°C. For the second round reaction (primed with 1 ⁄20 l of the first round products) the downstream primer was selected based on cDNA sequence about 300 bp downstream of the protein N terminus, 5Ј-TCT-GGC-GGC-GTG-GTA-AGC-TTA-3Ј, and the PCR protocol was 94°C for 2 min, 1 cycle; 62°C for 30 s, 68°C for 2 min, 94°C for 30 s, 30 cycles; 68°C for 7 min, 1 cycle; hold at 4°C.
Full-length Clones Obtained by PCR-The upstream primer encoded the N terminus of the mature protein, with an added methionine codon and Kozak consensus sequence for translation initiation (38)

Purification of the 8R-Lipoxygenase
Isolation of Active Lipoxygenase-This procedure is outlined in Fig. 1. Initially, preparation of an acetone powder of P. homomalla removes the endogenous prostaglandins and other lipophilic components and also serves to reduce the salt content of the coral extract. The lipoxygenase activity dissolves in cold pH 8 buffer (no detergent is required) and is recovered in the 100,000 ϫ g or 10,000 ϫ g supernatants. The enzyme is unstable in this black solution, and activity is lost completely after standing on ice overnight. The stability of the enzyme is enhanced greatly by fractionation over an open bed column of Q-Sepharose. This and other chromatographic procedures require the addition of detergent. The Q-Sepharose step yields a straw-colored extract that maintains activity for several days on ice. The solution (100 -150 ml) is dialyzed into very low ionic strength phosphate, and the activity is then retained on an open bed column of hydroxyapatite. The enzyme is eluted into a volume of 5-10 ml using 0.4 M sodium phosphate containing 0.3% E911.
Following dialysis into 20 mM phosphate and immediately prior to loading on the Mono-S cation exchanger, the sample is acidified to pH 5 by careful titration with dilute phosphoric acid (ϳ2.5 l/ml). The Mono-S column is run on a gradient up to 1.3 M NaCl and 0.5% E911, and the lipoxygenase elutes after most of the other proteins ( Fig. 2A). After this step, chromatography on the anion exchange Mono-Q column yields a smoothly rising 280-nm UV profile with a small UV peak coinciding with elu-tion of the active 8R-lipoxygenase (Fig. 2B). When fractions are collected across this UV peak and aliquots are examined by SDS-PAGE, it is apparent that the intensity of a protein band of ϳ100 -105 kDa closely parallels the activity of the 8R-lipoxygenase (Fig. 2B, inset). The molecular masses of known lipoxygenases are approximately 75 kDa for the mammalian enzymes and 94 -103 kDa for plant lipoxygenases (2,39,40).
Activity of Purified Protein-At this stage in the purification, the ϳ100-kDa component accounts for at least half of the protein in the sample (Fig. 3). The specific activity of the 8R-lipoxygenase is enriched approximately 30-fold compared with the Q-Sepharose eluant (protein concentration could not be measured in the initial black extract) (Table I). This enzyme preparation forms 8-HPETE product solely of the 8R configuration as shown by chiral column analysis of the corresponding HETE methyl ester (Fig. 4). The most concentrated solutions of 8R-lipoxygenase (Ն0.1 mg/ml protein) were colorless and gave a featureless UV-visible spectrum. The addition of heme had no effect on activity. The turnover number of the purified sample is (in the peak fractions) approximately 4000 min Ϫ1 , compared with reported values for pure mammalian lipoxygenases of 5000 -6000 min Ϫ1 (e.g. Refs. 41 and 42) and ϳ20,000 for the soybean lipoxygenase L-1 (1). These results confirm that the lipoxygenase constitutes a major component of the purified extract.
Lipoxygenase Homology-The ϳ100-kDa band was retrieved in denatured form by RP-HPLC on a C4 Vydac column using a water/acetonitrile, 0.1% trifluoroacetic acid system or, in some experiments, by preparative SDS-PAGE. The amino acid composition closely matched that reported for the soybean lipoxygenase L-1 and mammalian lipoxygenases (Fig. 5). The Nterminal sequence did not match other lipoxygenases, but a BLAST search on an internal peptide recovered following CNBr cleavage gave a match to the soybean lipoxygenase L-1, and significant homology to the same region of mammalian enzymes was evident (Fig. 6). These results provided the first evidence of the closely related primary structure of S-and R-lipoxygenases.

Molecular Cloning of the 8R-Lipoxygenase
Strategy for PCR Experiments-At first we were limited in the available peptide sequence to be used in molecular cloning. The main peptide fragments obtained by CNBr cleavage did not chromatograph well on RP-HPLC, and the main peaks were mixtures of three peptides as determined by Edman microsequencing. We also obtained mixed sequences when the CNBr fragments were run on a high percentage acrylamide Tricine SDS-PAGE system designed for resolution of small peptides (44). The one clear microsequencing result of an internal CNBr fragment gave a match to the S-lipoxygenases as mentioned above. Degenerate primers were designed for nested PCR experiments based on this sequence and also on the Nterminal sequence of the protein. Also, "guessmers" were designed based on conserved elements of the C-terminal sequences of plant lipoxygenases. The plant and mammalian enzymes share an identical C-terminal isoleucine residue (a ligand to the iron), but the consensus sequences differ beyond the first three or four residues into the protein. The plant consensus sequence was selected based on the apparent similarity in size of the 8R-lipoxygenase to plant lipoxygenases.
Obtaining the Initial PCR Clone-A number of different PCR reactions were run. The first to give a positive result used oligonucleotides based on the CNBr peptide sequence on the upstream side and downstream primers based on the conserved plant lipoxygenase C terminus (see "Experimental Procedures"). Using a nested PCR protocol we obtained a single band 406 bp in size. Sequencing of this fragment showed the conserved histidine and nearby asparagine residues corresponding to the most downstream of the characteristic iron ligands of the lipoxygenases, in addition to other conserved sequences.
Additional PCR Clones-The initial 406-bp clone contained one of the CNBr peptides that comprised a mixture of three peptides by Edman microsequencing. With this sequence revealed by cDNA cloning, and the second corresponding to the protein N terminus, we could deduce the third peptide sequence (MRGSRAPI . . . ) by a simple process of elimination. This allowed the use of a fresh set of degenerate PCR primers, which we used together with degenerate primers encoding the protein N terminus (see "Experimental Procedures"), and thus we obtained the 5Ј-sequence up to the protein N terminus. The middle of the coding sequence was then cloned as a 780-bp PCR  a The enzyme activity is measured in a 1-ml reaction volume as absorbance units (AU) increase at 235 nm per min in an aliquot of the sample. In this table, after the Mono-Q atep, 352 AU/min/mg for the whole sample corresponds to a turnover number of 1160 min Ϫ1 b ND, not determined (protein assay is unreliable in the initial black supernatant).

FIG. 4. Chiral column analysis of 8-H(P)ETE formed by the purified 8R-lipoxygenase.
An aliquot of enzyme purified by Mono-Q anion exchange chromatography was incubated with 50 M [ 14 C]arachidonic acid for 5 min at room temperature in 50 mM Tris, pH 8. Following extraction with methylene chloride, the HPETE product was isolated by RP-HPLC, methylated with diazomethane, reduced with triphenylphosphine, and repurified by SP-HPLC. The chirality of the resulting [ 14 C]8-HETE methyl ester was determined using a Chiralcel OD column (25 ϫ 0.46 cm) with a solvent of hexane/isopropyl alcohol (100:2, v/v) and a flow rate of 1 ml/min. The column eluant was monitored using a Hewlett-Packard 1040A diode array detector and by on-line liquid scintillation counting using a Flo-One radioactivity instrument. Arrows indicate the retention times of authentic 8R-and 8S-HETE methyl esters.

FIG. 5. Amino acid composition of the P. homomalla 8R-lipoxygenase compared with representative animal and plant lipoxygenases (soybean lipoxygenase L-1 and human 5S-lipoxygenase)
. Amino acids are indicated by the standard one-letter codes; B represents aspartate plus asparagine, and Z represents glutamate plus glutamine. 8R-lipoxygenase (q) determined on the purified protein (after the C4 Vydac column). The other values are based on the cDNA as follows: soybean lipoxygenase L-1 (å) (39), human 5S-lipoxygenase (Ⅺ) (43), and the 8R-lipoxygenase (---) (data from Fig. 7). product using gene-specific primers. The remaining 5Ј-sequence was obtained by 5Ј-RACE, and the remaining 3Ј-end of the cDNA sequence was readily cloned by 3Ј-RACE (see "Experimental Procedures").
Obtaining the Full-length Clone-cDNAs encoding the mature 8R-lipoxygenase protein were obtained by PCR using a proofreading mixture of Taq and Pwo DNA polymerases (see "Experimental Procedures"). One cDNA had the protein C terminus encoded in the downstream primer, and a second differed in using the 3Ј-untranslated region as downstream primer. The second product, therefore, did not predetermine the C-terminal cDNA sequence in the primer. This latter cDNA and the clones obtained by 3Ј-RACE had identical sequences at the C terminus, each showing that the C-terminal amino acid of the 8R-lipoxygenase is threonine (Fig. 7). The open reading frame encodes 715 amino acids. A signal peptide of 52 amino acids is cleaved to give the mature protein of 663 amino acids.

DISCUSSION
Homology with Other Lipoxygenases-One of the main conclusions of this study is that the P. homomalla 8R-lipoxygenase is a member of the same family of proteins as the known plant and animal S-lipoxygenases. This finding almost certainly extends to the other R-lipoxygenases. Initial evidence of the relatedness of the 8R-lipoxygenase comes from enzyme purification, which shows that the 8R-lipoxygenase protein is colorless, an observation compatible with a nonheme iron oxygenase. The amino acid composition closes matches other lipoxygenases, and one of the CNBr peptides showed homology to the soybean lipoxygenase and the mammalian enzymes. The relationship was confirmed by molecular cloning of the cDNA, allowing firm assignment of the 8R-lipoxygenase to the same gene family as the known S-lipoxygenases.
It is the location of key amino acids in precisely the correct positions that establishes the relatedness of the 8R-lipoxygenase. The overall sequence identity of the P. homomalla 8Rlipoxygenase to other lipoxygenases is low, on the order of 15-20% in amino acid identity. However, the percentage of identity is higher (ϳ25-30%), and many conservative substitutions of amino acids are evident in the parts of the enzyme encompassing the iron-binding ligands. All of the absolutely conserved histidine residues of S-lipoxygenases are present in the 8R-lipoxygenase in the correct positions. Three of these histidines are ligands to the nonheme iron at the active site (6 -9). The fifth heme ligand seen in one crystal structure of the soybean lipoxygenase (7), an asparagine, is also present in exactly the correct position in the 8R-lipoxygenase. Over 70% of the conserved residues of the S-lipoxygenases are retained in the P. homomalla enzyme (Fig. 8).
Features of the 8R-Lipoxygenase Protein-Our initial attempts at purification of the P. homomalla 8R-lipoxygenase were foiled by the strong tendency of the protein to chromatograph in protein microaggregates. It was only after development of the pH 5 cation exchange procedure that a significant enrichment in specific activity could be achieved. This low pH step seemed to "break" the microaggregates, something that could not be accomplished using detergents alone, and allow separation of the 8R-lipoxygenase. Purification yielded a protein that runs on SDS-PAGE with an apparent molecular mass of ϳ100 -105 kDa. This appeared compatible with the size of plant lipoxygenases, which have molecular masses of 94 -103 kDa (39,40). Based on the cDNA, however, the predicted size of the mature coral 8R-lipoxygenase protein is 76 kDa. This is typical of the only other class of animal lipoxygenase to be characterized, the mammalian enzymes. The presence of a presequence on the enzyme is unusual among lipoxygenases, to our knowledge reported only in the plant Lox2 genes exemplified by the inducible lipoxygenases of Arabidopsis and rice (40,46).
The basis of the extra size estimated by SDS-PAGE has not been examined directly, although there are several reasons to deduce that it is related to post-translational modifications of the protein. High level expression of the mature protein in bacteria shows a strong protein band of ϳ90 kDa (data not shown), corresponding more closely to the size anticipated from the cDNA. Post-translational modifications such as glycosylation might also account for the poor chromatographic characteristics of the natural P. homomalla protein. There are several potential sites for modification on the protein; for example, there are four asparagines with the NX(S/T) consensus sequence for N-glycosylation, there are eight protein kinase C phosphorylation sites, and there are four possible sites for myristoylation (not shown).
An alternative explanation for the discrepancy in size observed by SDS-PAGE and molecular cloning is that the cDNA encodes a different protein. The extensive match of amino acids from peptide microsequencing argues strongly against this possibility. The established peptide sequence covers parts of the enzyme from the N terminus of the mature protein to near the carboxyl tail of the polypeptide. This includes over 70 amino acids, all of which are in agreement with the predicted sequence from the cDNA. Since not one of the individual peptides gives a match to another protein in the Swiss-Prot protein sequence data base, we can be confident that the match in over 70 amino acids indicates that we have cloned the correct cDNA.
Differences from Other Lipoxygenases-One of the striking features of the P. homomalla 8R-lipoxygenase is the presence of a threonine residue at the C terminus. This is unique among reported lipoxygenase sequences, all of which have an isoleucine in this position. We know from the crystal structure of the soybean lipoxygenases that the carboxyl of this isoleucine is a ligand to the active site iron (6 -9), and site-directed mutagenesis studies confirm the importance of this residue. Using a murine 12S-lipoxygenase as the model enzyme, Chen et al. (47) showed that deletion of the C-terminal isoleucine yields a catalytically inactive protein. Substitution with valine was well tolerated, while only 10 -20% activity was retained on changing to leucine, asparagine, or serine, less than 2% with arginine and glycine, and no activity on substitution with lysine or aspartate. A threonine substitution was not tested (47) and clearly is an inviting target for mutagenesis studies in the P. homomalla enzyme; the histidine in the second to last position is also of interest as a potential iron ligand. However, we have yet to develop a system for expression of active enzyme. We FIG. 6. Alignment of an 8R-lipoxygenase CNBr peptide fragment with typical S-lipoxygenases. Homology to the soybean lipoxygenase L-1 was detected in a BLAST search of the Swiss-Prot protein sequence data base, and similarity to the same position in the mammalian sequences was evident by inspection. Amino acids identical to the coral 8R-lipoxygenase are underlined. The predicted position of the coral peptide was confirmed subsequently by molecular cloning (Fig. 7).
were unsuccessful using our standard transient expression system, human embryonic kidney 293 cells. The explanation could be PCR-induced mutations in the cDNA, the need for posttranslational modification, or, another possibility, the fact that the human embryonic kidney cells are incubated at 37°C and that this is unsuitable for expression of the P. homomalla 8R-lipoxygenase. Development of a prokaryotic expression system will be used to address this issue.
Based on the information available from this study alone, it is not possible to identify individual residues or sequences that constitute the structural basis for R and S stereospecificity among lipoxygenases. The 8R-lipoxygenase has less than 20% identity to the other enzymes; therefore, it is not possible to define the essential differences. We have available about 40 S-lipoxygenase sequences from animals and plants, and more R-lipoxygenase sequences are required for comparison.
A Basis for R and S Stereospecificity-While details of the basis of R and S stereospecificity remain to be determined, an important principle is established. The R-and S-lipoxygenases are related proteins. The histidine iron ligands and many other essential amino acids are conserved, and it is to be expected that the overall tertiary structures of the enzymes are related. The possibility that enantiomeric products, for example 8R-HPETE and 8S-HPETE, are formed by lipoxygenases with "mirror image" active sites is hardly credible. The basis of R or S stereospecificity must lie in the control of oxygenation of the FIG. 7. Nucleotide sequence of the P. homomalla cDNA and the deduced amino acid sequence of the enzyme. Two PCR products encoding the mature 8R-lipoxygenase protein (FLGWL . . . NSIHT) were fully sequenced (see "Experimental Procedures"). The additional sequence shown here from the 5Ј-end of the cDNA to the N terminus of the mature protein was obtained by 5Ј-RACE and the 3Ј-untranslated region by 3Ј-RACE. N-terminal and peptide sequence obtained by microsequencing of the purified protein is underlined. The peptide presequence (amino acids 1-52), which is not present in the mature protein, is shown in boldface type. The two fully sequenced clones of the mature protein had five single nucleotide differences that changed the encoded amino acid: at nucleotide position 575, A or G (Lys or Arg); at position 773, G or A (Gly or Asp); at position 820, A or G (Arg or Gly); at position 1603, G or T (Ala or Ser); at position 1634, C or A (Ala or Glu). Sequencing of these regions of an additional eight full-length clones showed all eight encoded A (Lys) at nucleotide position 575, G (Asp) at 773, A (Arg) at 820, T (Ser) at 1603, and G (Glu) at 1634. The consensus sequence is shown. Additionally, seven of the extra eight clones had a change of A to G (Asn to Asp) at nucleotide position 739. substrate in related active sites, Fig. 9.
In this line of thinking ( Fig. 9), there is a relation between 8R and 12S oxygenation, and similarly between that of 8S and 12R. Lipoxygenase catalysis to form either 8R-or 12S-HPETE involves an identical initial step, removal of the pro-S hydrogen from carbon 10 ( Fig. 9, top). Reaction with oxygen occurs according to the well established antarafacial rule (i.e. oxygenation occurs on the opposite face of the substrate from the initial hydrogen abstraction). Reaction at one end of the original 1,4-cis,cis-pentadiene gives 8R-HPETE, while reaction at the 12-carbon forms 12S-HPETE. How the position of the reaction with O 2 is controlled has yet to be established for any lipoxygenase. Catalysis to form 8S or 12R products follows the same principles except that binding of substrate is in the reverse orientation (Fig. 9, bottom). The concept that substrate can bind one way round or the other is well precedented in the lipoxygenase literature. It is the most straightforward explanation for the formation of 9S-hydroperoxylinoleic and 13Shydroperoxylinoleic acids by a single lipoxygenase enzyme (48,49) and also for the 5S-and 8S-oxygenase activity of 15Slipoxygenases (50).
This concept of the basis of R and S stereospecificity is supported by two examples of the formation of R configuration products by primarily S-lipoxygenases (51,52). One of these cases involves the type 2 lipoxygenases of soybeans and peas, which, at pH 9, convert linoleic acid into a mixture of 13Shydroperoxide together with 9-hydroperoxide composed of a 2-4-fold excess of the 9R enantiomer (51). The other example is a 12S-lipoxygenase of fish gills; it forms S configuration products in the first oxygenation, but it converts 15S-H(P)ETE to 8R,15S-DiH(P)ETE (52). Presumably, these enzymes have only one access channel for fatty acid into the active site, yet some R-configuration products can be formed. There is also a parallel to these concepts in the formation of the prostaglandin endoperoxide PGG 2 by the mammalian cyclooxygenases (53,54). This reaction involves two oxygenations of the fatty acid substrate, the first in the 11R configuration and the second in the 15S configuration. This is a double oxygenation parallel in many respects to the individual 8R and 12S oxygenations illustrated in Fig. 9. We conclude that R and S oxygenations in lipoxygenases involve different "fits" of the substrate and control of oxygenation and not a "mirror image" reaction at unre-lated active sites.
Concluding Remarks-Finally, since R-and S-lipoxygenases are members of the same gene family, it seems reasonable to expect that the R-lipoxygenases are distributed more widely than in invertebrate animals. The S-lipoxygenases occur throughout the plant and animal kingdoms (1)(2)(3)(4)23). In the plant kingdom, R-lipoxygenases could account for the occurrence of products such as the prostaglandins of marine algae and higher plants (e.g. Refs. 55 and 56). Similarly, in higher animals, R-lipoxygenases could well be involved in the synthesis of the R-HETEs reported in sources such as skin and cornea (19,(57)(58)(59)(60). Although a cytochrome P450 is usually implicated as the enzyme involved in these biosyntheses, R-specific lipoxygenases should also be considered a potential source of the R-HETEs in mammalian tissues. FIG. 8. Sequence alignment of representative plant and animal lipoxygenase with the P. homomalla 8R-lipoxygenase. The soybean L-1 enzyme and the human 12S-lipoxygenase are shown as representative of plant and animal lipoxygenases, respectively. The amino acids shown in these first two lines are the conserved amino acids in all animal and plant S-lipoxygenases, (with the exception of a sequence from the algae Porphyra purpura (45), which itself has fourteen changes to this consensus and was left out of this analysis). Each nonconserved amino acid is indicated by a dash (or the number of nonconserved residues is given in parentheses). The iron ligands are in boldface type. On the third line, the corresponding amino acid of the P. homomalla 8R-lipoxygenase is shown (whether conserved or changed).
FIG. 9. A basis for R or S stereospecificity in the lipoxygenase active site. Formation of four different products is represented in lipoxygenase active sites of related structure. Top, removal of the pro-S hydrogen from the carbon-10 of arachidonic acid occurs on one face of the substrate molecule, and oxygenation of the resulting radical intermediate from above can form either 8R-hydroperoxide (left) or 12Shydroperoxide (right). Thus, 8R and 12S oxygenations involve the same hydrogen abstraction followed by oxygenation on the other face of the substrate, either at C-8 or C-12. Below, for 8S and 12R oxygenations, the substrate binds in the reverse orientation, allowing removal of the pro-R hydrogen from C-10, followed by oxygenation on the other face of the substrate in either the 8S or 12R configuration.