Structural and functional criteria reveal a new nuclear import sequence on the 5-lipoxygenase protein.

Leukotrienes are lipid mediators with important roles in immunity. The enzyme 5-lipoxygenase initiates leukotriene synthesis; nuclear import of 5-lipoxygenase modulates leukotriene synthetic capacity. In this study, we used structural and functional criteria to identify potential nuclear import sequences. Specifically, we sought basic residues that 1) were common to different 5-lipoxygenases but not shared with other lipoxygenases, 2) were found on random coil/loop structures, and 3) could be replaced without eliminating catalytic activity. Application of these criteria to the putative bipartite nuclear import sequence of 5-lipoxygenase revealed that this region formed an alpha-helix rather than a random coil, that the critical residue arginine 651 serves a structural role, and that mutation of this residue eliminated catalytic activity. A previously unrecognized region corresponding to residues 518-530 on human 5-lipoxygenase was found to be unique to 5-lipoxygenase and on a random coil. This region alone was sufficient to drive import of green fluorescent protein to the same degree as complete 5-lipoxygenase. Replacement of basic residues in this region of the complete protein was capable of eliminating nuclear import without abolishing catalytic activity. Surprisingly, two subpopulations of cells expressing 5-lipoxygenase with this mutated region could be discerned: those with strongly impaired import and those with normal import. Taken together, these results show that the previously identified region with a bipartite motif is not a functional import sequence, whereas the newly identified basic region constitutes a true nuclear import sequence. Moreover, we suggest that another sequence that can mediate nuclear import of 5-lipoxygenase remains to be identified.

ically dependent on processes that modulate 5-LO activity.
One such process is nuclear import of 5-LO. In leukocytes circulating in peripheral blood, 5-LO is found exclusively in the cytoplasm (2)(3)(4). However, import of 5-LO into the nucleus can be induced by recruitment of cells into sites of inflammation (2,5), by adherence to different surfaces (2,5,6), or by cytokine treatment (3). Import of 5-LO is linked to activity since nuclear import of 5-LO can strongly enhance (2,3) or suppress (5) leukotriene synthesis upon subsequent cell stimulation.
The regulation of nuclear import of 5-LO is poorly understood. In particular, our understanding of the regions on 5-LO that act as nuclear import sequences (NISs) is incomplete. Previously, Funk and colleagues (7) identified three candidate regions based on the abundance of the basic amino acids Arg or Lys. Partial mutagenesis of these basic regions (BR) indicated that none of these sites acted as an NIS (7). A more complete study of one of these BRs, which matched the motif for bipartite NISs, led to the conclusion that this region, and in particular Arg 651 , was necessary for nuclear import of 5-LO (8,9). However, further analysis of Arg 651 mutants of 5-LO indicated that these had limited enzymatic activity (9), suggesting that these mutants were misfolded and that this might be the basis for impaired nuclear import.
Until now, candidate nuclear import regions on 5-LO have been identified solely on the presence of basic residues, which facilitate binding with importin proteins. In this study, additional structural and functional criteria were used. First, 5-LO from mouse, rat, and human leukocytes all show nuclear import (10), but the related 12-LO and 15-LO do not (11). As a result, the primary sequence structure of candidate NISs should be both unique to 5-LO and common across species. Second, to interact with importin proteins, both monopartite and bipartite NISs must conform to an appropriate secondary structure. Resolved crystal structures indicate that both the monopartite SV40 large T antigen NIS and the bipartite nucleoplasmin NIS must reside on random coil/loop structures to interact with importin ␣ (12)(13)(14). Thus, candidate regions on 5-LO should have a random coil or loop secondary structure. Finally, Funk and colleagues (7) demonstrated that mutations that eliminated enzymatic function could also abolish nuclear import. This showed that import can be inhibited nonspecifically, by general changes in protein structure, as well as specifically, by eliminating NIS binding to importin. More importantly, it showed that enzymatic function may be used to distinguish specific and nonspecific effects of mutagenesis on import.
This study applies these structural and functional criteria for NISs to 5-LO. Re-examination of the putative bipartite NIS shows that it fails to meet both structural and functional criteria and cannot be a true NIS. This contrasts with a novel region, spanning residues 518 -530 on human 5-LO, that meets these new criteria and is both sufficient and necessary for nuclear import. Interestingly, we find that although this region functions as an NIS, at least one other NIS must also exist on 5-LO.

EXPERIMENTAL PROCEDURES
Sequence and Structural Analysis-Amino acid sequences were obtained from Swiss-Prot from the ExPaSy (Expert Protein Analysis System) proteomics server of the Swiss Institute of Bioinformatics. Primary accession numbers for proteins are: for 5-LOs, human P09917, mouse P48999, rat P12527, hamster P51399; for 15-LOs, rabbit P12630, human P16050; for human platelet-type 12-LO P18054. Alignment of protein sequences was performed using CLUSTALW (15). Structural analysis utilized the resolved structures of rabbit 15-LO (Protein Data Bank identifier 1lox) and the theoretical model of the 5-LO catalytic region (16).
Plasmids and Mutagenesis-To construct a fusion peptide joining BR 518 to green fluorescent protein (GFP), complementary oligonucleotides encoding Val 514 -Leu 535 (VYGMRGRKSSGFPKSVKSREQL) were annealed and ligated to the BamHI and HindIII sites of pEGFP-C1. To construct GFP/BR/BR, two copies of the above mentioned oligonucleotides were ligated in tandem to the BamHI and HindIII sites of pEGFP-C1.
Specific amino acids within the putative 5-LO NISs were substituted in the pEGFP-C1/5-LO template (8)  Cell Culture, Transfection, and Imaging-NIH 3T3 cells were obtained from American Type Culture Collection (Manassas, VA) and grown under 5% CO 2 in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% calf serum, 100 units/ml penicillin, and 100 units/ml streptomycin. Cells were transfected using FuGENE 6 (Roche Molecular Biochemicals) or Polyfect (Qiagen, Inc.) transfection reagents according to the manufacturer's specifications. Transient transfectants were evaluated microscopically, live or after fixation with 4% paraformaldehyde, 16 -20 h after transfection. Comparable results were obtained when cells were examined as early as 9 h after transfection.
Immunoblotting-As described previously (8), cells were disrupted by sonication on ice, and protein concentrations were determined by a modified Coomassie dye binding assay (Pierce). Samples containing 10 g of protein were separated by SDS-polyacrylamide gel electrophoresis under reducing conditions and transferred to nitrocellulose. Membranes were probed with a rabbit polyclonal antibody raised against purified human leukocyte 5-LO (a generous gift from Dr. J. Evans, Merck Frosst Centre for Therapeutic Research) (17) or rabbit polyclonal anti-GFP (Santa Cruz Biotechnology, Inc.; titer 1:500) followed by peroxidase-conjugated secondary antibody and enhanced chemiluminescence detection (Amersham Biosciences).
Cell Stimulation and Analysis of Leukotriene Synthesis-To stimulate 5-LO activity, cells transfected with various 5-LO constructs were incubated for 30 min at 37°C in serum-free medium containing 10 M calcium ionophore A23187 and 10 M arachidonic acid. Immunoreactive leukotriene B 4 in conditioned media was quantitated by enzyme immunoassay (Cayman Chemical, Ann Arbor, MI) according to the supplier's instructions.
Quantitation of Subcellular Distribution-As an initial approach to quantitation, slides were fixed 16 h after transfection, and 100 positive cells were scored as to whether nuclear fluorescence was greater than, equal to, or less than cytosolic fluorescence. Care was taken to avoid damaged, dead, or autofluorescent cells. Results from at least three independent transfections per construct were used for statistical analysis. As a second approach, 100 individual cells per construct were scored for cytosolic and nuclear fluorescence intensity: using Adobe PhotoShop 5.5, grayscale digital images were adjusted to include the full black-to-white range, and representative gray values, from 0 (white) to 100 (black), were obtained for the cytoplasm and nucleoplasm. Cytoplasmic and nuclear values for each cell were summed to give total cellular fluorescence, and the percentage of fluorescence values for the nuclear compartment were calculated.
Statistical Analysis-Statistical significance was evaluated by oneway analysis of variance using p Ͻ 0.05 as indicative of statistical significance. Pairs of group means were analyzed using the Tukey-Kramer post test.

RESULTS
Reassessment of BR 638 as a Functional NIS-As noted above, nuclear import of 5-LO is common across species but does not occur in other LOs. This suggested that the primary sequence should be strongly conserved in 5-LO from different species and should not be found in other LOs. The putative bipartite NIS, characterized as RKX 11 RNKKK in human 5-LO (residues 638 -655), is largely conserved across different species of 5-LO, although the terminal grouping, RNKKK, changes to RNKNK in rodent 5-LOs (Fig. 1A). In other LOs, only the initial arginine as well as the critical arginine (Arg 651 in human 5-LO) are conserved. By primary structure alone, this region would remain a reasonable candidate functional NIS.
Interaction with importin requires that both simple and bipartite NISs have a coil or loop secondary structure. On 5-LOs and other LOs, almost the entire region corresponding to BR 638 forms an ␣-helix (Fig. 1B). This suggested that this region does not have an appropriate structure to mediate interaction with importin. One consequence of having the amino acids arranged in a helix is that the positively charged side groups of basic residues may not be oriented for accessibility to interact with importin. Most notably, the critical residue Arg 651 projects inward with its side chain apparently interacting with the negatively charge side group of Asp 473 (Fig. 1C). Further examination of the 628 -654 helix (Fig. 1B) revealed that the majority of the residues that are conserved across different LOs (Fig. 1A) are hydrophobic and oriented toward hydrophobic residues on the sister 470 -489 helix (data not shown). This indicated that the interaction between Arg 651 and Asp 473 serves a critical role in maintaining the correct positioning of the two helices. These structural features indicated that mutation of Arg 651 stopped nuclear import not because it prevented interaction of 5-LO with importin but rather because it eliminated the key Arg 651 -Asp 473 association. If so, then mutation of Asp 473 should similarly stop nuclear import. Consistent with this, replacement of Asp 473 with Gln, like the R651Q mutation, eliminated nuclear import of GFP/5-LO (Fig. 2). Importantly, both the R651Q mutant and the D473Q mutant of GFP/5-LO were catalytically inactive, both when tested in intact 3T3 cells and when tested in cell-free assays (data not shown). This result was consistent with these residues serving critical structural functions.
Evaluation of BR 518 as a Functional NIS-A novel basic region, beginning at Arg 518 on human 5-LO, was identified after aligning numerous mammalian 5-, 12-, and 15-LOs simultaneously using CLUSTALW. Correct alignment was suggested by high levels of amino acid similarity on both sides of the region as well as alignment of the GFPXS core (Fig. 3A). This region, BR 518 , contained three clustered basic residues at one end (RGR/KK) and three at the other (KSI/VKSR), which were conserved across 5-LOs and not found in 12-or 15-LOs (Fig. 3A). This region was found on a random coil of the catalytic domain of 5-LO (Fig. 3B). The presence of conserved glycine and proline residues, which can serve as "helix breakers," also indicates that this region would retain a random coil structure. Since the region was conserved across different 5-LOs, not found on other LOs, and was on a coiled structural element, it met our primary and secondary structural criteria for a good candidate NIS.
To determine whether this region was sufficient to mediate nuclear import, a sequence encoding the entire BR 518 region was appended to GFP. As published previously (8), GFP alone was found to be equally distributed between the nucleus and cytoplasm of transfected 3T3 cells, presumably because its small size allowed free diffusion into and out of the nucleus of 3T3 cells. As shown in Fig. 4, the GFP/5-LO fusion protein, as demonstrated previously (8), showed strong nuclear accumulation. Fusion of the BR 518 peptide to GFP (GFP/BRpeptide) produced nuclear import that mimicked that of GFP/5-LO, although the small size should still allow free diffusion of the protein back out of the nucleus. Fusion of two sequential BR 518 peptides to GFP (GFP/BR/BR) produced stronger nuclear import.
To determine whether either part of the BR 518 region was necessary for nuclear import, site-directed replacement of basic residues on GFP/5-LO was performed. The clusters, designated BR 518 a and BR 518 b, on either side of the Gly-Phe-Pro spacer were analyzed separately or together. Since the terminal argi-nine of BR 518 b is located on a helix (16), suggesting that it might serve a structural function, BR 518 b was first examined without this residue. The mutation BR 518 a was R518Q/G519G/ R520Q/K521Q and BR 518 b was K527Q/S528S/V529V/K530Q. Mutated proteins of the appropriate size were expressed in similar amounts as wild type GFP/5-LO in transfected 3T3 cells (Fig. 5A). This result was obtained using antibodies to either GFP (Fig. 5A) or 5-LO (data not shown). Surprisingly, several distinct patterns of subcellular distribution of the mutants were observed in a given transfection: some cells showed clear nuclear accumulation of the expressed protein, other cells showed a balanced distribution, and others showed cytosolic accumulation (Fig. 5B). This result was quantitated by scoring cells individually as having nuclear fluorescence greater than, equal to, or less than the cytosolic fluorescence. Representative images and numbers are given in Fig. 6. Although the majority (78%) of wild type GFP/5-LO cells had nuclear accumulation, a significant number (22%) had a balanced distribution. A balanced distribution of 5-LO, associated with nuclear envelope breakdown during mitosis, has been described (18) and quantitated (9). Mutation at either the BR 518 a or the BR 518 b site reduced the number of cells with nuclear import and increased those with cytosolic fluorescence. Mutation at both sites appeared to be additive.
Statistical analysis of results from multiple transfections showed that both the reduction of cells with nuclear import and the increase in cells with cytosolic fluorescence were statistically significant (Table I). This analysis also showed that mutation of both basic regions produced statistically greater changes in both distribution groups. No statistically significant change in the group showing balanced distribution was found for any mutation. Significantly, all mutants were functionally active, producing leukotriene B 4 when stimulated with the calcium ionophore A23187 in the presence of 10 M arachidonic acid (Table I). The double mutant produced marginally less LTB 4 than the single mutants or WT GFP/5-LO. These results indicated that mutation of the BR 518 (aϩb) region significantly, but only partially, impaired nuclear import of 5-LO.
Further Analysis of Accessory Regions-Since nuclear import was only partially eliminated by the BR 518 (aϩb) mutation, it was possible that additional residues remained in this region to mediate import. For example, the terminal residue in this region, Arg 532 , might be important. Indeed, mutation of this residue, when combined with the BR 518 (aϩb) mutation, completely eliminated nuclear import of 5-LO (Fig. 7A). However, mutation of Arg 532 alone also greatly reduced nuclear import (Fig. 7B). Importantly, mutation of Arg 532 , either alone or with BR 518 (aϩb), eliminated enzymatic activity (Table II). These results were reminiscent of our analysis of Arg 651 (Figs. 1 and 2) in which Arg 651 was found to interact with an acidic amino acid. Closer examination of Arg 532 indicated that its side chain appears to be positioned between those of Asp 490 and Asp 496 (not shown). These results suggested that Arg 532 serves a structural role and is not part of an NIS.
The possibility that basic amino acids on the surface of 5-LO, close to BR 518 , might cooperate in import was examined. Surprisingly, a cluster of basic residues, recognized previously by Chen et al. (7) as BR 128 , is near BR 518 . This region, 128 KQHRRK, was tested previously: the R131Q/R132Q mutant eliminated nuclear import but also was inactive, suggesting that these residues were structurally important. Indeed, Arg 131 is oriented toward the center of 5-LO and is conserved across different LOs, supporting a structural role. However, Arg 132 and Lys 133 are oriented outward and are unique to the 5-LOs, suggesting that they might be relevant to nuclear import. Mutation of these two residues had no effect on import when compared with WT 5-LO (Table III). Also, mutation of these  residues, when combined with mutation of BR 518 (aϩb), was without further effect. Finally, these substitutions had no effect on catalytic activity (Table III). These results demonstrated that the residues Arg 132 and Lys 133 are not involved in regulating nuclear import. Quantitative Evaluation of Nuclear Import-Although the above analyses demonstrated that mutation of the BR 518 site affected the subcellular distribution of 5-LO, they do not clearly define the nature of that effect. In particular, they did not clearly indicate whether the mutation simply reduced the efficiency of nuclear import or whether the mutation resulted in distinct subpopulations of cells. To address this question, nuclear and cytoplasmic fluorescence levels were quantitatively analyzed as described under "Experimental Procedures" so that the percentage in each compartment in individual cells could be charted. By this analysis, it was found that there were (at least) two subpopulations exhibiting nuclear import of GFP/ 5-LO when expressed in 3T3 cells (Fig. 8A). A major peak, consistently found in multiple transfections, consisted of cells with ϳ65% of total fluorescence in the nucleus (peak N 1 ). A shoulder, associated with 80 -90% nuclear fluorescence, was also consistently observed (peak N 2 ).
Mutation of BR 518 (aϩb), as shown in Fig. 8B, did not simply reduce the rate of nuclear import, which would have been manifested as a shift in peak position to the left. Instead, this mutation produced a new population with 35% nuclear fluorescence (Fig. 8B, peak C 1 ). It should be noted that these cells, with representative cells shown in Figs. 4 -6, had little or no nuclear import; the relatively high value of 35% nuclear fluorescence reflects the conservative scoring of this quantitative method. Offsetting this increase was a selective decrease in a subset of the cells with nuclear import. Although the N 1 nu-clear peak was slightly reduced, there was substantial reduction in the N 2 peak. In general terms, these results indicated that mutation of BR 518 reduced the capacity for the strong nuclear import that produced the N 2 peak. The resulting protein was still capable of pronounced nuclear import in some cells, as evidenced by the persistent N 1 peak. However, the resulting protein did not direct nuclear import in those cells comprising the C 1 peak.

DISCUSSION
The nuclear import of 5-LO has repeatedly been shown to be a mechanism for regulating the leukotriene synthetic capacity of leukocytes (2,3,5). Previous studies have provided evidence that BR 638 -655 is a bipartite NIS. More specifically, analysis by our group and others has found that Arg 651 is absolutely critical for nuclear import (8,9). In this study, we present evidence that, unlike other bipartite NISs, residues 638 -654 reside on a structural ␣-helix and that Arg 651 acts to link that structure to a neighboring helix, indicating that this region does not serve an NIS function. We have now for the first time identified a new region on 5-LO, involving residues 518 -530 on human 5-LO, that meets a more rigorous set of criteria as a functional NIS. Specifically, this region 1) is conserved on other mammalian 5-LOs and is not found on 12-or 15-LOs, 2) forms a random coil secondary structure, 3) can, as a peptide, direct the import of GFP, and 4) retains enzymatic activity when site-directed mutation impairs nuclear import of the full-length protein.
These results indicate that this region is a functional NIS on 5-LO. However, mutation of this region produces (at least) two distinct populations. The persistence of nuclear import in some cells, even when BR 518 is intensively mutated, suggests that another NIS must exist on 5-LO.
Our current understanding regarding 5-LO action includes acceptance of BR 638 -655 as a functional bipartite NIS (for example, see Ref. 19). The primary reason for this must be its strong similarity to the signature pattern of bipartite NISs, as indicated by its identification by such tools as PROSITE or MotifScan. However, this pattern has a high probability of occurrence in proteins, leading to false positives: about 4.2% of non-nuclear proteins have such patterns (20). The central function of basic residues in an NIS is to interact with importins, which mediate the import process (21). Structurally, the side chains of these critical residues must extend outward to facilitate this protein-protein interaction. However, structural analysis indicated that the critical residue Arg 651 (on 5-LO) is conserved across different LOs and that its side chain is oriented inward. Also, this basic residue was found to interact with a conserved acidic residue, Asp 473 , on 5-LO. In this study, we found that replacement of either residue with an uncharged amino acid eliminated both nuclear import and enzymatic activity. These findings strongly indicate that this region is not an NIS.
Such an analysis showed that, in addition to the presence of basic residues, structural criteria could be useful in identifying potential NISs. Also, the observation by Chen and Funk (11) that 12-LOs and 15-LOs do not show nuclear import suggested   Table I; results are reported as percentage of cells with the given distribution of fluorescence. Data are means (S.E.) of n ϭ 3 experiments; *, p Ͻ 0.05 versus wild type; **, p Ͻ 0.05 versus mBR 518 (a ϩ b). Activity was assessed as amount of leukotriene B 4 produced by transfected cells, adjusted for protein expression, as measured by enzyme immunoassay. that the 5-LO NIS should not be found on these closely related proteins. Using both primary and secondary structural information, we identified the BR 518 -530 segment as a potential NIS. Addition of a single copy of this peptide to GFP strongly directed nuclear import; two copies produced greater nuclear accumulation. Replacement of basic residues in this region of the GFP/5-LO fusion protein strongly impaired nuclear import. Significantly, these mutants retained their catalytic activity.
These results indicate that this region is a functional NIS. Surprisingly, elimination of basic residues in BR 518 -530 of 5-LO resulted in a protein that could be imported in some cells but not others within a given population of transfectants. This result was very different from the essentially total elimination of nuclear import observed when either Arg 651 or Asp 473 was replaced (Fig. 2) or when protein structure was intentionally disrupted (7). As shown in Figs. 5 and 6, the subpopulations of distributions observed in mBR 518 (aϩb) 5-LO transfectants are visually striking. As shown in Fig. 8B, they are quantifiably distinguishable. The subpopulation demonstrating unquestionable nuclear import (designated as NucϾCyto in Fig. 6 and as N 1 in Fig. 8B) can only mean that a functional NIS remains on this mutated 5-LO protein. A weak NIS was identified in the opposite end of the protein, in the N-terminal domain (7). On the other hand, the absence of strong nuclear import in other cells in the same population (designated as NucϽCyto in Fig. 6 and C 1 in Fig. 8B) may indicate that the alternative NIS can be inactivated. Alternatively, the export process may be important in determining the subcellular distribution of this protein. It is possible that the export process may also be regulated, that peak N 1 corresponds to inactivated export, and that peak C1 and/or the balanced distribution peak (at 50% nuclear fluorescence) corresponds to activated export.
The conclusion that a second NIS, in addition to BR 518 , may exist and may be inactivated allows additional interpretation of the quantitative data for WT 5-LO, as shown in Fig. 8A. Considering the import process alone, the greater nuclear fluorescence at peak N 2 may reflect the action of two active NISs, whereas the lesser peak N 1 may result from a single active NIS. This interpretation is consistent with the observation that neutralization of one NIS (BR 518 ) reduces peak N 2 more than N 1 (Fig. 8B). Alternatively, it is possible that, for WT 5-LO, the export process, when activated, may only be able to diminish nuclear import from N 2 to N 1 .
The nuclear import process is commonly regulated by phosphorylation (for review, see Ref. 22). There is evidence that the metabolic activity of 5-LO can be stimulated by phosphorylation, by protein kinase C (23, 24), a protein tyrosine kinase (25), and by MAP kinase kinase (26), but these studies did not show direct phosphorylation of 5-LO. More recently, two groups have been able to show direct phosphorylation of 5-LO by a tyrosine kinase (27) and by MAPKAP kinase 2 (28). Both groups also found that phosphorylation was necessary for translocation to the nucleus (27,29). However, it is not known whether any of these phosphorylation events might also modulate nuclear import of 5-LO.
It has been shown previously that when the nuclear envelope breaks down during mitosis, the nuclear accumulation of 5-LO observed in cells in culture is lost (18). Izumi and colleagues (9) found that following the completion of mitosis and re-formation of the nuclear envelope, nuclear import was activated to regenerate the nuclear accumulation of 5-LO. This suggests that at least one of the NISs of 5-LO may be activated during the G 1 phase of the cell cycle.
In summary, we have used a combination of structural and functional criteria to re-examine the bipartite NIS-like domain of 5-LO and found that it is not a functional nuclear import sequence. However, we have identified a novel site on 5-LO that is conserved across species but is not found on other lipoxygenases, is distributed over a random coil of the catalytic domain, is sufficient to direct nuclear import of GFP, is necessary for normal import of 5-LO, and whose mutation does not abolish enzymatic activity. These results indicate that the 518 -530 basic region of human 5-LO is a functional NIS. Sur-TABLE III Effect of mutations in the BR 128 region on the subcellular distribution of GFP/5-LO and GFP/5-LO mBR 518 in 3T3 cells Cell preparation is as described in Table I; results are reported as percentage of cells with the given distribution of fluorescence. Mutations in BR128 were R132Q/K133Q. Data are means (S.E.) of n ϭ 5 experiments. Activity was assessed as amount of leukotriene B 4 produced by transfected cells, adjusted for protein expression, as measured by enzyme immunoassay.  prisingly, our results indicate that at least one other NIS exists on 5-LO. We expect that a similar search strategy, based on the use of structural and functional criteria, will reveal the other NIS(s).