The activity of 5-LO (5-lipoxygenase), which catalyses two initial steps in the biosynthesis of pro-inflammatory LTs (leukotrienes), is strictly regulated. One recently discovered factor, CLP (coactosin-like protein), binds 5-LO and promotes LT formation. In the present paper we report that CLP also stabilizes 5-LO and prevents non-turnover inactivation of the enzyme in vitro. Mutagenesis of tryptophan residues in the 5-LO β-sandwich showed that 5-LO-Trp102 is essential for binding to CLP, and for CLP to support 5-LO activity. In addition, the stabilizing effect also depended on binding between CLP and 5-LO. After mutations which prevent interaction (5-LO-W102A or CLP-K131A), the protective effect of CLP was absent. A calculated 5-LO–CLP docking model indicates that CLP may bind to additional residues in both domains of 5-LO, thus possibly stabilizing the 5-LO structure. To obtain further support for binding between CLP and 5-LO in a living cell, subcellular localization of CLP and 5-LO in the monocytic cell line Mono Mac 6 was determined. In these cells, 5-LO associates with a nuclear fraction only when differentiated cells are primed with phorbol ester and stimulated with ionophore. The same pattern of redistribution was found for CLP, indicating that the two proteins associate with the nucleus in a co-ordinated fashion. The results of the present study support a role for CLP as a chaperoning scaffold factor, influencing both the stability and the activity of 5-LO.
5-LO (5-lipoxygenase) catalyses two initial steps in LT (leukotriene) biosynthesis, oxygenation of AA (arachidonic acid) to 5-HPETE [5(S)-hydroperoxy-6-trans-8,11,14-cis-eicosatetraenoic acid] and subsequent dehydration into the epoxide LTA4 . LTs are inflammatory lipid mediators which cause leucocyte chemotaxis and increased vascular permeability. The effects of LTs are well established in the pathogenesis of asthma, and accumulating data also indicate a role for LTs in atherosclerosis . In addition, 5-LO products are implicated in cancer cell survival , recently for leukaemia stem cells .
A model of the 5-LO structure, based on the crystal structure of the ferrous form of rabbit reticulocyte 15-LO [5,6], consists of an N-terminal β-sandwich (residues 1–114) and a larger C-terminal catalytic domain containing prosthetic iron (residues 121–673). The validity of this 5-LO model structure is supported by various mutagenesis studies (reviewed in ). Several factors which influence 5-LO enzyme activity bind to the C2-like β-sandwich, e.g. Ca2+ and PC (phosphatidylcholine), and when cells are stimulated to produce LTs, 5-LO is typically associated with the nuclear membrane . However, recent observations indicate a different subcellular distribution in neutrophils from males .
CLP (coactosin-like protein) is similar to coactosin , a member of the ADF (actin-depolymerizing factor)/cofilin group of actin-binding proteins. By two-hybrid screening with 5-LO as bait, we found that CLP can bind to 5-LO . Binding was also demonstrated by co-immunoprecipitation from lysates of transfected cells, and in vitro assays [GST (glutathione transferase) pull-down assay, native PAGE and chemical cross-linking] showed binding with a 1:1 molar stoichiometry . Human CLP also binds F-actin, and was found to co-localize with actin stress fibres in transfected CHO (Chinese-hamster ovary) and COS-7 cells . Also mouse CLP was found to bind 5-LO with a 1:1 stoichiometry . Mutagenesis showed the involvement of CLP-Lys75 and -Lys131 in binding to F-actin and 5-LO respectively . In the CLP structure  Lys75 and Lys131 are close, indicating overlapping binding sites, which could explain why a ternary complex of F-actin–CLP–5-LO has not been found. We have previously described that CLP can serve as a scaffold for Ca2+-induced activation of 5-LO . CLP has appeared in several array/proteome analyses, connecting CLP with cancer and inflammatory disease, e.g. rheumatoid arthritis . Recently, hyperforin, an anti-inflammatory compound from St John's wort which inhibits 5-LO activity, was found to impair the binding between CLP and 5-LO .
In the present study we show that CLP stabilizes 5-LO against non-turnover inactivation, and that one particular residue in the 5-LO β-sandwich (Trp102) is essential for binding of CLP to 5-LO, and for the effects of CLP on 5-LO activity and stability. A model of the docking complex was calculated (DOT algorithm) taking the available mutagenesis results into consideration.
MATERIALS AND METHODS
Expression of CLP and 5-LO
Recombinant human CLP was expressed as a GST fusion protein, using the plasmid pGEX-5X-1-CLP [12,15]. The fusion partner was removed by digestion with Factor Xa, and anion-exchange chromatography on MonoQ. Recombinant human 5-LO was expressed from the plasmid pT3-5-LO and purified on ATP-agarose (Sigma A2767) [19,20]. Two Escherichia coli strains were used. Expression in BL21 at 27 °C using a rich medium (TB) gave 5-LO preparations with specific activities of approx. 10 μmol of 5-H(P)ETE/mg, whereas expression in MV1190 at 27 °C using minimal medium resulted in specific activities of approx. 20 μmol of 5-H(P)ETE/mg. Mutated plasmids (W13A, W75A and W102A) were constructed from pT3-5-LO using the QuikChange® kit (Stratagene).
Time and heat inactivation
Solutions of purified 5-LO (14 μg/ml, total volume 500 μl), with or without purified wt (wild-type)-CLP or CLP mutants (1:1 stoichiometry), were prepared in AB+ buffer [50 mM Tris/HCl (pH 7.5), 2 mM EDTA and 0.1% 2-mercaptoethanol], and kept in closed Eppendorf tubes. For time-dependent inactivation at room temperature (22 °C), the tubes were kept on the laboratory bench, and at the indicated intervals, aliquots (10 μl) were removed for the 5-LO activity assay (described below). For heat inactivation, tubes were immersed in a water bath at 50, 55 and 60 °C. At the indicated intervals, aliquots (10 μl) were removed and assayed for 5-LO activity.
HPLC assay of 5-LO enzyme activity
Incubations were performed in Eppendorf tubes at room temperature for 10 min. Buffer AB [50 mM Tris/HCl (pH 7.5) and 2 mM EDTA] was added to tubes, followed by the addition of substrate mix containing Tris/HCl (pH 7.5), CaCl2, PC (Sigma P-3556), AA (Nu-Chek Prep), 13-HPODE [13(S)-hydroperoxy-9-cis-11-trans-octadecadienoic acid] and ATP. Tubes were immersed in a sonication bath for 1 min. The reaction was initiated by addition of aliquots of 5-LO from inactivation experiments containing approx. 150 ng of 5-LO. When CLP (1:1 stoichiometry) was present during inactivation experiments, it was also present in the assay. Final concentrations in the 100 μl incubation volume were: 77.4 mM Tris/HCl (pH 7.5), 100 μM AA, 1.2 mM EDTA, 1.9 mM Ca2+, 25 μg/ml PC, 10 μM 13-HPODE and 4.9 mM ATP.
Incubations were terminated with 300 μl of ice-cold stop solution (acetonitrile/water/acetic acid; 60:40:0.2, by vol) containing 3.3 μM 17(S)-hydroxy-(7Z,10Z,13Z,15E)-docosatetraenoic acid (internal standard at 235 nm, a gift from Dr Mats Hamberg, Department of Medical Biochemistry and Biophysics, Karolinska Institute, Stockholm, Sweden). Aliquots (100 μl) were injected on to a C18 HPLC column (Waters Nova Radial Pak) and AA metabolites were isocratically eluted with acetonitrile/water/acetic acid (60:40:0.2, v/v) at a flow rate of 1.2 ml/min. The eluate was monitored at 235 nm for 5-HPETE and 5-HETE [5(S)-hydroxy-6-trans-8,11,14-cis-eicosatetraenoic acid]. 5-LO enzymatic activity was calculated from the sum of peak areas, relative to internal standards.
GST pull-down assay
For binding studies in vitro, 20 μg of the GST–CLP fusion protein linked to glutathione–Sepharose 4B beads was incubated with purified 5-LO protein (5 μg) in the presence of BSA (50 μg) in 200 μl of buffer A [2 mM Tris/HCl (pH 8.0), 0.2 mM ATP, 0.2 mM CaCl2, 2 mM MgCl2, 50 mM KCl and 0.5 mM 2-mercaptoethanol]. After a 30-min gentle rotation at room temperature, beads were washed five times in buffer A (without BSA). Bound proteins were eluted with 150 μl of elution buffer [10 mM GSH in 50 mM Tris/HCl (pH 8.0)] during 60–90 min rotation at room temperature. Beads were sedimented, and aliquots of the supernatant eluate were assayed by SDS/PAGE followed by 5-LO Western blot analysis.
Pulse proteolysis with thermolysin
A stock solution of thermolysin (Sigma T7902) was prepared in 2.5 M NaCl and 10 mM CaCl2 and the concentration was determined by spectroscopy as described previously . Purified 5-LO (4 μg) with or without CLP (0.8 μg, 1:1 stoichiometry) was pre-incubated for 10 min at room temperature in PBS containing 10 mM CaCl2 and 50 mM NaCl. Thermolysin (4 μg) was added, and after 20–120 s incubations were stopped by the addition of 3.5 μl of EDTA (50 mM) and cooling on ice. Sample loading buffer (10× concentration) was added and 4 μl aliquots were subjected to SDS/PAGE on a Pharmacia FAST system, followed by Coomassie Blue staining.
Subcellular fractionation after detergent lysis
MM6 (Mono Mac 6) cells were grown in cell culture, and differentated with TGF-β (transforming growth factor-β; 5 ng/ml) and calcitriol (50 nM) for 96 h as described previously . Cells (10×106) were resuspended in 1 ml of PGC buffer (PBS containing 1 mg/ml glucose and 1 mM CaCl2). After pre-incubation for 10 min at 37 °C, cells were primed with PMA (100 nM for 10 min) and activated with ionophore (5 μM) and AA (40 μM) for another 10 min, as indicated. Cells were then chilled on ice for 5 min, lysed with Nonidet P40, and subcellular fractionation was performed as described previously . Aliquots (45 μl, derived from approx. 1 million cells) of pair-wise nuclear and non-nuclear fractions were immediately mixed with 9 μl of concentrated SDS loading buffer, heated for 5 min at 95 °C, and analysed for 5-LO and CLP protein by SDS/PAGE and immunoblotting. In-house antisera for 5-LO (1551) and CLP were used. Antibodies against lamin B (nuclear membrane marker, a control for the fractionation procedure) and against β-actin were from Santa Cruz Biotechnology.
Protein–protein docking of CLP and 5-LO
Protein–protein docking of CLP and 5-LO was performed using the ClusPro server  and applying the DOT algorithm [24,25]. Since an experimentally determined three-dimensional structure for 5-LO is not available, we used our previous homology model based on the crystal structure of rabbit 15-LO . For CLP the averaged structure of 20 NMR determinations (PDB: 1WNJ; ) was used as the starting conformation for protein–protein docking. The flexible ends (residues 1–4 and 137–145) were truncated. Parameters for the docking algorithm were: radius of clustering=9, number of electrostatic hits to be clustered=1500, and number of retained output-structures=10. Hydrogen atoms were added to both input structures and the proteins were minimized with a constrained backbone using the CHARMM force field . DOT performs a systematic rigid-body search of one molecule translated and rotated relative to a fixed second molecule by evaluating conformations during docking with a shape complementarity function. Intermolecular energies of all configurations generated are calculated as the sum of electrostatic and van-der-Waals energies. These energy terms are evaluated as correlation functions and used to rank the docking solution . The obtained docked complexes were filtered by our experimental data retaining only those complexes where CLP-Lys131 could interact with 5-LO and 5-LO-Trp102 with CLP. CLP-Lys131, 5-LO-Trp102 and nearby amino acids in the resulting complex were finally relaxed using the CHARMM force field to optimize the modelled interacting geometry.
Origin 8 was used for data analysis, and a Student's t test for two independent samples was performed to determine statistical differences between the means of groups of independent experiments. P values <0.05 were considered statistically significant.
Enzyme activity of wt-5-LO and three tryptophan mutants in the presence of CLP and/or PC
Efficient Ca2+-induced dioxygenase and LTA4 synthase activities of wt-5-LO requires the presence of a scaffold factor to which 5-LO can bind. Binding of 5-LO to PC has been shown to involve three surface-exposed tryptophan residues in the 5-LO β-sandwich (Trp13, Trp75 and Trp102). When these were exchanged for alanine, mutagenesis of Trp102 gave the most prominent effect, reducing the affinity of the isolated 5-LO C2-like domain to PC by approx. 20-fold . Because of the similar effects of PC and CLP (both support Ca2+-induced 5-LO activity) we performed the same three single-point mutations in intact 5-LO, and determined the effect on enzyme activity in the presence of CLP and/or PC. As shown in Figure 1(A), in the presence of PC (25 μg/ml) the formation of dioxygenase products (5-HPETE and 5-HETE) was rather similar for wt-5-LO and the mutants. Enzyme activity was much lower in the absence of scaffold factors. In assays from which PC was excluded, addition of CLP (1:1 stoichiometry) increased dioxygenase activity of wt-5-LO and two of the mutants (5-LO-W13A and 5-LO-W75A). In contrast, for 5-LO-W102A, CLP did not support the dioxygenase activity, which was barely detectable. When both PC and CLP were added, activities were again quite similar for all four proteins. Concomitantly, for wt-5-LO, 5-LO-W13A and 5-LO-W75A, addition of CLP resulted in an increased 5-HETE/5-HPETE ratio (results not shown).
Enzyme activity of wt-5-LO and three tryptophan mutants in the presence of CLP and/or PC
Formation of LTA4 was determined as the non-enzymatic hydrolysis products 6-trans-LTB4 and 12(S)-6-trans-LTB4. For wt-5-LO, inclusion of both CLP and PC led to a 6-fold up-regulation of LTA4 production, compared with incubations including only PC (Figure 1B). Similar up-regulation of LT production was found for 5-LO-W13A and 5-LO-W75A, but not for 5-LO-W102A. For all four 5-LO proteins, LT biosynthesis was small when PC was absent. Thus, as observed previously , CLP alone cannot support formation of LTA4, but CLP together with PC leads to a considerable increase in LTA4. Figure 1(C) shows the total 5-LO product formation, with and without PC and CLP. Taken together, the activity results show that Trp102 in 5-LO is essential for the effects of CLP on 5-LO activity.
Enzyme activity of wt-5-LO and three tryptophan mutants at reduced concentrations of PC and AA
To study the role of the three tryptophan residues on 5-LO activation (by Ca2+) in the presence of only PC as a scaffold factor, activity assays were performed at reduced concentrations of AA and PC (20 μM AA and 5 μg/ml PC). Only at such relatively low concentrations were differences in activities detected for the tryptophan mutants, in comparison with wt-5-LO (Figure 2). Thus at high PC and AA (our standard conditions: 100 μM AA and 25 μg/ml PC) no significant reduction in oxygenase activity was observed for the mutants in relation to wt-5-LO (compare with Figure 1A). However, reduced amounts of PC and AA resulted in a significantly lower formation of 5-HPETE and 5-HETE, which was observed for all three 5-LO mutants (Figure 2). Thus all three tryptophan residues contribute to PC support of Ca2+-induced 5-LO activation. None appears to be of particular importance. Accordingly, when all three tryptophan residues were mutated to alanine, there was a more pronounced reduction in 5-LO activity at reduced concentrations of PC and AA .
Enzyme activity of wt-5-LO and three tryptophan mutants at low concentrations of PC and AA
Mutant 5-LO-W102A does not bind GST–CLP
To ascertain whether the effects of CLP on 5-LO enzyme activity mirror specific binding of 5-LO proteins to CLP, pull-down assays, using the GST–CLP fusion protein, were performed. wt-5-LO, 5-LO-W13A and 5-LO-W75A all bound to GST–CLP, whereas 5-LO-W102A did not (Figure 3). There was no association of 5-LO to GST, used as a negative control. The enzyme activity of 5-LO-W102A was about the same as for wt-5-LO when assayed in the presence of PC (25 μg/ml) (Figure 1). This indicates that the loss of binding of 5-LO-W102A to CLP should not be due to compromised overall structure of this 5-LO mutant.
Mutant 5-LO-W102A does not bind GST–CLP
CLP prevents inactivation of 5-LO over time
In addition to supporting Ca2+-induced enzyme activity, we found that CLP prevents inactivation of 5-LO. Solutions of purified 5-LO were kept at room temperature (on the laboratory bench) for up to 5 days in sealed Eppendorf tubes under normal atmosphere. At intervals, aliquots were removed and subjected to the enzyme activity assay. 5-LO alone was gradually inactivated, with half activity remaining after 24 h, and approx. 20% after 120 h. However, in the presence of CLP (1:1 molar stoichiometry), enzyme activity was preserved, considering that approx. 2/3 of the initial activity remained after 120 h (Figure 4A). Since no substrate was present during the 5 days, CLP protected 5-LO against non-turnover inactivation. Exposure to oxygen is an important factor previously shown to lead to non-turnover inactivation of 5-LO (see  for a review).
CLP protects 5-LO activity
The effect of CLP was also tested with regard to inactivation of 5-LO during turnover. For this purpose 5-LO, with or without CLP (1:1), was subjected to repeated addition of substrate. 5-LO was incubated under standard assay conditions (see the Materials and methods section), including 100 μM AA. After 10 min, half of the assay mixture was removed, and analysed for 5-HETE and 5-HPETE. To the remaining half, AA (100 μM) was added again, and incubated for an additional 10 min. Prominent formation of 5-HETE and 5-HPETE was observed after the first addition of substrate (24 and 23.7 μmol/mg of protein) in both cases (samples with and without CLP). There were only minor increases in products after the second substrate additions, going up to 27 and 24 μmol/mg of protein respectively. This result showed that CLP could not prevent turnover-related inactivation of 5-LO.
CLP prevents heat-inactivation of 5-LO
Samples of 5-LO, with or without CLP (1:1 stoichiometry), were subjected to different temperatures for 10 min. Samples were then cooled on ice, and aliquots were taken to perform a 5-LO standard assay. At 37 °C there was no inactivation of 5-LO, whereas exposure to 80 °C resulted in complete inactivation. CLP could not protect at 80 °C, probably due to denaturation of CLP itself. However, at 50, 55 and at 60 °C, CLP had a protective effect. The protective effect was most prominent at 55 °C, and this temperature was chosen for a time curve. In this experiment, samples of 5-LO, with or without the presence of an equimolar (1:1) amount of CLP, were kept at 55 °C for up to 90 min. Aliquots were removed at intervals for an activity assay. As shown in Figure 4(B), the presence of CLP gave a considerable reduction of heat-inactivation of 5-LO at 55 °C.
Mutant 5-LO-W102A is not protected by CLP
The protective effect of CLP against heat inactivation was also investigated for the three tryptophan mutants. Heat treatment (55 °C) of 5-LO proteins for 80 min resulted in reduced 5-LO product formation in subsequent oxygenase activity assays (formation of 5-HPETE and 5-HETE) compared with room temperature controls (Figure 5A). These heat-induced reductions in activity were statistically significant [P<0.001 for all four proteins (Figure 5A)]. However, in the presence of CLP, heat treatment did not have any effect on the enzyme activities of wt-5-LO, 5-LO-W13A and 5-LO-W75A. It should be observed that the formation of 5-HPETE and 5-HETE for wt-5-LO, 5-LO-W13A and 5-LO-W75A was generally reduced in the presence of CLP, due to a shift of the 5-LO product profile in favour of LTs (compare with Figure 1). In contrast, CLP failed to protect the 5-LO-W102A mutant against heat-inactivation. In addition, among the four 5-LO proteins, this 5-LO mutant was the most sensitive to heat treatment, with a 90% reduction in activity. In comparison, heat exposure reduced the activity of wt-5-LO by 65%, but only by 35% for 5-LO-W75A, suggesting that mutation of Trp75 renders 5-LO less susceptible.
Mutant 5-LO-W102A is not protected by CLP
Inactivation at room temperature over time was determined for the mutant 5-LO-W102A, in comparison with wt-5-LO. As shown in Figure 5(B), without CLP both 5-LO enzymes lost activity similarly over 5 days. On the other hand, in the presence of CLP (1:1 stoichiometry) wt-5-LO maintained approx. 90% of its starting activity after 5 days (similar to Figure 4B), whereas CLP did not prevent inactivation of 5-LO-W102A over time.
CLP mutant K131A does not protect 5-LO
Two CLP mutants, K131A with reduced binding to 5-LO and K75A which binds to 5-LO (but not to F-actin), were tested with regard to protection of 5-LO against non-turnover inactivation at room temperature. Unlike wt-CLP and CLP-K75A which reduced 5-LO inactivation over time, the mutant CLP-K131A did not have this protective effect (Figure 6). The two CLP mutants were also tested with regard to protection of 5-LO against heat inactivation (55 °C for 80 min). Again it was found that CLP-K75A prevented inactivation of 5-LO, whereas CLP-K131A was ineffective (results not shown).
Effects of CLP mutants on inactivation of 5-LO over time
CLP prevents digestion of 5-LO by thermolysin
The thermostable protease thermolysin (from Bacillus thermoproteolyticus) can be used in so called ‘pulse proteolysis’ to determine effects of ligands on protein stability . 5-LO, which is known as an unstable enzyme, was rapidly cleaved by this protease, already after 20 s (Figure 7A, lane 1). On the other hand, CLP was quite resistant to thermolysin. A major part of CLP remained intact after 20 or 120 s incubations with the protease (Figure 7B). For ‘pulse proteolysis’, 20 s was chosen as the time used. 5-LO was pre-incubated with CLP (1:1 stoichiometry) for 10 min, and the sample was subjected to thermolysin for 20 s. Subsequent SDS/PAGE of the samples showed that 5-LO was preserved in the presence of CLP (Figure 7A, lanes 1 and 2). However, when the proteolysis time was increased to 90 s, 5-LO was also digested in the presence of CLP (Figure 7A, lanes 3 and 4). These results indicate that proteolytic cleavage of 5-LO can be delayed by CLP. A similar protective effect was shown for the Trp13 and Trp75 mutants, but not for the Trp102 mutant of 5-LO (results not shown). Several weak bands ranging in size between approx. 15 and 70 kDa appeared when 5-LO was incubated with thermolysin. Bands larger than 34.6 kDa (size of thermolysin) can be attributed to large 5-LO fragments, whereas smaller bands could also stem from autoproteolysis of thermolysin. In Figure 7(A), only 0.24 μg of CLP were applied to the FAST gel (compared with 1.2 μg of 5-LO and thermolysin), whereas in Figure 7(B), 2.4 μg of CLP was applied. This explains the absence of visible CLP bands in Figure 7(A).
Effect of CLP on thermolysin digestion of 5-LO
Distribution of CLP and 5-LO in nuclear/non-nuclear fractions from MM6 cells
One way to study possible association between 5-LO and CLP in a cell is to determine whether the two proteins translocate in a similar fashion upon cell stimulation. Subcellular fractions were prepared from MM6 cells which had been subjected to various treatments. First, undifferentiated MM6 cells (expressing CLP, but not 5-LO) were analysed. Cells were treated in four different ways: ionophore-stimulated only, primed with PMA and stimulated with ionophore, primed with PMA and stimulated with ionophore and AA, and control cells. In all cases, most CLP was found in the non-nuclear fractions. Only weak Western blot bands were detected in nuclear fractions (Figure 8A). As a loading control, β-actin appeared with similar band intensities for all non-nuclear fractions. Weaker bands were observed for the nuclear fractions. Thus for undifferentiated MM6 cells lacking 5-LO, association of CLP with the nucleus was small.
Distribution of CLP and 5-LO in nuclear/non-nuclear fractions from MM6 cells
Next, differentiated MM6 cells (expressing 5-LO and CLP) were analysed. It has been found previously that these cells require priming with PMA for 5-LO to associate with the nuclear membrane upon ionophore stimulation . Accordingly, when differentiated MM6 cells were challenged with ionophore only, 5-LO was recovered primarily in the non-nuclear fraction, and the same pattern was found for CLP (Figure 8B). However, following priming with PMA and subsequent activation with ionophore, both 5-LO and CLP were redistributed to the nuclear fraction. The same result was seen after PMA-priming, and activation with ionophore and AA (Figure 8B). These observations indicate that, in differentiated MM6 cells, trafficking of CLP and 5-LO occur by similar patterns. CLP has not been observed to change subcellular localization, without 5-LO doing that as well. Also, with the samples from the differentiated MM6 cells, β-actin gave similar bands for the non-nuclear fractions, However, in the nuclear fractions from PMA-primed cells, β-actin was increased, similar to 5-LO and CLP. Interestingly, β-actin can bind both to CLP  and to 5-LO (reviewed in ).
Protein–protein docking of the CLP–5-LO interaction
The combination of computational docking results from DOT and experimental data has been established as a useful tool for understanding potential molecular interactions . To better understand the interaction between CLP and 5-LO, docking was performed using our 5-LO model structure and the CLP NMR structure (PDB: 1WNJ; ). Assuming that the impairment of complex formation by the two mutations CLP-K131A and 5-LO-W102A points to a close interaction of CLP-Lys131 with 5-LO, and of 5-LO-Trp102 with CLP, only one model complex remained (Figure 9). In particular, in this model CLP-Lys131 is directed towards 5-LO and 5-LO-Trp102 points towards CLP. This model suggests a direct interaction via a cation–pi interaction between the side chains of CLP-Lys131 and 5-LO-Phe14. 5-LO-Trp102 could be involved in an interaction network also including 5-LO-Arg165, and the carbonyl oxygen of CLP-Lys131. In such an interaction network, the 5-LO-Trp102 side chain might engage in a cationic–pi interaction with 5-LO-Arg165 which in turn would form a hydrogen bond with the backbone carbonyl oxygen of CLP-Lys131 (Figure 9).
Protein–protein docking of the CLP–5-LO complex
Several other amino acids, located both in the 5-LO β-sandwich and in the catalytic domain, may potentially form hydrogen bonds stabilizing the complex [5-LO-Glu70–CLP-Arg91, 5-LO-Asp106–CLP-Val113, 5-LO-Glu108–CLP-Arg91, 5-LO-Thr137–CLP-Lys102 and 5-LO-Lys140–CLP-Gln106]. Four of these five residues are identical in other mammalian 5-LOs, one is conserved. All predicted interacting residues of CLP are identical in mouse CLP, which is also known to bind 5-LO .
5-LO has always been considered to be an unstable enzyme and 5-LO is inactivated both during turnover and during storage (non-turnover inactivation) (for a review see ). Inactivation of 5-LO during turnover can be due to reactions with the enzyme products 5-HPETE and LTA4. Non-turnover inactivation is thought to depend on oxygen, and deliberate exposure to oxygen inactivated 5-LO, due to loss of the prosthetic iron. Also treatment with H2O2 inactivated purified 5-LO, and catalase or glutathione peroxidase can protect against such inactivation.
CLP was previously found to bind 5-LO by co-immunoprecipitation from lysates of transfected cells. In addition, GST pull-down assays, native PAGE and chemical cross-linking showed binding in vitro with a 1:1 molar stoichiometry [14,12]. CLP serves as a scaffold for 5-LO activity , apparently by replacing or complementing membrane (PC) in this role. When CLP was present in the in vitro assays (no PC added) Ca2+ activation led to formation of 5-HPETE plus 5-HETE, whereas the presence of CLP together with PC gave considerable up-regulation of LTA4 formation. In the present study, we confirm these effects of CLP on 5-LO activity and we show that CLP also has another effect, i.e. to prevent non-turnover inactivation of 5-LO. Thus CLP, present in a 1:1 molar stoichiometry, could prevent both inactivation over time and during heat treatment. This stabilizing effect of CLP was found to be specific, since it strictly depended on binding to 5-LO. The CLP mutant K131A, with considerably reduced binding ability to 5-LO, did not protect 5-LO. Also, the 5-LO mutant W102A, which did not bind CLP, was neither stabilized nor activated by CLP. The 5-LO tryptophan mutants studied in the present report were as active as wt-5-LO (in the presence of PC; Figure 1) indicating that these mutations did not compromise the overall structure of 5-LO.
PC is the major phospholipid constituent of the nuclear membrane. In addition to its role as a scaffold for Ca2+-induced 5-LO activity, PC can also stabilize 5-LO. In fact, PC has been used as a stabilizing agent during purification of 5-LO [32,33], and membrane binding of 5-LO stabilized the structures of both 5-LO and the membrane . Addition of leucocyte protein fractions improved stability and activity of purified 5-LO [32,35]. In addition to the ‘macromolecular crowding’ provided by mixed proteins at relatively high concentrations, it appears possible that these protein fractions contained CLP. The observation that CLP is effective already at a 1:1 stoichiometry supports a specific interaction between 5-LO and CLP, different in nature from the protective effect of proteins in general. Also, the dependence on particular amino acid residues (both in 5-LO and in CLP) strongly suggests that the protective effect of CLP is due to specific interaction between these two proteins.
Binding of PC to the 5-LO β-sandwich involves three tryptophan residues (Trp13, Trp75 and Trp102) . Mutagenesis of these residues (to alanine) reduced the affinity of the isolated 5-LO C2-like domain to PC, substitution of Trp102 gave the most prominent effect with a 20-fold reduced affinity . When all of these residues were mutated to alanine in intact 5-LO, Ca2+-induced enzyme activity in the presence of relatively low concentrations of PC and AA was reduced to approx. 25% of wt-5-LO . In the present study we show that, for the three single tryptophan mutants, the activity at low PC and AA was approx. 50% of wt-5-LO. Thus all three residues seem to contribute, to about the same extent, for PC to support 5-LO enzyme activity. Similarly to rabbit 15-LO  it has been suggested that for intact 5-LO residues also in the catalytic domain contribute to membrane binding, and it was noted that the Kd for binding of intact 5-LO to PC-rich vesicles was considerably lower than for binding of the isolated β-sandwich . Thus it appears that, for intact 5-LO, Trp102 is not as influential for PC binding, as for the isolated 5-LO β-sandwich. Also, it has been proposed that 5-LO can bind to membranes in different modes, non-productive and productive .
Presumably, our determinations of Ca2+-induced enzyme activity in the presence of PC reflect a productive binding mode. In view of this model, it thus appears that Trp102 is more important for non-productive binding of 5-LO to PC, in comparison with the productive binding. Similar considerations can also apply to the role of 5-LO Trp75 for binding to PC. In the 5-LO model structure  this residue is located on the very tip of a surface-exposed loop of the β-sandwich, and in a model of a 5-LO–phospholipid interaction, this residue inserts into the hydrophobic part of the membrane . However, the present mutagenesis data (Figure 2) as well as previous findings (Trp75 mutated also to arginine, phenylalanine and serine)  do not point to a particular role for Trp75 with regard to PC-supported 5-LO activity.
On the other hand, for CLP to support 5-LO, Trp102 in 5-LO is essential, and mutagenesis of this residue obliterated binding of CLP to 5-LO. When both PC and CLP were added to 5-LO, total 5-LO product formation was about the same as with only PC present (Figure 1C), but strikingly the LT production was prominently increased (approx. 5-fold; Figure 1B). As suggested previously , this indicates a three-partner complex, comprising 5-LO, PC and CLP. Our findings show that Trp102 in 5-LO is important for binding to CLP. Apparently, Trp13 and Trp75 are free to bind PC, possibly Trp102 can also bind to PC (at the same time as binding CLP).
Trp102 is part of a stretch of conserved surface residues (FPCYRW) , and in the 5-LO model structure Trp102 is partially hidden in the cleft between the two domains, as previously demonstrated for the corresponding residue (Trp100) in 12/15-LO . Thus it appears possible that binding of CLP to 5-LO may have an allosteric effect on the association of the two domains in 5-LO, these may open up for CLP to directly access Trp102. Alternatively, as suggested by our model of the CLP–5-LO complex, 5-LO-Trp102 may influence CLP binding via an interaction network involving 5-LO-Arg165. The model also suggests a direct pi–cation interaction of CLP-Lys131 with 5-LO-Phe14, as well as six hydrogen bonds between 5-LO and CLP, three of which involve residues in the 5-LO catalytic domain, whereas the other three involve residues in the β-sandwich. In either case, binding of CLP to 5-LO may influence the relative positions of the two domains in 5-LO. We would like to point out that, in our modelling of the CLP–5-LO complex, Ca2+ was not included. Ca2+ is known to bind the 5-LO β-sandwich and different sets of Ca2+ ligands have been suggested [19,28,39,41]. Although CLP can bind to 5-LO in the absence of Ca2+ , the mode of binding will possibly be different in the presence of Ca2+.
CLP could protect 5-LO against thermolysin-catalysed proteolysis. Thermolysin is a Ca2+-dependent endopeptidase with preference for hydrophobic target sites, which can be used to study protein stability and ligand binding by so called ‘pulse proteolysis’ . In short time incubations (20 s), CLP protected 5-LO against thermolysin; however, when the incubation time was extended to 90 s, protection was not observed. This indicates that proteolysis of 5-LO was delayed in the presence of CLP, and not entirely prevented. An attractive explanation for this effect of CLP is that, by binding to 5-LO, it renders the 5-LO structure more compact, and consequently bulky hydrophobic residues (preferred cleavage sites of thermolysin; ) may become less accessible. In soy bean LO-1, hydrophobic interactions contribute to association of the two domains , and hydrophobic residues are present between the two domains in 5-LO model structures . The importance of 5-LO-Trp102 suggests that CLP binds at or near the domain interface of 5-LO. Possibly, CLP prevents access of thermolysin to a target site on hydrophobic interdomain areas of 5-LO.
When cells are activated to produce LTs, 5-LO typically migrates to the nuclear membrane. In previous studies with human neutrophils, CLP displayed the same changes in subcellular distribution as 5-LO [9,16], i.e. following ionophore stimulation both 5-LO and CLP were associated with a nuclear fraction. Recently, a gender difference regarding subcellular localization of 5-LO was found . It was shown that for male neutrophils a substantial amount of 5-LO was associated with the nucleus already in non-stimulated cells, and the same was found for CLP. In the present study we analysed subcellular fractions from MM6 cells, which offer some experimental options. First, undifferentiated MM6 cells express CLP, but not 5-LO. Regardless of stimulation, for these undifferentiated cells almost all CLP was recovered in the non-nuclear fractions. Secondly, differentiated MM6 cells (expressing 5-LO) required priming with PMA for 5-LO to associate with the nuclear membrane in response to ionophore , in accordance with a role for kinases in 5-LO translocation . Accordingly, when differentiated MM6 cells were stimulated with ionophore only, 5-LO was recovered primarily in the non-nuclear fraction, and the same was found for CLP (Figure 8B). However, following priming with PMA and subsequent activation with ionophore, substantial amounts of both 5-LO and CLP were found in the nuclear fraction. Thus, also in MM6 cells, it appears that migration of CLP is connected with migration of 5-LO.
In summary, the findings of the present study show that Trp102 in the 5-LO β-sandwich is important for binding of CLP. Our results also provide further support for CLP functioning as a chaperone for 5-LO, stabilizing 5-LO in the resting cell and functioning as a scaffold during Ca2+-induced 5-LO activity.
Mono Mac 6
- PGC buffer
PBS containing 1 mg/ml glucose and 1 mM CaCl2; TGF-β, transforming growth factor-β
Julia Esser, Marija Rakonjac, Gisbert Schneider, Dieter Steinhilber, Bengt Samuelsson and Olof Rådmark designed the research. Julia Esser, Marija Rakonjac, Bettina Hofmann and Lutz Fischer performed the studies. Patrick Provost provided analytical tools. Julia Esser, Marija Rakonjac, Bettina Hofmann, Gisbert Schneider and Olof Rådmark analysed the results. Julia Esser, Marija Rakonjac, Bettina Hofmann, Gisbert Schneider, Patrick Provost and Olof Rådmark wrote the paper.
We thank Dr Anders Wetterholm (Department of Medical Biochemistry and Biophysics, Karolinska Institute, Stockholm, Sweden) for all of his help.
This work was supported by the Swedish Research Council [grant number 03X-217]; by the European Union [grant number LSHM-CT-2004–00533, FP7-Health-201668]; and by the Karolinska Institutet. P.P. is a Senior Scholar from the Fonds de la Recherche en Santé du Québec.
These authors contributed equally.