Molecular Basis of the Specific Subcellular Localization of the C2-like Domain of 5-Lipoxygenase*

The activation of 5-lipoxygenase (5-LO) involves its calcium-dependent translocation to the nuclear envelope, where it catalyzes the two-step transformation of arachidonic acid into leukotriene A4, leading to the synthesis of various leukotrienes. To understand the mechanism by which 5-LO is specifically targeted to the nuclear envelope, we studied the membrane binding properties of the amino-terminal domain of 5-LO, which has been proposed to have a C2 domain-like structure. The model building, electrostatic potential calculation, and in vitro membrane binding studies of the isolated C2-like domain of 5-LO and selected mutants show that this Ca2+-dependent domain selectively binds zwitterionic phosphatidylcholine, which is conferred by tryptophan residues (Trp13, Trp75, and Trp102) located in the putative Ca2+-binding loops. The spatiotemporal dynamics of the enhanced green fluorescence protein-tagged C2-like domain of 5-LO and mutants in living cells also show that the phosphatidylcholine selectivity of the C2-like domain accounts for the specific targeting of 5-LO to the nuclear envelope. Together, these results show that the C2-like domain of 5-LO is a genuine Ca2+-dependent membrane-targeting domain and that the subcellular localization of the domain is governed in large part by its membrane binding properties.

Leukotrienes are potent lipid mediators of inflammation and allergic responses (1). 5-Lipoxygenase (5-LO) 1 catalyzes the two-step transformation of arachidonic acid into leukotriene A 4 , which then leads to the synthesis of all leukotrienes (2,3). Because of its critical role in controlling leukotriene production and the potential to block the production of all leukotrienes by specific inhibitors, 5-LO has been the subject of intense investigation. The cellular regulation of 5-LO activity is regulated by a complex mechanism involving calcium, adenosine triphos-phate, and phosphorylation as well as gene transcription (2,3). The subcellular localization of 5-LO in resting cells varies with the type of cell: it is present primarily in the cytoplasm of neutrophils, monocytes, and peritoneal macrophages, whereas it is predominantly located in the nuclei of rat basophilic leukemia cells and mouse bone marrow-derived mast cells and alveolar macrophages (4). However, cell activation leads to the translocation of 5-LO to the nuclear envelope, where 5-LOactivating protein is located (4). Group IV cytosolic phospholipase A 2 (cPLA 2 ), which is critically involved in the production of arachidonic acid, also translocates to the perinuclear region upon activation by calcium (5). Molecular modeling (6) predicted that the amino-terminal region of 5-LO (ϳ130 amino acids) might form the structure similar to the C2 domain that has been found in many cellular proteins involved in signaling and membrane trafficking (7-10); hereafter, it will be referred to as the 5-LO C2-like domain.
C2 domains share a common fold consisting of an eightstrand antiparallel ␤-sandwich connected by variable loops, which at one end of the domain form the binding sites for multiple Ca 2ϩ ions (7)(8)(9)(10). A prototype C2 domain binds Ca 2ϩ and mediates Ca 2ϩ -dependent membrane targeting of proteins. A recent study using the detergent-solubilized inclusion body of the glutathione S-transferase-tagged 5-LO C2-like domain and its mutants indicated that the 5-LO C2-like domain binds calcium ions via several ligands located in the putative calcium-binding loops (6). Also, a cell study using the green fluorescence protein-tagged 5-LO C2-like domain showed that the 5-LO C2-like domain drives the translocation of 5-LO to the nuclear envelope (11). Although these recent reports suggest that the 5-LO C2-like domain might function as a Ca 2ϩ -dependent membrane-targeting domain, Ca 2ϩ -dependent membrane binding properties of the isolated 5-LO C2-like domain have not been demonstrated. Furthermore, the mechanism by which the 5-LO C2-like domain specifically targets 5-LO to the nuclear envelope is unknown. This study was undertaken to fully characterize the membrane binding properties of the 5-LO C2-like domain and to identify the structural determinants of its specific nuclear envelope targeting. The model building, electrostatic potential calculation, and in vitro membrane binding studies of the isolated 5-LO C2-like domain and selected mutants establish that the 5-LO C2-like domain is a genuine Ca 2ϩ -dependent membrane-targeting domain with unique selectivity for zwitterionic phosphatidylcholine (PC), which is conferred by tryptophan residues located in the putative Ca 2ϩbinding loops. The spatiotemporal dynamics of the enhanced green fluorescence protein (EGFP)-tagged 5-LO C2-like domain and mutants in living cells also indicate that the PC selectivity of the domain accounts for the specific targeting of 5-LO to the nuclear envelope.
Construction of Expression Vectors and Mutagenesis-For bacterial expression, the cDNA of the 5-LO C2-like domain (residues 1-115) was cloned into the modified pET22d vector with an amino-terminal His 6 tag. Site-directed mutagenesis was carried out by the overlap extension PCR method.
Bacterial Expression of the 5-LO C2-like Domain-Escherichia coli strain BL21(DE3) was used as the host for protein expression. One liter of Luria broth supplemented with 100 g/ml ampicillin was inoculated with 5 ml of overnight culture grown at 37°C. Cells were grown at 37°C until the absorbance at 600 nm reached ϳ0.6, and then protein expression was induced with 0.5 mM isopropyl-1-thio-␤-D-galactopyranoside (Research Products International Corp., Mount Prospect, IL). After overnight incubation at room temperature, cells were harvested by centrifugation at 5000 ϫ g for 10 min at 4°C. Cells were resuspended in 50 ml of 50 mM Tris-HCl (pH 8.0) containing 1 mM EDTA, 1% (v/v) Triton X-100, 0.1 mM benzamidine, and 1 mM phenylmethylsulfonyl fluoride. The cell suspension was treated with 10 cycles of 15 s of sonication, followed by 45 s of incubation on ice. After the suspension was sonicated, the inclusion body pellet was obtained by centrifugation at 50,000 ϫ g for 15 min at 4°C. The pellet was resuspended in the same buffer, recentrifuged, and resuspended with stirring (ϳ2 h) in 50 ml of 50 mM Tris-HCl (pH 8.0) containing 8 M urea at room temperature. The solubilized protein was purified using a Ni 2ϩ -nitrilotriacetic acid column (QIAGEN Inc.) according to the manufacturer's instructions. The eluted protein fractions were pooled and dialyzed first against 25 mM Tris-HCl (pH 8.0) containing 2 M urea, followed by the same buffer containing 0.5 M urea and no urea, respectively, and finally against deionized water. The refolded protein was lyophilized and stored at Ϫ20°C. The lyophilized protein was resuspended in 10 mM HEPES (pH 7.4) for analysis. The protein concentration was measured using the enhanced BCA protocol (Pierce). The purity of all protein samples was Ͼ90% electrophoretically.
Equilibrium Dialysis-Equilibrium dialysis was performed using a two-chamber Micro-Equilibrium Dialyzer Surface Plasmon Resonance (SPR) Measurements-Preparation of the vesicle-coated Pioneer L1 sensor chip has been described in detail elsewhere (13). The sensor surface was coated with POPC, POPC/POPG (1:1), or POPC/POPS (1:1) vesicles. In control experiments, the fluorescence intensity of the flow buffer after rinsing the sensor chip coated with vesicles incorporating 10 mM 5-carboxyfluorescein (Molecular Probes, Inc.) was monitored to confirm that the vesicles remained intact on the chip. All experiments were performed with the uncoated sensor chip in the control cell because the 5-LO C2-like domain showed negligible binding to the sensor chip. The drift in signal for both sample and control flow cells was allowed to stabilize below 0.3 resonance unit/min before any kinetic experiments were performed. All kinetic experiments were performed at 24°C and at a flow rate of 60 l/min in 10 mM HEPES (pH 7.4) containing 0.1 M NaCl and varying concentrations of Ca 2ϩ . A high flow rate was used to circumvent mass transport effects. Association was monitored for 90 s, and dissociation for 4 min. The immobilized vesicle surface was then regenerated for subsequent measurements using 10 l of 50 mM NaOH. The regeneration solution was passed over the immobilized vesicle surface until the SPR signal reached the initial background value before protein injection. For data acquisition, five or more different concentrations (typically within a 10-fold range above or below the K d ) of each protein were used. After each set of measurements, the entire immobilized vesicles were removed by injection of 25 l of 40 mM CHAPS, followed by 25 l of octyl glucoside at 5 l/min, and the sensor chip was re-coated with a fresh vesicle solution for the next set of measurements. All data were evaluated using BIAevaluation Version 3.0 software (BIAcore). For each trial, the signal was corrected against the control surface response to eliminate any refractive index changes due to buffer change. Furthermore, the derivative plot was used to monitor potential mass transport effects. Once these factors were checked for each set of data, the association and dissociation phases of all sensorgrams were globally fit to a 1:1 Langmuir binding model: [protein⅐vesicle] 7 protein ϩ vesicle. The association phase was analyzed using the following equation: R ϭ (k a C/(k a C ϩ k d ))R max (1 Ϫ e Ϫ(kaCϩkd) )( tϪt0 )) ϩ RI, where RI is the refractive index change, R max is the theoretical binding capacity, C is analyte concentration, k a is the association rate constant, and t 0 is the initial time. The dissociation phase was analyzed using the equation R ϭ R 0 e Ϫkd ( tϪt0 ), where k d is the dissociation rate constant and R 0 is the initial response. The curve fitting efficiency was checked by residual plots and 2 . The dissociation constant (K d ) was then calculated from the equation Monolayer Measurements-The surface pressure () of solution in a circular Teflon trough (4 cm diameter ϫ 1 cm deep) was measured using a Wilhelmy plate attached to a computer-controlled Cahn electrobalance (Model C-32) as described previously (14,15). Five to ten microliters of phospholipid solution in ethanol/hexane (1:9 (v/v)) or chloroform was spread onto 10 ml of the subphase (20 mM Tris-HCl (pH 7.5) containing 0.1 M KCl and 0.1 mM Ca 2ϩ ) to form a monolayer with a given initial surface pressure ( 0 ). The subphase was continuously stirred at 60 rpm with a magnetic stir bar. Once the surface pressure reading of the monolayer had been stabilized (after ϳ5 min), the protein solution (typically 30 l) was injected into the subphase, and the change in surface pressure (⌬) was measured as a function of time. Typically, the ⌬ value reached a maximum after 35 min. The maximal ⌬ value depended on the protein concentration and reached a saturation value when [5-LO C2-like domain] Ն 1 g/ml. Protein concentrations in the subphase were therefore maintained above such values to ensure that the observed ⌬ represented a maximal value. The ⌬ versus 0 plots were constructed from these measurements.
Cell Culture-A stable HEK293 cell line expressing the ecdysone receptor (Invitrogen) was used for all experiments. Briefly, cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum at 37°C in 5% CO 2 and 98% humidity until 90% confluent. Cells were passaged into eight wells of a Lab-Tech TM chambered cover glass for later transfection and visualization. Only cells between passages 5 and 20 were used.
Construction of Gene Constructs of the 5-LO C2-like Domain Fused with EGFP-EGFP in the pEGFP vector (CLONTECH) was modified by PCR to remove the first methionine and to add two amino-terminal glycines and an EcoRI site. The modified EGFP gene was inserted into a modified pIND vector (Invitrogen) between the EcoRI and XhoI sites to yield plasmid pIND/EGFP. The 5-LO C2-like domain was cloned by PCR to remove the stop codon and to add two carboxyl-terminal glycines and an EcoRI site. The PCR product was digested with NotI and EcoRI and ligated into pIND/EGFP to generate a carboxyl-terminal EGFPfused 5-LO C2-like domain with spacer sequence GGNSGG.
Transfection and Protein Production-Cells (80 -90% confluent) in Lab-Tech TM chambered cover glass wells were exposed to 150 l of unsupplemented Dulbecco's modified Eagle's medium containing 0.5 g of endotoxin-free DNA and 1 l of LipofectAMINE TM reagent (Invitrogen) for 7-8 h at 37°C. After exposure, the transfection medium was removed, and the cells were washed once with fetal bovine serumsupplemented Dulbecco's modified Eagle's medium and overlaid with fetal bovine serum-supplemented Dulbecco's modified Eagle's medium containing Zeocin TM and 140 g/ml pronasterone A to induce protein production.
Confocal Microscopy-Cell imaging was performed using a fourchannel Zeiss LSM 510 laser scanning confocal microscope. To trigger the membrane translocation of EGFP-tagged 5-LO C2-like domains, the cell medium was removed, and the cells were washed with 150 l of 2 mM EGTA and overlaid with 300 l of HEK buffer (1 mM HEPES (pH 7.4) containing 2.5 mM MgCl 2 , 1 mM NaCl, 0.6 mM KCl, 0.67 mM D-glucose, and 6.4 mM sucrose) containing 10 M ionomycin and 2 mM CaCl 2 . Control experiments were done with dimethyl sulfoxide in place of ionomycin. EGFP was excited using the 488-nm line of an argon/ krypton laser. All experiments were carried out at the same laser power, which was found to induce minimal photobleaching over 30 scans, and at the same gain and offset settings on the photomultiplier tube. A line-pass 505-nm filter was used for all experiments. A 63ϫ 1.2 numerical aperture water immersion objective was used for all experiments. Cells for imaging were selected based on their initial intensity, which needed to fall in the upper third of the photomultiplier tube's range. The LSM 510 imaging software provides an option for time series imaging and was used to control the time intervals for imaging.
Homology Models and Electrostatic Potential Computation-Using a variety of search methods against the Protein Data Bank (16), including protein threading (123Dϩ), Smith-Waterman sequence alignment, and sequence-to-profile alignment (3D-PSSM) (17), we selected the structures for 15-lipoxygenase (Protein Data Bank code 1lox) (18) and an ␣-toxin from Clostridium perfringens (Protein Data Bank codes 1ca1 and 1qmd) (19,20) as high significance sequence and structural matches for the 5-LO C2-like domain. Homology models for the 5-LO C2-like domain were constructed with the program Modeler (21), and the model based on the ␣-toxin (code 1qmd) as structural template was used in the subsequent electrostatic potential calculations. In the modeling alignment, the 5-LO and ␣-toxin C2-like domain sequences shared 27% sequence identity and 53% sequence similarity as judged by favorable substitutions in the BLOSUM62 matrix (22). The structural template and homology model were evaluated using the program Verify 3D (23), which scores structures according to how well each residue fits into its structural environment based on criteria derived from statistical analyses of the Protein Data Bank. The homology model produced Verify 3D scores that were good relative to the structural template, an indication that the model is reliable for the type of qualitative analysis performed here. A model constructed using 15-lipoxygenase (code 1lox) as a template gave similar results. The calcium ions were modeled into the homology model using structure superposition of the model and the ␣-toxin C2-like domain structure with three calcium ions bound (code 1qmd). The electrostatic properties of the 5-LO C2-like domain homology model were calculated and visualized in the program GRASP (24). In Fig. 7 (A and B), the red and blue meshes represent the Ϫ25 and ϩ25 mV electrostatic equipotential contours, respectively, in 0.1 M KCl.

Molecular Modeling and Calcium Binding of the 5-LO C2-
like Domain-Among various C2 and C2-like domains with known tertiary structures, we found two good potential structural templates for the 5-LO C2-like domain: 15-lipoxygenase (Protein Data Bank code 1lox) (18) and the C2-like domain of C. perfringens ␣-toxin (Protein Data Bank code 1qmd) (19,20). The second template is especially appealing because this domain binds three calcium ions, and a sequence alignment between the C2-like domains of code 1qmd and 5-LO shows that 5-LO has most of the calcium-ligating residues conserved. The best alignment, as judged by sequence similarity, was achieved by making a ClustalW (25) multiple alignment of all three C2-like domains and then extracting the pairwise alignment for the C2-like domain of C. perfringens ␣-toxin and the 5-LO C2-like domain (Fig. 1). Fig. 2 shows the homology model for the 5-LO C2-like domain using the C2-like domain of C. perfringens ␣-toxin as a structural template. Although this model structure has three potential calcium-binding sites, two calcium ions are included in the model because one site should have greatly decreased calcium affinity due to the presence of a Lys residue in place of Asn 294 in the C2-like domain of C. perfringens ␣-toxin (see Fig. 1). This notion is consistent with the finding that 5-LO binds two Ca 2ϩ ions (26). The results of the electrostatic calculations (see below) are independent of which two sites are assumed to be occupied.
It has been reported that 5-LO independently binds two Ca 2ϩ ions with an apparent dissociation constant of ϳ6 M both in the presence and absence of PC (26). However, the calcium binding of the isolated 5-LO C2-like domain has been only qualitatively estimated using a detergent-solubilized inclusion body of the glutathione S-transferase-tagged 5-LO C2-like do- main (6). We therefore measured the calcium affinity of the isolated 5-LO C2-like domain by equilibrium dialysis. The His 6 -tagged 5-LO C2-like domain was expressed in E. coli as an inclusion body, which was solubilized, refolded, and purified to homogeneity. As shown in Fig. 3, the isolated 5-LO C2-like domain bound about two Ca 2ϩ ions both in the presence and absence of POPC vesicles under our experimental conditions. Although it has been shown that calcium ions bind to some C2 domains in a cooperative manner (27,28), the calcium binding curves shown in Fig. 3 were nicely fit to a Langmuir-type equation to yield K Ca values (7-9 M) that are comparable to that of full-length 5-LO (26). This confirms that the calcium sensitivity of 5-LO originates from the 5-LO C2-like domain.
Membrane Binding Properties of the 5-LO C2-like Domain-To establish that the 5-LO C2-like domain is a genuine calcium-dependent membrane-targeting domain, we measured the membrane binding properties of the isolated 5-LO C2-like domain by SPR analysis. SPR analysis allows direct determination of membrane association (k a ) and dissociation (k d ) rate constants for peripheral proteins (13,15). We first measured the binding of the 5-LO C2-like domain to immobilized vesicles with different compositions at 0.1 mM Ca 2ϩ . As shown in Table  I, the 5-LO C2-like domain had higher affinity (in terms of K d ) for zwitterionic PC vesicles than for anionic PS and phosphatidylglycerol vesicles. In this regard, the 5-LO C2-like domain is similar to the cPLA 2 C2 domain, which also has unique PC selectivity (10,29). This is also consistent with the previous finding that 5-LO binds PC vesicles and shows higher activity in the presence of PC vesicles than in the presence of anionic vesicles (30). We then measured the binding of the 5-LO C2-like domain to immobilized PC vesicles as a function of Ca 2ϩ . In the absence of Ca 2ϩ (i.e. 0.1 mM EGTA), no appreciable binding was detected with protein concentration up to 1 M, indicating that the 5-LO C2-like domain has greater than M affinity for POPC under this condition. The increase in calcium concentration from 1 M to 0.1 mM resulted in a 500-fold increase in affinity (K d ), demonstrating its Ca 2ϩ -dependent membrane binding property. As summarized in Table I, the rise in Ca 2ϩ increased k a (ϳ34-fold) and decreased k d (ϳ16-fold) to a comparable degree.
To further characterize the membrane binding properties of the 5-LO C2-like domain, we measured its interactions with phospholipid monolayers. The monolayer technique has been shown to be a sensitive tool for assessing the relative membrane penetrating ability of peripheral proteins (15,31). We first measured the penetration of the 5-LO C2-like domain into the POPC monolayer in the presence and absence of Ca 2ϩ in comparison with the cPLA 2 C2 domain. We previously showed that Ca 2ϩ drastically increases the monolayer penetration of the cPLA 2 C2 domain by exposing aliphatic and aromatic residues in the Ca 2ϩ -binding loops (12). As illustrated in Fig. 4, Ca 2ϩ also enhanced the monolayer penetration of the 5-LO C2-like domain, albeit not as drastically as seen with the cPLA 2 C2 domain. This suggests that a role of Ca 2ϩ in the membrane binding of the 5-LO C2-like domain is to induce a local conformational change in the Ca 2ϩ -binding loops to expose hydrophobic residues for membrane insertion. We then measured the interactions of the 5-LO C2-like domain with phospholipid monolayers with different compositions (Fig. 5). In agreement with the PC selectivity seen in the SPR binding data, the 5-LO C2-like domain displayed significantly reduced penetration into anionic phospholipid monolayers in the presence of 0.1 mM Ca 2ϩ .
Membrane Binding of 5-LO C2-like Domain Mutants-The model structure of the 5-LO C2-like domain suggests the presence of three surface-exposed tryptophan residues (Trp 13 , Trp 75 , and Trp 102 ) in the Ca 2ϩ -binding loops. Our recent study on the cPLA 2 C2 domain showed that aromatic residues in the Ca 2ϩ -binding loops are involved in its PC selectivity. 2 We thus mutated the three tryptophan residues to Ala and measured the membrane interactions of mutants by SPR and monolayer analyses to determine whether or not these residues are involved in the PC selectivity of the 5-LO C2-like domain. As shown in Table I, all mutations reduced the PC affinity of the domain, albeit to different degrees. W13A, W75A, and W102A had 4.5-, 9-, and 20-fold lower affinities for PC in terms of K d , respectively. For W13A and W75A, the mutations affected k a and k d to a comparable degree (2-4-fold), whereas for W102A, the mutation changed k a (10-fold) much more significantly than k d (2-fold). We previously reported that the mutations of interfacial tryptophan residues influence both k a and k d values, although changes in k a are in general larger than those in k d (32). Thus, our mutation data indicate that the three Trp residues (Trp 102 in particular) are involved in the membrane binding of the 5-LO C2-like domain, playing a dual role of accelerating the membrane association and elongating the membrane residence time. For anionic POPC/POPS (1:1) membranes, the mutations of the three Trp residues had modest effects, with a Ͻ2-fold change in affinity. As a result, three mutants (W13A, W75A, and W102A) showed essentially no preference for PC. The PC selectivity expressed in terms of the ratio of K d for PC to K d for PC/PS (1:1) varied from 0.5 to 1 for the mutants, whereas it was 4.5 for the wild-type 5-LO C2-like domain. Thus, it is evident that the three Trp residues are involved in the PC selectivity of the 5-LO C2-like domain.
We then measured the monolayer penetration of the mutants. As shown in Fig. 6, W75A and W102A exhibited significantly reduced penetration into the POPC monolayer, whereas W102A had slightly lower penetration compared with the wildtype 5-LO C2-like domain; however, none showed a drastic decrease in monolayer penetration. These data thus indicate that the three Trp residues (Trp 75 and Trp 102 in particular) are involved in partial penetration into the membrane. Unlike the wild-type 5-LO C2-like domain, which clearly preferred the POPC monolayer to the POPC/POPS (1:1) monolayer, the three mutants did not have PC selectivity in monolayer penetration (data not shown), further supporting the notion that the three Trp residues are involved in specific PC binding.
Calculation of the Electrostatic Potential-Our SPR and monolayer measurements described above indicated that Ca 2ϩinduced interactions of surface Trp residues with PC play a major role in the membrane binding of the 5-LO C2-like domain. To account for the origin of this Ca 2ϩ -induced effect, we calculated the electrostatic equipotential profiles of calciumfree and calcium-bound 5-LO C2-like domains. As shown Fig.  7A, the domain has a net charge of Ϫ9 and is thus highly negatively charged in the absence of Ca 2ϩ . This highly negative potential would provide a high energetic barrier for the domain to bind to membranes due to the high desolvation penalty associated with bringing the domain close to the membrane surface. For binding to anionic membranes, electrostatic repul-sion between the negative potential of the domain and anionic lipid head groups would also disfavor the binding. Fig. 7B illustrates that the negative potential in the Ca 2ϩ -binding region is significantly neutralized by the Ca 2ϩ ions so that the desolvation penalty associated with bringing the domain close to the membrane surface is decreased. In particular, the negative potential surrounding the region encompassing Trp 13 , Trp 75 , and Trp 102 is dramatically reduced. This would allow Trp residues to penetrate more easily into the membrane interface, and because of the electrostatic considerations described above, the penetration is expected to be more significant for PC than for anionic membranes. As is the case with the cPLA 2 C2 domain (12), the Ca 2ϩ binding can also induce the conformational changes in these surface-exposed Trp side chains to orient themselves for more productive membrane insertion. Last, the 5-LO C2-like domain is still negatively charged even in the calcium-bound state. Thus, it is expected that there will be significant repulsion from anionic membranes, and this could partially contribute to the observed PC selectivity of the domain.   the subcellular localization of 5-LO and also to assess the physiological relevance of our in vitro measurements, we transfected the 5-LO C2-like domain and mutants tagged with EGFP into HEK293 cells and measured their spatiotemporal dynamics by time-lapse confocal microscopy. As shown in Fig.  8, the EGFP-tagged 5-LO C2-like domain was dispersed in the cytoplasm and the nucleus in the resting state. When the cells were activated by the Ca 2ϩ ionophore ionomycin, the 5-LO C2-like domain rapidly translocated to the nuclear envelope; the translocation was completed within 5 min. Some residual EGFP-tagged protein signal in the nucleus appears to be due to the saturation of the nuclear envelope by the overexpressed protein. It has been shown that PS is rich in the inner plasma membranes of mammalian cells (33,34), whereas PC is abundantly present in the nuclear membranes (35). We and others have also shown that PS-selective C2 domains of protein kinase C-␣ 2 and phospholipase C-␦ (36) translocate to the plasma membrane, whereas the PC-selective cPLA 2 C2 domain translocates to the perinuclear region (37,38). Thus, it would seem that the subcellular localization pattern of the 5-LO C2-like domain is consistent with its PC selectivity. This notion is corroborated by the altered subcellular localization patterns of mutants W75A and W102A, which have no PC selectivity. In contrast to the wild-type 5-LO C2-like domain, these mutants translocate to both the plasma membrane and nuclear envelope in response to Ca 2ϩ import. Together, these cell data show that the 5-LO C2-like domain is a genuine Ca 2ϩ -dependent membrane-targeting domain, the subcellular localization of which is governed in large part by its membrane binding properties. DISCUSSION This work represents the systematic in vitro and cell studies on the isolated C2-like domain of 5-LO. Our SPR and monolayer measurements, electrostatic potential calculation, and cell translocation studies show that the 5-LO C2-like domain is a genuine Ca 2ϩ -dependent membrane-targeting domain that binds two calcium ions and has PC selectivity. In this regard, the 5-LO C2-like domain is very similar to the cPLA 2 C2 domain. The cPLA 2 C2 domain cooperatively binds two Ca 2ϩ ions with an apparent dissociation constant of ϳ10 M (27,28). Our homology modeling suggests the presence of two high affinity Ca 2ϩ -binding sites in the 5-LO C2-like domain. It was previously reported that 5-LO binds two Ca 2ϩ ions with an apparent dissociation constant of ϳ6 M in the presence and absence of PC (26). Our equilibrium dialysis measurements show that the isolated 5-LO C2-like domain also binds two Ca 2ϩ ions with comparable affinity (K Ca ϭ 7-9 M) under similar conditions. Furthermore, our results indicate that Ca 2ϩ ions play similar roles in the membrane binding of the cPLA 2 C2 and 5-LO C2-like domains. At least three roles have been proposed for the C2 domain-bound Ca 2ϩ ions (9, 10): i.e. negative-to-positive electrostatic potential switch, formation of a protein-Ca 2ϩ -anionic lipid complex, and induction of conformational changes. For the cPLA 2 C2 domain, Ca 2ϩ has been shown to change the side chain orientations of aliphatic and aromatic residues in the Ca 2ϩ -binding loops, thereby leading to their membrane insertion and hydrophobic interactions (12,39,40). The electrostatic potential calculation also suggested that Ca 2ϩ might enhance the binding of the cPLA 2 C2 domain to the PC membrane by charge neutralization, which reduces the desolvation costs associated with bringing the anionic and hence highly solvated domain to the PC membrane surface (41). The present study indicates that Ca 2ϩ ions play the same dual role for the 5-LO C2-like domain. Because the 5-LO C2like domain is highly negatively charged in the absence of Ca 2ϩ , the charge neutralization by the Ca 2ϩ ions is essential for reducing the desolvation penalty associated with bringing the domain close to the membrane surface. Moreover, our monolayer measurements indicate that Ca 2ϩ binding induces the side conformation of the surface-exposed Trp side chains to orient themselves for more productive membrane insertion. A main difference between the two C2 domains is that Ca 2ϩ is absolutely required for the monolayer penetration of the cPLA 2 C2 domain, whereas the 5-LO C2-like domain has lower but definite monolayer penetrating power in the absence of Ca 2ϩ (see Fig. 4). Thus, it appears that interfacial aliphatic and aromatic residues in the 5-LO C2-like domain are in a semiproductive orientation in the calcium-free state and that the calcium binding causes a smaller scale reorientation. Our previous study showed that interfacial aliphatic residues and Phe largely slow the membrane dissociation, whereas interfacial Tyr and Trp residues affect membrane association more than membrane dissociation (13). In the case of the cPLA 2 C2 domain, calcium largely decreases k d , as it primarily promotes the membrane penetration of aliphatic residues and Phe in the calcium-binding loops (32). In contrast, calcium affects k a more than k d , presumably because it mainly affects the membrane binding of three Trp residues.
A majority of Ca 2ϩ -dependent membrane-binding C2 domains prefer anionic phospholipids to zwitterionic ones because calcium generates positive electrostatic potential on the membrane-binding surface (36,41), or calcium itself coordinates to an anionic phospholipid(s) (42). To our knowledge, the cPLA 2 C2 and 5-LO C2-like domains are the only C2 domains with PC selectivity. Our recent structure-function study showed that aliphatic and aromatic side chains located in the Ca 2ϩ -binding loops are critically involved in the PC selectivity of the cPLA 2 C2 domain. 2 A recent electrostatic potential calculation also showed that the PC selectivity of the cPLA 2 C2 domain can be explained by a significantly higher electrostatic repulsion of anionic cPLA 2 C2 domain molecules at the surface of anionic membranes than at the surface of PC membranes (41). Our results indicate that the PC selectivity of the 5-LO C2-like domain originates from similar factors. The complete loss of PC selectivity by the mutation of Trp 13 , Trp 75 , or Trp 102 points to their direct interactions with the PC head group. It is also possible, however, that the PC selectivity derives from the fact that it is easier for Trp residues to partially insert their side chains into PC than into anionic phospholipids. Because the bulkier zwitterionic PC head groups are less hydrated and not tethered by intermolecular hydrogen bonds, it is likely that the penetration of the Trp residues into the PC membrane is easier than that into an anionic membrane. As with the cPLA 2 C2 domain, the 5-LO C2-like domain is anionic even in the calcium-bound state, and it would be repelled from the anionic membrane surface; this should also contribute to the PC selectivity of the domain.
Taken together, the results described herein show great similarities between the cPLA 2 C2 and 5-LO C2-like domains. We therefore propose that the 5-LO C2-like domain binds to the membrane in such an orientation to optimize the partial membrane insertion of Trp residues, as proposed for the cPLA 2 C2 domain (10,12). Although this investigation focuses on three Trp residues, our model structure suggests the presence of other hydrophobic residues, including Phe 14 , Leu 76 , and Tyr 74 , on the putative membrane-binding surface (see Fig. 2) that might also play an important role in membrane binding of the domain. Further structure-function studies are in progress to assess the contributions of these residues to the energetics of membrane binding of the 5-LO C2-like domain.
The mechanisms by which 5-LO is regulated in the cell remain unclear. This study indicates that Ca 2ϩ -dependent membrane binding properties of the 5-LO C2-like domain (PC selectivity in particular) govern its subcellular localization behaviors. Although the exact lipid composition of different cellular membranes of HEK293 cells has not been determined yet, it is expected from the known lipid compositions of mammalian subcellular membranes that the inner plasma membranes of HEK293 cells are rich in PS and that the perinuclear membranes, including the nuclear envelope, contain higher PC content and lower anionic lipids (33)(34)(35). In response to Ca 2ϩ import, the 5-LO C2-like domain, with PC selectivity, translocates to the PC-abundant nuclear envelope. W75A and W102A, which have little PC selectivity, are localized to both the plasma membrane and the nuclear envelope, supporting the notion that the specific targeting of the 5-LO C2-like domain to the nuclear envelope is due to its PC selectivity.
In summary, this work establishes the 5-LO C2-like domain as a genuine Ca 2ϩ -dependent membrane-targeting module that has distinct membrane binding properties and that plays a major role in the subcellular localization behaviors of 5-LO. As such, this study lays the foundation for further investigation of the complex mechanism of membrane targeting and activation of 5-LO, which involve calcium, ATP, protein phosphorylation, and presumably protein-protein interactions. FIG. 7. Electrostatic potential of the 5-LO C2-like domain. The electrostatic potentials for the 5-LO C2-like domain homology model were calculated and visualized in the program GRASP. A corresponds to the calcium-free state, and B depicts the same structure with two calcium sites being occupied. The calciumbinding sites are located at the top of molecule, and the locations of the mutated Trp residues are indicated by arrows. The red meshes correspond to Ϫ25 mV equipotential profiles, and the blue meshes to ϩ25 mV profiles in 100 mM KCl.