Expression of cloned human reticulocyte 15-lipoxygenase and immunological evidence that 15-lipoxygenases of different cell types are related.

Cloned 15-lipoxygenase has been expressed for the first time in eukaryotic and prokaryotic cells. Transfection of osteosarcoma cells with a mammalian expression plasmid containing the cDNA for human reticulocyte 15-lipoxygenase resulted in cell lines that were capable of oxidizing body arachidonic acid and linoleic acid. The lipoxygenase metabolites were identified by reverse-phase and straight-phase high pressure liquid chromatography, ultraviolet spectroscopy, and direct mass spectrometry, verifying that the cDNA for 15-lipoxygenase encodes an enzyme with authentic 15-lipoxygenase activity. Incubation of the transformed cells with arachidonic acid generated 15-hydroxyeicosatetraenoic acid (HETE) and 12-HETE in a ratio of 8.6 to 1, demonstrating that 15-lipoxygenase can also perform 12-lipoxygenation. Lesser amounts of 15-keto-ETE, four isomers of 8,15-diHETE, and one isomer of 14,15-diHETE were observed. Incubation with linoleic acid generated predominantly 13-hydroxy linoleic acid. The reaction was inhibited by eicosatetraynoic acid but not by indomethacin. Antibodies to a peptide corresponding to a unique region of the predicted amino acid sequence were generated and shown to react with one major band of 70 kDa on immunoblots of human leukocyte 15-lipoxygenase. To obtain antibodies to the full length enzyme, the cDNA was subcloned into a bacterial expression vector and was expressed as a fusion with the CheY protein. The overexpressed protein was readily purified from bacteria and was shown to be immunoreactive to the peptide-derived antibody. Antibodies raised to this recombinant enzyme did not cross-react with human leukocyte 5-lipoxygenase but did identify 15-lipoxygenase in rabbit reticulocytes, human leukocytes, and tracheal epithelial cells, suggesting that the 15-lipoxygenases from these different cell types are structurally related.

Antibodies to a peptide corresponding to a unique region of the predicted amino acid sequence were generated and shown to react with one major band of 70 kDa on immunoblots of human leukocyte 15-lipoxygenase. To obtain antibodies to the full length enzyme, the cDNA was subcloned into a bacterial expression vector and was expressed as a fusion with the CheY protein.
The overexpressed protein was readily purified from bacteria and was shown to be immunoreactive to the peptide-derived antibody. Antibodies raised to this recombinant enzyme did not cross-react with human leukocyte 5-lipoxygenase but did identify 15-lipoxygenase in rabbit reticulocytes, human leukocytes, and tracheal epithelial cells, suggesting that the 15-lipoxygenases from these different cell types are structurally related.
The enzyme 15-lipoxygenase catalyzes the insertion of molecular oxygen into arachidonic acid at carbon 15 (1). The enzyme can also oxygenate other polyenoic free fatty acids (2, 3) as well as a variety of phospholipids (4). This ability to perform lipid peroxidation is manifested in multiple biological systems. For example, 15-lipoxygenase appears to contribute to cellular differentiation in the reticulocyte (2), and to the generation of inflammatory mediators in leukocytes (1, 5), and in human airway epithelial cells (6, 7). These potential biological actions for 15-lipoxygenase have led to increasing interest in understanding the molecular mechanism of the enzyme. Soybean 15-lipoxygenase has been studied extensively, and cDNAs for three different isozymes have been reported (8-10). The mammalian form of the enzyme has been purified to homogeneity from rabbit reticulocytes (2) and from human leukocytes (11). Using protein sequence information from the mammalian enzymes, we recently isolated a cDNA encoding 15-lipoxygenase from a human reticulocyte library (12). The nucleotide sequence of this cDNA has allowed us to deduce the amino acid sequence of the enzyme and to identify potential domains critical to enzymatic function based on sequence similarity to other lipoxygenases. The nucleotide sequence from the rabbit reticulocyte lipoxygenase has since been reported (13), and the predicted amino acid sequence is 81% identical to the human enzyme. However, expression of cloned 15-lipoxygenase from any species has not been described. Active expression of the cDNA is critical to establishing authenticity of the isolated clones and to investigating relationships between the enzyme structure and its function.
Investigators of the rabbit reticulocyte 15-lipoxygenase have presented immunological evidence that the enzyme is specific to red cells (2). Whether the 15-lipoxygenase activity detected in various human tissues is due to the same enzyme or isozymes is unknown. Antibodies to human 15-lipoxygenase will be useful in defining these relationships, as well as in studying enzyme regulation and cellular localization. The biological investigation of 15-lipoxygenase in general has been hampered by the lack of antibodies to the human enzyme. This deficiency has been due to the difficulty in obtaining adequate quantities of pure human enzyme. Our prior success in isolating human 15-lipoxygenase was due to the observation that patients treated with interleukin-2 resulted in large quantities of eosinophil-enriched leukocytes (11,14). Because changes in clinical protocols preclude ready availability of this source, molecular biological techniques were essential to provide reagent levels of the enzyme.
In this paper we describe the active expression of cloned Analysis-SDS-PAGE was performed on 10% polyacrylamide gels overlaid with a 3% polyacrylamide stacking gel as described (22). Separated proteins were then transferred to nitrocellulose filters, and immunoblot analysis was performed (23) with either the peptide-derived or the CheY-15lipoxygenase antibody as a primary antibody.
The secondary antibody was goat anti-rabbit IgG linked to horseradish peroxidase (Bio-Rad), and the blots were developed as described (24).

Creation of 15-Lipoxygenase-containing
Cell Lines-The cDNA for 15-lipoxygenase was subcloned into the expression vector pR135 for mammalian expression (17) (Fig. 1). This vector contains the hygromycin-selectable marker for isolation of the transformed cells and the cytomegalovirus immediate-early promoter for control of transcriptional expression. Plasmids with and without the 15-lipoxygenase cDNA were transfected into the EBNAl-positive human osteocarcoma cell line 143.98.2 by the calcium-phosphate method. Individual hygromycin-resistant clones were screened for 15-lipoxygenase activity. Enzyme activity was not detected in the nontransfected cells or in any of the colonies (n = 5) transfected with the vector alone. In contrast, 6 out of 10 clones transfected with the cDNA for 15-lipoxygenase exhibited enzyme activity. The formation of lipoxygenase products varied from clone to clone and ranged from 230 to 4530 pmol of 13-HODE generated/lo6 cells in 15 min. The variability in expression levels presumably results because the expression plasmid is episomal, and its gene copy number varies more A 0 from cell to cell than that of an integrative system (17). Clones containing active 15-lipoxygenase grew at less than 50% the rate of cells not expressing 15-lipoxygenase. As 15lipoxygenase activity decreased in these cells over time, their growth rates approached that of the nonexpressing clones. Passage of cells one to two times/week over a 5-month period gradually resulted in substantial loss of activity. Cultures were used at low passage number to minimize variability. Storage of low-passage-number cells permitted ongoing analysis of 15lipoxygenase in this expression system. The most highly expressing clone, OS15LOX(+), was selected for further characterization, and clone OS15LOX(-), which contained the vector but no insert, was used to determine basal levels of lipoxygenase expression.

Characterization of Expressed 15-Lipoxygenase
Actiuity-Purified mammalian 15-lipoxygenases have been shown to form a variety of hydroperoxides from arachidonic acid and linoleic acid (26,27) and cellular peroxidases presumably convert these to their hydroxy derivatives (6,7,28). To identify the enzyme activity expressed by clone OSlSLOX(+), cells were incubated with exogenous arachidonic acid or linoleic acid, and the resulting oxygenation products were extracted and analyzed by HPLC. Representative chromatograms are shown in Fig. 2. Incubation of OS15LOX(+) cells with arachidonic acid resulted in the predominant generation of 15-HETE, with lesser amounts of 12-HETE and 15-keto-ETE, four isomers of 8,15-diHETE, and one isomer of 14,15-diHETE (Fig. 2, panel A, and Table I). Incubation of cells with linoleic acid resulted in predominant generation of 13-HODE and lesser amounts of 13-keto-ODE (Fig. 2, panel B). Cells transformed with the vector alone did not metabolize either linoleic or arachidonic acid (Fig. 2, panel C). All compounds were identified by elution times on RP-HPLC. The amounts of product generated in eight experiments from maximal concentrations of arachidonic acid (160 PM) are shown in Table I. The average ratio of 15-HETE to 12-HETE formed was 8.6 to 1. In addition, the major peaks coeluting with 15-HETE, 12-HETE, SS,lSS-diHETE, and 13-HODE were subjected to further analysis. Their ultraviolet spectra and elution on SP-HPLC were identical to those of reference standards (Table  I) Several characteristics of product formation by 15-lipoxygenase in OSlSLOX(+) cells were determined. Maximal and half-maximal generation of 15-HETE occurred at 160 and 40 pM arachidonic acid, respectively. Product generation was complete within 15 min at 37 "C. Low levels of product were released without loss of vital dye exclusion, although maximal release occurred only after cell viability was significantly decreased. In addition, the same chromatographic profile of products was observed if cells were incubated with substrate in the culture flasks, indicating that the trypsinization before assay did not alter product formation. For example, incubation of cells in monolayer with 20 pM arachidonic acid generated 15% of maximal 15-HETE formation without loss of cell viability. These characteristics are similar to those described for the 15-lipoxygenase of freshly isolated epithelial cells (6).
The formation of 15-HETE was inhibited by preincubation with the inhibitor ETYA in a dose-dependent manner. Approximately 50% of maximal 15-HETE formation occurred at 200 nM ETYA, and complete inhibition occurred at 10 FM. There was no inhibition of enzyme activity seen with indomethacin (10m4 M) which rules out the possibility of HETE formation via a cyclooxygenase mechanism. The effects of both inhibitors are consistent with the described characteristics of native rabbit leukocyte (29) and rabbit reticulocyte 15-lipoxygenase (27).
The transformed cells did not generate 15-lipoxygenase products when challenged with calcium ionophore A23187 (10 PM).
Because this ionophore is believed to increase endogenous levels of arachidonic acid, the absence of 15-lipoxygenase activity may be due to the human osteosarcoma cell line not containing enzymes required for endogenous release of substrate (17) or, less likely, to the inability of the methods used in this study to detect minute amounts of product. Cell suspensions were incubated with 160 pM arachidonic acid. Cell supernatants were extracted, and products were separated by HPLC and quantified by ultraviolet absorbance. Ultraviolet spectra and SP-HPLC retention times were determined for select comnounds as describedunder "Materials and Methods." Peak numbers correspond to the peaks in Fig. 2 ase (Fig. 4), indicating that the gene products of the 15LOX cDNA and leukocyte 15-lipoxygenase were antigenically related. The basis for the occasional occurrence of a higher molecular weight band is unknown (Fig. 4). The specificity of the immunoreactivity to both bands was evident by the fact that preincubation with synthetic peptide (0.5 mg) but not with an equimolar concentration of bovine serum albumin inhibited the binding. Furthermore, normal rabbit serum in the same dilution did not react with human leukocyte 15. lipoxygenase. The peptide-derived antibody did not hybridize with human 5-lipoxygenase (Fig. 4) or with rabbit reticulocyte lipoxygenase (Fig. 6), reflecting the differences of amino acid sequence in this region of these proteins and confirming the specificity of the resulting antibody.
To obtain antibodies to multiple epitopes of the enzyme, the cDNA was subcloned into the bacterial expression vector pJC264 (Fig. 1B). The bacterial proteins from multiple transformants were analyzed by SDS-PAGE.
A band with the expected molecular mass for the CheY-15.lipoxygenase fusion protein (82 kDa) was present in bacteria containing the 15LOX cDNA but not in bacteria transformed with the vector alone or with the vector containing the cDNA in the incorrect orientation (Fig. 5A). This heterologously expressed protein was immunoreactive to the peptide-derived antibody on immunoblots (Fig. 5B). The multiple lower molecular weight bands may be degradation products or partial translation products. The immunoreaction could be inhibited by prein- CheYI5LOX(-), or with 15LOX in the incorrect orientation (not shown).
R, an immunoblot using the peptide-derived antisera (1:lOO) demonstrated immunoreactivity to the fusion-protein and to multiple degradation products.
No immunoreactivity was observed in CheY15LOX (-) or in CheY15LOX(+) if nonimmune sera were used or if immune sera were preincubated with the peptide that was used to generate the antibodies. The recombinant 15-lipoxygenase was purified easily by differential solubility and was used to generate polyclonal antibodies. C, immunoblots were performed with these antibodies on the proteins shown in Fig. 4. The antibodies (1:3OUO dilution) recognized one major band in human leukocyte 15lipoxygenase and did not cross-react to human 5-lipoxygenase. D, the peptide-derived antisera recognized the same band in leukocyte 15lipoxygenase as did the antibodies to the fusion protein.
cubation with peptide and the peptide-derived antibody did not hybridize to extracts from bacteria that did not contain the 15LOX expression plasmid (Fig. 5B).
A polyclonal antibody raised to the fusion protein hybridized (1:3000 dilution) to the same band (70 kDa) of human leukocyte 15-lipoxygenase as the peptide-derived antibody (Fig. 5,C and D). Normal rabbit immunoglobulin gave no reaction at similar dilutions, and all reactions were decreased by preincubation with purified rabbit reticulocyte 15lipoxygenase and were completely inhibited by preincubation with the fusion protein. Hybridization to human 5lipoxygenase did not occur even at dilutions of 1:lOO.
The antibodies were used to compare the immunoreactivity of recombinant 15-lipoxygenase with that of several cell types. Because reticulocyte lipoxygenase is expressed at significant levels in acute, severe anemia (30) and because patients with this disorder are rare, we used a model of experimental anemia in rabbits. Rabbit reticulocyte lipoxygenase was prepared by ammonium sulfate precipitation as described (2). Sonicates of human tracheal epithelial cells (95% purity), and the osteosarcoma clones OSl5LOX(+) and OSl5LOX(-) were centrifuged, and the 10,000 X g supernatants were analyzed by immunoblots as described under "Materials and Methods." The antibody to the fusion protein hybridized to one major band (70 kDa) in the extract from the human osteosarcoma cell line, OSl5LOS(+), and in the extract from the human leukocytes, the extract from the human tracheal epithelial cells, and purified rabbit reticulocyte 15-lipoxygenase (Fig.  6B) where known species-dependent sequence differences account for the nonreactivity. cyte lipoxygenase is unclear, but an underestimate of this enzyme's molecular weight (75,000) on SDS-PAGE has been previously noted (13). The lower molecular weight bands were determined to be degradation products. All immunoreactions were inhibited by the fuson protein, and nonimmune IgG gave no reaction. The antibody to the full length recombinant enzyme did not hybridize to the cell clone OS15LOX( -) which did not express lipoxygenase activity. Finally, the peptidederived antibody identified similar major bands in all preparations except rabbit reticulocyte lipoxygenase (Fig. 6C) where the differences in the amino acid sequence are sufficient to account for the nonreactivity (12, 13).

DISCUSSION
Establishing the authenticity of cDNA that encodes a given enzyme is an essential step in analyzing the biochemical and biological features of the recombinant protein.
We have expressed the gene for 15-lipoxygenase in a human osteosarcoma cell line and have demonstrated active catalytic function. The expressed protein is indistinguishable from the native 15lipoxygenase in terms of its molecular size, its antigenicity, its oxygenation of both arachidonic acid and linoleic acid, and its sensitivity to a lipoxygenase inhibitor. Furthermore, the major metabolites produced by the recombinant enzyme have been confirmed by RP-HPLC, SP-HPLC, ultraviolet spectroscopy, and mass spectrometry.
These data confirm that the 15LOX cDNA obtained from the human reticulocyte cDNA library does indeed encode human 15.lipoxygenase. The demonstration of active catalytic function of cloned 1.5 lipoxygenase in an expression system will now permit the testing of several hypotheses that may relate the enzyme's activity to its structure.
The expressed 15-lipoxygenase possesses the ability to per-form 12-lipoxygenation and generates 12-HETE from arachidonic acid. This suggests that the enzyme is multifunctional and is independent confirmation of an earlier observation with purified rabbit reticulocyte lipoxygenase (26); however, these prior experiments could not rule out the possibility of contaminating 12-lipoxygenase. Furthermore, the finding that recombinant 15lipoxygenase generates 15HETE and 12-HETE in a ratio of approximately 9:l suggests that this single enzyme may be responsible for the 12-HETE generated from human airway cells (6) and eosinophils (28). In both cell types and in the OSl!jLOX(+) cells, arachidonic acid is converted into 15 and 12-HETE in the same ratio. We observed slower growth of the cell lines actively expressing 15lipoxygenase. This effect on cell growth may be due to the degradation of mitochondria, as described in reticulocytes (2), or to another effect of lipid peroxidation on other cellular organelles, or to an undescribed action of 15-lipoxygenase. Further study of the OS15LOX(+) cell line may give insight into the intracellular role of 15-lipoxygenase. Furthermore, investigation of how the enzyme is processed in the OS15LOX(+) cells and how it is activated to release 15lipoxygenase metabolites will yield important information that has been lacking due to the paucity of cell culture systems expressing 15-lipoxygenase. Airway epithelial cells lose enzyme activity rapidly during the first passage of cells.' Eosinophils have only recently been maintained in culture for l-2 weeks, but their 15-lipoxygenase activity has not been described (31). Keratinocytes express 15-lipoxygenase activity only when cocultured with fibroblasts and only in sonicated form (32). Umbilical vein endothelial cells can be cultured for a low passage number and contain small amounts of 15lipoxygenase activity (33). In contrast, the activity of 15lipoxygenase in the OS15LOX(+) cells is approximately equal to that of freshly isolated eosinophils, which is 50% that of airway cells and lo-100 times that of other cell types. The OS15LOX(+) cell line, therefore, represents a potential cell system for studying the cellular processing and biological action of human 15-lipoxygenase. reticulocytes (2, 30), and airway cells (6, 7). The newly developed antibodies described here suggest that the 15-lipoxygenases of these different cell types are antigenically related. These results extend our prior observation, based on Nterminal sequence data, that the leukocyte and reticulocyte 15-lipoxygenases are structurally related (11). Determination of the full extent of structural similarity among the 15lipoxygenases of different tissues must await cloning of the cDNAs from tissue-specific libraries but should be facilitated by the human nucleic acid probes and antibodies characterized here. The exact cell types that contain the enzyme within the airway epithelium and inflammatory cells need to be determined by immunocytochemical methods, and these studies can now proceed using the developed antibodies. The antibodies described here are capable of distinguishing cell types with different levels of 15-lipoxygenase expression using immunofluorescent staining.3 Finally, the extensive information collected on the biology of the reticulocyte 15-lipoxygenase needs to be examined in terms of its relevance to other cell types. It is likely that the reticulocyte enzyme's action on membrane lipids during cellular degradation and differentiation relates to the biological role of the 15-lipoxygenase in epithelial cells and leukocytes.

2.
The relationship among 15-lipoxygenases from different tissues is unknown. Isozymes of the soybean 15-lipoxygenase are well described. In addition, the 12-lipoxygenase of bovine platelets has been shown to be immunologically and biochemically distinct from that of leukocytes (34). In contrast, 5lipoxygenase has been cloned from human placenta (35), HL-60 cells (36), and rat basophilic leukemic cells (37). The primary structure is the same in both human tissues, and there is 90% identity between the rat and the human enzymes.

9.
The most extensively studied mammalian 15-lipoxygenase is the rabbit reticulocyte enzyme which is expressed during the reticulocytosis of anemia (2). While investigators of the rabbit reticulocyte enzyme have presented immunologic and catalytic data that suggest that 15-lipoxygenase is unique to rabbit reticulocytes, it remains possible that in other species the enzyme is less tissue-specific. Alternatively, the tissue distribution of the enzyme may be such that the prior studies on whole lung (2) may not have been sensitive enough to detect the low levels of specific tissue expression. We have found that the 15LOX cDNA hybridizes to RNA from different tissues within human lung to reveal a 2.7-kilobase band in human bronchi and human airway epithelium but not to RNA obtained from whole lung preparations that contain mainly lung parenchyma (data not shown). This suggests that 15-lipoxygenase is preferentially distributed within lung tissue along the airway and its epithelium.