The N-terminal “β-Barrel” Domain of 5-Lipoxygenase Is Essential for Nuclear Membrane Translocation*

5-Lipoxygenase is the key enzyme in the formation of leukotrienes, which are potent lipid mediators of asthma pathophysiology. This enzyme translocates to the nuclear envelope in a calcium-dependent manner for leukotriene biosynthesis. Eight green fluorescent protein (GFP)-lipoxygenase constructs, representing the major human and mouse enzymes within this family, were constructed and their cDNAs transfected into human embryonic kidney 293 cells. Of these eight lipoxygenases, only the 5-lipoxygenase was clearly nuclear localized and translocated to the nuclear envelope upon stimulation with the calcium ionophore A23187. The N-terminal “β -barrel” domain of 5-lipoxygenase, but not the catalytic domain, was necessary and sufficient for nuclear envelope translocation. The GFP-N-terminal 5-lipoxygenase domain translocated faster than GFP-5-lipoxygenase. β-Barrel/catalytic domain chimeras with 12- and 15-lipoxygenase indicated that only the N-terminal domain of 5-lipoxygenase could carry out this translocation function. Mutations of iron atom binding ligands (His550 or deletion of C-terminal isoleucine) that disrupt nuclear localization do not alter translocation capacity indicating distinct determinants of nuclear localization and translocation. Moreover, data show that GFP-5-lipoxygenase β-barrel containing constructs can translocate to the nuclear membrane whether cytoplasmic or nuclear localized. Thus, the predicted β-barrel domain of 5-lipoxygenase may function like the C2 domain within protein kinase C and cytosolic phospholipase A2 with unique determinants that direct its localization to the nuclear envelope.

Leukotrienes are potent lipid mediators of inflammation and anaphylaxis (1). They are generated by an initial reaction with the enzyme 5-lipoxygenase from arachidonic acid that has been liberated from membrane lipids (1)(2)(3). 5-Lipoxygenase is a 78-kDa protein found predominantly within inflammatory cell types (e.g. macrophages, mast cells, neutrophils, and eosinophils). The location of the enzyme is cell-type specific. In unstimulated neutrophils, it is found in the cytoplasm, whereas in alveolar macrophages and bone marrow-derived mast cells it is situated mainly in the nucleus (4 -7). Transfection of 5-lipoxygenase cDNA leads to nuclear localized protein in HEK 1 293, COS, NIH-3T3, and Chinese hamster ovary cells, as well as bone marrow-derived mast cells and RAW macrophages (8,9).
Regardless of the cellular localization, 5-lipoxygenase undergoes a calcium-dependent translocation event to the nuclear membrane in activated inflammatory cells (5). Early studies documented a reversible, soluble to membrane compartment transition in neutrophils challenged with the calcium ionophore A23187 (10,11). Experiments with RBL-2H3 cells demonstrated that the translocation to membranes required influx of extracellular calcium, which could be induced either with ionomycin or cross-linking of IgE receptors (12)(13)(14). The neutrophil 5-lipoxygenase cytosol to membrane translocation was associated with a loss of activity (11). However, in alveolar macrophages and RBL cells the translocation to membrane fractions activated 5-lipoxygenase activity (12,15). It is generally regarded that the translocation event to the nuclear envelope is a necessary event in leukotriene formation. At this site there is an apparent transfer of substrate to 5-lipoxygenase by unknown mechanisms, via the integral membrane protein referred to as 5-lipoxygenase activating protein (16,17). 5-Lipoxygenase was recently shown to directly bind two calcium ions (18). It has long been known that this enzyme is unique among lipoxygenases in its ability to have its activity stimulated by calcium (19,20). Although this calcium stimulatory effect is not an absolute requirement for the purified enzyme, it is essential when the enzyme is present in intact cells, or in isolated preparations incubated with membranes or phospholipids (21,22). 5-Lipoxygenase also possesses a nucleotide-binding site of unknown function and ATP is known to stimulate activity (23,24).
There has been little, if any, insight to document the domain(s) within mammalian lipoxygenases that govern membrane translocation. The lipoxygenases are known to possess two domains based on the crystal structures of two soybean enzymes and the rabbit reticulocyte 15-lipoxygenase (25)(26)(27)(28). The N terminus contains a ␤-barrel region of ϳ110 -115 (mammals) and 150 (plants) amino acids, in addition to the large non-heme iron containing catalytic domain at the C terminus. Sequence alignments between mammalian 15-and 5-lipoxygenases indicate ϳ33% identity in the predicted N-terminal domain and only around 10% homology between the corresponding region of plants and mammals. The function of the ␤-barrel of lipoxygenases is unknown but it has been suggested that this domain, which bears resemblance to the C-terminal domain of certain lipases, is important for lipid binding (27). In fact, a recent study showed that this region within a cucumber lipoxygenase is important for binding to liposomes and lipid bodies (29).
Here, we show strong proof that the 5-lipoxygenase putative 2 ␤-barrel domain is unique among the mammalian lipoxygenase members in its ability to direct nuclear membrane translocation. Using green fluorescent protein-lipoxygenase fusions we present evidence for its necessity and sufficiency in nuclear translocation in real time using transfected cells.
pEGFP-12LO-The XbaI insert from pcDNA1-6His12LX (31) was blunt end ligated into the blunted HindIII site of pEGFP-C2. The six-histidine tag was shown previously not to affect enzyme activity (31). The EGFP-tagged protein displayed high level 12-lipoxygenase activity in HEK 293-transfected cells (exclusive formation of 12-HPETE from arachidonic acid).
pEGFP-15LO-1-An EcoRI/BglII fragment of the coding region for human 15-lipoxygenase (32) was cloned into the EcoRI and BamHI sites of pEGFP-C2. This construct yielded both 15-HETE and 12-HETE (9:1 ratio) from arachidonic acid in transfected cells similar to studies without the GFP tag (32).
pEGFP-15LO-2-The coding region for human 15-lipoxygenase-2 was amplified by RT-PCR using primers and conditions based on the published sequence (33) starting with human hair RNA template (34). The PCR products were first cloned in pCR2.1 (Invitrogen) and sequences verified. Two fragments, a 0.9-kilobase EcoRI/AseI piece encoding the N-terminal region, and a 1.1-kilobase AseI/BamHI insert encoding the C terminus, were ligated in a three-fragment reaction with EcoRI/BamHI-digested pEGFP-C2. The construct when transfected into HEK 293 cells and subsequently incubated with arachidonic acid synthesized exclusively 15-HPETE.
pEGFP-12(R)LO-The construction and characterization of this plasmid was described previously (34).
pEGFP-e12LO-The original expression construct described previously (35) was cut with EcoRI, filled in with a Klenow reaction, and blunt end ligated into the SmaI site of pEGFP-C2.
pEGFP-8LO-The coding region for murine 8-lipoxygenase was amplified by RT-PCR using primers and conditions based on the published sequence (36) starting with phorbol ester-treated epidermal RNA template from a 6-day-old mouse (35). The PCR products were first cloned in pCR2.1 (Invitrogen) and sequences verified. The EcoRI fragment with the correct sequence was first cloned into the expression vector pcDNA3 (Invitrogen) and then in pEGFP-C2. When transfected into HEK 293 cells 8-lipoxygenase activity was detected.
pEGP-eLO-3-The sequence for this novel lipoxygenase was cloned exactly as done for 8-lipoxygenase described above based on the published sequence (37). No enzyme activity was detected with arachidonic acid substrate as described, presumably since this enzyme utilizes a different, as of yet undetermined, substrate.

Cell Culture and Transfection
HEK 293 cells were cultured in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum. Cells growing in culture dishes, cover glass-bottomed culture dishes (for photography of living cells) or glass chamber slides were transfected with FuGENE 6 transfection reagent (Roche Molecular Biochemicals) according to the manufacturer's instructions. The transfected cells (15-20 h post-transfection) were used for translocation studies, fluorescence microscopy, and protein preparation for Western blot and lipoxygenase activity assays.

Translocation Studies and Fluorescence Microscopy
The HEK 293 cells on chamber slides, 15-20 h post-transfection, were washed twice with pre-warmed Dulbecco's modified Eagle's medium. For data documentation, the washed cells, 20 -30 min postincubation with 10 M A23187 (Sigma) or 0.1% Me 2 SO (vehicle) in serum-free Dulbecco's modified Eagle's medium, were fixed with 2% paraformaldehyde in phosphate-buffered saline for 20 min. Slides were mounted with Gel/Mount and kept at 4°C. Cell images were examined as described previously (8) for conventional fluorescence microscopy or using a Nikon E600 upright microscope equipped with Bio-Rad 1024 confocal imaging system. For time course analysis of translocation, the GFP fluorescence of living cells incubated in a 37°C stage chamber, 15-20 h post-transfection, was recorded before and after addition of 10 M A23187 using a Nikon TE300 inverted microscope equipped with Bio-Rad 1024 confocal system. Raw data photoimages were acquired by LazerSharp software (Bio-Rad) and processed further by Confocal Assistant and Adobe Photoshop programs.

Western Blot and Activity Assays
Cytosol proteins from transfected cells for Western blot and lipoxygenase activity assay were prepared as described previously (6). A mouse monoclonal antibody against GFP (Berkeley Antibody Co., 1:1000 dilution) was used for enhanced chemiluminescence detection. Activity assay of 12-and 15-lipoxygenases was carried out with 10,000 ϫ g supernatant proteins of transfected cells in phosphatebuffered saline. The supernatants were incubated with 100 M arachidonic acid for 15 min at 37°C and the reactions were terminated with 2 volumes of stop solution (8). Activity assay for 5-lipoxygenase with addition of Ca 2ϩ and ATP and reverse phase-high pressure liquid chromatography analysis of arachidonate metabolites (HETEs and HPETEs) by lipoxygenases were done as described previously (8)

5-Lipoxygenase Is Unique among Human and Murine Lipoxygenases in Its
Ability to Translocate to the Nuclear Envelope-Eight different GFP-lipoxygenase fusion constructs representing the major lipoxygenase forms unique to man and mouse were prepared and introduced into HEK 293 cells. The cells were visualized 15-20 h post-transfection either with or without calcium ionophore A23187 stimulation and were fixed for data documentation. In unstimulated cells, GFP-5LO was localized primarily within the nucleus as we had demonstrated previously (8). All other GFP-lipoxygenases were localized primarily to the cytosol of HEK 293 cells (left panels, Fig. 1, A and  C). The lone exception, perhaps, was GFP-8LO, which distributed throughout the cell, in a pattern more reminiscent of GFP alone. All GFP-lipoxygenases were expressed as Ϸ110-kDa proteins by Western blot analysis using an anti-GFP antibody (Fig. 1B). Each lipoxygenase fusion yielded enzyme activity 2 Since the crystal structure of 5-lipoxygenase has not been determined, it can only be assumed that this enzyme has a two-domain structure similar to other lipoxygenases with elucidated structures. Instead of using "putative" throughout the manuscript it is assumed that there is a N-terminal ␤-barrel-like domain and a C-terminal catalytic domain and the word putative will not be used for the remainder of the text for simplicity.
When the cells were stimulated with A23187, there was a clear translocation to the nuclear envelope in nearly all GFP-5LO expressing cells as evidenced by a "ring-like" pattern (right panel, Fig. 1, A and C). There was little evidence, if any, for translocation to the nuclear membrane for the other seven GFP-lipoxygenase fusions (right panels, Fig. 1A).

5-Lipoxygenase ␤-Barrel and Translocation
The ␤-Barrel Domain, but Not the Catalytic Domain, of 5-Lipoxygenase Is Essential for Nuclear Membrane Translocation-Since lipoxygenases are known to possess two distinct domains (25-28), we prepared GFP-5LO fusions of the N-terminal ␤-barrel and the C-terminal catalytic domains. The precise demarcation between the two domains is somewhat arbitrary because there is a "linking" region of about 15 amino acids between the last ␤-strand of the ␤-barrel and the first helical structure of the catalytic domain (residues 110 -125) based on the rabbit reticulocyte 15-lipoxygenase structure (27). We chose a division between residues 114 and 115 of 5-lipoxygenase. In unstimulated cells, the GFP-5LO-(1-114) ␤-barrel construct was localized primarily to the nucleus ( Fig. 2A, top left panel) consistent with our previous finding of weak nuclear localizing signal (NLS) sequences within the first 80 residues of the protein (8). The GFP-5LO-(115-673) catalytic domain fusion protein was distributed throughout the cell ( Fig. 2A, bottom left panel). Only GFP-5LO-(1-114) translocated to the nuclear membrane ( Fig. 2A, top right panel) upon ionophore challenge indicating that the ␤-barrel was essential for translocation. Extending the ␤-barrel region (residues 1-127 or 1-166) yielded a similar translocation pattern but shortening the region from the C-terminal side (residues 1-80) abolished the ability to translocate (Fig. 2B). Truncation at the N terminus (N-6 deletion; 6 residues) did not affect translocation capacity but truncation of 17 amino acids did (Fig. 2B). Both truncations abolished enzyme activity consistent with data showing that lipoxygenases can tolerate N-terminal additions but not truncations (31,38). The data for eight different constructs are summarized in Fig. 2C. The results suggest that determinants both near the N terminus (residues 6 -17) and the C terminus (residues 80 -114) of the ␤-barrel are either directly involved or related to folding determinants that regulate interaction with the nuclear membrane.
The time course of translocation for both the GFP full-length and ␤-barrel 5-LO fusions was monitored in living cells (Fig. 3). Both nuclear localized fusions translocated, with earliest detectable events around 5 min after A23187 addition for the GFP-5LO ␤-barrel construct. Complete "ring-like" patterns were discernable by 20 -30 min for both. The translocation was faster with the GFP-␤-barrel construct than the full-length fusion.
Translocation Patterns and Enzymatic Activity of Chimeric ␤-Barrel/Catalytic Domain GFP-lipoxygenases-To assess fur-ther the importance of the 5-lipoxygenase ␤-barrel with respect to translocation ability, eight distinct chimeric lipoxygenase constructs were made with 5-LO and either 15-LO-1 or 12-LO (the three so-called classical lipoxygenases) (Fig. 4). First, all chimeric constructs were cytoplasmic localized. The NLS sequences in 5-LO are complex and context dependent (i.e. there may be signals at both ends of the 5-LO molecule and not all fusions to different reporter proteins (e.g. GFP versus pyruvate kinase) are transported to the nucleus) (8,9). Second, only chimeras with the complete ␤-barrel of 5-LO translocated to the nuclear envelope. If the ␤-barrel was truncated (1-80 residues), the chimeric protein no longer translocated. These particular chimeras, however, displayed enzymatic activity of the corresponding 12-or 15-lipoxygenase catalytic domains (Figs. 4B and 5). Third, chimeras with either the ␤-barrel of 12-LO or 15-LO did not translocate to the nuclear envelope. Enzyme activity of these chimeras was dependent on the two-domain fusion boundary that, in general, had to be shifted toward the N terminus (Fig. 5). GFP-5LO and chimeras with the extended 5-LO catalytic domain were able to make 5-HPETE, 5-HETE, and products from LTA 4 (LTB 4 and 6-trans-LTB 4 isomers) from arachidonic acid.
Translocation Patterns of GFP-5LO Constructs with Ironbinding Ligand Mutations-Mammalian lipoxygenases contain a non-heme iron atom bound by three histidine residues and an oxygen molecule on the C-terminal isoleucine (27). Point mutation of these critical histidines, or truncation of the C terminus, can severely cripple iron binding and abolishes enzyme activity (39). Most of these mutations also severely disrupt nuclear targeting indicative of specific folding requirements for nuclear localization (8). We tested the same mutants we made previously and found that the ability to translocate was unaffected by these inactivating mutations, in stark contrast to our results with nuclear targeting (Fig. 6). DISCUSSION We have demonstrated using GFP-lipoxygenase fusion constructs transfected into HEK 293 cells and stimulated with calcium ionophore A23187 the following conclusions: 1) 5-lipoxygenase is unique in its nuclear localization and ability to translocate to the nuclear envelope when compared with other human and murine lipoxygenase family members; 2) the ␤-barrel region of 5-lipoxygenase is necessary and sufficient for nuclear membrane translocation; 3) determinants at both ends fluorescence was visualized in unstimulated (left panels) and A23187 challenged (right panels) cells as described in the legend to Fig. 1 and under "Experimental Procedures." B, the GFP-5LO ␤-barrel domain was shortened from the N terminus by six residues (N6-deletion), 17 residues (N-17 deletion), or C terminus (1-80) or extended to include additional residues linking the two domains (1-127) or including part of the catalytic domain (1-166). Fluorescence was visualized as mentioned above. Note that separate fields of cells are depicted in the left and right panels for all images. Experiments were repeated three times with similar results. C, summary of results in tabular form depicting various constructs tested. FIG. 3. Time course of A23187-induced translocation of GFP-5LO full-length (top panels) and (1-114)-␤-barrel (bottom panels) fusion proteins in HEK 293 transfected cells. The cells were maintained at 37°C and data collected on living cells using a Bio-Rad 1024 confocal imaging system.

5-Lipoxygenase ␤-Barrel and Translocation
of the ␤-barrel region are important for translocation; 4) catalytically active functional chimeras of 5-lipoxygenase, 12-lipoxygenase, and 15-lipoxygenase type 1 can be created by correct association of the two domains at or close to the junction of the two domains; 5) 5-lipoxygenase nuclear targeting and translocation determinants are distinct; and 6) nuclear membrane translocation can occur from either the nuclear or cytoplasmic side.
Lipoxygenases possess a two-domain structure (25)(26)(27)(28). The N-terminal ␤-barrel domain of rabbit reticulocyte 15-lipoxyge-nase revealed striking homology with the C-terminal domain of various lipases. These domains on the distinct proteins share approximately the same size (115-125 amino acids) and both need to access lipid substrates at a membrane surface. A detailed characterization of the role of the ␤-barrel in mammalian lipoxygenase function for substrate or membrane interaction has not been addressed previously. However, a recent report with a cucumber lipoxygenase indicated that the ␤-barrel was necessary for transport from cytosol to lipid bodies (29). Among the mammalian lipoxygenase members identified so far, only the 5-lipoxygenase has a definite binding requirement for calcium ions (18). Calcium is essential for membrane association and has recently been shown to bind to the ␤-barrel of 5-lipoxygenase (40). 3 Thus, in many respects this domain appears to mimic the C2 domains of protein kinase C and cytosolic phospholipase A 2 , both which bind calcium and translocate to membranes (41,42). What directs the 5-lipoxygenase, and not other lipoxygenases, specifically to the nuclear membrane is not clear. Besides the clear calcium binding and dependence on this The hydroperoxy compound, 5-HPETE, and its reduction product, 5-HETE (analysis at 235 nm) were detected in addition to LTB 4 (formed by endogenous leukotriene A 4 hydrolase present in incubations) and the two 6-trans-LTB 4 isomers (nonenzymatic breakdown products of LTA 4 ). The three latter products did not separate using the aqueous acetonitrile reverse phase-high pressure liquid chromatography system. IIB inset, using another system (methanol:H 2 O, 70:30 in 10 mM ammonium acetate, pH 7.8) better separation was achieved. Results are representative from at least three experiments.

5-Lipoxygenase ␤-Barrel and Translocation
cation in vivo, the 5-lipoxygenase appears unique among other lipoxygenases in having an accessory protein (5-lipoxygenase activating protein) that is important for substrate presentation. Moreover, certain interactions with other cytoskeletal and signaling molecules may represent another mode of directing 5-lipoxygenase translocation specifically to the nuclear envelope (43)(44)(45). The 5-lipoxygenase contains an "insertion" sequence of 5 residues in the ␤-barrel that is not present in most mammalian 12-and 15-lipoxygenases (46). An aspartic acid residue within this stretch, as well as other determinants at both ends of the domain, could be important for calcium binding, and proper protein folding for optimum interactions with membrane sites analogous to cytosolic phospholipase A 2 (47).
Although other lipoxygenases did not translocate to the nu-clear envelope with A23187 stimulation in the transfected HEK 293 cells, this finding does not exclude translocation to other membrane sites for access of substrate. It is likely these other lipoxygenases interact in defined ways with membranes that is not evident in this system. Indeed, it has been shown that rat and human 12-lipoxygenases in platelets and tumor cells, as well as 15-lipoxygenase in reticulocytes and interleukin 4-treated monocytes translocate from cytosol fractions to undefined membrane sites (48 -50).
In this study and a previous one (8), we were able to visualize 5-lipoxygenase cellular localization in living cells in real time or fixed with paraformaldehyde. The enzyme demonstrated an intrinsic capability to enter the nucleus of transfected cells indicative of specific NLS sequences. We showed that there were weak nontraditional NLS sequences somewhere within the first 80 residues of the protein (8) and Healy et al. (9) showed divergent results with weak determinants residing in a C-terminal basic region that were important for nuclear localization. The NLS determinants in the ␤-barrel are definitely distinct from the translocation determinants. Mutations introduced to strong iron binding ligands (His-550 or terminal Ile-673 truncation) (39) changed the cellular localization from nuclear to cytoplasmic, whereas they had no effect on translocation. We interpret these results to mean that proper folding of the catalytic domain for appropriate contact with the ␤-barrel and other potential chaperone proteins is essential for nuclear entry but is not important for translocation and association with the attendant membrane lipid and protein partners.
One other study has attempted to address the translocation of 5-and 15-lipoxygenases using truncation mutants and a chimeric approach with transfection into RAW264.7 macrophages (51). This study, unlike ours, was unable to assign a specific region within 5-lipoxygenase as important for membrane targeting. We are not exactly sure why divergent results were obtained but some points to consider are differences in cell type, transfection methods, cellular localization, and choice of construct design. We 3 and Healy et al. (9) found that GFP-5LO localizes to the nucleus of this macrophage cell line instead of the cytosol as found in the study of Christmas et al. (51). However, here and in other studies in alveolar macrophages it has been shown that both cytoplasmic and nuclear pools of 5-lipoxygenase are capable of translocation to the nuclear envelope (5). We demonstrated with our chimeric approach that it was critical to carefully assign the junction between domains for obtaining active proteins. This finding indicates that important contacts between the two domains must be maintained for proper functional responses. Significant variability in enzyme activity and translocation capacity was evident with junction shifts either direction (Ϯ30 -40 residues) from the initial chosen ␤-barrel/catalytic domain junction (residue 115). Using this approach, translocation determinants could be definitely localized to the ␤-barrel region in our studies. Previous studies in our laboratory showed that it was difficult to construct functional chimeras with 5-lipoxygenase and either 12-or 15-lipoxygenase in several locations within the center of the molecule within the catalytic domain (38). However, we and several other groups have been able to successfully alter oxygen insertional activity and product profiles with limited point mutations or small substitutions (38,(52)(53)(54).
Although we did not investigate in detail parameters affecting the time course of translocation, this process appeared to be slower than expected if functionally correlated to 5-HPETE or leukotriene biosynthesis in stimulated inflammatory cells. The GFP-5LO ␤-barrel showed evidence of translocation by at least 5 min at 37°C after ionophore stimulation and continued to proceed over the next 10 -20 min. It is possible that the earliest FIG. 6. Mutations to iron-binding ligands of 5-lipoxygenase do not affect ability to translocate to the nuclear envelope. A, mutations to disrupt iron ligands in 5-lipoxygenase H367Q (top panels), H550Q (middle panels), and deletion of the six C-terminal residues including the isoleucine oxygen ligand were prepared as described previously (8). A23187-induced translocation was assessed as in the legend to Fig. 1. Results are representative from at least three determinations. B, summary of results in tabular form depicting these constructs. events after ionophore stimulation may not be detectable by fluorescence assay of GFP due to sensitivity problems. Alternatively, the GFP-tag may actually slow the translocation process relative to nonfusion protein. Another aspect to consider is that HEK 293 cells do not normally express 5-lipoxygenase or 5-lipoxygenase activating protein and cellular machinery may not function analogously as in activated inflammatory cells. For instance, 5-lipoxygenase phosphorylation, which can occur in vitro (55), may not occur to the same extent or rapidity in these cells as in neutrophils.
In summary, we have shown that the ␤-barrel domain of 5-lipoxygenase is necessary for nuclear envelope translocation. Addressing the concerted actions of calcium and ATP binding, in addition to phosphorylation status, with respect to translocation and functional correlation with leukotriene biosynthesis and enzyme inactivation at membrane sites will be important topics for future studies.